task_mmu.c 38.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
#include <linux/mm.h>
Davidlohr Bueso's avatar
Davidlohr Bueso committed
2
#include <linux/vmacache.h>
Linus Torvalds's avatar
Linus Torvalds committed
3
#include <linux/hugetlb.h>
4
#include <linux/huge_mm.h>
Linus Torvalds's avatar
Linus Torvalds committed
5
6
#include <linux/mount.h>
#include <linux/seq_file.h>
Mauricio Lin's avatar
Mauricio Lin committed
7
#include <linux/highmem.h>
Kees Cook's avatar
Kees Cook committed
8
#include <linux/ptrace.h>
9
#include <linux/slab.h>
10
11
#include <linux/pagemap.h>
#include <linux/mempolicy.h>
12
#include <linux/rmap.h>
13
14
#include <linux/swap.h>
#include <linux/swapops.h>
15
#include <linux/mmu_notifier.h>
16
#include <linux/page_idle.h>
Mauricio Lin's avatar
Mauricio Lin committed
17

Linus Torvalds's avatar
Linus Torvalds committed
18
19
#include <asm/elf.h>
#include <asm/uaccess.h>
Mauricio Lin's avatar
Mauricio Lin committed
20
#include <asm/tlbflush.h>
Linus Torvalds's avatar
Linus Torvalds committed
21
22
#include "internal.h"

23
void task_mem(struct seq_file *m, struct mm_struct *mm)
Linus Torvalds's avatar
Linus Torvalds committed
24
{
25
	unsigned long data, text, lib, swap, ptes, pmds;
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
	unsigned long hiwater_vm, total_vm, hiwater_rss, total_rss;

	/*
	 * Note: to minimize their overhead, mm maintains hiwater_vm and
	 * hiwater_rss only when about to *lower* total_vm or rss.  Any
	 * collector of these hiwater stats must therefore get total_vm
	 * and rss too, which will usually be the higher.  Barriers? not
	 * worth the effort, such snapshots can always be inconsistent.
	 */
	hiwater_vm = total_vm = mm->total_vm;
	if (hiwater_vm < mm->hiwater_vm)
		hiwater_vm = mm->hiwater_vm;
	hiwater_rss = total_rss = get_mm_rss(mm);
	if (hiwater_rss < mm->hiwater_rss)
		hiwater_rss = mm->hiwater_rss;
Linus Torvalds's avatar
Linus Torvalds committed
41
42
43
44

	data = mm->total_vm - mm->shared_vm - mm->stack_vm;
	text = (PAGE_ALIGN(mm->end_code) - (mm->start_code & PAGE_MASK)) >> 10;
	lib = (mm->exec_vm << (PAGE_SHIFT-10)) - text;
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
45
	swap = get_mm_counter(mm, MM_SWAPENTS);
46
47
	ptes = PTRS_PER_PTE * sizeof(pte_t) * atomic_long_read(&mm->nr_ptes);
	pmds = PTRS_PER_PMD * sizeof(pmd_t) * mm_nr_pmds(mm);
48
	seq_printf(m,
49
		"VmPeak:\t%8lu kB\n"
Linus Torvalds's avatar
Linus Torvalds committed
50
51
		"VmSize:\t%8lu kB\n"
		"VmLck:\t%8lu kB\n"
52
		"VmPin:\t%8lu kB\n"
53
		"VmHWM:\t%8lu kB\n"
Linus Torvalds's avatar
Linus Torvalds committed
54
55
56
57
58
		"VmRSS:\t%8lu kB\n"
		"VmData:\t%8lu kB\n"
		"VmStk:\t%8lu kB\n"
		"VmExe:\t%8lu kB\n"
		"VmLib:\t%8lu kB\n"
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
59
		"VmPTE:\t%8lu kB\n"
60
		"VmPMD:\t%8lu kB\n"
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
61
		"VmSwap:\t%8lu kB\n",
62
		hiwater_vm << (PAGE_SHIFT-10),
63
		total_vm << (PAGE_SHIFT-10),
Linus Torvalds's avatar
Linus Torvalds committed
64
		mm->locked_vm << (PAGE_SHIFT-10),
65
		mm->pinned_vm << (PAGE_SHIFT-10),
66
67
		hiwater_rss << (PAGE_SHIFT-10),
		total_rss << (PAGE_SHIFT-10),
Linus Torvalds's avatar
Linus Torvalds committed
68
69
		data << (PAGE_SHIFT-10),
		mm->stack_vm << (PAGE_SHIFT-10), text, lib,
70
71
		ptes >> 10,
		pmds >> 10,
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
72
		swap << (PAGE_SHIFT-10));
73
	hugetlb_report_usage(m, mm);
Linus Torvalds's avatar
Linus Torvalds committed
74
75
76
77
78
79
80
}

unsigned long task_vsize(struct mm_struct *mm)
{
	return PAGE_SIZE * mm->total_vm;
}

81
82
83
unsigned long task_statm(struct mm_struct *mm,
			 unsigned long *shared, unsigned long *text,
			 unsigned long *data, unsigned long *resident)
Linus Torvalds's avatar
Linus Torvalds committed
84
{
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
85
	*shared = get_mm_counter(mm, MM_FILEPAGES);
Linus Torvalds's avatar
Linus Torvalds committed
86
87
88
	*text = (PAGE_ALIGN(mm->end_code) - (mm->start_code & PAGE_MASK))
								>> PAGE_SHIFT;
	*data = mm->total_vm - mm->shared_vm;
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
89
	*resident = *shared + get_mm_counter(mm, MM_ANONPAGES);
Linus Torvalds's avatar
Linus Torvalds committed
90
91
92
	return mm->total_vm;
}

93
94
#ifdef CONFIG_NUMA
/*
95
 * Save get_task_policy() for show_numa_map().
96
97
98
99
100
101
 */
static void hold_task_mempolicy(struct proc_maps_private *priv)
{
	struct task_struct *task = priv->task;

	task_lock(task);
102
	priv->task_mempolicy = get_task_policy(task);
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
	mpol_get(priv->task_mempolicy);
	task_unlock(task);
}
static void release_task_mempolicy(struct proc_maps_private *priv)
{
	mpol_put(priv->task_mempolicy);
}
#else
static void hold_task_mempolicy(struct proc_maps_private *priv)
{
}
static void release_task_mempolicy(struct proc_maps_private *priv)
{
}
#endif

119
static void vma_stop(struct proc_maps_private *priv)
120
{
121
122
123
124
125
	struct mm_struct *mm = priv->mm;

	release_task_mempolicy(priv);
	up_read(&mm->mmap_sem);
	mmput(mm);
126
}
127

128
129
130
131
132
133
134
135
static struct vm_area_struct *
m_next_vma(struct proc_maps_private *priv, struct vm_area_struct *vma)
{
	if (vma == priv->tail_vma)
		return NULL;
	return vma->vm_next ?: priv->tail_vma;
}

136
137
138
139
140
141
static void m_cache_vma(struct seq_file *m, struct vm_area_struct *vma)
{
	if (m->count < m->size)	/* vma is copied successfully */
		m->version = m_next_vma(m->private, vma) ? vma->vm_start : -1UL;
}

142
static void *m_start(struct seq_file *m, loff_t *ppos)
Mauricio Lin's avatar
Mauricio Lin committed
143
{
144
	struct proc_maps_private *priv = m->private;
145
	unsigned long last_addr = m->version;
146
	struct mm_struct *mm;
147
148
	struct vm_area_struct *vma;
	unsigned int pos = *ppos;
149

150
151
152
153
	/* See m_cache_vma(). Zero at the start or after lseek. */
	if (last_addr == -1UL)
		return NULL;

154
	priv->task = get_proc_task(priv->inode);
155
	if (!priv->task)
156
		return ERR_PTR(-ESRCH);
157

158
159
160
	mm = priv->mm;
	if (!mm || !atomic_inc_not_zero(&mm->mm_users))
		return NULL;
161

162
	down_read(&mm->mmap_sem);
163
	hold_task_mempolicy(priv);
164
	priv->tail_vma = get_gate_vma(mm);
165

166
167
168
169
170
171
172
	if (last_addr) {
		vma = find_vma(mm, last_addr);
		if (vma && (vma = m_next_vma(priv, vma)))
			return vma;
	}

	m->version = 0;
173
	if (pos < mm->map_count) {
174
175
		for (vma = mm->mmap; pos; pos--) {
			m->version = vma->vm_start;
176
			vma = vma->vm_next;
177
		}
178
		return vma;
179
	}
180

181
	/* we do not bother to update m->version in this case */
182
183
	if (pos == mm->map_count && priv->tail_vma)
		return priv->tail_vma;
184
185
186

	vma_stop(priv);
	return NULL;
187
188
189
190
191
}

static void *m_next(struct seq_file *m, void *v, loff_t *pos)
{
	struct proc_maps_private *priv = m->private;
192
	struct vm_area_struct *next;
193
194

	(*pos)++;
195
	next = m_next_vma(priv, v);
196
197
198
	if (!next)
		vma_stop(priv);
	return next;
199
200
201
202
203
204
}

static void m_stop(struct seq_file *m, void *v)
{
	struct proc_maps_private *priv = m->private;

205
206
	if (!IS_ERR_OR_NULL(v))
		vma_stop(priv);
207
	if (priv->task) {
208
		put_task_struct(priv->task);
209
210
		priv->task = NULL;
	}
211
212
}

213
214
215
216
217
218
219
220
static int proc_maps_open(struct inode *inode, struct file *file,
			const struct seq_operations *ops, int psize)
{
	struct proc_maps_private *priv = __seq_open_private(file, ops, psize);

	if (!priv)
		return -ENOMEM;

221
	priv->inode = inode;
222
223
224
225
226
227
228
229
	priv->mm = proc_mem_open(inode, PTRACE_MODE_READ);
	if (IS_ERR(priv->mm)) {
		int err = PTR_ERR(priv->mm);

		seq_release_private(inode, file);
		return err;
	}

230
231
232
	return 0;
}

233
234
235
236
237
238
239
240
241
242
243
static int proc_map_release(struct inode *inode, struct file *file)
{
	struct seq_file *seq = file->private_data;
	struct proc_maps_private *priv = seq->private;

	if (priv->mm)
		mmdrop(priv->mm);

	return seq_release_private(inode, file);
}

244
static int do_maps_open(struct inode *inode, struct file *file,
245
			const struct seq_operations *ops)
246
{
247
248
	return proc_maps_open(inode, file, ops,
				sizeof(struct proc_maps_private));
249
}
Mauricio Lin's avatar
Mauricio Lin committed
250

251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
static pid_t pid_of_stack(struct proc_maps_private *priv,
				struct vm_area_struct *vma, bool is_pid)
{
	struct inode *inode = priv->inode;
	struct task_struct *task;
	pid_t ret = 0;

	rcu_read_lock();
	task = pid_task(proc_pid(inode), PIDTYPE_PID);
	if (task) {
		task = task_of_stack(task, vma, is_pid);
		if (task)
			ret = task_pid_nr_ns(task, inode->i_sb->s_fs_info);
	}
	rcu_read_unlock();

	return ret;
}

270
271
static void
show_map_vma(struct seq_file *m, struct vm_area_struct *vma, int is_pid)
Linus Torvalds's avatar
Linus Torvalds committed
272
{
Mauricio Lin's avatar
Mauricio Lin committed
273
274
	struct mm_struct *mm = vma->vm_mm;
	struct file *file = vma->vm_file;
275
	struct proc_maps_private *priv = m->private;
276
	vm_flags_t flags = vma->vm_flags;
Linus Torvalds's avatar
Linus Torvalds committed
277
	unsigned long ino = 0;
278
	unsigned long long pgoff = 0;
279
	unsigned long start, end;
Linus Torvalds's avatar
Linus Torvalds committed
280
	dev_t dev = 0;
281
	const char *name = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
282
283

	if (file) {
Al Viro's avatar
Al Viro committed
284
		struct inode *inode = file_inode(vma->vm_file);
Linus Torvalds's avatar
Linus Torvalds committed
285
286
		dev = inode->i_sb->s_dev;
		ino = inode->i_ino;
287
		pgoff = ((loff_t)vma->vm_pgoff) << PAGE_SHIFT;
Linus Torvalds's avatar
Linus Torvalds committed
288
289
	}

290
291
	/* We don't show the stack guard page in /proc/maps */
	start = vma->vm_start;
292
293
294
295
296
	if (stack_guard_page_start(vma, start))
		start += PAGE_SIZE;
	end = vma->vm_end;
	if (stack_guard_page_end(vma, end))
		end -= PAGE_SIZE;
297

298
299
	seq_setwidth(m, 25 + sizeof(void *) * 6 - 1);
	seq_printf(m, "%08lx-%08lx %c%c%c%c %08llx %02x:%02x %lu ",
300
			start,
301
			end,
Linus Torvalds's avatar
Linus Torvalds committed
302
303
304
305
			flags & VM_READ ? 'r' : '-',
			flags & VM_WRITE ? 'w' : '-',
			flags & VM_EXEC ? 'x' : '-',
			flags & VM_MAYSHARE ? 's' : 'p',
306
			pgoff,
307
			MAJOR(dev), MINOR(dev), ino);
Linus Torvalds's avatar
Linus Torvalds committed
308
309
310
311
312

	/*
	 * Print the dentry name for named mappings, and a
	 * special [heap] marker for the heap:
	 */
Mauricio Lin's avatar
Mauricio Lin committed
313
	if (file) {
314
		seq_pad(m, ' ');
315
		seq_file_path(m, file, "\n");
316
317
318
		goto done;
	}

319
320
321
322
323
324
	if (vma->vm_ops && vma->vm_ops->name) {
		name = vma->vm_ops->name(vma);
		if (name)
			goto done;
	}

325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
	name = arch_vma_name(vma);
	if (!name) {
		pid_t tid;

		if (!mm) {
			name = "[vdso]";
			goto done;
		}

		if (vma->vm_start <= mm->brk &&
		    vma->vm_end >= mm->start_brk) {
			name = "[heap]";
			goto done;
		}

340
		tid = pid_of_stack(priv, vma, is_pid);
341
342
343
344
345
346
347
348
		if (tid != 0) {
			/*
			 * Thread stack in /proc/PID/task/TID/maps or
			 * the main process stack.
			 */
			if (!is_pid || (vma->vm_start <= mm->start_stack &&
			    vma->vm_end >= mm->start_stack)) {
				name = "[stack]";
349
			} else {
350
				/* Thread stack in /proc/PID/maps */
351
				seq_pad(m, ' ');
352
				seq_printf(m, "[stack:%d]", tid);
Linus Torvalds's avatar
Linus Torvalds committed
353
			}
354
		}
355
356
357
358
	}

done:
	if (name) {
359
		seq_pad(m, ' ');
360
		seq_puts(m, name);
Linus Torvalds's avatar
Linus Torvalds committed
361
362
	}
	seq_putc(m, '\n');
363
364
}

365
static int show_map(struct seq_file *m, void *v, int is_pid)
366
{
367
	show_map_vma(m, v, is_pid);
368
	m_cache_vma(m, v);
Linus Torvalds's avatar
Linus Torvalds committed
369
370
371
	return 0;
}

372
373
374
375
376
377
378
379
380
381
static int show_pid_map(struct seq_file *m, void *v)
{
	return show_map(m, v, 1);
}

static int show_tid_map(struct seq_file *m, void *v)
{
	return show_map(m, v, 0);
}

382
static const struct seq_operations proc_pid_maps_op = {
383
384
385
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
386
387
388
389
390
391
392
393
	.show	= show_pid_map
};

static const struct seq_operations proc_tid_maps_op = {
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
	.show	= show_tid_map
394
395
};

396
static int pid_maps_open(struct inode *inode, struct file *file)
397
398
399
400
{
	return do_maps_open(inode, file, &proc_pid_maps_op);
}

401
402
403
404
405
406
407
408
409
static int tid_maps_open(struct inode *inode, struct file *file)
{
	return do_maps_open(inode, file, &proc_tid_maps_op);
}

const struct file_operations proc_pid_maps_operations = {
	.open		= pid_maps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
410
	.release	= proc_map_release,
411
412
413
414
};

const struct file_operations proc_tid_maps_operations = {
	.open		= tid_maps_open,
415
416
	.read		= seq_read,
	.llseek		= seq_lseek,
417
	.release	= proc_map_release,
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
};

/*
 * Proportional Set Size(PSS): my share of RSS.
 *
 * PSS of a process is the count of pages it has in memory, where each
 * page is divided by the number of processes sharing it.  So if a
 * process has 1000 pages all to itself, and 1000 shared with one other
 * process, its PSS will be 1500.
 *
 * To keep (accumulated) division errors low, we adopt a 64bit
 * fixed-point pss counter to minimize division errors. So (pss >>
 * PSS_SHIFT) would be the real byte count.
 *
 * A shift of 12 before division means (assuming 4K page size):
 * 	- 1M 3-user-pages add up to 8KB errors;
 * 	- supports mapcount up to 2^24, or 16M;
 * 	- supports PSS up to 2^52 bytes, or 4PB.
 */
#define PSS_SHIFT 12

439
#ifdef CONFIG_PROC_PAGE_MONITOR
Peter Zijlstra's avatar
Peter Zijlstra committed
440
struct mem_size_stats {
441
442
443
444
445
446
	unsigned long resident;
	unsigned long shared_clean;
	unsigned long shared_dirty;
	unsigned long private_clean;
	unsigned long private_dirty;
	unsigned long referenced;
447
	unsigned long anonymous;
448
	unsigned long anonymous_thp;
Peter Zijlstra's avatar
Peter Zijlstra committed
449
	unsigned long swap;
450
451
	unsigned long shared_hugetlb;
	unsigned long private_hugetlb;
452
	u64 pss;
453
	u64 swap_pss;
454
455
};

456
457
458
459
460
461
462
463
464
465
static void smaps_account(struct mem_size_stats *mss, struct page *page,
		unsigned long size, bool young, bool dirty)
{
	int mapcount;

	if (PageAnon(page))
		mss->anonymous += size;

	mss->resident += size;
	/* Accumulate the size in pages that have been accessed. */
466
	if (young || page_is_young(page) || PageReferenced(page))
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
		mss->referenced += size;
	mapcount = page_mapcount(page);
	if (mapcount >= 2) {
		u64 pss_delta;

		if (dirty || PageDirty(page))
			mss->shared_dirty += size;
		else
			mss->shared_clean += size;
		pss_delta = (u64)size << PSS_SHIFT;
		do_div(pss_delta, mapcount);
		mss->pss += pss_delta;
	} else {
		if (dirty || PageDirty(page))
			mss->private_dirty += size;
		else
			mss->private_clean += size;
		mss->pss += (u64)size << PSS_SHIFT;
	}
}
487

488
489
static void smaps_pte_entry(pte_t *pte, unsigned long addr,
		struct mm_walk *walk)
490
491
{
	struct mem_size_stats *mss = walk->private;
492
	struct vm_area_struct *vma = walk->vma;
493
	struct page *page = NULL;
494

495
496
497
498
	if (pte_present(*pte)) {
		page = vm_normal_page(vma, addr, *pte);
	} else if (is_swap_pte(*pte)) {
		swp_entry_t swpent = pte_to_swp_entry(*pte);
499

500
501
502
		if (!non_swap_entry(swpent)) {
			int mapcount;

503
			mss->swap += PAGE_SIZE;
504
505
506
507
508
509
510
511
512
513
			mapcount = swp_swapcount(swpent);
			if (mapcount >= 2) {
				u64 pss_delta = (u64)PAGE_SIZE << PSS_SHIFT;

				do_div(pss_delta, mapcount);
				mss->swap_pss += pss_delta;
			} else {
				mss->swap_pss += (u64)PAGE_SIZE << PSS_SHIFT;
			}
		} else if (is_migration_entry(swpent))
514
515
			page = migration_entry_to_page(swpent);
	}
516
517
518

	if (!page)
		return;
519
	smaps_account(mss, page, PAGE_SIZE, pte_young(*pte), pte_dirty(*pte));
520
521
}

522
523
524
525
526
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
static void smaps_pmd_entry(pmd_t *pmd, unsigned long addr,
		struct mm_walk *walk)
{
	struct mem_size_stats *mss = walk->private;
527
	struct vm_area_struct *vma = walk->vma;
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
	struct page *page;

	/* FOLL_DUMP will return -EFAULT on huge zero page */
	page = follow_trans_huge_pmd(vma, addr, pmd, FOLL_DUMP);
	if (IS_ERR_OR_NULL(page))
		return;
	mss->anonymous_thp += HPAGE_PMD_SIZE;
	smaps_account(mss, page, HPAGE_PMD_SIZE,
			pmd_young(*pmd), pmd_dirty(*pmd));
}
#else
static void smaps_pmd_entry(pmd_t *pmd, unsigned long addr,
		struct mm_walk *walk)
{
}
#endif

545
static int smaps_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
546
			   struct mm_walk *walk)
Mauricio Lin's avatar
Mauricio Lin committed
547
{
548
	struct vm_area_struct *vma = walk->vma;
549
	pte_t *pte;
550
	spinlock_t *ptl;
Mauricio Lin's avatar
Mauricio Lin committed
551

552
	if (pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
553
		smaps_pmd_entry(pmd, addr, walk);
554
		spin_unlock(ptl);
555
		return 0;
556
	}
557
558
559

	if (pmd_trans_unstable(pmd))
		return 0;
560
561
562
563
564
	/*
	 * The mmap_sem held all the way back in m_start() is what
	 * keeps khugepaged out of here and from collapsing things
	 * in here.
	 */
565
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
566
	for (; addr != end; pte++, addr += PAGE_SIZE)
567
		smaps_pte_entry(pte, addr, walk);
568
569
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();
570
	return 0;
Mauricio Lin's avatar
Mauricio Lin committed
571
572
}

573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
static void show_smap_vma_flags(struct seq_file *m, struct vm_area_struct *vma)
{
	/*
	 * Don't forget to update Documentation/ on changes.
	 */
	static const char mnemonics[BITS_PER_LONG][2] = {
		/*
		 * In case if we meet a flag we don't know about.
		 */
		[0 ... (BITS_PER_LONG-1)] = "??",

		[ilog2(VM_READ)]	= "rd",
		[ilog2(VM_WRITE)]	= "wr",
		[ilog2(VM_EXEC)]	= "ex",
		[ilog2(VM_SHARED)]	= "sh",
		[ilog2(VM_MAYREAD)]	= "mr",
		[ilog2(VM_MAYWRITE)]	= "mw",
		[ilog2(VM_MAYEXEC)]	= "me",
		[ilog2(VM_MAYSHARE)]	= "ms",
		[ilog2(VM_GROWSDOWN)]	= "gd",
		[ilog2(VM_PFNMAP)]	= "pf",
		[ilog2(VM_DENYWRITE)]	= "dw",
595
596
597
#ifdef CONFIG_X86_INTEL_MPX
		[ilog2(VM_MPX)]		= "mp",
#endif
598
599
600
601
602
603
604
605
606
607
608
		[ilog2(VM_LOCKED)]	= "lo",
		[ilog2(VM_IO)]		= "io",
		[ilog2(VM_SEQ_READ)]	= "sr",
		[ilog2(VM_RAND_READ)]	= "rr",
		[ilog2(VM_DONTCOPY)]	= "dc",
		[ilog2(VM_DONTEXPAND)]	= "de",
		[ilog2(VM_ACCOUNT)]	= "ac",
		[ilog2(VM_NORESERVE)]	= "nr",
		[ilog2(VM_HUGETLB)]	= "ht",
		[ilog2(VM_ARCH_1)]	= "ar",
		[ilog2(VM_DONTDUMP)]	= "dd",
609
610
611
#ifdef CONFIG_MEM_SOFT_DIRTY
		[ilog2(VM_SOFTDIRTY)]	= "sd",
#endif
612
613
614
615
		[ilog2(VM_MIXEDMAP)]	= "mm",
		[ilog2(VM_HUGEPAGE)]	= "hg",
		[ilog2(VM_NOHUGEPAGE)]	= "nh",
		[ilog2(VM_MERGEABLE)]	= "mg",
616
617
		[ilog2(VM_UFFD_MISSING)]= "um",
		[ilog2(VM_UFFD_WP)]	= "uw",
618
619
620
621
622
623
624
625
626
627
628
629
630
	};
	size_t i;

	seq_puts(m, "VmFlags: ");
	for (i = 0; i < BITS_PER_LONG; i++) {
		if (vma->vm_flags & (1UL << i)) {
			seq_printf(m, "%c%c ",
				   mnemonics[i][0], mnemonics[i][1]);
		}
	}
	seq_putc(m, '\n');
}

631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
#ifdef CONFIG_HUGETLB_PAGE
static int smaps_hugetlb_range(pte_t *pte, unsigned long hmask,
				 unsigned long addr, unsigned long end,
				 struct mm_walk *walk)
{
	struct mem_size_stats *mss = walk->private;
	struct vm_area_struct *vma = walk->vma;
	struct page *page = NULL;

	if (pte_present(*pte)) {
		page = vm_normal_page(vma, addr, *pte);
	} else if (is_swap_pte(*pte)) {
		swp_entry_t swpent = pte_to_swp_entry(*pte);

		if (is_migration_entry(swpent))
			page = migration_entry_to_page(swpent);
	}
	if (page) {
		int mapcount = page_mapcount(page);

		if (mapcount >= 2)
			mss->shared_hugetlb += huge_page_size(hstate_vma(vma));
		else
			mss->private_hugetlb += huge_page_size(hstate_vma(vma));
	}
	return 0;
}
#endif /* HUGETLB_PAGE */

660
static int show_smap(struct seq_file *m, void *v, int is_pid)
Mauricio Lin's avatar
Mauricio Lin committed
661
662
663
{
	struct vm_area_struct *vma = v;
	struct mem_size_stats mss;
664
665
	struct mm_walk smaps_walk = {
		.pmd_entry = smaps_pte_range,
666
667
668
#ifdef CONFIG_HUGETLB_PAGE
		.hugetlb_entry = smaps_hugetlb_range,
#endif
669
670
671
		.mm = vma->vm_mm,
		.private = &mss,
	};
Mauricio Lin's avatar
Mauricio Lin committed
672
673

	memset(&mss, 0, sizeof mss);
674
	/* mmap_sem is held in m_start */
675
	walk_page_vma(vma, &smaps_walk);
676

677
	show_map_vma(m, vma, is_pid);
678
679
680
681
682
683
684
685
686

	seq_printf(m,
		   "Size:           %8lu kB\n"
		   "Rss:            %8lu kB\n"
		   "Pss:            %8lu kB\n"
		   "Shared_Clean:   %8lu kB\n"
		   "Shared_Dirty:   %8lu kB\n"
		   "Private_Clean:  %8lu kB\n"
		   "Private_Dirty:  %8lu kB\n"
Peter Zijlstra's avatar
Peter Zijlstra committed
687
		   "Referenced:     %8lu kB\n"
688
		   "Anonymous:      %8lu kB\n"
689
		   "AnonHugePages:  %8lu kB\n"
690
691
		   "Shared_Hugetlb: %8lu kB\n"
		   "Private_Hugetlb: %7lu kB\n"
692
		   "Swap:           %8lu kB\n"
693
		   "SwapPss:        %8lu kB\n"
694
		   "KernelPageSize: %8lu kB\n"
695
696
		   "MMUPageSize:    %8lu kB\n"
		   "Locked:         %8lu kB\n",
697
698
699
700
701
702
703
		   (vma->vm_end - vma->vm_start) >> 10,
		   mss.resident >> 10,
		   (unsigned long)(mss.pss >> (10 + PSS_SHIFT)),
		   mss.shared_clean  >> 10,
		   mss.shared_dirty  >> 10,
		   mss.private_clean >> 10,
		   mss.private_dirty >> 10,
Peter Zijlstra's avatar
Peter Zijlstra committed
704
		   mss.referenced >> 10,
705
		   mss.anonymous >> 10,
706
		   mss.anonymous_thp >> 10,
707
708
		   mss.shared_hugetlb >> 10,
		   mss.private_hugetlb >> 10,
709
		   mss.swap >> 10,
710
		   (unsigned long)(mss.swap_pss >> (10 + PSS_SHIFT)),
711
		   vma_kernel_pagesize(vma) >> 10,
712
713
714
		   vma_mmu_pagesize(vma) >> 10,
		   (vma->vm_flags & VM_LOCKED) ?
			(unsigned long)(mss.pss >> (10 + PSS_SHIFT)) : 0);
715

716
	show_smap_vma_flags(m, vma);
717
	m_cache_vma(m, vma);
718
	return 0;
Mauricio Lin's avatar
Mauricio Lin committed
719
720
}

721
722
723
724
725
726
727
728
729
730
static int show_pid_smap(struct seq_file *m, void *v)
{
	return show_smap(m, v, 1);
}

static int show_tid_smap(struct seq_file *m, void *v)
{
	return show_smap(m, v, 0);
}

731
static const struct seq_operations proc_pid_smaps_op = {
732
733
734
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
735
736
737
738
739
740
741
742
	.show	= show_pid_smap
};

static const struct seq_operations proc_tid_smaps_op = {
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
	.show	= show_tid_smap
743
744
};

745
static int pid_smaps_open(struct inode *inode, struct file *file)
746
747
748
749
{
	return do_maps_open(inode, file, &proc_pid_smaps_op);
}

750
751
752
753
754
755
756
757
758
static int tid_smaps_open(struct inode *inode, struct file *file)
{
	return do_maps_open(inode, file, &proc_tid_smaps_op);
}

const struct file_operations proc_pid_smaps_operations = {
	.open		= pid_smaps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
759
	.release	= proc_map_release,
760
761
762
763
};

const struct file_operations proc_tid_smaps_operations = {
	.open		= tid_smaps_open,
764
765
	.read		= seq_read,
	.llseek		= seq_lseek,
766
	.release	= proc_map_release,
767
768
};

769
770
771
772
enum clear_refs_types {
	CLEAR_REFS_ALL = 1,
	CLEAR_REFS_ANON,
	CLEAR_REFS_MAPPED,
773
	CLEAR_REFS_SOFT_DIRTY,
774
	CLEAR_REFS_MM_HIWATER_RSS,
775
776
777
	CLEAR_REFS_LAST,
};

778
struct clear_refs_private {
779
	enum clear_refs_types type;
780
781
};

782
#ifdef CONFIG_MEM_SOFT_DIRTY
783
784
785
786
787
788
789
790
791
792
static inline void clear_soft_dirty(struct vm_area_struct *vma,
		unsigned long addr, pte_t *pte)
{
	/*
	 * The soft-dirty tracker uses #PF-s to catch writes
	 * to pages, so write-protect the pte as well. See the
	 * Documentation/vm/soft-dirty.txt for full description
	 * of how soft-dirty works.
	 */
	pte_t ptent = *pte;
793
794
795

	if (pte_present(ptent)) {
		ptent = pte_wrprotect(ptent);
796
		ptent = pte_clear_soft_dirty(ptent);
797
798
799
800
	} else if (is_swap_pte(ptent)) {
		ptent = pte_swp_clear_soft_dirty(ptent);
	}

801
802
803
	set_pte_at(vma->vm_mm, addr, pte, ptent);
}

804
805
806
807
808
809
static inline void clear_soft_dirty_pmd(struct vm_area_struct *vma,
		unsigned long addr, pmd_t *pmdp)
{
	pmd_t pmd = *pmdp;

	pmd = pmd_wrprotect(pmd);
810
	pmd = pmd_clear_soft_dirty(pmd);
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830

	if (vma->vm_flags & VM_SOFTDIRTY)
		vma->vm_flags &= ~VM_SOFTDIRTY;

	set_pmd_at(vma->vm_mm, addr, pmdp, pmd);
}

#else

static inline void clear_soft_dirty(struct vm_area_struct *vma,
		unsigned long addr, pte_t *pte)
{
}

static inline void clear_soft_dirty_pmd(struct vm_area_struct *vma,
		unsigned long addr, pmd_t *pmdp)
{
}
#endif

831
static int clear_refs_pte_range(pmd_t *pmd, unsigned long addr,
832
				unsigned long end, struct mm_walk *walk)
833
{
834
	struct clear_refs_private *cp = walk->private;
835
	struct vm_area_struct *vma = walk->vma;
836
837
838
839
	pte_t *pte, ptent;
	spinlock_t *ptl;
	struct page *page;

840
841
842
843
844
845
846
847
848
849
	if (pmd_trans_huge_lock(pmd, vma, &ptl) == 1) {
		if (cp->type == CLEAR_REFS_SOFT_DIRTY) {
			clear_soft_dirty_pmd(vma, addr, pmd);
			goto out;
		}

		page = pmd_page(*pmd);

		/* Clear accessed and referenced bits. */
		pmdp_test_and_clear_young(vma, addr, pmd);
850
		test_and_clear_page_young(page);
851
852
853
854
855
856
		ClearPageReferenced(page);
out:
		spin_unlock(ptl);
		return 0;
	}

857
858
	if (pmd_trans_unstable(pmd))
		return 0;
859

860
861
862
863
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	for (; addr != end; pte++, addr += PAGE_SIZE) {
		ptent = *pte;

864
865
866
867
868
		if (cp->type == CLEAR_REFS_SOFT_DIRTY) {
			clear_soft_dirty(vma, addr, pte);
			continue;
		}

869
870
871
		if (!pte_present(ptent))
			continue;

872
873
874
875
876
877
		page = vm_normal_page(vma, addr, ptent);
		if (!page)
			continue;

		/* Clear accessed and referenced bits. */
		ptep_test_and_clear_young(vma, addr, pte);
878
		test_and_clear_page_young(page);
879
880
881
882
883
884
885
		ClearPageReferenced(page);
	}
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();
	return 0;
}

886
887
888
889
890
891
static int clear_refs_test_walk(unsigned long start, unsigned long end,
				struct mm_walk *walk)
{
	struct clear_refs_private *cp = walk->private;
	struct vm_area_struct *vma = walk->vma;

892
893
894
	if (vma->vm_flags & VM_PFNMAP)
		return 1;

895
896
897
898
899
900
901
902
903
904
905
906
907
	/*
	 * Writing 1 to /proc/pid/clear_refs affects all pages.
	 * Writing 2 to /proc/pid/clear_refs only affects anonymous pages.
	 * Writing 3 to /proc/pid/clear_refs only affects file mapped pages.
	 * Writing 4 to /proc/pid/clear_refs affects all pages.
	 */
	if (cp->type == CLEAR_REFS_ANON && vma->vm_file)
		return 1;
	if (cp->type == CLEAR_REFS_MAPPED && !vma->vm_file)
		return 1;
	return 0;
}

908
909
static ssize_t clear_refs_write(struct file *file, const char __user *buf,
				size_t count, loff_t *ppos)
910
{
911
	struct task_struct *task;
912
	char buffer[PROC_NUMBUF];
913
	struct mm_struct *mm;
914
	struct vm_area_struct *vma;
915
916
	enum clear_refs_types type;
	int itype;
917
	int rv;
918

919
920
921
922
923
	memset(buffer, 0, sizeof(buffer));
	if (count > sizeof(buffer) - 1)
		count = sizeof(buffer) - 1;
	if (copy_from_user(buffer, buf, count))
		return -EFAULT;
924
	rv = kstrtoint(strstrip(buffer), 10, &itype);
925
926
	if (rv < 0)
		return rv;
927
928
	type = (enum clear_refs_types)itype;
	if (type < CLEAR_REFS_ALL || type >= CLEAR_REFS_LAST)
929
		return -EINVAL;
930

Al Viro's avatar
Al Viro committed
931
	task = get_proc_task(file_inode(file));
932
933
934
935
	if (!task)
		return -ESRCH;
	mm = get_task_mm(task);
	if (mm) {
936
		struct clear_refs_private cp = {
937
			.type = type,
938
		};
939
940
		struct mm_walk clear_refs_walk = {
			.pmd_entry = clear_refs_pte_range,
941
			.test_walk = clear_refs_test_walk,
942
			.mm = mm,
943
			.private = &cp,
944
		};
945
946
947
948
949
950
951
952
953
954
955
956

		if (type == CLEAR_REFS_MM_HIWATER_RSS) {
			/*
			 * Writing 5 to /proc/pid/clear_refs resets the peak
			 * resident set size to this mm's current rss value.
			 */
			down_write(&mm->mmap_sem);
			reset_mm_hiwater_rss(mm);
			up_write(&mm->mmap_sem);
			goto out_mm;
		}

957
		down_read(&mm->mmap_sem);
958
959
960
961
962
963
964
965
966
967
968
969
970
		if (type == CLEAR_REFS_SOFT_DIRTY) {
			for (vma = mm->mmap; vma; vma = vma->vm_next) {
				if (!(vma->vm_flags & VM_SOFTDIRTY))
					continue;
				up_read(&mm->mmap_sem);
				down_write(&mm->mmap_sem);
				for (vma = mm->mmap; vma; vma = vma->vm_next) {
					vma->vm_flags &= ~VM_SOFTDIRTY;
					vma_set_page_prot(vma);
				}
				downgrade_write(&mm->mmap_sem);
				break;
			}
971
			mmu_notifier_invalidate_range_start(mm, 0, -1);
972
		}
973
		walk_page_range(0, ~0UL, &clear_refs_walk);
974
975
		if (type == CLEAR_REFS_SOFT_DIRTY)
			mmu_notifier_invalidate_range_end(mm, 0, -1);
976
977
		flush_tlb_mm(mm);
		up_read(&mm->mmap_sem);
978
out_mm:
979
980
981
		mmput(mm);
	}
	put_task_struct(task);
982
983

	return count;
984
985
}

986
987
const struct file_operations proc_clear_refs_operations = {
	.write		= clear_refs_write,
988
	.llseek		= noop_llseek,
989
990
};

991
992
993
994
typedef struct {
	u64 pme;
} pagemap_entry_t;

995
struct pagemapread {
996
	int pos, len;		/* units: PM_ENTRY_BYTES, not bytes */
997
	pagemap_entry_t *buffer;
998
	bool show_pfn;
999
1000
};

1001
1002
1003
#define PAGEMAP_WALK_SIZE	(PMD_SIZE)
#define PAGEMAP_WALK_MASK	(PMD_MASK)

1004
1005
1006
1007
#define PM_ENTRY_BYTES		sizeof(pagemap_entry_t)
#define PM_PFRAME_BITS		55
#define PM_PFRAME_MASK		GENMASK_ULL(PM_PFRAME_BITS - 1, 0)
#define PM_SOFT_DIRTY		BIT_ULL(55)
1008
#define PM_MMAP_EXCLUSIVE	BIT_ULL(56)
1009
1010
1011
1012
#define PM_FILE			BIT_ULL(61)
#define PM_SWAP			BIT_ULL(62)
#define PM_PRESENT		BIT_ULL(63)

1013
1014
#define PM_END_OF_BUFFER    1

1015
static inline pagemap_entry_t make_pme(u64 frame, u64 flags)
1016
{
1017
	return (pagemap_entry_t) { .pme = (frame & PM_PFRAME_MASK) | flags };
1018
1019
1020
}

static int add_to_pagemap(unsigned long addr, pagemap_entry_t *pme,
1021
1022
			  struct pagemapread *pm)
{
1023
	pm->buffer[pm->pos++] = *pme;
1024
	if (pm->pos >= pm->len)
1025
		return PM_END_OF_BUFFER;
1026
1027
1028
1029
	return 0;
}

static int pagemap_pte_hole(unsigned long start, unsigned long end,
1030
				struct mm_walk *walk)
1031
{
1032
	struct pagemapread *pm = walk->private;
1033
	unsigned long addr = start;
1034
	int err = 0;
1035

1036
1037
	while (addr < end) {
		struct vm_area_struct *vma = find_vma(walk->mm, addr);
1038
		pagemap_entry_t pme = make_pme(0, 0);
1039
1040
		/* End of address space hole, which we mark as non-present. */
		unsigned long hole_end;
1041

1042
1043
1044
1045
1046
1047
1048
1049
1050
		if (vma)
			hole_end = min(end, vma->vm_start);
		else
			hole_end = end;

		for (; addr < hole_end; addr += PAGE_SIZE) {
			err = add_to_pagemap(addr, &pme, pm);
			if (err)
				goto out;
1051
1052
		}

1053
1054
1055
1056
1057
		if (!vma)
			break;

		/* Addresses in the VMA. */
		if (vma->vm_flags & VM_SOFTDIRTY)
1058
			pme = make_pme(0, PM_SOFT_DIRTY);
1059
		for (; addr < min(end, vma->vm_end); addr += PAGE_SIZE) {
1060
1061
1062
1063
			err = add_to_pagemap(addr, &pme, pm);
			if (err)
				goto out;
		}
1064
	}
1065
out:
1066
1067
1068
	return err;
}

1069
static pagemap_entry_t pte_to_pagemap_entry(struct pagemapread *pm,
1070
		struct vm_area_struct *vma, unsigned long addr, pte_t pte)
1071
{
1072
	u64 frame = 0, flags = 0;
1073
	struct page *page = NULL;
1074

1075
	if (pte_present(pte)) {
1076
1077
		if (pm->show_pfn)
			frame = pte_pfn(pte);
1078
		flags |= PM_PRESENT;
1079
		page = vm_normal_page(vma, addr, pte);
1080
		if (pte_soft_dirty(pte))
1081
			flags |= PM_SOFT_DIRTY;
1082
	} else if (is_swap_pte(pte)) {
1083
1084
		swp_entry_t entry;
		if (pte_swp_soft_dirty(pte))
1085
			flags |= PM_SOFT_DIRTY;
1086
		entry = pte_to_swp_entry(pte);
1087
1088
		frame = swp_type(entry) |
			(swp_offset(entry) << MAX_SWAPFILES_SHIFT);
1089
		flags |= PM_SWAP;
1090
1091
1092
1093
1094
1095
		if (is_migration_entry(entry))
			page = migration_entry_to_page(entry);
	}

	if (page && !PageAnon(page))
		flags |= PM_FILE;
1096
1097
	if (page && page_mapcount(page) == 1)
		flags |= PM_MMAP_EXCLUSIVE;
1098
1099
	if (vma->vm_flags & VM_SOFTDIRTY)
		flags |= PM_SOFT_DIRTY;
1100

1101
	return make_pme(frame, flags);
1102
1103
}

1104
static int pagemap_pmd_range(pmd_t *pmdp, unsigned long addr, unsigned long end,
1105
			     struct mm_walk *walk)
1106
{
1107
	struct vm_area_struct *vma = walk->vma;
1108
	struct pagemapread *pm = walk->private;
1109
	spinlock_t *ptl;
1110
	pte_t *pte, *orig_pte;
1111
1112
	int err = 0;

1113
1114
1115
1116
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	if (pmd_trans_huge_lock(pmdp, vma, &ptl) == 1) {
		u64 flags = 0, frame = 0;
		pmd_t pmd = *pmdp;
1117

1118
		if ((vma->vm_flags & VM_SOFTDIRTY) || pmd_soft_dirty(pmd))
1119
			flags |= PM_SOFT_DIRTY;