task_mmu.c 45.8 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2
#include <linux/pagewalk.h>
D
Davidlohr Bueso 已提交
3
#include <linux/vmacache.h>
L
Linus Torvalds 已提交
4
#include <linux/hugetlb.h>
5
#include <linux/huge_mm.h>
L
Linus Torvalds 已提交
6 7
#include <linux/mount.h>
#include <linux/seq_file.h>
M
Mauricio Lin 已提交
8
#include <linux/highmem.h>
K
Kees Cook 已提交
9
#include <linux/ptrace.h>
10
#include <linux/slab.h>
11 12
#include <linux/pagemap.h>
#include <linux/mempolicy.h>
13
#include <linux/rmap.h>
14
#include <linux/swap.h>
15
#include <linux/sched/mm.h>
16
#include <linux/swapops.h>
17
#include <linux/mmu_notifier.h>
18
#include <linux/page_idle.h>
19
#include <linux/shmem_fs.h>
M
Minchan Kim 已提交
20
#include <linux/uaccess.h>
21
#include <linux/pkeys.h>
M
Mauricio Lin 已提交
22

L
Linus Torvalds 已提交
23
#include <asm/elf.h>
M
Minchan Kim 已提交
24
#include <asm/tlb.h>
M
Mauricio Lin 已提交
25
#include <asm/tlbflush.h>
L
Linus Torvalds 已提交
26 27
#include "internal.h"

28 29
#define SEQ_PUT_DEC(str, val) \
		seq_put_decimal_ull_width(m, str, (val) << (PAGE_SHIFT-10), 8)
30
void task_mem(struct seq_file *m, struct mm_struct *mm)
L
Linus Torvalds 已提交
31
{
32
	unsigned long text, lib, swap, anon, file, shmem;
33 34
	unsigned long hiwater_vm, total_vm, hiwater_rss, total_rss;

35 36 37 38
	anon = get_mm_counter(mm, MM_ANONPAGES);
	file = get_mm_counter(mm, MM_FILEPAGES);
	shmem = get_mm_counter(mm, MM_SHMEMPAGES);

39 40 41 42 43 44 45 46 47 48
	/*
	 * 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;
49
	hiwater_rss = total_rss = anon + file + shmem;
50 51
	if (hiwater_rss < mm->hiwater_rss)
		hiwater_rss = mm->hiwater_rss;
L
Linus Torvalds 已提交
52

53 54 55 56 57
	/* split executable areas between text and lib */
	text = PAGE_ALIGN(mm->end_code) - (mm->start_code & PAGE_MASK);
	text = min(text, mm->exec_vm << PAGE_SHIFT);
	lib = (mm->exec_vm << PAGE_SHIFT) - text;

K
KAMEZAWA Hiroyuki 已提交
58
	swap = get_mm_counter(mm, MM_SWAPENTS);
59 60 61
	SEQ_PUT_DEC("VmPeak:\t", hiwater_vm);
	SEQ_PUT_DEC(" kB\nVmSize:\t", total_vm);
	SEQ_PUT_DEC(" kB\nVmLck:\t", mm->locked_vm);
62
	SEQ_PUT_DEC(" kB\nVmPin:\t", atomic64_read(&mm->pinned_vm));
63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
	SEQ_PUT_DEC(" kB\nVmHWM:\t", hiwater_rss);
	SEQ_PUT_DEC(" kB\nVmRSS:\t", total_rss);
	SEQ_PUT_DEC(" kB\nRssAnon:\t", anon);
	SEQ_PUT_DEC(" kB\nRssFile:\t", file);
	SEQ_PUT_DEC(" kB\nRssShmem:\t", shmem);
	SEQ_PUT_DEC(" kB\nVmData:\t", mm->data_vm);
	SEQ_PUT_DEC(" kB\nVmStk:\t", mm->stack_vm);
	seq_put_decimal_ull_width(m,
		    " kB\nVmExe:\t", text >> 10, 8);
	seq_put_decimal_ull_width(m,
		    " kB\nVmLib:\t", lib >> 10, 8);
	seq_put_decimal_ull_width(m,
		    " kB\nVmPTE:\t", mm_pgtables_bytes(mm) >> 10, 8);
	SEQ_PUT_DEC(" kB\nVmSwap:\t", swap);
	seq_puts(m, " kB\n");
78
	hugetlb_report_usage(m, mm);
L
Linus Torvalds 已提交
79
}
80
#undef SEQ_PUT_DEC
L
Linus Torvalds 已提交
81 82 83 84 85 86

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

87 88 89
unsigned long task_statm(struct mm_struct *mm,
			 unsigned long *shared, unsigned long *text,
			 unsigned long *data, unsigned long *resident)
L
Linus Torvalds 已提交
90
{
91 92
	*shared = get_mm_counter(mm, MM_FILEPAGES) +
			get_mm_counter(mm, MM_SHMEMPAGES);
L
Linus Torvalds 已提交
93 94
	*text = (PAGE_ALIGN(mm->end_code) - (mm->start_code & PAGE_MASK))
								>> PAGE_SHIFT;
95
	*data = mm->data_vm + mm->stack_vm;
K
KAMEZAWA Hiroyuki 已提交
96
	*resident = *shared + get_mm_counter(mm, MM_ANONPAGES);
L
Linus Torvalds 已提交
97 98 99
	return mm->total_vm;
}

100 101
#ifdef CONFIG_NUMA
/*
102
 * Save get_task_policy() for show_numa_map().
103 104 105 106 107 108
 */
static void hold_task_mempolicy(struct proc_maps_private *priv)
{
	struct task_struct *task = priv->task;

	task_lock(task);
109
	priv->task_mempolicy = get_task_policy(task);
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
	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

126 127 128 129 130 131 132 133
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;
}

134 135 136
static void m_cache_vma(struct seq_file *m, struct vm_area_struct *vma)
{
	if (m->count < m->size)	/* vma is copied successfully */
137
		m->version = m_next_vma(m->private, vma) ? vma->vm_end : -1UL;
138 139
}

140
static void *m_start(struct seq_file *m, loff_t *ppos)
M
Mauricio Lin 已提交
141
{
142
	struct proc_maps_private *priv = m->private;
143
	unsigned long last_addr = m->version;
144
	struct mm_struct *mm;
145 146
	struct vm_area_struct *vma;
	unsigned int pos = *ppos;
147

148 149 150 151
	/* See m_cache_vma(). Zero at the start or after lseek. */
	if (last_addr == -1UL)
		return NULL;

152
	priv->task = get_proc_task(priv->inode);
153
	if (!priv->task)
A
Al Viro 已提交
154
		return ERR_PTR(-ESRCH);
155

156
	mm = priv->mm;
157 158 159
	if (!mm || !mmget_not_zero(mm)) {
		put_task_struct(priv->task);
		priv->task = NULL;
160
		return NULL;
161
	}
162

163 164
	if (down_read_killable(&mm->mmap_sem)) {
		mmput(mm);
165 166
		put_task_struct(priv->task);
		priv->task = NULL;
167 168 169
		return ERR_PTR(-EINTR);
	}

170
	hold_task_mempolicy(priv);
171
	priv->tail_vma = get_gate_vma(mm);
172

173
	if (last_addr) {
174 175 176 177
		vma = find_vma(mm, last_addr - 1);
		if (vma && vma->vm_start <= last_addr)
			vma = m_next_vma(priv, vma);
		if (vma)
178 179 180 181
			return vma;
	}

	m->version = 0;
182
	if (pos < mm->map_count) {
183 184
		for (vma = mm->mmap; pos; pos--) {
			m->version = vma->vm_start;
185
			vma = vma->vm_next;
186
		}
187
		return vma;
188
	}
189

190
	/* we do not bother to update m->version in this case */
191 192
	if (pos == mm->map_count && priv->tail_vma)
		return priv->tail_vma;
193 194

	return NULL;
195 196 197 198 199
}

static void *m_next(struct seq_file *m, void *v, loff_t *pos)
{
	struct proc_maps_private *priv = m->private;
200
	struct vm_area_struct *next;
201 202

	(*pos)++;
203
	next = m_next_vma(priv, v);
204
	return next;
205 206 207 208 209
}

static void m_stop(struct seq_file *m, void *v)
{
	struct proc_maps_private *priv = m->private;
210
	struct mm_struct *mm = priv->mm;
211

212 213 214 215 216 217 218 219
	if (!priv->task)
		return;

	release_task_mempolicy(priv);
	up_read(&mm->mmap_sem);
	mmput(mm);
	put_task_struct(priv->task);
	priv->task = NULL;
220 221
}

222 223 224 225 226 227 228 229
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;

230
	priv->inode = inode;
231 232 233 234 235 236 237 238
	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;
	}

239 240 241
	return 0;
}

242 243 244 245 246 247 248 249 250 251 252
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);
}

253
static int do_maps_open(struct inode *inode, struct file *file,
254
			const struct seq_operations *ops)
255
{
256 257
	return proc_maps_open(inode, file, ops,
				sizeof(struct proc_maps_private));
258
}
M
Mauricio Lin 已提交
259

260 261 262 263
/*
 * Indicate if the VMA is a stack for the given task; for
 * /proc/PID/maps that is the stack of the main task.
 */
264
static int is_stack(struct vm_area_struct *vma)
265
{
266 267 268 269 270 271 272
	/*
	 * We make no effort to guess what a given thread considers to be
	 * its "stack".  It's not even well-defined for programs written
	 * languages like Go.
	 */
	return vma->vm_start <= vma->vm_mm->start_stack &&
		vma->vm_end >= vma->vm_mm->start_stack;
273 274
}

275 276 277 278 279 280
static void show_vma_header_prefix(struct seq_file *m,
				   unsigned long start, unsigned long end,
				   vm_flags_t flags, unsigned long long pgoff,
				   dev_t dev, unsigned long ino)
{
	seq_setwidth(m, 25 + sizeof(void *) * 6 - 1);
281 282 283 284 285 286 287 288 289 290 291 292
	seq_put_hex_ll(m, NULL, start, 8);
	seq_put_hex_ll(m, "-", end, 8);
	seq_putc(m, ' ');
	seq_putc(m, flags & VM_READ ? 'r' : '-');
	seq_putc(m, flags & VM_WRITE ? 'w' : '-');
	seq_putc(m, flags & VM_EXEC ? 'x' : '-');
	seq_putc(m, flags & VM_MAYSHARE ? 's' : 'p');
	seq_put_hex_ll(m, " ", pgoff, 8);
	seq_put_hex_ll(m, " ", MAJOR(dev), 2);
	seq_put_hex_ll(m, ":", MINOR(dev), 2);
	seq_put_decimal_ull(m, " ", ino);
	seq_putc(m, ' ');
293 294
}

295
static void
296
show_map_vma(struct seq_file *m, struct vm_area_struct *vma)
L
Linus Torvalds 已提交
297
{
M
Mauricio Lin 已提交
298 299
	struct mm_struct *mm = vma->vm_mm;
	struct file *file = vma->vm_file;
300
	vm_flags_t flags = vma->vm_flags;
L
Linus Torvalds 已提交
301
	unsigned long ino = 0;
302
	unsigned long long pgoff = 0;
303
	unsigned long start, end;
L
Linus Torvalds 已提交
304
	dev_t dev = 0;
305
	const char *name = NULL;
L
Linus Torvalds 已提交
306 307

	if (file) {
A
Al Viro 已提交
308
		struct inode *inode = file_inode(vma->vm_file);
L
Linus Torvalds 已提交
309 310
		dev = inode->i_sb->s_dev;
		ino = inode->i_ino;
311
		pgoff = ((loff_t)vma->vm_pgoff) << PAGE_SHIFT;
L
Linus Torvalds 已提交
312 313
	}

314
	start = vma->vm_start;
315
	end = vma->vm_end;
316
	show_vma_header_prefix(m, start, end, flags, pgoff, dev, ino);
L
Linus Torvalds 已提交
317 318 319 320 321

	/*
	 * Print the dentry name for named mappings, and a
	 * special [heap] marker for the heap:
	 */
M
Mauricio Lin 已提交
322
	if (file) {
323
		seq_pad(m, ' ');
M
Miklos Szeredi 已提交
324
		seq_file_path(m, file, "\n");
325 326 327
		goto done;
	}

328 329 330 331 332 333
	if (vma->vm_ops && vma->vm_ops->name) {
		name = vma->vm_ops->name(vma);
		if (name)
			goto done;
	}

334 335 336 337 338 339 340 341 342 343 344 345 346
	name = arch_vma_name(vma);
	if (!name) {
		if (!mm) {
			name = "[vdso]";
			goto done;
		}

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

347
		if (is_stack(vma))
348
			name = "[stack]";
349 350 351 352
	}

done:
	if (name) {
353
		seq_pad(m, ' ');
354
		seq_puts(m, name);
L
Linus Torvalds 已提交
355 356
	}
	seq_putc(m, '\n');
357 358
}

359
static int show_map(struct seq_file *m, void *v)
360
{
361
	show_map_vma(m, v);
362
	m_cache_vma(m, v);
L
Linus Torvalds 已提交
363 364 365
	return 0;
}

366
static const struct seq_operations proc_pid_maps_op = {
367 368 369
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
370
	.show	= show_map
371 372
};

373
static int pid_maps_open(struct inode *inode, struct file *file)
374 375 376 377
{
	return do_maps_open(inode, file, &proc_pid_maps_op);
}

378 379 380 381
const struct file_operations proc_pid_maps_operations = {
	.open		= pid_maps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
382
	.release	= proc_map_release,
383 384
};

385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
/*
 * 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

404
#ifdef CONFIG_PROC_PAGE_MONITOR
P
Peter Zijlstra 已提交
405
struct mem_size_stats {
406 407 408 409 410 411
	unsigned long resident;
	unsigned long shared_clean;
	unsigned long shared_dirty;
	unsigned long private_clean;
	unsigned long private_dirty;
	unsigned long referenced;
412
	unsigned long anonymous;
413
	unsigned long lazyfree;
414
	unsigned long anonymous_thp;
415
	unsigned long shmem_thp;
S
Song Liu 已提交
416
	unsigned long file_thp;
P
Peter Zijlstra 已提交
417
	unsigned long swap;
418 419
	unsigned long shared_hugetlb;
	unsigned long private_hugetlb;
420
	u64 pss;
421 422 423
	u64 pss_anon;
	u64 pss_file;
	u64 pss_shmem;
424
	u64 pss_locked;
425
	u64 swap_pss;
426
	bool check_shmem_swap;
427 428
};

429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
static void smaps_page_accumulate(struct mem_size_stats *mss,
		struct page *page, unsigned long size, unsigned long pss,
		bool dirty, bool locked, bool private)
{
	mss->pss += pss;

	if (PageAnon(page))
		mss->pss_anon += pss;
	else if (PageSwapBacked(page))
		mss->pss_shmem += pss;
	else
		mss->pss_file += pss;

	if (locked)
		mss->pss_locked += pss;

	if (dirty || PageDirty(page)) {
		if (private)
			mss->private_dirty += size;
		else
			mss->shared_dirty += size;
	} else {
		if (private)
			mss->private_clean += size;
		else
			mss->shared_clean += size;
	}
}

458
static void smaps_account(struct mem_size_stats *mss, struct page *page,
459
		bool compound, bool young, bool dirty, bool locked)
460
{
461
	int i, nr = compound ? compound_nr(page) : 1;
462
	unsigned long size = nr * PAGE_SIZE;
463

464 465 466 467
	/*
	 * First accumulate quantities that depend only on |size| and the type
	 * of the compound page.
	 */
468
	if (PageAnon(page)) {
469
		mss->anonymous += size;
470 471 472
		if (!PageSwapBacked(page) && !dirty && !PageDirty(page))
			mss->lazyfree += size;
	}
473 474 475

	mss->resident += size;
	/* Accumulate the size in pages that have been accessed. */
476
	if (young || page_is_young(page) || PageReferenced(page))
477 478
		mss->referenced += size;

479
	/*
480 481 482
	 * Then accumulate quantities that may depend on sharing, or that may
	 * differ page-by-page.
	 *
483 484 485 486 487
	 * page_count(page) == 1 guarantees the page is mapped exactly once.
	 * If any subpage of the compound page mapped with PTE it would elevate
	 * page_count().
	 */
	if (page_count(page) == 1) {
488 489
		smaps_page_accumulate(mss, page, size, size << PSS_SHIFT, dirty,
			locked, true);
490 491 492 493
		return;
	}
	for (i = 0; i < nr; i++, page++) {
		int mapcount = page_mapcount(page);
494 495 496 497 498
		unsigned long pss = PAGE_SIZE << PSS_SHIFT;
		if (mapcount >= 2)
			pss /= mapcount;
		smaps_page_accumulate(mss, page, PAGE_SIZE, pss, dirty, locked,
				      mapcount < 2);
499 500
	}
}
501

502 503
#ifdef CONFIG_SHMEM
static int smaps_pte_hole(unsigned long addr, unsigned long end,
504
			  __always_unused int depth, struct mm_walk *walk)
505 506 507
{
	struct mem_size_stats *mss = walk->private;

508 509
	mss->swap += shmem_partial_swap_usage(
			walk->vma->vm_file->f_mapping, addr, end);
510 511 512

	return 0;
}
513 514 515
#else
#define smaps_pte_hole		NULL
#endif /* CONFIG_SHMEM */
516

517 518
static void smaps_pte_entry(pte_t *pte, unsigned long addr,
		struct mm_walk *walk)
519 520
{
	struct mem_size_stats *mss = walk->private;
521
	struct vm_area_struct *vma = walk->vma;
522
	bool locked = !!(vma->vm_flags & VM_LOCKED);
523
	struct page *page = NULL;
524

525 526 527 528
	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);
529

530 531 532
		if (!non_swap_entry(swpent)) {
			int mapcount;

533
			mss->swap += PAGE_SIZE;
534 535 536 537 538 539 540 541 542 543
			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))
544
			page = migration_entry_to_page(swpent);
545 546
		else if (is_device_private_entry(swpent))
			page = device_private_entry_to_page(swpent);
547 548
	} else if (unlikely(IS_ENABLED(CONFIG_SHMEM) && mss->check_shmem_swap
							&& pte_none(*pte))) {
549 550 551 552 553
		page = find_get_entry(vma->vm_file->f_mapping,
						linear_page_index(vma, addr));
		if (!page)
			return;

554
		if (xa_is_value(page))
555 556
			mss->swap += PAGE_SIZE;
		else
557
			put_page(page);
558 559

		return;
560
	}
561 562 563

	if (!page)
		return;
564

565
	smaps_account(mss, page, false, pte_young(*pte), pte_dirty(*pte), locked);
566 567
}

568 569 570 571 572
#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;
573
	struct vm_area_struct *vma = walk->vma;
574
	bool locked = !!(vma->vm_flags & VM_LOCKED);
575 576 577 578 579 580
	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;
581 582 583 584
	if (PageAnon(page))
		mss->anonymous_thp += HPAGE_PMD_SIZE;
	else if (PageSwapBacked(page))
		mss->shmem_thp += HPAGE_PMD_SIZE;
585 586
	else if (is_zone_device_page(page))
		/* pass */;
587
	else
S
Song Liu 已提交
588
		mss->file_thp += HPAGE_PMD_SIZE;
589
	smaps_account(mss, page, true, pmd_young(*pmd), pmd_dirty(*pmd), locked);
590 591 592 593 594 595 596 597
}
#else
static void smaps_pmd_entry(pmd_t *pmd, unsigned long addr,
		struct mm_walk *walk)
{
}
#endif

598
static int smaps_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
D
Dave Hansen 已提交
599
			   struct mm_walk *walk)
M
Mauricio Lin 已提交
600
{
601
	struct vm_area_struct *vma = walk->vma;
602
	pte_t *pte;
603
	spinlock_t *ptl;
M
Mauricio Lin 已提交
604

605 606
	ptl = pmd_trans_huge_lock(pmd, vma);
	if (ptl) {
607 608
		if (pmd_present(*pmd))
			smaps_pmd_entry(pmd, addr, walk);
609
		spin_unlock(ptl);
610
		goto out;
611
	}
612 613

	if (pmd_trans_unstable(pmd))
614
		goto out;
615 616 617 618 619
	/*
	 * The mmap_sem held all the way back in m_start() is what
	 * keeps khugepaged out of here and from collapsing things
	 * in here.
	 */
620
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
621
	for (; addr != end; pte++, addr += PAGE_SIZE)
622
		smaps_pte_entry(pte, addr, walk);
623
	pte_unmap_unlock(pte - 1, ptl);
624
out:
625
	cond_resched();
626
	return 0;
M
Mauricio Lin 已提交
627 628
}

629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650
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",
651 652 653
#ifdef CONFIG_X86_INTEL_MPX
		[ilog2(VM_MPX)]		= "mp",
#endif
654 655 656 657 658 659 660 661 662
		[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",
J
Jan Kara 已提交
663
		[ilog2(VM_SYNC)]	= "sf",
664
		[ilog2(VM_ARCH_1)]	= "ar",
665
		[ilog2(VM_WIPEONFORK)]	= "wf",
666
		[ilog2(VM_DONTDUMP)]	= "dd",
667 668 669
#ifdef CONFIG_MEM_SOFT_DIRTY
		[ilog2(VM_SOFTDIRTY)]	= "sd",
#endif
670 671 672 673
		[ilog2(VM_MIXEDMAP)]	= "mm",
		[ilog2(VM_HUGEPAGE)]	= "hg",
		[ilog2(VM_NOHUGEPAGE)]	= "nh",
		[ilog2(VM_MERGEABLE)]	= "mg",
674 675
		[ilog2(VM_UFFD_MISSING)]= "um",
		[ilog2(VM_UFFD_WP)]	= "uw",
676
#ifdef CONFIG_ARCH_HAS_PKEYS
677 678 679 680 681
		/* These come out via ProtectionKey: */
		[ilog2(VM_PKEY_BIT0)]	= "",
		[ilog2(VM_PKEY_BIT1)]	= "",
		[ilog2(VM_PKEY_BIT2)]	= "",
		[ilog2(VM_PKEY_BIT3)]	= "",
682 683
#if VM_PKEY_BIT4
		[ilog2(VM_PKEY_BIT4)]	= "",
684
#endif
685
#endif /* CONFIG_ARCH_HAS_PKEYS */
686 687 688 689 690
	};
	size_t i;

	seq_puts(m, "VmFlags: ");
	for (i = 0; i < BITS_PER_LONG; i++) {
691 692
		if (!mnemonics[i][0])
			continue;
693
		if (vma->vm_flags & (1UL << i)) {
694 695 696
			seq_putc(m, mnemonics[i][0]);
			seq_putc(m, mnemonics[i][1]);
			seq_putc(m, ' ');
697 698 699 700 701
		}
	}
	seq_putc(m, '\n');
}

702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
#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);
718 719
		else if (is_device_private_entry(swpent))
			page = device_private_entry_to_page(swpent);
720 721 722 723 724 725 726 727 728 729 730
	}
	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;
}
731 732
#else
#define smaps_hugetlb_range	NULL
733 734
#endif /* HUGETLB_PAGE */

735 736 737 738 739 740 741 742 743 744 745
static const struct mm_walk_ops smaps_walk_ops = {
	.pmd_entry		= smaps_pte_range,
	.hugetlb_entry		= smaps_hugetlb_range,
};

static const struct mm_walk_ops smaps_shmem_walk_ops = {
	.pmd_entry		= smaps_pte_range,
	.hugetlb_entry		= smaps_hugetlb_range,
	.pte_hole		= smaps_pte_hole,
};

746 747
static void smap_gather_stats(struct vm_area_struct *vma,
			     struct mem_size_stats *mss)
M
Mauricio Lin 已提交
748
{
749
#ifdef CONFIG_SHMEM
750 751
	/* In case of smaps_rollup, reset the value from previous vma */
	mss->check_shmem_swap = false;
752
	if (vma->vm_file && shmem_mapping(vma->vm_file->f_mapping)) {
753 754 755 756 757 758 759 760 761 762 763 764 765 766
		/*
		 * For shared or readonly shmem mappings we know that all
		 * swapped out pages belong to the shmem object, and we can
		 * obtain the swap value much more efficiently. For private
		 * writable mappings, we might have COW pages that are
		 * not affected by the parent swapped out pages of the shmem
		 * object, so we have to distinguish them during the page walk.
		 * Unless we know that the shmem object (or the part mapped by
		 * our VMA) has no swapped out pages at all.
		 */
		unsigned long shmem_swapped = shmem_swap_usage(vma);

		if (!shmem_swapped || (vma->vm_flags & VM_SHARED) ||
					!(vma->vm_flags & VM_WRITE)) {
767
			mss->swap += shmem_swapped;
768
		} else {
769
			mss->check_shmem_swap = true;
770 771
			walk_page_vma(vma, &smaps_shmem_walk_ops, mss);
			return;
772
		}
773 774
	}
#endif
775
	/* mmap_sem is held in m_start */
776
	walk_page_vma(vma, &smaps_walk_ops, mss);
777 778 779 780
}

#define SEQ_PUT_DEC(str, val) \
		seq_put_decimal_ull_width(m, str, (val) >> 10, 8)
781 782

/* Show the contents common for smaps and smaps_rollup */
783 784
static void __show_smap(struct seq_file *m, const struct mem_size_stats *mss,
	bool rollup_mode)
785 786 787
{
	SEQ_PUT_DEC("Rss:            ", mss->resident);
	SEQ_PUT_DEC(" kB\nPss:            ", mss->pss >> PSS_SHIFT);
788 789 790 791 792 793 794 795 796 797 798 799
	if (rollup_mode) {
		/*
		 * These are meaningful only for smaps_rollup, otherwise two of
		 * them are zero, and the other one is the same as Pss.
		 */
		SEQ_PUT_DEC(" kB\nPss_Anon:       ",
			mss->pss_anon >> PSS_SHIFT);
		SEQ_PUT_DEC(" kB\nPss_File:       ",
			mss->pss_file >> PSS_SHIFT);
		SEQ_PUT_DEC(" kB\nPss_Shmem:      ",
			mss->pss_shmem >> PSS_SHIFT);
	}
800 801 802 803 804 805 806 807 808
	SEQ_PUT_DEC(" kB\nShared_Clean:   ", mss->shared_clean);
	SEQ_PUT_DEC(" kB\nShared_Dirty:   ", mss->shared_dirty);
	SEQ_PUT_DEC(" kB\nPrivate_Clean:  ", mss->private_clean);
	SEQ_PUT_DEC(" kB\nPrivate_Dirty:  ", mss->private_dirty);
	SEQ_PUT_DEC(" kB\nReferenced:     ", mss->referenced);
	SEQ_PUT_DEC(" kB\nAnonymous:      ", mss->anonymous);
	SEQ_PUT_DEC(" kB\nLazyFree:       ", mss->lazyfree);
	SEQ_PUT_DEC(" kB\nAnonHugePages:  ", mss->anonymous_thp);
	SEQ_PUT_DEC(" kB\nShmemPmdMapped: ", mss->shmem_thp);
S
Song Liu 已提交
809
	SEQ_PUT_DEC(" kB\nFilePmdMapped: ", mss->file_thp);
810 811 812 813 814 815 816 817 818 819 820
	SEQ_PUT_DEC(" kB\nShared_Hugetlb: ", mss->shared_hugetlb);
	seq_put_decimal_ull_width(m, " kB\nPrivate_Hugetlb: ",
				  mss->private_hugetlb >> 10, 7);
	SEQ_PUT_DEC(" kB\nSwap:           ", mss->swap);
	SEQ_PUT_DEC(" kB\nSwapPss:        ",
					mss->swap_pss >> PSS_SHIFT);
	SEQ_PUT_DEC(" kB\nLocked:         ",
					mss->pss_locked >> PSS_SHIFT);
	seq_puts(m, " kB\n");
}

821 822 823
static int show_smap(struct seq_file *m, void *v)
{
	struct vm_area_struct *vma = v;
824 825 826 827 828 829 830 831 832 833 834 835 836
	struct mem_size_stats mss;

	memset(&mss, 0, sizeof(mss));

	smap_gather_stats(vma, &mss);

	show_map_vma(m, vma);

	SEQ_PUT_DEC("Size:           ", vma->vm_end - vma->vm_start);
	SEQ_PUT_DEC(" kB\nKernelPageSize: ", vma_kernel_pagesize(vma));
	SEQ_PUT_DEC(" kB\nMMUPageSize:    ", vma_mmu_pagesize(vma));
	seq_puts(m, " kB\n");

837
	__show_smap(m, &mss, false);
838

839 840
	seq_printf(m, "THPeligible:		%d\n",
		   transparent_hugepage_enabled(vma));
841

842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
	if (arch_pkeys_enabled())
		seq_printf(m, "ProtectionKey:  %8u\n", vma_pkey(vma));
	show_smap_vma_flags(m, vma);

	m_cache_vma(m, vma);

	return 0;
}

static int show_smaps_rollup(struct seq_file *m, void *v)
{
	struct proc_maps_private *priv = m->private;
	struct mem_size_stats mss;
	struct mm_struct *mm;
	struct vm_area_struct *vma;
	unsigned long last_vma_end = 0;
858 859
	int ret = 0;

860 861 862
	priv->task = get_proc_task(priv->inode);
	if (!priv->task)
		return -ESRCH;
863

864 865 866 867
	mm = priv->mm;
	if (!mm || !mmget_not_zero(mm)) {
		ret = -ESRCH;
		goto out_put_task;
868
	}
869

870
	memset(&mss, 0, sizeof(mss));
871

872 873 874 875
	ret = down_read_killable(&mm->mmap_sem);
	if (ret)
		goto out_put_mm;

876
	hold_task_mempolicy(priv);
877

878 879 880
	for (vma = priv->mm->mmap; vma; vma = vma->vm_next) {
		smap_gather_stats(vma, &mss);
		last_vma_end = vma->vm_end;
881
	}
882 883 884 885 886 887

	show_vma_header_prefix(m, priv->mm->mmap->vm_start,
			       last_vma_end, 0, 0, 0, 0);
	seq_pad(m, ' ');
	seq_puts(m, "[rollup]\n");

888
	__show_smap(m, &mss, true);
889 890 891 892

	release_task_mempolicy(priv);
	up_read(&mm->mmap_sem);

893 894
out_put_mm:
	mmput(mm);
895 896 897 898
out_put_task:
	put_task_struct(priv->task);
	priv->task = NULL;

899
	return ret;
M
Mauricio Lin 已提交
900
}
901
#undef SEQ_PUT_DEC
M
Mauricio Lin 已提交
902

903
static const struct seq_operations proc_pid_smaps_op = {
904 905 906
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
907
	.show	= show_smap
908 909
};

910
static int pid_smaps_open(struct inode *inode, struct file *file)
911 912 913 914
{
	return do_maps_open(inode, file, &proc_pid_smaps_op);
}

915
static int smaps_rollup_open(struct inode *inode, struct file *file)
916
{
917
	int ret;
918
	struct proc_maps_private *priv;
919 920 921

	priv = kzalloc(sizeof(*priv), GFP_KERNEL_ACCOUNT);
	if (!priv)
922
		return -ENOMEM;
923 924 925 926 927 928 929 930 931 932 933 934

	ret = single_open(file, show_smaps_rollup, priv);
	if (ret)
		goto out_free;

	priv->inode = inode;
	priv->mm = proc_mem_open(inode, PTRACE_MODE_READ);
	if (IS_ERR(priv->mm)) {
		ret = PTR_ERR(priv->mm);

		single_release(inode, file);
		goto out_free;
935
	}
936

937
	return 0;
938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953

out_free:
	kfree(priv);
	return ret;
}

static int smaps_rollup_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);

	kfree(priv);
	return single_release(inode, file);
954 955
}

956 957 958 959
const struct file_operations proc_pid_smaps_operations = {
	.open		= pid_smaps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
960
	.release	= proc_map_release,
961 962
};

963
const struct file_operations proc_pid_smaps_rollup_operations = {
964
	.open		= smaps_rollup_open,
965 966
	.read		= seq_read,
	.llseek		= seq_lseek,
967
	.release	= smaps_rollup_release,
968 969
};

970 971 972 973
enum clear_refs_types {
	CLEAR_REFS_ALL = 1,
	CLEAR_REFS_ANON,
	CLEAR_REFS_MAPPED,
974
	CLEAR_REFS_SOFT_DIRTY,
975
	CLEAR_REFS_MM_HIWATER_RSS,
976 977 978
	CLEAR_REFS_LAST,
};

979
struct clear_refs_private {
980
	enum clear_refs_types type;
981 982
};

983
#ifdef CONFIG_MEM_SOFT_DIRTY
984 985 986 987 988 989
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
990
	 * Documentation/admin-guide/mm/soft-dirty.rst for full description
991 992 993
	 * of how soft-dirty works.
	 */
	pte_t ptent = *pte;
994 995

	if (pte_present(ptent)) {
996 997 998 999
		pte_t old_pte;

		old_pte = ptep_modify_prot_start(vma, addr, pte);
		ptent = pte_wrprotect(old_pte);
1000
		ptent = pte_clear_soft_dirty(ptent);
1001
		ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent);
1002 1003
	} else if (is_swap_pte(ptent)) {
		ptent = pte_swp_clear_soft_dirty(ptent);
1004
		set_pte_at(vma->vm_mm, addr, pte, ptent);
1005
	}
1006
}
1007 1008 1009 1010 1011 1012
#else
static inline void clear_soft_dirty(struct vm_area_struct *vma,
		unsigned long addr, pte_t *pte)
{
}
#endif
1013

1014
#if defined(CONFIG_MEM_SOFT_DIRTY) && defined(CONFIG_TRANSPARENT_HUGEPAGE)
1015 1016 1017
static inline void clear_soft_dirty_pmd(struct vm_area_struct *vma,
		unsigned long addr, pmd_t *pmdp)
{
1018
	pmd_t old, pmd = *pmdp;
1019

1020 1021
	if (pmd_present(pmd)) {
		/* See comment in change_huge_pmd() */
1022 1023
		old = pmdp_invalidate(vma, addr, pmdp);
		if (pmd_dirty(old))
1024
			pmd = pmd_mkdirty(pmd);
1025
		if (pmd_young(old))
1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
			pmd = pmd_mkyoung(pmd);

		pmd = pmd_wrprotect(pmd);
		pmd = pmd_clear_soft_dirty(pmd);

		set_pmd_at(vma->vm_mm, addr, pmdp, pmd);
	} else if (is_migration_entry(pmd_to_swp_entry(pmd))) {
		pmd = pmd_swp_clear_soft_dirty(pmd);
		set_pmd_at(vma->vm_mm, addr, pmdp, pmd);
	}
1036 1037 1038 1039 1040 1041 1042 1043
}
#else
static inline void clear_soft_dirty_pmd(struct vm_area_struct *vma,
		unsigned long addr, pmd_t *pmdp)
{
}
#endif

1044
static int clear_refs_pte_range(pmd_t *pmd, unsigned long addr,
D
Dave Hansen 已提交
1045
				unsigned long end, struct mm_walk *walk)
1046
{
1047
	struct clear_refs_private *cp = walk->private;
1048
	struct vm_area_struct *vma = walk->vma;
1049 1050 1051 1052
	pte_t *pte, ptent;
	spinlock_t *ptl;
	struct page *page;

1053 1054
	ptl = pmd_trans_huge_lock(pmd, vma);
	if (ptl) {
1055 1056 1057 1058 1059
		if (cp->type == CLEAR_REFS_SOFT_DIRTY) {
			clear_soft_dirty_pmd(vma, addr, pmd);
			goto out;
		}

1060 1061 1062
		if (!pmd_present(*pmd))
			goto out;

1063 1064 1065 1066
		page = pmd_page(*pmd);

		/* Clear accessed and referenced bits. */
		pmdp_test_and_clear_young(vma, addr, pmd);
1067
		test_and_clear_page_young(page);
1068 1069 1070 1071 1072 1073
		ClearPageReferenced(page);
out:
		spin_unlock(ptl);
		return 0;
	}

1074 1075
	if (pmd_trans_unstable(pmd))
		return 0;
1076

1077 1078 1079 1080
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	for (; addr != end; pte++, addr += PAGE_SIZE) {
		ptent = *pte;

1081 1082 1083 1084 1085
		if (cp->type == CLEAR_REFS_SOFT_DIRTY) {
			clear_soft_dirty(vma, addr, pte);
			continue;
		}

1086 1087 1088
		if (!pte_present(ptent))
			continue;

1089 1090 1091 1092 1093 1094
		page = vm_normal_page(vma, addr, ptent);
		if (!page)
			continue;

		/* Clear accessed and referenced bits. */
		ptep_test_and_clear_young(vma, addr, pte);
1095
		test_and_clear_page_young(page);
1096 1097 1098 1099 1100 1101 1102
		ClearPageReferenced(page);
	}
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();
	return 0;
}

1103 1104 1105 1106 1107 1108
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;

1109 1110 1111
	if (vma->vm_flags & VM_PFNMAP)
		return 1;

1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
	/*
	 * 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;
}

1125 1126 1127 1128 1129
static const struct mm_walk_ops clear_refs_walk_ops = {
	.pmd_entry		= clear_refs_pte_range,
	.test_walk		= clear_refs_test_walk,
};

1130 1131
static ssize_t clear_refs_write(struct file *file, const char __user *buf,
				size_t count, loff_t *ppos)
1132
{
1133
	struct task_struct *task;
1134
	char buffer[PROC_NUMBUF];
1135
	struct mm_struct *mm;
1136
	struct vm_area_struct *vma;
1137
	enum clear_refs_types type;
M
Minchan Kim 已提交
1138
	struct mmu_gather tlb;
1139
	int itype;
A
Alexey Dobriyan 已提交
1140
	int rv;
1141

1142 1143 1144 1145 1146
	memset(buffer, 0, sizeof(buffer));
	if (count > sizeof(buffer) - 1)
		count = sizeof(buffer) - 1;
	if (copy_from_user(buffer, buf, count))
		return -EFAULT;
1147
	rv = kstrtoint(strstrip(buffer), 10, &itype);
A
Alexey Dobriyan 已提交
1148 1149
	if (rv < 0)
		return rv;
1150 1151
	type = (enum clear_refs_types)itype;
	if (type < CLEAR_REFS_ALL || type >= CLEAR_REFS_LAST)
1152
		return -EINVAL;
1153

A
Al Viro 已提交
1154
	task = get_proc_task(file_inode(file));
1155 1156 1157 1158
	if (!task)
		return -ESRCH;
	mm = get_task_mm(task);
	if (mm) {
1159
		struct mmu_notifier_range range;
1160
		struct clear_refs_private cp = {
1161
			.type = type,
1162
		};
1163 1164

		if (type == CLEAR_REFS_MM_HIWATER_RSS) {
1165 1166 1167 1168 1169
			if (down_write_killable(&mm->mmap_sem)) {
				count = -EINTR;
				goto out_mm;
			}

1170 1171 1172 1173 1174 1175 1176 1177 1178
			/*
			 * Writing 5 to /proc/pid/clear_refs resets the peak
			 * resident set size to this mm's current rss value.
			 */
			reset_mm_hiwater_rss(mm);
			up_write(&mm->mmap_sem);
			goto out_mm;
		}

1179 1180 1181 1182
		if (down_read_killable(&mm->mmap_sem)) {
			count = -EINTR;
			goto out_mm;
		}
M
Minchan Kim 已提交
1183
		tlb_gather_mmu(&tlb, mm, 0, -1);
1184 1185 1186 1187 1188
		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);
1189 1190 1191 1192
				if (down_write_killable(&mm->mmap_sem)) {
					count = -EINTR;
					goto out_mm;
				}
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210
				/*
				 * Avoid to modify vma->vm_flags
				 * without locked ops while the
				 * coredump reads the vm_flags.
				 */
				if (!mmget_still_valid(mm)) {
					/*
					 * Silently return "count"
					 * like if get_task_mm()
					 * failed. FIXME: should this
					 * function have returned
					 * -ESRCH if get_task_mm()
					 * failed like if
					 * get_proc_task() fails?
					 */
					up_write(&mm->mmap_sem);
					goto out_mm;
				}
1211 1212 1213 1214 1215 1216 1217
				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;
			}
1218

1219 1220
			mmu_notifier_range_init(&range, MMU_NOTIFY_SOFT_DIRTY,
						0, NULL, mm, 0, -1UL);
1221
			mmu_notifier_invalidate_range_start(&range);
1222
		}
1223 1224
		walk_page_range(mm, 0, mm->highest_vm_end, &clear_refs_walk_ops,
				&cp);
1225
		if (type == CLEAR_REFS_SOFT_DIRTY)
1226
			mmu_notifier_invalidate_range_end(&range);
M
Minchan Kim 已提交
1227
		tlb_finish_mmu(&tlb, 0, -1);
1228
		up_read(&mm->mmap_sem);
1229
out_mm:
1230 1231 1232
		mmput(mm);
	}
	put_task_struct(task);
1233 1234

	return count;
1235 1236
}

1237 1238
const struct file_operations proc_clear_refs_operations = {
	.write		= clear_refs_write,
1239
	.llseek		= noop_llseek,
1240 1241
};

1242 1243 1244 1245
typedef struct {
	u64 pme;
} pagemap_entry_t;

1246
struct pagemapread {
1247
	int pos, len;		/* units: PM_ENTRY_BYTES, not bytes */
1248
	pagemap_entry_t *buffer;
1249
	bool show_pfn;
1250 1251
};

1252 1253 1254
#define PAGEMAP_WALK_SIZE	(PMD_SIZE)
#define PAGEMAP_WALK_MASK	(PMD_MASK)

1255 1256 1257 1258
#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)
1259
#define PM_MMAP_EXCLUSIVE	BIT_ULL(56)
1260 1261 1262 1263
#define PM_FILE			BIT_ULL(61)
#define PM_SWAP			BIT_ULL(62)
#define PM_PRESENT		BIT_ULL(63)

1264 1265
#define PM_END_OF_BUFFER    1

1266
static inline pagemap_entry_t make_pme(u64 frame, u64 flags)
1267
{
1268
	return (pagemap_entry_t) { .pme = (frame & PM_PFRAME_MASK) | flags };
1269 1270 1271
}

static int add_to_pagemap(unsigned long addr, pagemap_entry_t *pme,
1272 1273
			  struct pagemapread *pm)
{
1274
	pm->buffer[pm->pos++] = *pme;
1275
	if (pm->pos >= pm->len)
1276
		return PM_END_OF_BUFFER;
1277 1278 1279 1280
	return 0;
}

static int pagemap_pte_hole(unsigned long start, unsigned long end,
1281
			    __always_unused int depth, struct mm_walk *walk)
1282
{
D
Dave Hansen 已提交
1283
	struct pagemapread *pm = walk->private;
1284
	unsigned long addr = start;
1285
	int err = 0;
1286

1287 1288
	while (addr < end) {
		struct vm_area_struct *vma = find_vma(walk->mm, addr);
1289
		pagemap_entry_t pme = make_pme(0, 0);
1290 1291
		/* End of address space hole, which we mark as non-present. */
		unsigned long hole_end;
1292

1293 1294 1295 1296 1297 1298 1299 1300 1301
		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;
1302 1303
		}

1304 1305 1306 1307 1308
		if (!vma)
			break;

		/* Addresses in the VMA. */
		if (vma->vm_flags & VM_SOFTDIRTY)
1309
			pme = make_pme(0, PM_SOFT_DIRTY);
1310
		for (; addr < min(end, vma->vm_end); addr += PAGE_SIZE) {
1311 1312 1313 1314
			err = add_to_pagemap(addr, &pme, pm);
			if (err)
				goto out;
		}
1315
	}
1316
out:
1317 1318 1319
	return err;
}

1320
static pagemap_entry_t pte_to_pagemap_entry(struct pagemapread *pm,
1321
		struct vm_area_struct *vma, unsigned long addr, pte_t pte)
1322
{
1323
	u64 frame = 0, flags = 0;
1324
	struct page *page = NULL;
1325

1326
	if (pte_present(pte)) {
1327 1328
		if (pm->show_pfn)
			frame = pte_pfn(pte);
1329
		flags |= PM_PRESENT;
1330
		page = vm_normal_page(vma, addr, pte);
1331
		if (pte_soft_dirty(pte))
1332
			flags |= PM_SOFT_DIRTY;
1333
	} else if (is_swap_pte(pte)) {
1334 1335
		swp_entry_t entry;
		if (pte_swp_soft_dirty(pte))
1336
			flags |= PM_SOFT_DIRTY;
1337
		entry = pte_to_swp_entry(pte);
1338 1339 1340
		if (pm->show_pfn)
			frame = swp_type(entry) |
				(swp_offset(entry) << MAX_SWAPFILES_SHIFT);
1341
		flags |= PM_SWAP;
1342 1343
		if (is_migration_entry(entry))
			page = migration_entry_to_page(entry);
1344 1345 1346

		if (is_device_private_entry(entry))
			page = device_private_entry_to_page(entry);
1347 1348 1349 1350
	}

	if (page && !PageAnon(page))
		flags |= PM_FILE;
1351 1352
	if (page && page_mapcount(page) == 1)
		flags |= PM_MMAP_EXCLUSIVE;
1353 1354
	if (vma->vm_flags & VM_SOFTDIRTY)
		flags |= PM_SOFT_DIRTY;
1355

1356
	return make_pme(frame, flags);
1357 1358
}

1359
static int pagemap_pmd_range(pmd_t *pmdp, unsigned long addr, unsigned long end,
D
Dave Hansen 已提交
1360
			     struct mm_walk *walk)
1361
{
1362
	struct vm_area_struct *vma = walk->vma;
D
Dave Hansen 已提交
1363
	struct pagemapread *pm = walk->private;
1364
	spinlock_t *ptl;
1365
	pte_t *pte, *orig_pte;
1366 1367
	int err = 0;

1368
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
1369 1370
	ptl = pmd_trans_huge_lock(pmdp, vma);
	if (ptl) {
1371 1372
		u64 flags = 0, frame = 0;
		pmd_t pmd = *pmdp;
1373
		struct page *page = NULL;
1374

1375
		if (vma->vm_flags & VM_SOFTDIRTY)
1376
			flags |= PM_SOFT_DIRTY;
1377

1378
		if (pmd_present(pmd)) {
1379
			page = pmd_page(pmd);
1380

1381
			flags |= PM_PRESENT;
1382 1383
			if (pmd_soft_dirty(pmd))
				flags |= PM_SOFT_DIRTY;
1384 1385 1386
			if (pm->show_pfn)
				frame = pmd_pfn(pmd) +
					((addr & ~PMD_MASK) >> PAGE_SHIFT);
1387
		}
1388 1389 1390
#ifdef CONFIG_ARCH_ENABLE_THP_MIGRATION
		else if (is_swap_pmd(pmd)) {
			swp_entry_t entry = pmd_to_swp_entry(pmd);
1391
			unsigned long offset;
1392

1393 1394 1395 1396 1397 1398
			if (pm->show_pfn) {
				offset = swp_offset(entry) +
					((addr & ~PMD_MASK) >> PAGE_SHIFT);
				frame = swp_type(entry) |
					(offset << MAX_SWAPFILES_SHIFT);
			}
1399
			flags |= PM_SWAP;
1400 1401
			if (pmd_swp_soft_dirty(pmd))
				flags |= PM_SOFT_DIRTY;
1402 1403 1404 1405 1406 1407 1408
			VM_BUG_ON(!is_pmd_migration_entry(pmd));
			page = migration_entry_to_page(entry);
		}
#endif

		if (page && page_mapcount(page) == 1)
			flags |= PM_MMAP_EXCLUSIVE;
1409

1410
		for (; addr != end; addr += PAGE_SIZE) {
1411
			pagemap_entry_t pme = make_pme(frame, flags);
1412

1413
			err = add_to_pagemap(addr, &pme, pm);
1414 1415
			if (err)
				break;
1416 1417 1418 1419 1420 1421
			if (pm->show_pfn) {
				if (flags & PM_PRESENT)
					frame++;
				else if (flags & PM_SWAP)
					frame += (1 << MAX_SWAPFILES_SHIFT);
			}
1422
		}
1423
		spin_unlock(ptl);
1424
		return err;
1425 1426
	}

1427
	if (pmd_trans_unstable(pmdp))
1428
		return 0;
1429
#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
1430

1431 1432 1433 1434
	/*
	 * We can assume that @vma always points to a valid one and @end never
	 * goes beyond vma->vm_end.
	 */
1435
	orig_pte = pte = pte_offset_map_lock(walk->mm, pmdp, addr, &ptl);
1436 1437
	for (; addr < end; pte++, addr += PAGE_SIZE) {
		pagemap_entry_t pme;
1438

1439
		pme = pte_to_pagemap_entry(pm, vma, addr, *pte);
1440
		err = add_to_pagemap(addr, &pme, pm);
1441
		if (err)
1442
			break;
1443
	}
1444
	pte_unmap_unlock(orig_pte, ptl);
1445 1446 1447 1448 1449 1450

	cond_resched();

	return err;
}

1451
#ifdef CONFIG_HUGETLB_PAGE
1452
/* This function walks within one hugetlb entry in the single call */
1453
static int pagemap_hugetlb_range(pte_t *ptep, unsigned long hmask,
1454 1455
				 unsigned long addr, unsigned long end,
				 struct mm_walk *walk)
1456 1457
{
	struct pagemapread *pm = walk->private;
1458
	struct vm_area_struct *vma = walk->vma;
1459
	u64 flags = 0, frame = 0;
1460
	int err = 0;
1461
	pte_t pte;
1462

1463
	if (vma->vm_flags & VM_SOFTDIRTY)
1464
		flags |= PM_SOFT_DIRTY;
1465

1466 1467 1468 1469 1470 1471 1472
	pte = huge_ptep_get(ptep);
	if (pte_present(pte)) {
		struct page *page = pte_page(pte);

		if (!PageAnon(page))
			flags |= PM_FILE;

1473 1474 1475
		if (page_mapcount(page) == 1)
			flags |= PM_MMAP_EXCLUSIVE;

1476
		flags |= PM_PRESENT;
1477 1478 1479
		if (pm->show_pfn)
			frame = pte_pfn(pte) +
				((addr & ~hmask) >> PAGE_SHIFT);
1480 1481
	}

1482
	for (; addr != end; addr += PAGE_SIZE) {
1483 1484
		pagemap_entry_t pme = make_pme(frame, flags);

1485
		err = add_to_pagemap(addr, &pme, pm);
1486 1487
		if (err)
			return err;
1488
		if (pm->show_pfn && (flags & PM_PRESENT))
1489
			frame++;
1490 1491 1492 1493 1494 1495
	}

	cond_resched();

	return err;
}
1496 1497
#else
#define pagemap_hugetlb_range	NULL
1498
#endif /* HUGETLB_PAGE */
1499

1500 1501 1502 1503 1504 1505
static const struct mm_walk_ops pagemap_ops = {
	.pmd_entry	= pagemap_pmd_range,
	.pte_hole	= pagemap_pte_hole,
	.hugetlb_entry	= pagemap_hugetlb_range,
};

1506 1507 1508
/*
 * /proc/pid/pagemap - an array mapping virtual pages to pfns
 *
1509 1510 1511
 * For each page in the address space, this file contains one 64-bit entry
 * consisting of the following:
 *
1512
 * Bits 0-54  page frame number (PFN) if present
1513
 * Bits 0-4   swap type if swapped
1514
 * Bits 5-54  swap offset if swapped
1515
 * Bit  55    pte is soft-dirty (see Documentation/admin-guide/mm/soft-dirty.rst)
1516 1517
 * Bit  56    page exclusively mapped
 * Bits 57-60 zero
1518
 * Bit  61    page is file-page or shared-anon
1519 1520 1521 1522 1523 1524
 * Bit  62    page swapped
 * Bit  63    page present
 *
 * If the page is not present but in swap, then the PFN contains an
 * encoding of the swap file number and the page's offset into the
 * swap. Unmapped pages return a null PFN. This allows determining
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534
 * precisely which pages are mapped (or in swap) and comparing mapped
 * pages between processes.
 *
 * Efficient users of this interface will use /proc/pid/maps to
 * determine which areas of memory are actually mapped and llseek to
 * skip over unmapped regions.
 */
static ssize_t pagemap_read(struct file *file, char __user *buf,
			    size_t count, loff_t *ppos)
{
1535
	struct mm_struct *mm = file->private_data;
1536
	struct pagemapread pm;
1537 1538 1539 1540
	unsigned long src;
	unsigned long svpfn;
	unsigned long start_vaddr;
	unsigned long end_vaddr;
1541
	int ret = 0, copied = 0;
1542

V
Vegard Nossum 已提交
1543
	if (!mm || !mmget_not_zero(mm))
1544 1545 1546 1547
		goto out;

	ret = -EINVAL;
	/* file position must be aligned */
1548
	if ((*ppos % PM_ENTRY_BYTES) || (count % PM_ENTRY_BYTES))
1549
		goto out_mm;
1550 1551

	ret = 0;
1552
	if (!count)
1553
		goto out_mm;
1554

1555 1556 1557
	/* do not disclose physical addresses: attack vector */
	pm.show_pfn = file_ns_capable(file, &init_user_ns, CAP_SYS_ADMIN);

1558
	pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
1559
	pm.buffer = kmalloc_array(pm.len, PM_ENTRY_BYTES, GFP_KERNEL);
1560
	ret = -ENOMEM;
1561
	if (!pm.buffer)
1562
		goto out_mm;
1563

1564 1565 1566
	src = *ppos;
	svpfn = src / PM_ENTRY_BYTES;
	start_vaddr = svpfn << PAGE_SHIFT;
1567
	end_vaddr = mm->task_size;
1568 1569

	/* watch out for wraparound */
1570
	if (svpfn > mm->task_size >> PAGE_SHIFT)
1571 1572 1573 1574 1575 1576 1577 1578
		start_vaddr = end_vaddr;

	/*
	 * The odds are that this will stop walking way
	 * before end_vaddr, because the length of the
	 * user buffer is tracked in "pm", and the walk
	 * will stop when we hit the end of the buffer.
	 */
1579 1580 1581 1582 1583 1584
	ret = 0;
	while (count && (start_vaddr < end_vaddr)) {
		int len;
		unsigned long end;

		pm.pos = 0;
1585
		end = (start_vaddr + PAGEMAP_WALK_SIZE) & PAGEMAP_WALK_MASK;
1586 1587 1588
		/* overflow ? */
		if (end < start_vaddr || end > end_vaddr)
			end = end_vaddr;
1589 1590 1591
		ret = down_read_killable(&mm->mmap_sem);
		if (ret)
			goto out_free;
1592
		ret = walk_page_range(mm, start_vaddr, end, &pagemap_ops, &pm);
1593 1594 1595 1596
		up_read(&mm->mmap_sem);
		start_vaddr = end;

		len = min(count, PM_ENTRY_BYTES * pm.pos);
1597
		if (copy_to_user(buf, pm.buffer, len)) {
1598
			ret = -EFAULT;
1599
			goto out_free;
1600 1601 1602 1603
		}
		copied += len;
		buf += len;
		count -= len;
1604
	}
1605 1606 1607 1608
	*ppos += copied;
	if (!ret || ret == PM_END_OF_BUFFER)
		ret = copied;

1609 1610
out_free:
	kfree(pm.buffer);
1611 1612
out_mm:
	mmput(mm);
1613 1614 1615 1616
out:
	return ret;
}

1617 1618
static int pagemap_open(struct inode *inode, struct file *file)
{
1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633
	struct mm_struct *mm;

	mm = proc_mem_open(inode, PTRACE_MODE_READ);
	if (IS_ERR(mm))
		return PTR_ERR(mm);
	file->private_data = mm;
	return 0;
}

static int pagemap_release(struct inode *inode, struct file *file)
{
	struct mm_struct *mm = file->private_data;

	if (mm)
		mmdrop(mm);
1634 1635 1636
	return 0;
}

1637 1638 1639
const struct file_operations proc_pagemap_operations = {
	.llseek		= mem_lseek, /* borrow this */
	.read		= pagemap_read,
1640
	.open		= pagemap_open,
1641
	.release	= pagemap_release,
1642
};
1643
#endif /* CONFIG_PROC_PAGE_MONITOR */
1644

1645 1646
#ifdef CONFIG_NUMA

1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
struct numa_maps {
	unsigned long pages;
	unsigned long anon;
	unsigned long active;
	unsigned long writeback;
	unsigned long mapcount_max;
	unsigned long dirty;
	unsigned long swapcache;
	unsigned long node[MAX_NUMNODES];
};

1658 1659 1660 1661 1662
struct numa_maps_private {
	struct proc_maps_private proc_maps;
	struct numa_maps md;
};

1663 1664
static void gather_stats(struct page *page, struct numa_maps *md, int pte_dirty,
			unsigned long nr_pages)
1665 1666 1667
{
	int count = page_mapcount(page);

1668
	md->pages += nr_pages;
1669
	if (pte_dirty || PageDirty(page))
1670
		md->dirty += nr_pages;
1671 1672

	if (PageSwapCache(page))
1673
		md->swapcache += nr_pages;
1674 1675

	if (PageActive(page) || PageUnevictable(page))
1676
		md->active += nr_pages;
1677 1678

	if (PageWriteback(page))
1679
		md->writeback += nr_pages;
1680 1681

	if (PageAnon(page))
1682
		md->anon += nr_pages;
1683 1684 1685 1686

	if (count > md->mapcount_max)
		md->mapcount_max = count;

1687
	md->node[page_to_nid(page)] += nr_pages;
1688 1689
}

1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
static struct page *can_gather_numa_stats(pte_t pte, struct vm_area_struct *vma,
		unsigned long addr)
{
	struct page *page;
	int nid;

	if (!pte_present(pte))
		return NULL;

	page = vm_normal_page(vma, addr, pte);
	if (!page)
		return NULL;

	if (PageReserved(page))
		return NULL;

	nid = page_to_nid(page);
1707
	if (!node_isset(nid, node_states[N_MEMORY]))
1708 1709 1710 1711 1712
		return NULL;

	return page;
}

1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
static struct page *can_gather_numa_stats_pmd(pmd_t pmd,
					      struct vm_area_struct *vma,
					      unsigned long addr)
{
	struct page *page;
	int nid;

	if (!pmd_present(pmd))
		return NULL;

	page = vm_normal_page_pmd(vma, addr, pmd);
	if (!page)
		return NULL;

	if (PageReserved(page))
		return NULL;

	nid = page_to_nid(page);
	if (!node_isset(nid, node_states[N_MEMORY]))
		return NULL;

	return page;
}
#endif

1739 1740 1741
static int gather_pte_stats(pmd_t *pmd, unsigned long addr,
		unsigned long end, struct mm_walk *walk)
{
1742 1743
	struct numa_maps *md = walk->private;
	struct vm_area_struct *vma = walk->vma;
1744 1745 1746 1747
	spinlock_t *ptl;
	pte_t *orig_pte;
	pte_t *pte;

1748
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
1749 1750
	ptl = pmd_trans_huge_lock(pmd, vma);
	if (ptl) {
1751 1752
		struct page *page;

1753
		page = can_gather_numa_stats_pmd(*pmd, vma, addr);
1754
		if (page)
1755
			gather_stats(page, md, pmd_dirty(*pmd),
1756
				     HPAGE_PMD_SIZE/PAGE_SIZE);
1757
		spin_unlock(ptl);
1758
		return 0;
1759 1760
	}

1761 1762
	if (pmd_trans_unstable(pmd))
		return 0;
1763
#endif
1764 1765
	orig_pte = pte = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
	do {
1766
		struct page *page = can_gather_numa_stats(*pte, vma, addr);
1767 1768
		if (!page)
			continue;
1769
		gather_stats(page, md, pte_dirty(*pte), 1);
1770 1771 1772

	} while (pte++, addr += PAGE_SIZE, addr != end);
	pte_unmap_unlock(orig_pte, ptl);
1773
	cond_resched();
1774 1775 1776
	return 0;
}
#ifdef CONFIG_HUGETLB_PAGE
1777
static int gather_hugetlb_stats(pte_t *pte, unsigned long hmask,
1778 1779
		unsigned long addr, unsigned long end, struct mm_walk *walk)
{
1780
	pte_t huge_pte = huge_ptep_get(pte);
1781 1782 1783
	struct numa_maps *md;
	struct page *page;

1784
	if (!pte_present(huge_pte))
1785 1786
		return 0;

1787
	page = pte_page(huge_pte);
1788 1789 1790 1791
	if (!page)
		return 0;

	md = walk->private;
1792
	gather_stats(page, md, pte_dirty(huge_pte), 1);
1793 1794 1795 1796
	return 0;
}

#else
1797
static int gather_hugetlb_stats(pte_t *pte, unsigned long hmask,
1798 1799 1800 1801 1802 1803
		unsigned long addr, unsigned long end, struct mm_walk *walk)
{
	return 0;
}
#endif

1804 1805 1806 1807 1808
static const struct mm_walk_ops show_numa_ops = {
	.hugetlb_entry = gather_hugetlb_stats,
	.pmd_entry = gather_pte_stats,
};

1809 1810 1811
/*
 * Display pages allocated per node and memory policy via /proc.
 */
1812
static int show_numa_map(struct seq_file *m, void *v)
1813
{
1814 1815
	struct numa_maps_private *numa_priv = m->private;
	struct proc_maps_private *proc_priv = &numa_priv->proc_maps;
1816
	struct vm_area_struct *vma = v;
1817
	struct numa_maps *md = &numa_priv->md;
1818 1819 1820
	struct file *file = vma->vm_file;
	struct mm_struct *mm = vma->vm_mm;
	struct mempolicy *pol;
1821 1822
	char buffer[64];
	int nid;
1823 1824 1825 1826

	if (!mm)
		return 0;

1827 1828
	/* Ensure we start with an empty set of numa_maps statistics. */
	memset(md, 0, sizeof(*md));
1829

1830 1831 1832 1833 1834 1835 1836
	pol = __get_vma_policy(vma, vma->vm_start);
	if (pol) {
		mpol_to_str(buffer, sizeof(buffer), pol);
		mpol_cond_put(pol);
	} else {
		mpol_to_str(buffer, sizeof(buffer), proc_priv->task_mempolicy);
	}
1837 1838 1839 1840

	seq_printf(m, "%08lx %s", vma->vm_start, buffer);

	if (file) {
1841
		seq_puts(m, " file=");
M
Miklos Szeredi 已提交
1842
		seq_file_path(m, file, "\n\t= ");
1843
	} else if (vma->vm_start <= mm->brk && vma->vm_end >= mm->start_brk) {
1844
		seq_puts(m, " heap");
1845
	} else if (is_stack(vma)) {
1846
		seq_puts(m, " stack");
1847 1848
	}

1849
	if (is_vm_hugetlb_page(vma))
1850
		seq_puts(m, " huge");
1851

1852
	/* mmap_sem is held by m_start */
1853
	walk_page_vma(vma, &show_numa_ops, md);
1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878

	if (!md->pages)
		goto out;

	if (md->anon)
		seq_printf(m, " anon=%lu", md->anon);

	if (md->dirty)
		seq_printf(m, " dirty=%lu", md->dirty);

	if (md->pages != md->anon && md->pages != md->dirty)
		seq_printf(m, " mapped=%lu", md->pages);

	if (md->mapcount_max > 1)
		seq_printf(m, " mapmax=%lu", md->mapcount_max);

	if (md->swapcache)
		seq_printf(m, " swapcache=%lu", md->swapcache);

	if (md->active < md->pages && !is_vm_hugetlb_page(vma))
		seq_printf(m, " active=%lu", md->active);

	if (md->writeback)
		seq_printf(m, " writeback=%lu", md->writeback);

1879 1880 1881
	for_each_node_state(nid, N_MEMORY)
		if (md->node[nid])
			seq_printf(m, " N%d=%lu", nid, md->node[nid]);
1882 1883

	seq_printf(m, " kernelpagesize_kB=%lu", vma_kernel_pagesize(vma) >> 10);
1884 1885
out:
	seq_putc(m, '\n');
1886
	m_cache_vma(m, vma);
1887 1888
	return 0;
}
1889

1890
static const struct seq_operations proc_pid_numa_maps_op = {
1891 1892 1893
	.start  = m_start,
	.next   = m_next,
	.stop   = m_stop,
1894
	.show   = show_numa_map,
1895
};
1896

1897 1898
static int pid_numa_maps_open(struct inode *inode, struct file *file)
{
1899 1900
	return proc_maps_open(inode, file, &proc_pid_numa_maps_op,
				sizeof(struct numa_maps_private));
1901 1902 1903 1904 1905 1906
}

const struct file_operations proc_pid_numa_maps_operations = {
	.open		= pid_numa_maps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
1907
	.release	= proc_map_release,
1908 1909
};

1910
#endif /* CONFIG_NUMA */