inode.c 24.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * hugetlbpage-backed filesystem.  Based on ramfs.
 *
 * William Irwin, 2002
 *
 * Copyright (C) 2002 Linus Torvalds.
 */

#include <linux/module.h>
#include <linux/thread_info.h>
#include <asm/current.h>
#include <linux/sched.h>		/* remove ASAP */
#include <linux/fs.h>
#include <linux/mount.h>
#include <linux/file.h>
16
#include <linux/kernel.h>
L
Linus Torvalds 已提交
17 18 19 20 21
#include <linux/writeback.h>
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include <linux/init.h>
#include <linux/string.h>
22
#include <linux/capability.h>
23
#include <linux/ctype.h>
L
Linus Torvalds 已提交
24 25 26
#include <linux/backing-dev.h>
#include <linux/hugetlb.h>
#include <linux/pagevec.h>
27
#include <linux/parser.h>
28
#include <linux/mman.h>
L
Linus Torvalds 已提交
29 30 31 32
#include <linux/slab.h>
#include <linux/dnotify.h>
#include <linux/statfs.h>
#include <linux/security.h>
M
Mimi Zohar 已提交
33
#include <linux/ima.h>
L
Linus Torvalds 已提交
34 35 36 37 38 39

#include <asm/uaccess.h>

/* some random number */
#define HUGETLBFS_MAGIC	0x958458f6

40
static const struct super_operations hugetlbfs_ops;
41
static const struct address_space_operations hugetlbfs_aops;
42
const struct file_operations hugetlbfs_file_operations;
43 44
static const struct inode_operations hugetlbfs_dir_inode_operations;
static const struct inode_operations hugetlbfs_inode_operations;
L
Linus Torvalds 已提交
45 46 47

static struct backing_dev_info hugetlbfs_backing_dev_info = {
	.ra_pages	= 0,	/* No readahead */
48
	.capabilities	= BDI_CAP_NO_ACCT_AND_WRITEBACK,
L
Linus Torvalds 已提交
49 50 51 52
};

int sysctl_hugetlb_shm_group;

53 54 55
enum {
	Opt_size, Opt_nr_inodes,
	Opt_mode, Opt_uid, Opt_gid,
56
	Opt_pagesize,
57 58 59
	Opt_err,
};

60
static const match_table_t tokens = {
61 62 63 64 65
	{Opt_size,	"size=%s"},
	{Opt_nr_inodes,	"nr_inodes=%s"},
	{Opt_mode,	"mode=%o"},
	{Opt_uid,	"uid=%u"},
	{Opt_gid,	"gid=%u"},
66
	{Opt_pagesize,	"pagesize=%s"},
67 68 69
	{Opt_err,	NULL},
};

70 71 72 73 74 75 76 77 78 79
static void huge_pagevec_release(struct pagevec *pvec)
{
	int i;

	for (i = 0; i < pagevec_count(pvec); ++i)
		put_page(pvec->pages[i]);

	pagevec_reinit(pvec);
}

L
Linus Torvalds 已提交
80 81
static int hugetlbfs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
82
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
83 84
	loff_t len, vma_len;
	int ret;
85
	struct hstate *h = hstate_file(file);
L
Linus Torvalds 已提交
86

87
	/*
88 89 90 91 92 93
	 * vma address alignment (but not the pgoff alignment) has
	 * already been checked by prepare_hugepage_range.  If you add
	 * any error returns here, do so after setting VM_HUGETLB, so
	 * is_vm_hugetlb_page tests below unmap_region go the right
	 * way when do_mmap_pgoff unwinds (may be important on powerpc
	 * and ia64).
94 95 96
	 */
	vma->vm_flags |= VM_HUGETLB | VM_RESERVED;
	vma->vm_ops = &hugetlb_vm_ops;
L
Linus Torvalds 已提交
97

98
	if (vma->vm_pgoff & ~(huge_page_mask(h) >> PAGE_SHIFT))
99 100
		return -EINVAL;

L
Linus Torvalds 已提交
101 102
	vma_len = (loff_t)(vma->vm_end - vma->vm_start);

103
	mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
104 105 106 107 108
	file_accessed(file);

	ret = -ENOMEM;
	len = vma_len + ((loff_t)vma->vm_pgoff << PAGE_SHIFT);

109
	if (hugetlb_reserve_pages(inode,
110
				vma->vm_pgoff >> huge_page_order(h),
111 112
				len >> huge_page_shift(h), vma,
				vma->vm_flags))
113
		goto out;
114

A
Adam Litke 已提交
115 116
	ret = 0;
	hugetlb_prefault_arch_hook(vma->vm_mm);
117
	if (vma->vm_flags & VM_WRITE && inode->i_size < len)
L
Linus Torvalds 已提交
118 119
		inode->i_size = len;
out:
120
	mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
121 122 123 124 125

	return ret;
}

/*
126
 * Called under down_write(mmap_sem).
L
Linus Torvalds 已提交
127 128
 */

129
#ifndef HAVE_ARCH_HUGETLB_UNMAPPED_AREA
L
Linus Torvalds 已提交
130 131 132 133 134 135 136
static unsigned long
hugetlb_get_unmapped_area(struct file *file, unsigned long addr,
		unsigned long len, unsigned long pgoff, unsigned long flags)
{
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	unsigned long start_addr;
137
	struct hstate *h = hstate_file(file);
L
Linus Torvalds 已提交
138

139
	if (len & ~huge_page_mask(h))
L
Linus Torvalds 已提交
140 141 142 143
		return -EINVAL;
	if (len > TASK_SIZE)
		return -ENOMEM;

144
	if (flags & MAP_FIXED) {
145
		if (prepare_hugepage_range(file, addr, len))
146 147 148 149
			return -EINVAL;
		return addr;
	}

L
Linus Torvalds 已提交
150
	if (addr) {
151
		addr = ALIGN(addr, huge_page_size(h));
L
Linus Torvalds 已提交
152 153 154 155 156 157 158 159
		vma = find_vma(mm, addr);
		if (TASK_SIZE - len >= addr &&
		    (!vma || addr + len <= vma->vm_start))
			return addr;
	}

	start_addr = mm->free_area_cache;

160 161 162
	if (len <= mm->cached_hole_size)
		start_addr = TASK_UNMAPPED_BASE;

L
Linus Torvalds 已提交
163
full_search:
164
	addr = ALIGN(start_addr, huge_page_size(h));
L
Linus Torvalds 已提交
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181

	for (vma = find_vma(mm, addr); ; vma = vma->vm_next) {
		/* At this point:  (!vma || addr < vma->vm_end). */
		if (TASK_SIZE - len < addr) {
			/*
			 * Start a new search - just in case we missed
			 * some holes.
			 */
			if (start_addr != TASK_UNMAPPED_BASE) {
				start_addr = TASK_UNMAPPED_BASE;
				goto full_search;
			}
			return -ENOMEM;
		}

		if (!vma || addr + len <= vma->vm_start)
			return addr;
182
		addr = ALIGN(vma->vm_end, huge_page_size(h));
L
Linus Torvalds 已提交
183 184 185 186
	}
}
#endif

B
Badari Pulavarty 已提交
187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232
static int
hugetlbfs_read_actor(struct page *page, unsigned long offset,
			char __user *buf, unsigned long count,
			unsigned long size)
{
	char *kaddr;
	unsigned long left, copied = 0;
	int i, chunksize;

	if (size > count)
		size = count;

	/* Find which 4k chunk and offset with in that chunk */
	i = offset >> PAGE_CACHE_SHIFT;
	offset = offset & ~PAGE_CACHE_MASK;

	while (size) {
		chunksize = PAGE_CACHE_SIZE;
		if (offset)
			chunksize -= offset;
		if (chunksize > size)
			chunksize = size;
		kaddr = kmap(&page[i]);
		left = __copy_to_user(buf, kaddr + offset, chunksize);
		kunmap(&page[i]);
		if (left) {
			copied += (chunksize - left);
			break;
		}
		offset = 0;
		size -= chunksize;
		buf += chunksize;
		copied += chunksize;
		i++;
	}
	return copied ? copied : -EFAULT;
}

/*
 * Support for read() - Find the page attached to f_mapping and copy out the
 * data. Its *very* similar to do_generic_mapping_read(), we can't use that
 * since it has PAGE_CACHE_SIZE assumptions.
 */
static ssize_t hugetlbfs_read(struct file *filp, char __user *buf,
			      size_t len, loff_t *ppos)
{
233
	struct hstate *h = hstate_file(filp);
B
Badari Pulavarty 已提交
234 235
	struct address_space *mapping = filp->f_mapping;
	struct inode *inode = mapping->host;
236 237
	unsigned long index = *ppos >> huge_page_shift(h);
	unsigned long offset = *ppos & ~huge_page_mask(h);
B
Badari Pulavarty 已提交
238 239 240 241 242 243 244 245 246 247 248 249 250 251
	unsigned long end_index;
	loff_t isize;
	ssize_t retval = 0;

	mutex_lock(&inode->i_mutex);

	/* validate length */
	if (len == 0)
		goto out;

	isize = i_size_read(inode);
	if (!isize)
		goto out;

252
	end_index = (isize - 1) >> huge_page_shift(h);
B
Badari Pulavarty 已提交
253 254
	for (;;) {
		struct page *page;
255
		unsigned long nr, ret;
256
		int ra;
B
Badari Pulavarty 已提交
257 258

		/* nr is the maximum number of bytes to copy from this page */
259
		nr = huge_page_size(h);
B
Badari Pulavarty 已提交
260 261 262
		if (index >= end_index) {
			if (index > end_index)
				goto out;
263
			nr = ((isize - 1) & ~huge_page_mask(h)) + 1;
B
Badari Pulavarty 已提交
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
			if (nr <= offset) {
				goto out;
			}
		}
		nr = nr - offset;

		/* Find the page */
		page = find_get_page(mapping, index);
		if (unlikely(page == NULL)) {
			/*
			 * We have a HOLE, zero out the user-buffer for the
			 * length of the hole or request.
			 */
			ret = len < nr ? len : nr;
			if (clear_user(buf, ret))
279 280 281
				ra = -EFAULT;
			else
				ra = 0;
B
Badari Pulavarty 已提交
282 283 284 285
		} else {
			/*
			 * We have the page, copy it to user space buffer.
			 */
286 287
			ra = hugetlbfs_read_actor(page, offset, buf, len, nr);
			ret = ra;
B
Badari Pulavarty 已提交
288
		}
289
		if (ra < 0) {
B
Badari Pulavarty 已提交
290
			if (retval == 0)
291
				retval = ra;
B
Badari Pulavarty 已提交
292 293 294 295 296 297 298 299
			if (page)
				page_cache_release(page);
			goto out;
		}

		offset += ret;
		retval += ret;
		len -= ret;
300 301
		index += offset >> huge_page_shift(h);
		offset &= ~huge_page_mask(h);
B
Badari Pulavarty 已提交
302 303 304 305 306 307 308 309 310

		if (page)
			page_cache_release(page);

		/* short read or no more work */
		if ((ret != nr) || (len == 0))
			break;
	}
out:
311
	*ppos = ((loff_t)index << huge_page_shift(h)) + offset;
B
Badari Pulavarty 已提交
312 313 314 315
	mutex_unlock(&inode->i_mutex);
	return retval;
}

N
Nick Piggin 已提交
316 317 318 319
static int hugetlbfs_write_begin(struct file *file,
			struct address_space *mapping,
			loff_t pos, unsigned len, unsigned flags,
			struct page **pagep, void **fsdata)
L
Linus Torvalds 已提交
320 321 322 323
{
	return -EINVAL;
}

N
Nick Piggin 已提交
324 325 326
static int hugetlbfs_write_end(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned copied,
			struct page *page, void *fsdata)
L
Linus Torvalds 已提交
327
{
N
Nick Piggin 已提交
328
	BUG();
L
Linus Torvalds 已提交
329 330 331 332 333
	return -EINVAL;
}

static void truncate_huge_page(struct page *page)
{
334
	cancel_dirty_page(page, /* No IO accounting for huge pages? */0);
L
Linus Torvalds 已提交
335 336 337 338 339
	ClearPageUptodate(page);
	remove_from_page_cache(page);
	put_page(page);
}

340
static void truncate_hugepages(struct inode *inode, loff_t lstart)
L
Linus Torvalds 已提交
341
{
342
	struct hstate *h = hstate_inode(inode);
343
	struct address_space *mapping = &inode->i_data;
344
	const pgoff_t start = lstart >> huge_page_shift(h);
L
Linus Torvalds 已提交
345 346
	struct pagevec pvec;
	pgoff_t next;
347
	int i, freed = 0;
L
Linus Torvalds 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367

	pagevec_init(&pvec, 0);
	next = start;
	while (1) {
		if (!pagevec_lookup(&pvec, mapping, next, PAGEVEC_SIZE)) {
			if (next == start)
				break;
			next = start;
			continue;
		}

		for (i = 0; i < pagevec_count(&pvec); ++i) {
			struct page *page = pvec.pages[i];

			lock_page(page);
			if (page->index > next)
				next = page->index;
			++next;
			truncate_huge_page(page);
			unlock_page(page);
368
			freed++;
L
Linus Torvalds 已提交
369 370 371 372
		}
		huge_pagevec_release(&pvec);
	}
	BUG_ON(!lstart && mapping->nrpages);
373
	hugetlb_unreserve_pages(inode, start, freed);
L
Linus Torvalds 已提交
374 375 376 377
}

static void hugetlbfs_delete_inode(struct inode *inode)
{
378
	truncate_hugepages(inode, 0);
379 380 381
	clear_inode(inode);
}

382
static void hugetlbfs_forget_inode(struct inode *inode) __releases(inode_lock)
L
Linus Torvalds 已提交
383
{
384 385 386
	struct super_block *sb = inode->i_sb;

	if (!hlist_unhashed(&inode->i_hash)) {
J
Joern Engel 已提交
387
		if (!(inode->i_state & (I_DIRTY|I_SYNC)))
388 389 390 391 392 393 394
			list_move(&inode->i_list, &inode_unused);
		inodes_stat.nr_unused++;
		if (!sb || (sb->s_flags & MS_ACTIVE)) {
			spin_unlock(&inode_lock);
			return;
		}
		inode->i_state |= I_WILL_FREE;
L
Linus Torvalds 已提交
395
		spin_unlock(&inode_lock);
396 397 398 399 400 401 402 403 404
		/*
		 * write_inode_now is a noop as we set BDI_CAP_NO_WRITEBACK
		 * in our backing_dev_info.
		 */
		write_inode_now(inode, 1);
		spin_lock(&inode_lock);
		inode->i_state &= ~I_WILL_FREE;
		inodes_stat.nr_unused--;
		hlist_del_init(&inode->i_hash);
L
Linus Torvalds 已提交
405 406 407 408 409 410
	}
	list_del_init(&inode->i_list);
	list_del_init(&inode->i_sb_list);
	inode->i_state |= I_FREEING;
	inodes_stat.nr_inodes--;
	spin_unlock(&inode_lock);
411
	truncate_hugepages(inode, 0);
L
Linus Torvalds 已提交
412 413 414 415 416 417 418
	clear_inode(inode);
	destroy_inode(inode);
}

static void hugetlbfs_drop_inode(struct inode *inode)
{
	if (!inode->i_nlink)
419
		generic_delete_inode(inode);
L
Linus Torvalds 已提交
420 421 422 423 424
	else
		hugetlbfs_forget_inode(inode);
}

static inline void
H
Hugh Dickins 已提交
425
hugetlb_vmtruncate_list(struct prio_tree_root *root, pgoff_t pgoff)
L
Linus Torvalds 已提交
426 427 428 429
{
	struct vm_area_struct *vma;
	struct prio_tree_iter iter;

H
Hugh Dickins 已提交
430
	vma_prio_tree_foreach(vma, &iter, root, pgoff, ULONG_MAX) {
L
Linus Torvalds 已提交
431 432 433
		unsigned long v_offset;

		/*
H
Hugh Dickins 已提交
434 435 436 437
		 * Can the expression below overflow on 32-bit arches?
		 * No, because the prio_tree returns us only those vmas
		 * which overlap the truncated area starting at pgoff,
		 * and no vma on a 32-bit arch can span beyond the 4GB.
L
Linus Torvalds 已提交
438
		 */
H
Hugh Dickins 已提交
439 440 441
		if (vma->vm_pgoff < pgoff)
			v_offset = (pgoff - vma->vm_pgoff) << PAGE_SHIFT;
		else
L
Linus Torvalds 已提交
442 443
			v_offset = 0;

444
		__unmap_hugepage_range(vma,
445
				vma->vm_start + v_offset, vma->vm_end, NULL);
L
Linus Torvalds 已提交
446 447 448 449 450
	}
}

static int hugetlb_vmtruncate(struct inode *inode, loff_t offset)
{
H
Hugh Dickins 已提交
451
	pgoff_t pgoff;
L
Linus Torvalds 已提交
452
	struct address_space *mapping = inode->i_mapping;
453
	struct hstate *h = hstate_inode(inode);
L
Linus Torvalds 已提交
454

455
	BUG_ON(offset & ~huge_page_mask(h));
H
Hugh Dickins 已提交
456
	pgoff = offset >> PAGE_SHIFT;
L
Linus Torvalds 已提交
457

458
	i_size_write(inode, offset);
L
Linus Torvalds 已提交
459 460 461 462
	spin_lock(&mapping->i_mmap_lock);
	if (!prio_tree_empty(&mapping->i_mmap))
		hugetlb_vmtruncate_list(&mapping->i_mmap, pgoff);
	spin_unlock(&mapping->i_mmap_lock);
463
	truncate_hugepages(inode, offset);
L
Linus Torvalds 已提交
464 465 466 467 468 469
	return 0;
}

static int hugetlbfs_setattr(struct dentry *dentry, struct iattr *attr)
{
	struct inode *inode = dentry->d_inode;
470
	struct hstate *h = hstate_inode(inode);
L
Linus Torvalds 已提交
471 472 473 474 475 476 477 478 479 480 481
	int error;
	unsigned int ia_valid = attr->ia_valid;

	BUG_ON(!inode);

	error = inode_change_ok(inode, attr);
	if (error)
		goto out;

	if (ia_valid & ATTR_SIZE) {
		error = -EINVAL;
482
		if (!(attr->ia_size & ~huge_page_mask(h)))
L
Linus Torvalds 已提交
483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506
			error = hugetlb_vmtruncate(inode, attr->ia_size);
		if (error)
			goto out;
		attr->ia_valid &= ~ATTR_SIZE;
	}
	error = inode_setattr(inode, attr);
out:
	return error;
}

static struct inode *hugetlbfs_get_inode(struct super_block *sb, uid_t uid, 
					gid_t gid, int mode, dev_t dev)
{
	struct inode *inode;

	inode = new_inode(sb);
	if (inode) {
		struct hugetlbfs_inode_info *info;
		inode->i_mode = mode;
		inode->i_uid = uid;
		inode->i_gid = gid;
		inode->i_mapping->a_ops = &hugetlbfs_aops;
		inode->i_mapping->backing_dev_info =&hugetlbfs_backing_dev_info;
		inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
507
		INIT_LIST_HEAD(&inode->i_mapping->private_list);
L
Linus Torvalds 已提交
508
		info = HUGETLBFS_I(inode);
509
		mpol_shared_policy_init(&info->policy, NULL);
L
Linus Torvalds 已提交
510 511 512 513 514 515 516 517 518 519 520 521 522
		switch (mode & S_IFMT) {
		default:
			init_special_inode(inode, mode, dev);
			break;
		case S_IFREG:
			inode->i_op = &hugetlbfs_inode_operations;
			inode->i_fop = &hugetlbfs_file_operations;
			break;
		case S_IFDIR:
			inode->i_op = &hugetlbfs_dir_inode_operations;
			inode->i_fop = &simple_dir_operations;

			/* directory inodes start off with i_nlink == 2 (for "." entry) */
523
			inc_nlink(inode);
L
Linus Torvalds 已提交
524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
			break;
		case S_IFLNK:
			inode->i_op = &page_symlink_inode_operations;
			break;
		}
	}
	return inode;
}

/*
 * File creation. Allocate an inode, and we're done..
 */
static int hugetlbfs_mknod(struct inode *dir,
			struct dentry *dentry, int mode, dev_t dev)
{
	struct inode *inode;
	int error = -ENOSPC;
	gid_t gid;

	if (dir->i_mode & S_ISGID) {
		gid = dir->i_gid;
		if (S_ISDIR(mode))
			mode |= S_ISGID;
	} else {
548
		gid = current_fsgid();
L
Linus Torvalds 已提交
549
	}
550
	inode = hugetlbfs_get_inode(dir->i_sb, current_fsuid(), gid, mode, dev);
L
Linus Torvalds 已提交
551 552 553 554 555 556 557 558 559 560 561 562 563
	if (inode) {
		dir->i_ctime = dir->i_mtime = CURRENT_TIME;
		d_instantiate(dentry, inode);
		dget(dentry);	/* Extra count - pin the dentry in core */
		error = 0;
	}
	return error;
}

static int hugetlbfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
{
	int retval = hugetlbfs_mknod(dir, dentry, mode | S_IFDIR, 0);
	if (!retval)
564
		inc_nlink(dir);
L
Linus Torvalds 已提交
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
	return retval;
}

static int hugetlbfs_create(struct inode *dir, struct dentry *dentry, int mode, struct nameidata *nd)
{
	return hugetlbfs_mknod(dir, dentry, mode | S_IFREG, 0);
}

static int hugetlbfs_symlink(struct inode *dir,
			struct dentry *dentry, const char *symname)
{
	struct inode *inode;
	int error = -ENOSPC;
	gid_t gid;

	if (dir->i_mode & S_ISGID)
		gid = dir->i_gid;
	else
583
		gid = current_fsgid();
L
Linus Torvalds 已提交
584

585
	inode = hugetlbfs_get_inode(dir->i_sb, current_fsuid(),
L
Linus Torvalds 已提交
586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601
					gid, S_IFLNK|S_IRWXUGO, 0);
	if (inode) {
		int l = strlen(symname)+1;
		error = page_symlink(inode, symname, l);
		if (!error) {
			d_instantiate(dentry, inode);
			dget(dentry);
		} else
			iput(inode);
	}
	dir->i_ctime = dir->i_mtime = CURRENT_TIME;

	return error;
}

/*
602
 * mark the head page dirty
L
Linus Torvalds 已提交
603 604 605
 */
static int hugetlbfs_set_page_dirty(struct page *page)
{
606
	struct page *head = compound_head(page);
607 608

	SetPageDirty(head);
L
Linus Torvalds 已提交
609 610 611
	return 0;
}

612
static int hugetlbfs_statfs(struct dentry *dentry, struct kstatfs *buf)
L
Linus Torvalds 已提交
613
{
614
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(dentry->d_sb);
615
	struct hstate *h = hstate_inode(dentry->d_inode);
L
Linus Torvalds 已提交
616 617

	buf->f_type = HUGETLBFS_MAGIC;
618
	buf->f_bsize = huge_page_size(h);
L
Linus Torvalds 已提交
619 620
	if (sbinfo) {
		spin_lock(&sbinfo->stat_lock);
621 622 623 624 625 626 627 628
		/* If no limits set, just report 0 for max/free/used
		 * blocks, like simple_statfs() */
		if (sbinfo->max_blocks >= 0) {
			buf->f_blocks = sbinfo->max_blocks;
			buf->f_bavail = buf->f_bfree = sbinfo->free_blocks;
			buf->f_files = sbinfo->max_inodes;
			buf->f_ffree = sbinfo->free_inodes;
		}
L
Linus Torvalds 已提交
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644
		spin_unlock(&sbinfo->stat_lock);
	}
	buf->f_namelen = NAME_MAX;
	return 0;
}

static void hugetlbfs_put_super(struct super_block *sb)
{
	struct hugetlbfs_sb_info *sbi = HUGETLBFS_SB(sb);

	if (sbi) {
		sb->s_fs_info = NULL;
		kfree(sbi);
	}
}

645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669
static inline int hugetlbfs_dec_free_inodes(struct hugetlbfs_sb_info *sbinfo)
{
	if (sbinfo->free_inodes >= 0) {
		spin_lock(&sbinfo->stat_lock);
		if (unlikely(!sbinfo->free_inodes)) {
			spin_unlock(&sbinfo->stat_lock);
			return 0;
		}
		sbinfo->free_inodes--;
		spin_unlock(&sbinfo->stat_lock);
	}

	return 1;
}

static void hugetlbfs_inc_free_inodes(struct hugetlbfs_sb_info *sbinfo)
{
	if (sbinfo->free_inodes >= 0) {
		spin_lock(&sbinfo->stat_lock);
		sbinfo->free_inodes++;
		spin_unlock(&sbinfo->stat_lock);
	}
}


670
static struct kmem_cache *hugetlbfs_inode_cachep;
L
Linus Torvalds 已提交
671 672 673

static struct inode *hugetlbfs_alloc_inode(struct super_block *sb)
{
674
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(sb);
L
Linus Torvalds 已提交
675 676
	struct hugetlbfs_inode_info *p;

677 678
	if (unlikely(!hugetlbfs_dec_free_inodes(sbinfo)))
		return NULL;
679
	p = kmem_cache_alloc(hugetlbfs_inode_cachep, GFP_KERNEL);
680 681
	if (unlikely(!p)) {
		hugetlbfs_inc_free_inodes(sbinfo);
L
Linus Torvalds 已提交
682
		return NULL;
683
	}
L
Linus Torvalds 已提交
684 685 686 687 688
	return &p->vfs_inode;
}

static void hugetlbfs_destroy_inode(struct inode *inode)
{
689
	hugetlbfs_inc_free_inodes(HUGETLBFS_SB(inode->i_sb));
L
Linus Torvalds 已提交
690 691 692 693
	mpol_free_shared_policy(&HUGETLBFS_I(inode)->policy);
	kmem_cache_free(hugetlbfs_inode_cachep, HUGETLBFS_I(inode));
}

694
static const struct address_space_operations hugetlbfs_aops = {
N
Nick Piggin 已提交
695 696
	.write_begin	= hugetlbfs_write_begin,
	.write_end	= hugetlbfs_write_end,
L
Linus Torvalds 已提交
697 698 699
	.set_page_dirty	= hugetlbfs_set_page_dirty,
};

700

701
static void init_once(void *foo)
702 703 704
{
	struct hugetlbfs_inode_info *ei = (struct hugetlbfs_inode_info *)foo;

C
Christoph Lameter 已提交
705
	inode_init_once(&ei->vfs_inode);
706 707
}

708
const struct file_operations hugetlbfs_file_operations = {
B
Badari Pulavarty 已提交
709
	.read			= hugetlbfs_read,
L
Linus Torvalds 已提交
710 711 712 713 714
	.mmap			= hugetlbfs_file_mmap,
	.fsync			= simple_sync_file,
	.get_unmapped_area	= hugetlb_get_unmapped_area,
};

715
static const struct inode_operations hugetlbfs_dir_inode_operations = {
L
Linus Torvalds 已提交
716 717 718 719 720 721 722 723 724 725 726 727
	.create		= hugetlbfs_create,
	.lookup		= simple_lookup,
	.link		= simple_link,
	.unlink		= simple_unlink,
	.symlink	= hugetlbfs_symlink,
	.mkdir		= hugetlbfs_mkdir,
	.rmdir		= simple_rmdir,
	.mknod		= hugetlbfs_mknod,
	.rename		= simple_rename,
	.setattr	= hugetlbfs_setattr,
};

728
static const struct inode_operations hugetlbfs_inode_operations = {
L
Linus Torvalds 已提交
729 730 731
	.setattr	= hugetlbfs_setattr,
};

732
static const struct super_operations hugetlbfs_ops = {
L
Linus Torvalds 已提交
733 734 735
	.alloc_inode    = hugetlbfs_alloc_inode,
	.destroy_inode  = hugetlbfs_destroy_inode,
	.statfs		= hugetlbfs_statfs,
736
	.delete_inode	= hugetlbfs_delete_inode,
L
Linus Torvalds 已提交
737 738
	.drop_inode	= hugetlbfs_drop_inode,
	.put_super	= hugetlbfs_put_super,
M
Miklos Szeredi 已提交
739
	.show_options	= generic_show_options,
L
Linus Torvalds 已提交
740 741 742 743 744
};

static int
hugetlbfs_parse_options(char *options, struct hugetlbfs_config *pconfig)
{
745 746 747
	char *p, *rest;
	substring_t args[MAX_OPT_ARGS];
	int option;
748 749
	unsigned long long size = 0;
	enum { NO_SIZE, SIZE_STD, SIZE_PERCENT } setsize = NO_SIZE;
L
Linus Torvalds 已提交
750 751 752 753

	if (!options)
		return 0;

754 755
	while ((p = strsep(&options, ",")) != NULL) {
		int token;
756 757
		if (!*p)
			continue;
758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775

		token = match_token(p, tokens, args);
		switch (token) {
		case Opt_uid:
			if (match_int(&args[0], &option))
 				goto bad_val;
			pconfig->uid = option;
			break;

		case Opt_gid:
			if (match_int(&args[0], &option))
 				goto bad_val;
			pconfig->gid = option;
			break;

		case Opt_mode:
			if (match_octal(&args[0], &option))
 				goto bad_val;
776
			pconfig->mode = option & 01777U;
777 778 779 780 781 782 783
			break;

		case Opt_size: {
			/* memparse() will accept a K/M/G without a digit */
			if (!isdigit(*args[0].from))
				goto bad_val;
			size = memparse(args[0].from, &rest);
784 785 786
			setsize = SIZE_STD;
			if (*rest == '%')
				setsize = SIZE_PERCENT;
787 788
			break;
		}
L
Linus Torvalds 已提交
789

790 791 792 793 794 795 796
		case Opt_nr_inodes:
			/* memparse() will accept a K/M/G without a digit */
			if (!isdigit(*args[0].from))
				goto bad_val;
			pconfig->nr_inodes = memparse(args[0].from, &rest);
			break;

797 798 799 800 801 802 803 804 805 806 807 808 809
		case Opt_pagesize: {
			unsigned long ps;
			ps = memparse(args[0].from, &rest);
			pconfig->hstate = size_to_hstate(ps);
			if (!pconfig->hstate) {
				printk(KERN_ERR
				"hugetlbfs: Unsupported page size %lu MB\n",
					ps >> 20);
				return -EINVAL;
			}
			break;
		}

810
		default:
811 812 813
			printk(KERN_ERR "hugetlbfs: Bad mount option: \"%s\"\n",
				 p);
			return -EINVAL;
814 815
			break;
		}
L
Linus Torvalds 已提交
816
	}
817 818 819 820 821 822 823 824 825 826 827 828

	/* Do size after hstate is set up */
	if (setsize > NO_SIZE) {
		struct hstate *h = pconfig->hstate;
		if (setsize == SIZE_PERCENT) {
			size <<= huge_page_shift(h);
			size *= h->max_huge_pages;
			do_div(size, 100);
		}
		pconfig->nr_blocks = (size >> huge_page_shift(h));
	}

L
Linus Torvalds 已提交
829
	return 0;
830 831 832 833

bad_val:
 	printk(KERN_ERR "hugetlbfs: Bad value '%s' for mount option '%s'\n",
	       args[0].from, p);
834
 	return -EINVAL;
L
Linus Torvalds 已提交
835 836 837 838 839 840 841 842 843 844 845
}

static int
hugetlbfs_fill_super(struct super_block *sb, void *data, int silent)
{
	struct inode * inode;
	struct dentry * root;
	int ret;
	struct hugetlbfs_config config;
	struct hugetlbfs_sb_info *sbinfo;

M
Miklos Szeredi 已提交
846 847
	save_mount_options(sb, data);

L
Linus Torvalds 已提交
848 849
	config.nr_blocks = -1; /* No limit on size by default */
	config.nr_inodes = -1; /* No limit on number of inodes by default */
850 851
	config.uid = current_fsuid();
	config.gid = current_fsgid();
L
Linus Torvalds 已提交
852
	config.mode = 0755;
853
	config.hstate = &default_hstate;
L
Linus Torvalds 已提交
854 855 856 857 858 859 860 861
	ret = hugetlbfs_parse_options(data, &config);
	if (ret)
		return ret;

	sbinfo = kmalloc(sizeof(struct hugetlbfs_sb_info), GFP_KERNEL);
	if (!sbinfo)
		return -ENOMEM;
	sb->s_fs_info = sbinfo;
862
	sbinfo->hstate = config.hstate;
L
Linus Torvalds 已提交
863 864 865 866 867 868
	spin_lock_init(&sbinfo->stat_lock);
	sbinfo->max_blocks = config.nr_blocks;
	sbinfo->free_blocks = config.nr_blocks;
	sbinfo->max_inodes = config.nr_inodes;
	sbinfo->free_inodes = config.nr_inodes;
	sb->s_maxbytes = MAX_LFS_FILESIZE;
869 870
	sb->s_blocksize = huge_page_size(config.hstate);
	sb->s_blocksize_bits = huge_page_shift(config.hstate);
L
Linus Torvalds 已提交
871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890
	sb->s_magic = HUGETLBFS_MAGIC;
	sb->s_op = &hugetlbfs_ops;
	sb->s_time_gran = 1;
	inode = hugetlbfs_get_inode(sb, config.uid, config.gid,
					S_IFDIR | config.mode, 0);
	if (!inode)
		goto out_free;

	root = d_alloc_root(inode);
	if (!root) {
		iput(inode);
		goto out_free;
	}
	sb->s_root = root;
	return 0;
out_free:
	kfree(sbinfo);
	return -ENOMEM;
}

891
int hugetlb_get_quota(struct address_space *mapping, long delta)
L
Linus Torvalds 已提交
892 893 894 895 896 897
{
	int ret = 0;
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(mapping->host->i_sb);

	if (sbinfo->free_blocks > -1) {
		spin_lock(&sbinfo->stat_lock);
898 899
		if (sbinfo->free_blocks - delta >= 0)
			sbinfo->free_blocks -= delta;
L
Linus Torvalds 已提交
900 901 902 903 904 905 906 907
		else
			ret = -ENOMEM;
		spin_unlock(&sbinfo->stat_lock);
	}

	return ret;
}

908
void hugetlb_put_quota(struct address_space *mapping, long delta)
L
Linus Torvalds 已提交
909 910 911 912 913
{
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(mapping->host->i_sb);

	if (sbinfo->free_blocks > -1) {
		spin_lock(&sbinfo->stat_lock);
914
		sbinfo->free_blocks += delta;
L
Linus Torvalds 已提交
915 916 917 918
		spin_unlock(&sbinfo->stat_lock);
	}
}

919 920
static int hugetlbfs_get_sb(struct file_system_type *fs_type,
	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
L
Linus Torvalds 已提交
921
{
922
	return get_sb_nodev(fs_type, flags, data, hugetlbfs_fill_super, mnt);
L
Linus Torvalds 已提交
923 924 925 926 927 928 929 930 931 932 933 934
}

static struct file_system_type hugetlbfs_fs_type = {
	.name		= "hugetlbfs",
	.get_sb		= hugetlbfs_get_sb,
	.kill_sb	= kill_litter_super,
};

static struct vfsmount *hugetlbfs_vfsmount;

static int can_do_hugetlb_shm(void)
{
935
	return capable(CAP_IPC_LOCK) || in_group_p(sysctl_hugetlb_shm_group);
L
Linus Torvalds 已提交
936 937
}

938
struct file *hugetlb_file_setup(const char *name, size_t size, int acctflag)
L
Linus Torvalds 已提交
939 940
{
	int error = -ENOMEM;
941
	int unlock_shm = 0;
L
Linus Torvalds 已提交
942 943 944 945
	struct file *file;
	struct inode *inode;
	struct dentry *dentry, *root;
	struct qstr quick_string;
946
	struct user_struct *user = current_user();
L
Linus Torvalds 已提交
947

948 949 950
	if (!hugetlbfs_vfsmount)
		return ERR_PTR(-ENOENT);

951 952 953 954 955 956 957 958
	if (!can_do_hugetlb_shm()) {
		if (user_shm_lock(size, user)) {
			unlock_shm = 1;
			WARN_ONCE(1,
			  "Using mlock ulimits for SHM_HUGETLB deprecated\n");
		} else
			return ERR_PTR(-EPERM);
	}
L
Linus Torvalds 已提交
959 960

	root = hugetlbfs_vfsmount->mnt_root;
961
	quick_string.name = name;
L
Linus Torvalds 已提交
962 963 964 965 966 967 968
	quick_string.len = strlen(quick_string.name);
	quick_string.hash = 0;
	dentry = d_alloc(root, &quick_string);
	if (!dentry)
		goto out_shm_unlock;

	error = -ENOSPC;
969 970
	inode = hugetlbfs_get_inode(root->d_sb, current_fsuid(),
				current_fsgid(), S_IFREG | S_IRWXUGO, 0);
L
Linus Torvalds 已提交
971
	if (!inode)
972
		goto out_dentry;
L
Linus Torvalds 已提交
973

974
	error = -ENOMEM;
975
	if (hugetlb_reserve_pages(inode, 0,
976 977
			size >> huge_page_shift(hstate_inode(inode)), NULL,
			acctflag))
978 979
		goto out_inode;

L
Linus Torvalds 已提交
980 981 982
	d_instantiate(dentry, inode);
	inode->i_size = size;
	inode->i_nlink = 0;
983 984 985 986 987 988

	error = -ENFILE;
	file = alloc_file(hugetlbfs_vfsmount, dentry,
			FMODE_WRITE | FMODE_READ,
			&hugetlbfs_file_operations);
	if (!file)
989
		goto out_dentry; /* inode is already attached */
M
Mimi Zohar 已提交
990
	ima_counts_get(file);
991

L
Linus Torvalds 已提交
992 993
	return file;

994 995
out_inode:
	iput(inode);
L
Linus Torvalds 已提交
996 997 998
out_dentry:
	dput(dentry);
out_shm_unlock:
999 1000
	if (unlock_shm)
		user_shm_unlock(size, user);
L
Linus Torvalds 已提交
1001 1002 1003 1004 1005 1006 1007 1008
	return ERR_PTR(error);
}

static int __init init_hugetlbfs_fs(void)
{
	int error;
	struct vfsmount *vfsmount;

P
Peter Zijlstra 已提交
1009 1010 1011 1012
	error = bdi_init(&hugetlbfs_backing_dev_info);
	if (error)
		return error;

L
Linus Torvalds 已提交
1013 1014
	hugetlbfs_inode_cachep = kmem_cache_create("hugetlbfs_inode_cache",
					sizeof(struct hugetlbfs_inode_info),
1015
					0, 0, init_once);
L
Linus Torvalds 已提交
1016
	if (hugetlbfs_inode_cachep == NULL)
P
Peter Zijlstra 已提交
1017
		goto out2;
L
Linus Torvalds 已提交
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034

	error = register_filesystem(&hugetlbfs_fs_type);
	if (error)
		goto out;

	vfsmount = kern_mount(&hugetlbfs_fs_type);

	if (!IS_ERR(vfsmount)) {
		hugetlbfs_vfsmount = vfsmount;
		return 0;
	}

	error = PTR_ERR(vfsmount);

 out:
	if (error)
		kmem_cache_destroy(hugetlbfs_inode_cachep);
P
Peter Zijlstra 已提交
1035 1036
 out2:
	bdi_destroy(&hugetlbfs_backing_dev_info);
L
Linus Torvalds 已提交
1037 1038 1039 1040 1041 1042 1043
	return error;
}

static void __exit exit_hugetlbfs_fs(void)
{
	kmem_cache_destroy(hugetlbfs_inode_cachep);
	unregister_filesystem(&hugetlbfs_fs_type);
P
Peter Zijlstra 已提交
1044
	bdi_destroy(&hugetlbfs_backing_dev_info);
L
Linus Torvalds 已提交
1045 1046 1047 1048 1049 1050
}

module_init(init_hugetlbfs_fs)
module_exit(exit_hugetlbfs_fs)

MODULE_LICENSE("GPL");