inode.c 24.0 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>
N
Nick Black 已提交
33
#include <linux/magic.h>
N
Naoya Horiguchi 已提交
34
#include <linux/migrate.h>
L
Linus Torvalds 已提交
35 36 37

#include <asm/uaccess.h>

38
static const struct super_operations hugetlbfs_ops;
39
static const struct address_space_operations hugetlbfs_aops;
40
const struct file_operations hugetlbfs_file_operations;
41 42
static const struct inode_operations hugetlbfs_dir_inode_operations;
static const struct inode_operations hugetlbfs_inode_operations;
L
Linus Torvalds 已提交
43 44

static struct backing_dev_info hugetlbfs_backing_dev_info = {
45
	.name		= "hugetlbfs",
L
Linus Torvalds 已提交
46
	.ra_pages	= 0,	/* No readahead */
47
	.capabilities	= BDI_CAP_NO_ACCT_AND_WRITEBACK,
L
Linus Torvalds 已提交
48 49 50 51
};

int sysctl_hugetlb_shm_group;

52 53 54
enum {
	Opt_size, Opt_nr_inodes,
	Opt_mode, Opt_uid, Opt_gid,
55
	Opt_pagesize,
56 57 58
	Opt_err,
};

59
static const match_table_t tokens = {
60 61 62 63 64
	{Opt_size,	"size=%s"},
	{Opt_nr_inodes,	"nr_inodes=%s"},
	{Opt_mode,	"mode=%o"},
	{Opt_uid,	"uid=%u"},
	{Opt_gid,	"gid=%u"},
65
	{Opt_pagesize,	"pagesize=%s"},
66 67 68
	{Opt_err,	NULL},
};

69 70 71 72 73 74 75 76 77 78
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
79 80
static int hugetlbfs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
81
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
82 83
	loff_t len, vma_len;
	int ret;
84
	struct hstate *h = hstate_file(file);
L
Linus Torvalds 已提交
85

86
	/*
87 88 89 90 91 92
	 * 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).
93 94 95
	 */
	vma->vm_flags |= VM_HUGETLB | VM_RESERVED;
	vma->vm_ops = &hugetlb_vm_ops;
L
Linus Torvalds 已提交
96

97
	if (vma->vm_pgoff & (~huge_page_mask(h) >> PAGE_SHIFT))
98 99
		return -EINVAL;

L
Linus Torvalds 已提交
100 101
	vma_len = (loff_t)(vma->vm_end - vma->vm_start);

102
	mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
103 104 105 106 107
	file_accessed(file);

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

108
	if (hugetlb_reserve_pages(inode,
109
				vma->vm_pgoff >> huge_page_order(h),
110 111
				len >> huge_page_shift(h), vma,
				vma->vm_flags))
112
		goto out;
113

A
Adam Litke 已提交
114 115
	ret = 0;
	hugetlb_prefault_arch_hook(vma->vm_mm);
116
	if (vma->vm_flags & VM_WRITE && inode->i_size < len)
L
Linus Torvalds 已提交
117 118
		inode->i_size = len;
out:
119
	mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
120 121 122 123 124

	return ret;
}

/*
125
 * Called under down_write(mmap_sem).
L
Linus Torvalds 已提交
126 127
 */

128
#ifndef HAVE_ARCH_HUGETLB_UNMAPPED_AREA
L
Linus Torvalds 已提交
129 130 131 132 133 134 135
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;
136
	struct hstate *h = hstate_file(file);
L
Linus Torvalds 已提交
137

138
	if (len & ~huge_page_mask(h))
L
Linus Torvalds 已提交
139 140 141 142
		return -EINVAL;
	if (len > TASK_SIZE)
		return -ENOMEM;

143
	if (flags & MAP_FIXED) {
144
		if (prepare_hugepage_range(file, addr, len))
145 146 147 148
			return -EINVAL;
		return addr;
	}

L
Linus Torvalds 已提交
149
	if (addr) {
150
		addr = ALIGN(addr, huge_page_size(h));
L
Linus Torvalds 已提交
151 152 153 154 155 156 157 158
		vma = find_vma(mm, addr);
		if (TASK_SIZE - len >= addr &&
		    (!vma || addr + len <= vma->vm_start))
			return addr;
	}

	start_addr = mm->free_area_cache;

159 160 161
	if (len <= mm->cached_hole_size)
		start_addr = TASK_UNMAPPED_BASE;

L
Linus Torvalds 已提交
162
full_search:
163
	addr = ALIGN(start_addr, huge_page_size(h));
L
Linus Torvalds 已提交
164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180

	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;
181
		addr = ALIGN(vma->vm_end, huge_page_size(h));
L
Linus Torvalds 已提交
182 183 184 185
	}
}
#endif

B
Badari Pulavarty 已提交
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
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)
{
232
	struct hstate *h = hstate_file(filp);
B
Badari Pulavarty 已提交
233 234
	struct address_space *mapping = filp->f_mapping;
	struct inode *inode = mapping->host;
235 236
	unsigned long index = *ppos >> huge_page_shift(h);
	unsigned long offset = *ppos & ~huge_page_mask(h);
B
Badari Pulavarty 已提交
237 238 239 240 241 242 243 244 245 246 247 248 249 250
	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;

251
	end_index = (isize - 1) >> huge_page_shift(h);
B
Badari Pulavarty 已提交
252 253
	for (;;) {
		struct page *page;
254
		unsigned long nr, ret;
255
		int ra;
B
Badari Pulavarty 已提交
256 257

		/* nr is the maximum number of bytes to copy from this page */
258
		nr = huge_page_size(h);
B
Badari Pulavarty 已提交
259 260 261
		if (index >= end_index) {
			if (index > end_index)
				goto out;
262
			nr = ((isize - 1) & ~huge_page_mask(h)) + 1;
B
Badari Pulavarty 已提交
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
			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))
278 279 280
				ra = -EFAULT;
			else
				ra = 0;
B
Badari Pulavarty 已提交
281 282 283 284
		} else {
			/*
			 * We have the page, copy it to user space buffer.
			 */
285 286
			ra = hugetlbfs_read_actor(page, offset, buf, len, nr);
			ret = ra;
B
Badari Pulavarty 已提交
287
		}
288
		if (ra < 0) {
B
Badari Pulavarty 已提交
289
			if (retval == 0)
290
				retval = ra;
B
Badari Pulavarty 已提交
291 292 293 294 295 296 297 298
			if (page)
				page_cache_release(page);
			goto out;
		}

		offset += ret;
		retval += ret;
		len -= ret;
299 300
		index += offset >> huge_page_shift(h);
		offset &= ~huge_page_mask(h);
B
Badari Pulavarty 已提交
301 302 303 304 305 306 307 308 309

		if (page)
			page_cache_release(page);

		/* short read or no more work */
		if ((ret != nr) || (len == 0))
			break;
	}
out:
310
	*ppos = ((loff_t)index << huge_page_shift(h)) + offset;
B
Badari Pulavarty 已提交
311 312 313 314
	mutex_unlock(&inode->i_mutex);
	return retval;
}

N
Nick Piggin 已提交
315 316 317 318
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
319 320 321 322
{
	return -EINVAL;
}

N
Nick Piggin 已提交
323 324 325
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
326
{
N
Nick Piggin 已提交
327
	BUG();
L
Linus Torvalds 已提交
328 329 330 331 332
	return -EINVAL;
}

static void truncate_huge_page(struct page *page)
{
333
	cancel_dirty_page(page, /* No IO accounting for huge pages? */0);
L
Linus Torvalds 已提交
334
	ClearPageUptodate(page);
335
	delete_from_page_cache(page);
L
Linus Torvalds 已提交
336 337
}

338
static void truncate_hugepages(struct inode *inode, loff_t lstart)
L
Linus Torvalds 已提交
339
{
340
	struct hstate *h = hstate_inode(inode);
341
	struct address_space *mapping = &inode->i_data;
342
	const pgoff_t start = lstart >> huge_page_shift(h);
L
Linus Torvalds 已提交
343 344
	struct pagevec pvec;
	pgoff_t next;
345
	int i, freed = 0;
L
Linus Torvalds 已提交
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365

	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);
366
			freed++;
L
Linus Torvalds 已提交
367 368 369 370
		}
		huge_pagevec_release(&pvec);
	}
	BUG_ON(!lstart && mapping->nrpages);
371
	hugetlb_unreserve_pages(inode, start, freed);
L
Linus Torvalds 已提交
372 373
}

A
Al Viro 已提交
374
static void hugetlbfs_evict_inode(struct inode *inode)
L
Linus Torvalds 已提交
375
{
376
	truncate_hugepages(inode, 0);
A
Al Viro 已提交
377
	end_writeback(inode);
378 379
}

L
Linus Torvalds 已提交
380
static inline void
H
Hugh Dickins 已提交
381
hugetlb_vmtruncate_list(struct prio_tree_root *root, pgoff_t pgoff)
L
Linus Torvalds 已提交
382 383 384 385
{
	struct vm_area_struct *vma;
	struct prio_tree_iter iter;

H
Hugh Dickins 已提交
386
	vma_prio_tree_foreach(vma, &iter, root, pgoff, ULONG_MAX) {
L
Linus Torvalds 已提交
387 388 389
		unsigned long v_offset;

		/*
H
Hugh Dickins 已提交
390 391 392 393
		 * 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
394
		 */
H
Hugh Dickins 已提交
395 396 397
		if (vma->vm_pgoff < pgoff)
			v_offset = (pgoff - vma->vm_pgoff) << PAGE_SHIFT;
		else
L
Linus Torvalds 已提交
398 399
			v_offset = 0;

400
		__unmap_hugepage_range(vma,
401
				vma->vm_start + v_offset, vma->vm_end, NULL);
L
Linus Torvalds 已提交
402 403 404 405 406
	}
}

static int hugetlb_vmtruncate(struct inode *inode, loff_t offset)
{
H
Hugh Dickins 已提交
407
	pgoff_t pgoff;
L
Linus Torvalds 已提交
408
	struct address_space *mapping = inode->i_mapping;
409
	struct hstate *h = hstate_inode(inode);
L
Linus Torvalds 已提交
410

411
	BUG_ON(offset & ~huge_page_mask(h));
H
Hugh Dickins 已提交
412
	pgoff = offset >> PAGE_SHIFT;
L
Linus Torvalds 已提交
413

414
	i_size_write(inode, offset);
415
	mutex_lock(&mapping->i_mmap_mutex);
L
Linus Torvalds 已提交
416 417
	if (!prio_tree_empty(&mapping->i_mmap))
		hugetlb_vmtruncate_list(&mapping->i_mmap, pgoff);
418
	mutex_unlock(&mapping->i_mmap_mutex);
419
	truncate_hugepages(inode, offset);
L
Linus Torvalds 已提交
420 421 422 423 424 425
	return 0;
}

static int hugetlbfs_setattr(struct dentry *dentry, struct iattr *attr)
{
	struct inode *inode = dentry->d_inode;
426
	struct hstate *h = hstate_inode(inode);
L
Linus Torvalds 已提交
427 428 429 430 431 432 433
	int error;
	unsigned int ia_valid = attr->ia_valid;

	BUG_ON(!inode);

	error = inode_change_ok(inode, attr);
	if (error)
C
Christoph Hellwig 已提交
434
		return error;
L
Linus Torvalds 已提交
435 436 437

	if (ia_valid & ATTR_SIZE) {
		error = -EINVAL;
C
Christoph Hellwig 已提交
438 439 440
		if (attr->ia_size & ~huge_page_mask(h))
			return -EINVAL;
		error = hugetlb_vmtruncate(inode, attr->ia_size);
L
Linus Torvalds 已提交
441
		if (error)
C
Christoph Hellwig 已提交
442
			return error;
L
Linus Torvalds 已提交
443
	}
C
Christoph Hellwig 已提交
444 445 446 447

	setattr_copy(inode, attr);
	mark_inode_dirty(inode);
	return 0;
L
Linus Torvalds 已提交
448 449 450 451 452 453 454 455 456 457
}

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;
458
		inode->i_ino = get_next_ino();
L
Linus Torvalds 已提交
459 460 461 462 463 464
		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;
465
		INIT_LIST_HEAD(&inode->i_mapping->private_list);
L
Linus Torvalds 已提交
466
		info = HUGETLBFS_I(inode);
467 468 469 470 471 472 473
		/*
		 * The policy is initialized here even if we are creating a
		 * private inode because initialization simply creates an
		 * an empty rb tree and calls spin_lock_init(), later when we
		 * call mpol_free_shared_policy() it will just return because
		 * the rb tree will still be empty.
		 */
474
		mpol_shared_policy_init(&info->policy, NULL);
L
Linus Torvalds 已提交
475 476 477 478 479 480 481 482 483 484 485 486 487
		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) */
488
			inc_nlink(inode);
L
Linus Torvalds 已提交
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
			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 {
513
		gid = current_fsgid();
L
Linus Torvalds 已提交
514
	}
515
	inode = hugetlbfs_get_inode(dir->i_sb, current_fsuid(), gid, mode, dev);
L
Linus Torvalds 已提交
516 517 518 519 520 521 522 523 524 525 526 527 528
	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)
529
		inc_nlink(dir);
L
Linus Torvalds 已提交
530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
	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
548
		gid = current_fsgid();
L
Linus Torvalds 已提交
549

550
	inode = hugetlbfs_get_inode(dir->i_sb, current_fsuid(),
L
Linus Torvalds 已提交
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566
					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;
}

/*
567
 * mark the head page dirty
L
Linus Torvalds 已提交
568 569 570
 */
static int hugetlbfs_set_page_dirty(struct page *page)
{
571
	struct page *head = compound_head(page);
572 573

	SetPageDirty(head);
L
Linus Torvalds 已提交
574 575 576
	return 0;
}

N
Naoya Horiguchi 已提交
577 578 579 580 581 582 583 584 585 586 587 588 589
static int hugetlbfs_migrate_page(struct address_space *mapping,
				struct page *newpage, struct page *page)
{
	int rc;

	rc = migrate_huge_page_move_mapping(mapping, newpage, page);
	if (rc)
		return rc;
	migrate_page_copy(newpage, page);

	return 0;
}

590
static int hugetlbfs_statfs(struct dentry *dentry, struct kstatfs *buf)
L
Linus Torvalds 已提交
591
{
592
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(dentry->d_sb);
593
	struct hstate *h = hstate_inode(dentry->d_inode);
L
Linus Torvalds 已提交
594 595

	buf->f_type = HUGETLBFS_MAGIC;
596
	buf->f_bsize = huge_page_size(h);
L
Linus Torvalds 已提交
597 598
	if (sbinfo) {
		spin_lock(&sbinfo->stat_lock);
599 600 601 602 603 604 605 606
		/* 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
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
		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);
	}
}

623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
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);
	}
}


648
static struct kmem_cache *hugetlbfs_inode_cachep;
L
Linus Torvalds 已提交
649 650 651

static struct inode *hugetlbfs_alloc_inode(struct super_block *sb)
{
652
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(sb);
L
Linus Torvalds 已提交
653 654
	struct hugetlbfs_inode_info *p;

655 656
	if (unlikely(!hugetlbfs_dec_free_inodes(sbinfo)))
		return NULL;
657
	p = kmem_cache_alloc(hugetlbfs_inode_cachep, GFP_KERNEL);
658 659
	if (unlikely(!p)) {
		hugetlbfs_inc_free_inodes(sbinfo);
L
Linus Torvalds 已提交
660
		return NULL;
661
	}
L
Linus Torvalds 已提交
662 663 664
	return &p->vfs_inode;
}

N
Nick Piggin 已提交
665 666 667 668 669 670 671
static void hugetlbfs_i_callback(struct rcu_head *head)
{
	struct inode *inode = container_of(head, struct inode, i_rcu);
	INIT_LIST_HEAD(&inode->i_dentry);
	kmem_cache_free(hugetlbfs_inode_cachep, HUGETLBFS_I(inode));
}

L
Linus Torvalds 已提交
672 673
static void hugetlbfs_destroy_inode(struct inode *inode)
{
674
	hugetlbfs_inc_free_inodes(HUGETLBFS_SB(inode->i_sb));
L
Linus Torvalds 已提交
675
	mpol_free_shared_policy(&HUGETLBFS_I(inode)->policy);
N
Nick Piggin 已提交
676
	call_rcu(&inode->i_rcu, hugetlbfs_i_callback);
L
Linus Torvalds 已提交
677 678
}

679
static const struct address_space_operations hugetlbfs_aops = {
N
Nick Piggin 已提交
680 681
	.write_begin	= hugetlbfs_write_begin,
	.write_end	= hugetlbfs_write_end,
L
Linus Torvalds 已提交
682
	.set_page_dirty	= hugetlbfs_set_page_dirty,
N
Naoya Horiguchi 已提交
683
	.migratepage    = hugetlbfs_migrate_page,
L
Linus Torvalds 已提交
684 685
};

686

687
static void init_once(void *foo)
688 689 690
{
	struct hugetlbfs_inode_info *ei = (struct hugetlbfs_inode_info *)foo;

C
Christoph Lameter 已提交
691
	inode_init_once(&ei->vfs_inode);
692 693
}

694
const struct file_operations hugetlbfs_file_operations = {
B
Badari Pulavarty 已提交
695
	.read			= hugetlbfs_read,
L
Linus Torvalds 已提交
696
	.mmap			= hugetlbfs_file_mmap,
697
	.fsync			= noop_fsync,
L
Linus Torvalds 已提交
698
	.get_unmapped_area	= hugetlb_get_unmapped_area,
699
	.llseek		= default_llseek,
L
Linus Torvalds 已提交
700 701
};

702
static const struct inode_operations hugetlbfs_dir_inode_operations = {
L
Linus Torvalds 已提交
703 704 705 706 707 708 709 710 711 712 713 714
	.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,
};

715
static const struct inode_operations hugetlbfs_inode_operations = {
L
Linus Torvalds 已提交
716 717 718
	.setattr	= hugetlbfs_setattr,
};

719
static const struct super_operations hugetlbfs_ops = {
L
Linus Torvalds 已提交
720 721
	.alloc_inode    = hugetlbfs_alloc_inode,
	.destroy_inode  = hugetlbfs_destroy_inode,
A
Al Viro 已提交
722
	.evict_inode	= hugetlbfs_evict_inode,
L
Linus Torvalds 已提交
723 724
	.statfs		= hugetlbfs_statfs,
	.put_super	= hugetlbfs_put_super,
M
Miklos Szeredi 已提交
725
	.show_options	= generic_show_options,
L
Linus Torvalds 已提交
726 727 728 729 730
};

static int
hugetlbfs_parse_options(char *options, struct hugetlbfs_config *pconfig)
{
731 732 733
	char *p, *rest;
	substring_t args[MAX_OPT_ARGS];
	int option;
734 735
	unsigned long long size = 0;
	enum { NO_SIZE, SIZE_STD, SIZE_PERCENT } setsize = NO_SIZE;
L
Linus Torvalds 已提交
736 737 738 739

	if (!options)
		return 0;

740 741
	while ((p = strsep(&options, ",")) != NULL) {
		int token;
742 743
		if (!*p)
			continue;
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761

		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;
762
			pconfig->mode = option & 01777U;
763 764 765 766 767 768 769
			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);
770 771 772
			setsize = SIZE_STD;
			if (*rest == '%')
				setsize = SIZE_PERCENT;
773 774
			break;
		}
L
Linus Torvalds 已提交
775

776 777 778 779 780 781 782
		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;

783 784 785 786 787 788 789 790 791 792 793 794 795
		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;
		}

796
		default:
797 798 799
			printk(KERN_ERR "hugetlbfs: Bad mount option: \"%s\"\n",
				 p);
			return -EINVAL;
800 801
			break;
		}
L
Linus Torvalds 已提交
802
	}
803 804 805 806 807 808 809 810 811 812 813 814

	/* 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
815
	return 0;
816 817 818 819

bad_val:
 	printk(KERN_ERR "hugetlbfs: Bad value '%s' for mount option '%s'\n",
	       args[0].from, p);
820
 	return -EINVAL;
L
Linus Torvalds 已提交
821 822 823 824 825 826 827 828 829 830 831
}

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
832 833
	save_mount_options(sb, data);

L
Linus Torvalds 已提交
834 835
	config.nr_blocks = -1; /* No limit on size by default */
	config.nr_inodes = -1; /* No limit on number of inodes by default */
836 837
	config.uid = current_fsuid();
	config.gid = current_fsgid();
L
Linus Torvalds 已提交
838
	config.mode = 0755;
839
	config.hstate = &default_hstate;
L
Linus Torvalds 已提交
840 841 842 843 844 845 846 847
	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;
848
	sbinfo->hstate = config.hstate;
L
Linus Torvalds 已提交
849 850 851 852 853 854
	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;
855 856
	sb->s_blocksize = huge_page_size(config.hstate);
	sb->s_blocksize_bits = huge_page_shift(config.hstate);
L
Linus Torvalds 已提交
857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876
	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;
}

877
int hugetlb_get_quota(struct address_space *mapping, long delta)
L
Linus Torvalds 已提交
878 879 880 881 882 883
{
	int ret = 0;
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(mapping->host->i_sb);

	if (sbinfo->free_blocks > -1) {
		spin_lock(&sbinfo->stat_lock);
884 885
		if (sbinfo->free_blocks - delta >= 0)
			sbinfo->free_blocks -= delta;
L
Linus Torvalds 已提交
886 887 888 889 890 891 892 893
		else
			ret = -ENOMEM;
		spin_unlock(&sbinfo->stat_lock);
	}

	return ret;
}

894
void hugetlb_put_quota(struct address_space *mapping, long delta)
L
Linus Torvalds 已提交
895 896 897 898 899
{
	struct hugetlbfs_sb_info *sbinfo = HUGETLBFS_SB(mapping->host->i_sb);

	if (sbinfo->free_blocks > -1) {
		spin_lock(&sbinfo->stat_lock);
900
		sbinfo->free_blocks += delta;
L
Linus Torvalds 已提交
901 902 903 904
		spin_unlock(&sbinfo->stat_lock);
	}
}

A
Al Viro 已提交
905 906
static struct dentry *hugetlbfs_mount(struct file_system_type *fs_type,
	int flags, const char *dev_name, void *data)
L
Linus Torvalds 已提交
907
{
A
Al Viro 已提交
908
	return mount_nodev(fs_type, flags, data, hugetlbfs_fill_super);
L
Linus Torvalds 已提交
909 910 911 912
}

static struct file_system_type hugetlbfs_fs_type = {
	.name		= "hugetlbfs",
A
Al Viro 已提交
913
	.mount		= hugetlbfs_mount,
L
Linus Torvalds 已提交
914 915 916 917 918
	.kill_sb	= kill_litter_super,
};

static struct vfsmount *hugetlbfs_vfsmount;

919
static int can_do_hugetlb_shm(void)
L
Linus Torvalds 已提交
920
{
921
	return capable(CAP_IPC_LOCK) || in_group_p(sysctl_hugetlb_shm_group);
L
Linus Torvalds 已提交
922 923
}

924 925
struct file *hugetlb_file_setup(const char *name, size_t size,
				vm_flags_t acctflag,
926
				struct user_struct **user, int creat_flags)
L
Linus Torvalds 已提交
927 928 929 930
{
	int error = -ENOMEM;
	struct file *file;
	struct inode *inode;
931 932
	struct path path;
	struct dentry *root;
L
Linus Torvalds 已提交
933 934
	struct qstr quick_string;

935
	*user = NULL;
936 937 938
	if (!hugetlbfs_vfsmount)
		return ERR_PTR(-ENOENT);

939
	if (creat_flags == HUGETLB_SHMFS_INODE && !can_do_hugetlb_shm()) {
940 941
		*user = current_user();
		if (user_shm_lock(size, *user)) {
942
			printk_once(KERN_WARNING "Using mlock ulimits for SHM_HUGETLB is deprecated\n");
943 944
		} else {
			*user = NULL;
945
			return ERR_PTR(-EPERM);
946
		}
947
	}
L
Linus Torvalds 已提交
948 949

	root = hugetlbfs_vfsmount->mnt_root;
950
	quick_string.name = name;
L
Linus Torvalds 已提交
951 952
	quick_string.len = strlen(quick_string.name);
	quick_string.hash = 0;
953 954
	path.dentry = d_alloc(root, &quick_string);
	if (!path.dentry)
L
Linus Torvalds 已提交
955 956
		goto out_shm_unlock;

957
	path.mnt = mntget(hugetlbfs_vfsmount);
L
Linus Torvalds 已提交
958
	error = -ENOSPC;
959 960
	inode = hugetlbfs_get_inode(root->d_sb, current_fsuid(),
				current_fsgid(), S_IFREG | S_IRWXUGO, 0);
L
Linus Torvalds 已提交
961
	if (!inode)
962
		goto out_dentry;
L
Linus Torvalds 已提交
963

964
	error = -ENOMEM;
965
	if (hugetlb_reserve_pages(inode, 0,
966 967
			size >> huge_page_shift(hstate_inode(inode)), NULL,
			acctflag))
968 969
		goto out_inode;

970
	d_instantiate(path.dentry, inode);
L
Linus Torvalds 已提交
971 972
	inode->i_size = size;
	inode->i_nlink = 0;
973 974

	error = -ENFILE;
975
	file = alloc_file(&path, FMODE_WRITE | FMODE_READ,
976 977
			&hugetlbfs_file_operations);
	if (!file)
978
		goto out_dentry; /* inode is already attached */
979

L
Linus Torvalds 已提交
980 981
	return file;

982 983
out_inode:
	iput(inode);
L
Linus Torvalds 已提交
984
out_dentry:
985
	path_put(&path);
L
Linus Torvalds 已提交
986
out_shm_unlock:
987 988 989 990
	if (*user) {
		user_shm_unlock(size, *user);
		*user = NULL;
	}
L
Linus Torvalds 已提交
991 992 993 994 995 996 997 998
	return ERR_PTR(error);
}

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

P
Peter Zijlstra 已提交
999 1000 1001 1002
	error = bdi_init(&hugetlbfs_backing_dev_info);
	if (error)
		return error;

L
Linus Torvalds 已提交
1003 1004
	hugetlbfs_inode_cachep = kmem_cache_create("hugetlbfs_inode_cache",
					sizeof(struct hugetlbfs_inode_info),
1005
					0, 0, init_once);
L
Linus Torvalds 已提交
1006
	if (hugetlbfs_inode_cachep == NULL)
P
Peter Zijlstra 已提交
1007
		goto out2;
L
Linus Torvalds 已提交
1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024

	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
1025 1026
 out2:
	bdi_destroy(&hugetlbfs_backing_dev_info);
L
Linus Torvalds 已提交
1027 1028 1029 1030 1031 1032
	return error;
}

static void __exit exit_hugetlbfs_fs(void)
{
	kmem_cache_destroy(hugetlbfs_inode_cachep);
1033
	kern_unmount(hugetlbfs_vfsmount);
L
Linus Torvalds 已提交
1034
	unregister_filesystem(&hugetlbfs_fs_type);
P
Peter Zijlstra 已提交
1035
	bdi_destroy(&hugetlbfs_backing_dev_info);
L
Linus Torvalds 已提交
1036 1037 1038 1039 1040 1041
}

module_init(init_hugetlbfs_fs)
module_exit(exit_hugetlbfs_fs)

MODULE_LICENSE("GPL");