dir.c 42.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/* dir.c: AFS filesystem directory handling
 *
D
David Howells 已提交
3
 * Copyright (C) 2002, 2018 Red Hat, Inc. All Rights Reserved.
L
Linus Torvalds 已提交
4 5 6 7 8 9 10 11 12 13
 * Written by David Howells (dhowells@redhat.com)
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */

#include <linux/kernel.h>
#include <linux/fs.h>
14
#include <linux/namei.h>
L
Linus Torvalds 已提交
15
#include <linux/pagemap.h>
D
David Howells 已提交
16
#include <linux/swap.h>
D
David Howells 已提交
17
#include <linux/ctype.h>
A
Alexey Dobriyan 已提交
18
#include <linux/sched.h>
D
David Howells 已提交
19
#include <linux/task_io_accounting_ops.h>
L
Linus Torvalds 已提交
20
#include "internal.h"
21
#include "xdr_fs.h"
L
Linus Torvalds 已提交
22

23
static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
A
Al Viro 已提交
24
				 unsigned int flags);
L
Linus Torvalds 已提交
25
static int afs_dir_open(struct inode *inode, struct file *file);
A
Al Viro 已提交
26
static int afs_readdir(struct file *file, struct dir_context *ctx);
27
static int afs_d_revalidate(struct dentry *dentry, unsigned int flags);
N
Nick Piggin 已提交
28
static int afs_d_delete(const struct dentry *dentry);
29
static int afs_lookup_one_filldir(struct dir_context *ctx, const char *name, int nlen,
30
				  loff_t fpos, u64 ino, unsigned dtype);
31 32
static int afs_lookup_filldir(struct dir_context *ctx, const char *name, int nlen,
			      loff_t fpos, u64 ino, unsigned dtype);
A
Al Viro 已提交
33
static int afs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
A
Al Viro 已提交
34
		      bool excl);
35
static int afs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode);
36 37 38 39 40 41 42
static int afs_rmdir(struct inode *dir, struct dentry *dentry);
static int afs_unlink(struct inode *dir, struct dentry *dentry);
static int afs_link(struct dentry *from, struct inode *dir,
		    struct dentry *dentry);
static int afs_symlink(struct inode *dir, struct dentry *dentry,
		       const char *content);
static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
43 44
		      struct inode *new_dir, struct dentry *new_dentry,
		      unsigned int flags);
D
David Howells 已提交
45 46 47 48 49 50 51 52
static int afs_dir_releasepage(struct page *page, gfp_t gfp_flags);
static void afs_dir_invalidatepage(struct page *page, unsigned int offset,
				   unsigned int length);

static int afs_dir_set_page_dirty(struct page *page)
{
	BUG(); /* This should never happen. */
}
L
Linus Torvalds 已提交
53

54
const struct file_operations afs_dir_file_operations = {
L
Linus Torvalds 已提交
55
	.open		= afs_dir_open,
D
David Howells 已提交
56
	.release	= afs_release,
A
Al Viro 已提交
57
	.iterate_shared	= afs_readdir,
D
David Howells 已提交
58
	.lock		= afs_lock,
59
	.llseek		= generic_file_llseek,
L
Linus Torvalds 已提交
60 61
};

62
const struct inode_operations afs_dir_inode_operations = {
63 64 65 66 67 68 69
	.create		= afs_create,
	.lookup		= afs_lookup,
	.link		= afs_link,
	.unlink		= afs_unlink,
	.symlink	= afs_symlink,
	.mkdir		= afs_mkdir,
	.rmdir		= afs_rmdir,
70
	.rename		= afs_rename,
D
David Howells 已提交
71
	.permission	= afs_permission,
D
David Howells 已提交
72
	.getattr	= afs_getattr,
73
	.setattr	= afs_setattr,
D
David Howells 已提交
74
	.listxattr	= afs_listxattr,
L
Linus Torvalds 已提交
75 76
};

D
David Howells 已提交
77 78 79 80 81 82
const struct address_space_operations afs_dir_aops = {
	.set_page_dirty	= afs_dir_set_page_dirty,
	.releasepage	= afs_dir_releasepage,
	.invalidatepage	= afs_dir_invalidatepage,
};

A
Al Viro 已提交
83
const struct dentry_operations afs_fs_dentry_operations = {
L
Linus Torvalds 已提交
84 85
	.d_revalidate	= afs_d_revalidate,
	.d_delete	= afs_d_delete,
86
	.d_release	= afs_d_release,
87
	.d_automount	= afs_d_automount,
L
Linus Torvalds 已提交
88 89
};

90 91 92 93 94 95 96
struct afs_lookup_one_cookie {
	struct dir_context	ctx;
	struct qstr		name;
	bool			found;
	struct afs_fid		fid;
};

97
struct afs_lookup_cookie {
98 99 100 101 102 103 104 105
	struct dir_context	ctx;
	struct qstr		name;
	bool			found;
	bool			one_only;
	unsigned short		nr_fids;
	struct afs_file_status	*statuses;
	struct afs_callback	*callbacks;
	struct afs_fid		fids[50];
L
Linus Torvalds 已提交
106 107 108 109 110
};

/*
 * check that a directory page is valid
 */
D
David Howells 已提交
111 112
static bool afs_dir_check_page(struct afs_vnode *dvnode, struct page *page,
			       loff_t i_size)
L
Linus Torvalds 已提交
113
{
114
	struct afs_xdr_dir_page *dbuf;
D
David Howells 已提交
115
	loff_t latter, off;
L
Linus Torvalds 已提交
116 117
	int tmp, qty;

118 119 120 121 122 123 124 125
	/* Determine how many magic numbers there should be in this page, but
	 * we must take care because the directory may change size under us.
	 */
	off = page_offset(page);
	if (i_size <= off)
		goto checked;

	latter = i_size - off;
L
Linus Torvalds 已提交
126 127 128 129
	if (latter >= PAGE_SIZE)
		qty = PAGE_SIZE;
	else
		qty = latter;
130
	qty /= sizeof(union afs_xdr_dir_block);
L
Linus Torvalds 已提交
131 132

	/* check them */
133
	dbuf = kmap(page);
L
Linus Torvalds 已提交
134
	for (tmp = 0; tmp < qty; tmp++) {
135
		if (dbuf->blocks[tmp].hdr.magic != AFS_DIR_MAGIC) {
136
			printk("kAFS: %s(%lx): bad magic %d/%d is %04hx\n",
D
David Howells 已提交
137
			       __func__, dvnode->vfs_inode.i_ino, tmp, qty,
138
			       ntohs(dbuf->blocks[tmp].hdr.magic));
D
David Howells 已提交
139
			trace_afs_dir_check_failed(dvnode, off, i_size);
140
			kunmap(page);
141
			trace_afs_file_error(dvnode, -EIO, afs_file_error_dir_bad_magic);
L
Linus Torvalds 已提交
142 143
			goto error;
		}
144 145 146 147 148 149

		/* Make sure each block is NUL terminated so we can reasonably
		 * use string functions on it.  The filenames in the page
		 * *should* be NUL-terminated anyway.
		 */
		((u8 *)&dbuf->blocks[tmp])[AFS_DIR_BLOCK_SIZE - 1] = 0;
L
Linus Torvalds 已提交
150 151
	}

152 153
	kunmap(page);

154
checked:
D
David Howells 已提交
155
	afs_stat_v(dvnode, n_read_dir);
156
	return true;
L
Linus Torvalds 已提交
157

D
David Howells 已提交
158
error:
159
	return false;
D
David Howells 已提交
160
}
L
Linus Torvalds 已提交
161 162 163 164 165 166 167 168

/*
 * open an AFS directory file
 */
static int afs_dir_open(struct inode *inode, struct file *file)
{
	_enter("{%lu}", inode->i_ino);

169 170
	BUILD_BUG_ON(sizeof(union afs_xdr_dir_block) != 2048);
	BUILD_BUG_ON(sizeof(union afs_xdr_dirent) != 32);
L
Linus Torvalds 已提交
171

172
	if (test_bit(AFS_VNODE_DELETED, &AFS_FS_I(inode)->flags))
L
Linus Torvalds 已提交
173 174
		return -ENOENT;

D
David Howells 已提交
175
	return afs_open(inode, file);
D
David Howells 已提交
176
}
L
Linus Torvalds 已提交
177

D
David Howells 已提交
178 179 180 181 182 183
/*
 * Read the directory into the pagecache in one go, scrubbing the previous
 * contents.  The list of pages is returned, pinning them so that they don't
 * get reclaimed during the iteration.
 */
static struct afs_read *afs_read_dir(struct afs_vnode *dvnode, struct key *key)
D
David Howells 已提交
184
	__acquires(&dvnode->validate_lock)
D
David Howells 已提交
185 186 187 188 189 190 191 192 193
{
	struct afs_read *req;
	loff_t i_size;
	int nr_pages, nr_inline, i, n;
	int ret = -ENOMEM;

retry:
	i_size = i_size_read(&dvnode->vfs_inode);
	if (i_size < 2048)
194 195 196
		return ERR_PTR(afs_bad(dvnode, afs_file_error_dir_small));
	if (i_size > 2048 * 1024) {
		trace_afs_file_error(dvnode, -EFBIG, afs_file_error_dir_big);
D
David Howells 已提交
197
		return ERR_PTR(-EFBIG);
198
	}
D
David Howells 已提交
199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267

	_enter("%llu", i_size);

	/* Get a request record to hold the page list.  We want to hold it
	 * inline if we can, but we don't want to make an order 1 allocation.
	 */
	nr_pages = (i_size + PAGE_SIZE - 1) / PAGE_SIZE;
	nr_inline = nr_pages;
	if (nr_inline > (PAGE_SIZE - sizeof(*req)) / sizeof(struct page *))
		nr_inline = 0;

	req = kzalloc(sizeof(*req) + sizeof(struct page *) * nr_inline,
		      GFP_KERNEL);
	if (!req)
		return ERR_PTR(-ENOMEM);

	refcount_set(&req->usage, 1);
	req->nr_pages = nr_pages;
	req->actual_len = i_size; /* May change */
	req->len = nr_pages * PAGE_SIZE; /* We can ask for more than there is */
	req->data_version = dvnode->status.data_version; /* May change */
	if (nr_inline > 0) {
		req->pages = req->array;
	} else {
		req->pages = kcalloc(nr_pages, sizeof(struct page *),
				     GFP_KERNEL);
		if (!req->pages)
			goto error;
	}

	/* Get a list of all the pages that hold or will hold the directory
	 * content.  We need to fill in any gaps that we might find where the
	 * memory reclaimer has been at work.  If there are any gaps, we will
	 * need to reread the entire directory contents.
	 */
	i = 0;
	do {
		n = find_get_pages_contig(dvnode->vfs_inode.i_mapping, i,
					  req->nr_pages - i,
					  req->pages + i);
		_debug("find %u at %u/%u", n, i, req->nr_pages);
		if (n == 0) {
			gfp_t gfp = dvnode->vfs_inode.i_mapping->gfp_mask;

			if (test_and_clear_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
				afs_stat_v(dvnode, n_inval);

			ret = -ENOMEM;
			req->pages[i] = __page_cache_alloc(gfp);
			if (!req->pages[i])
				goto error;
			ret = add_to_page_cache_lru(req->pages[i],
						    dvnode->vfs_inode.i_mapping,
						    i, gfp);
			if (ret < 0)
				goto error;

			set_page_private(req->pages[i], 1);
			SetPagePrivate(req->pages[i]);
			unlock_page(req->pages[i]);
			i++;
		} else {
			i += n;
		}
	} while (i < req->nr_pages);

	/* If we're going to reload, we need to lock all the pages to prevent
	 * races.
	 */
D
David Howells 已提交
268 269 270
	ret = -ERESTARTSYS;
	if (down_read_killable(&dvnode->validate_lock) < 0)
		goto error;
D
David Howells 已提交
271

D
David Howells 已提交
272 273
	if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		goto success;
D
David Howells 已提交
274

D
David Howells 已提交
275 276 277 278 279
	up_read(&dvnode->validate_lock);
	if (down_write_killable(&dvnode->validate_lock) < 0)
		goto error;

	if (!test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags)) {
D
David Howells 已提交
280 281
		ret = afs_fetch_data(dvnode, key, req);
		if (ret < 0)
D
David Howells 已提交
282
			goto error_unlock;
D
David Howells 已提交
283 284 285 286 287 288 289 290 291 292 293

		task_io_account_read(PAGE_SIZE * req->nr_pages);

		if (req->len < req->file_size)
			goto content_has_grown;

		/* Validate the data we just read. */
		ret = -EIO;
		for (i = 0; i < req->nr_pages; i++)
			if (!afs_dir_check_page(dvnode, req->pages[i],
						req->actual_len))
D
David Howells 已提交
294
				goto error_unlock;
D
David Howells 已提交
295 296 297 298 299 300

		// TODO: Trim excess pages

		set_bit(AFS_VNODE_DIR_VALID, &dvnode->flags);
	}

D
David Howells 已提交
301
	downgrade_write(&dvnode->validate_lock);
D
David Howells 已提交
302 303 304 305
success:
	return req;

error_unlock:
D
David Howells 已提交
306
	up_write(&dvnode->validate_lock);
D
David Howells 已提交
307 308 309 310 311 312
error:
	afs_put_read(req);
	_leave(" = %d", ret);
	return ERR_PTR(ret);

content_has_grown:
D
David Howells 已提交
313
	up_write(&dvnode->validate_lock);
D
David Howells 已提交
314 315 316 317
	afs_put_read(req);
	goto retry;
}

L
Linus Torvalds 已提交
318 319 320
/*
 * deal with one block in an AFS directory
 */
321 322
static int afs_dir_iterate_block(struct afs_vnode *dvnode,
				 struct dir_context *ctx,
323
				 union afs_xdr_dir_block *block,
A
Al Viro 已提交
324
				 unsigned blkoff)
L
Linus Torvalds 已提交
325
{
326
	union afs_xdr_dirent *dire;
L
Linus Torvalds 已提交
327 328
	unsigned offset, next, curr;
	size_t nlen;
A
Al Viro 已提交
329
	int tmp;
L
Linus Torvalds 已提交
330

A
Al Viro 已提交
331
	_enter("%u,%x,%p,,",(unsigned)ctx->pos,blkoff,block);
L
Linus Torvalds 已提交
332

333
	curr = (ctx->pos - blkoff) / sizeof(union afs_xdr_dirent);
L
Linus Torvalds 已提交
334 335

	/* walk through the block, an entry at a time */
336 337
	for (offset = (blkoff == 0 ? AFS_DIR_RESV_BLOCKS0 : AFS_DIR_RESV_BLOCKS);
	     offset < AFS_DIR_SLOTS_PER_BLOCK;
L
Linus Torvalds 已提交
338 339 340 341 342
	     offset = next
	     ) {
		next = offset + 1;

		/* skip entries marked unused in the bitmap */
343
		if (!(block->hdr.bitmap[offset / 8] &
L
Linus Torvalds 已提交
344
		      (1 << (offset % 8)))) {
345
			_debug("ENT[%zu.%u]: unused",
346
			       blkoff / sizeof(union afs_xdr_dir_block), offset);
L
Linus Torvalds 已提交
347
			if (offset >= curr)
A
Al Viro 已提交
348
				ctx->pos = blkoff +
349
					next * sizeof(union afs_xdr_dirent);
L
Linus Torvalds 已提交
350 351 352 353 354 355 356
			continue;
		}

		/* got a valid entry */
		dire = &block->dirents[offset];
		nlen = strnlen(dire->u.name,
			       sizeof(*block) -
357
			       offset * sizeof(union afs_xdr_dirent));
L
Linus Torvalds 已提交
358

359
		_debug("ENT[%zu.%u]: %s %zu \"%s\"",
360
		       blkoff / sizeof(union afs_xdr_dir_block), offset,
L
Linus Torvalds 已提交
361 362 363 364
		       (offset < curr ? "skip" : "fill"),
		       nlen, dire->u.name);

		/* work out where the next possible entry is */
365
		for (tmp = nlen; tmp > 15; tmp -= sizeof(union afs_xdr_dirent)) {
366
			if (next >= AFS_DIR_SLOTS_PER_BLOCK) {
367
				_debug("ENT[%zu.%u]:"
L
Linus Torvalds 已提交
368
				       " %u travelled beyond end dir block"
369
				       " (len %u/%zu)",
370
				       blkoff / sizeof(union afs_xdr_dir_block),
L
Linus Torvalds 已提交
371
				       offset, next, tmp, nlen);
372
				return afs_bad(dvnode, afs_file_error_dir_over_end);
L
Linus Torvalds 已提交
373
			}
374
			if (!(block->hdr.bitmap[next / 8] &
L
Linus Torvalds 已提交
375
			      (1 << (next % 8)))) {
376 377
				_debug("ENT[%zu.%u]:"
				       " %u unmarked extension (len %u/%zu)",
378
				       blkoff / sizeof(union afs_xdr_dir_block),
L
Linus Torvalds 已提交
379
				       offset, next, tmp, nlen);
380
				return afs_bad(dvnode, afs_file_error_dir_unmarked_ext);
L
Linus Torvalds 已提交
381 382
			}

383
			_debug("ENT[%zu.%u]: ext %u/%zu",
384
			       blkoff / sizeof(union afs_xdr_dir_block),
L
Linus Torvalds 已提交
385 386 387 388 389 390 391 392 393
			       next, tmp, nlen);
			next++;
		}

		/* skip if starts before the current position */
		if (offset < curr)
			continue;

		/* found the next entry */
A
Al Viro 已提交
394
		if (!dir_emit(ctx, dire->u.name, nlen,
L
Linus Torvalds 已提交
395
			      ntohl(dire->u.vnode),
396 397
			      (ctx->actor == afs_lookup_filldir ||
			       ctx->actor == afs_lookup_one_filldir)?
A
Al Viro 已提交
398
			      ntohl(dire->u.unique) : DT_UNKNOWN)) {
L
Linus Torvalds 已提交
399 400 401 402
			_leave(" = 0 [full]");
			return 0;
		}

403
		ctx->pos = blkoff + next * sizeof(union afs_xdr_dirent);
L
Linus Torvalds 已提交
404 405 406 407
	}

	_leave(" = 1 [more]");
	return 1;
D
David Howells 已提交
408
}
L
Linus Torvalds 已提交
409 410

/*
411
 * iterate through the data blob that lists the contents of an AFS directory
L
Linus Torvalds 已提交
412
 */
A
Al Viro 已提交
413 414
static int afs_dir_iterate(struct inode *dir, struct dir_context *ctx,
			   struct key *key)
L
Linus Torvalds 已提交
415
{
D
David Howells 已提交
416
	struct afs_vnode *dvnode = AFS_FS_I(dir);
417 418
	struct afs_xdr_dir_page *dbuf;
	union afs_xdr_dir_block *dblock;
D
David Howells 已提交
419
	struct afs_read *req;
L
Linus Torvalds 已提交
420 421 422 423
	struct page *page;
	unsigned blkoff, limit;
	int ret;

A
Al Viro 已提交
424
	_enter("{%lu},%u,,", dir->i_ino, (unsigned)ctx->pos);
L
Linus Torvalds 已提交
425

426
	if (test_bit(AFS_VNODE_DELETED, &AFS_FS_I(dir)->flags)) {
L
Linus Torvalds 已提交
427 428 429 430
		_leave(" = -ESTALE");
		return -ESTALE;
	}

D
David Howells 已提交
431 432 433 434
	req = afs_read_dir(dvnode, key);
	if (IS_ERR(req))
		return PTR_ERR(req);

L
Linus Torvalds 已提交
435
	/* round the file position up to the next entry boundary */
436 437
	ctx->pos += sizeof(union afs_xdr_dirent) - 1;
	ctx->pos &= ~(sizeof(union afs_xdr_dirent) - 1);
L
Linus Torvalds 已提交
438 439 440

	/* walk through the blocks in sequence */
	ret = 0;
D
David Howells 已提交
441
	while (ctx->pos < req->actual_len) {
442
		blkoff = ctx->pos & ~(sizeof(union afs_xdr_dir_block) - 1);
L
Linus Torvalds 已提交
443

D
David Howells 已提交
444 445 446 447 448
		/* Fetch the appropriate page from the directory and re-add it
		 * to the LRU.
		 */
		page = req->pages[blkoff / PAGE_SIZE];
		if (!page) {
449
			ret = afs_bad(dvnode, afs_file_error_dir_missing_page);
L
Linus Torvalds 已提交
450 451
			break;
		}
D
David Howells 已提交
452
		mark_page_accessed(page);
L
Linus Torvalds 已提交
453 454 455

		limit = blkoff & ~(PAGE_SIZE - 1);

D
David Howells 已提交
456
		dbuf = kmap(page);
L
Linus Torvalds 已提交
457 458 459 460

		/* deal with the individual blocks stashed on this page */
		do {
			dblock = &dbuf->blocks[(blkoff % PAGE_SIZE) /
461
					       sizeof(union afs_xdr_dir_block)];
462
			ret = afs_dir_iterate_block(dvnode, ctx, dblock, blkoff);
L
Linus Torvalds 已提交
463
			if (ret != 1) {
D
David Howells 已提交
464
				kunmap(page);
L
Linus Torvalds 已提交
465 466 467
				goto out;
			}

468
			blkoff += sizeof(union afs_xdr_dir_block);
L
Linus Torvalds 已提交
469

A
Al Viro 已提交
470
		} while (ctx->pos < dir->i_size && blkoff < limit);
L
Linus Torvalds 已提交
471

D
David Howells 已提交
472
		kunmap(page);
L
Linus Torvalds 已提交
473 474 475
		ret = 0;
	}

D
David Howells 已提交
476
out:
D
David Howells 已提交
477
	up_read(&dvnode->validate_lock);
D
David Howells 已提交
478
	afs_put_read(req);
L
Linus Torvalds 已提交
479 480
	_leave(" = %d", ret);
	return ret;
D
David Howells 已提交
481
}
L
Linus Torvalds 已提交
482 483 484 485

/*
 * read an AFS directory
 */
A
Al Viro 已提交
486
static int afs_readdir(struct file *file, struct dir_context *ctx)
L
Linus Torvalds 已提交
487
{
488
	return afs_dir_iterate(file_inode(file), ctx, afs_file_key(file));
D
David Howells 已提交
489
}
L
Linus Torvalds 已提交
490 491

/*
492
 * Search the directory for a single name
L
Linus Torvalds 已提交
493 494 495
 * - if afs_dir_iterate_block() spots this function, it'll pass the FID
 *   uniquifier through dtype
 */
496 497
static int afs_lookup_one_filldir(struct dir_context *ctx, const char *name,
				  int nlen, loff_t fpos, u64 ino, unsigned dtype)
L
Linus Torvalds 已提交
498
{
499 500
	struct afs_lookup_one_cookie *cookie =
		container_of(ctx, struct afs_lookup_one_cookie, ctx);
L
Linus Torvalds 已提交
501

A
Al Viro 已提交
502 503
	_enter("{%s,%u},%s,%u,,%llu,%u",
	       cookie->name.name, cookie->name.len, name, nlen,
504
	       (unsigned long long) ino, dtype);
L
Linus Torvalds 已提交
505

506
	/* insanity checks first */
507 508
	BUILD_BUG_ON(sizeof(union afs_xdr_dir_block) != 2048);
	BUILD_BUG_ON(sizeof(union afs_xdr_dirent) != 32);
509

A
Al Viro 已提交
510 511
	if (cookie->name.len != nlen ||
	    memcmp(cookie->name.name, name, nlen) != 0) {
L
Linus Torvalds 已提交
512 513 514 515 516 517 518 519 520 521
		_leave(" = 0 [no]");
		return 0;
	}

	cookie->fid.vnode = ino;
	cookie->fid.unique = dtype;
	cookie->found = 1;

	_leave(" = -1 [found]");
	return -1;
D
David Howells 已提交
522
}
L
Linus Torvalds 已提交
523 524

/*
525
 * Do a lookup of a single name in a directory
526
 * - just returns the FID the dentry name maps to if found
L
Linus Torvalds 已提交
527
 */
528 529
static int afs_do_lookup_one(struct inode *dir, struct dentry *dentry,
			     struct afs_fid *fid, struct key *key)
L
Linus Torvalds 已提交
530
{
A
Al Viro 已提交
531
	struct afs_super_info *as = dir->i_sb->s_fs_info;
532 533
	struct afs_lookup_one_cookie cookie = {
		.ctx.actor = afs_lookup_one_filldir,
A
Al Viro 已提交
534 535 536
		.name = dentry->d_name,
		.fid.vid = as->volume->vid
	};
L
Linus Torvalds 已提交
537 538
	int ret;

A
Al Viro 已提交
539
	_enter("{%lu},%p{%pd},", dir->i_ino, dentry, dentry);
L
Linus Torvalds 已提交
540 541

	/* search the directory */
A
Al Viro 已提交
542
	ret = afs_dir_iterate(dir, &cookie.ctx, key);
L
Linus Torvalds 已提交
543
	if (ret < 0) {
544 545
		_leave(" = %d [iter]", ret);
		return ret;
L
Linus Torvalds 已提交
546 547 548 549
	}

	ret = -ENOENT;
	if (!cookie.found) {
550 551
		_leave(" = -ENOENT [not found]");
		return -ENOENT;
L
Linus Torvalds 已提交
552 553
	}

554
	*fid = cookie.fid;
555
	_leave(" = 0 { vn=%llu u=%u }", fid->vnode, fid->unique);
556 557 558
	return 0;
}

559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575
/*
 * search the directory for a name
 * - if afs_dir_iterate_block() spots this function, it'll pass the FID
 *   uniquifier through dtype
 */
static int afs_lookup_filldir(struct dir_context *ctx, const char *name,
			      int nlen, loff_t fpos, u64 ino, unsigned dtype)
{
	struct afs_lookup_cookie *cookie =
		container_of(ctx, struct afs_lookup_cookie, ctx);
	int ret;

	_enter("{%s,%u},%s,%u,,%llu,%u",
	       cookie->name.name, cookie->name.len, name, nlen,
	       (unsigned long long) ino, dtype);

	/* insanity checks first */
576 577
	BUILD_BUG_ON(sizeof(union afs_xdr_dir_block) != 2048);
	BUILD_BUG_ON(sizeof(union afs_xdr_dirent) != 32);
578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764

	if (cookie->found) {
		if (cookie->nr_fids < 50) {
			cookie->fids[cookie->nr_fids].vnode	= ino;
			cookie->fids[cookie->nr_fids].unique	= dtype;
			cookie->nr_fids++;
		}
	} else if (cookie->name.len == nlen &&
		   memcmp(cookie->name.name, name, nlen) == 0) {
		cookie->fids[0].vnode	= ino;
		cookie->fids[0].unique	= dtype;
		cookie->found = 1;
		if (cookie->one_only)
			return -1;
	}

	ret = cookie->nr_fids >= 50 ? -1 : 0;
	_leave(" = %d", ret);
	return ret;
}

/*
 * Do a lookup in a directory.  We make use of bulk lookup to query a slew of
 * files in one go and create inodes for them.  The inode of the file we were
 * asked for is returned.
 */
static struct inode *afs_do_lookup(struct inode *dir, struct dentry *dentry,
				   struct key *key)
{
	struct afs_lookup_cookie *cookie;
	struct afs_cb_interest *cbi = NULL;
	struct afs_super_info *as = dir->i_sb->s_fs_info;
	struct afs_iget_data data;
	struct afs_fs_cursor fc;
	struct afs_vnode *dvnode = AFS_FS_I(dir);
	struct inode *inode = NULL;
	int ret, i;

	_enter("{%lu},%p{%pd},", dir->i_ino, dentry, dentry);

	cookie = kzalloc(sizeof(struct afs_lookup_cookie), GFP_KERNEL);
	if (!cookie)
		return ERR_PTR(-ENOMEM);

	cookie->ctx.actor = afs_lookup_filldir;
	cookie->name = dentry->d_name;
	cookie->nr_fids = 1; /* slot 0 is saved for the fid we actually want */

	read_seqlock_excl(&dvnode->cb_lock);
	if (dvnode->cb_interest &&
	    dvnode->cb_interest->server &&
	    test_bit(AFS_SERVER_FL_NO_IBULK, &dvnode->cb_interest->server->flags))
		cookie->one_only = true;
	read_sequnlock_excl(&dvnode->cb_lock);

	for (i = 0; i < 50; i++)
		cookie->fids[i].vid = as->volume->vid;

	/* search the directory */
	ret = afs_dir_iterate(dir, &cookie->ctx, key);
	if (ret < 0) {
		inode = ERR_PTR(ret);
		goto out;
	}

	inode = ERR_PTR(-ENOENT);
	if (!cookie->found)
		goto out;

	/* Check to see if we already have an inode for the primary fid. */
	data.volume = dvnode->volume;
	data.fid = cookie->fids[0];
	inode = ilookup5(dir->i_sb, cookie->fids[0].vnode, afs_iget5_test, &data);
	if (inode)
		goto out;

	/* Need space for examining all the selected files */
	inode = ERR_PTR(-ENOMEM);
	cookie->statuses = kcalloc(cookie->nr_fids, sizeof(struct afs_file_status),
				   GFP_KERNEL);
	if (!cookie->statuses)
		goto out;

	cookie->callbacks = kcalloc(cookie->nr_fids, sizeof(struct afs_callback),
				    GFP_KERNEL);
	if (!cookie->callbacks)
		goto out_s;

	/* Try FS.InlineBulkStatus first.  Abort codes for the individual
	 * lookups contained therein are stored in the reply without aborting
	 * the whole operation.
	 */
	if (cookie->one_only)
		goto no_inline_bulk_status;

	inode = ERR_PTR(-ERESTARTSYS);
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
			if (test_bit(AFS_SERVER_FL_NO_IBULK,
				      &fc.cbi->server->flags)) {
				fc.ac.abort_code = RX_INVALID_OPERATION;
				fc.ac.error = -ECONNABORTED;
				break;
			}
			afs_fs_inline_bulk_status(&fc,
						  afs_v2net(dvnode),
						  cookie->fids,
						  cookie->statuses,
						  cookie->callbacks,
						  cookie->nr_fids, NULL);
		}

		if (fc.ac.error == 0)
			cbi = afs_get_cb_interest(fc.cbi);
		if (fc.ac.abort_code == RX_INVALID_OPERATION)
			set_bit(AFS_SERVER_FL_NO_IBULK, &fc.cbi->server->flags);
		inode = ERR_PTR(afs_end_vnode_operation(&fc));
	}

	if (!IS_ERR(inode))
		goto success;
	if (fc.ac.abort_code != RX_INVALID_OPERATION)
		goto out_c;

no_inline_bulk_status:
	/* We could try FS.BulkStatus next, but this aborts the entire op if
	 * any of the lookups fails - so, for the moment, revert to
	 * FS.FetchStatus for just the primary fid.
	 */
	cookie->nr_fids = 1;
	inode = ERR_PTR(-ERESTARTSYS);
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
			afs_fs_fetch_status(&fc,
					    afs_v2net(dvnode),
					    cookie->fids,
					    cookie->statuses,
					    cookie->callbacks,
					    NULL);
		}

		if (fc.ac.error == 0)
			cbi = afs_get_cb_interest(fc.cbi);
		inode = ERR_PTR(afs_end_vnode_operation(&fc));
	}

	if (IS_ERR(inode))
		goto out_c;

	for (i = 0; i < cookie->nr_fids; i++)
		cookie->statuses[i].abort_code = 0;

success:
	/* Turn all the files into inodes and save the first one - which is the
	 * one we actually want.
	 */
	if (cookie->statuses[0].abort_code != 0)
		inode = ERR_PTR(afs_abort_to_error(cookie->statuses[0].abort_code));

	for (i = 0; i < cookie->nr_fids; i++) {
		struct inode *ti;

		if (cookie->statuses[i].abort_code != 0)
			continue;

		ti = afs_iget(dir->i_sb, key, &cookie->fids[i],
			      &cookie->statuses[i],
			      &cookie->callbacks[i],
			      cbi);
		if (i == 0) {
			inode = ti;
		} else {
			if (!IS_ERR(ti))
				iput(ti);
		}
	}

out_c:
	afs_put_cb_interest(afs_v2net(dvnode), cbi);
	kfree(cookie->callbacks);
out_s:
	kfree(cookie->statuses);
out:
	kfree(cookie);
	return inode;
}

765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
/*
 * Look up an entry in a directory with @sys substitution.
 */
static struct dentry *afs_lookup_atsys(struct inode *dir, struct dentry *dentry,
				       struct key *key)
{
	struct afs_sysnames *subs;
	struct afs_net *net = afs_i2net(dir);
	struct dentry *ret;
	char *buf, *p, *name;
	int len, i;

	_enter("");

	ret = ERR_PTR(-ENOMEM);
	p = buf = kmalloc(AFSNAMEMAX, GFP_KERNEL);
	if (!buf)
		goto out_p;
	if (dentry->d_name.len > 4) {
		memcpy(p, dentry->d_name.name, dentry->d_name.len - 4);
		p += dentry->d_name.len - 4;
	}

	/* There is an ordered list of substitutes that we have to try. */
	read_lock(&net->sysnames_lock);
	subs = net->sysnames;
	refcount_inc(&subs->usage);
	read_unlock(&net->sysnames_lock);

	for (i = 0; i < subs->nr; i++) {
		name = subs->subs[i];
		len = dentry->d_name.len - 4 + strlen(name);
		if (len >= AFSNAMEMAX) {
			ret = ERR_PTR(-ENAMETOOLONG);
			goto out_s;
		}

		strcpy(p, name);
		ret = lookup_one_len(buf, dentry->d_parent, len);
		if (IS_ERR(ret) || d_is_positive(ret))
			goto out_s;
		dput(ret);
	}

	/* We don't want to d_add() the @sys dentry here as we don't want to
	 * the cached dentry to hide changes to the sysnames list.
	 */
	ret = NULL;
out_s:
	afs_put_sysnames(subs);
	kfree(buf);
out_p:
	key_put(key);
	return ret;
}

821 822 823
/*
 * look up an entry in a directory
 */
824
static struct dentry *afs_lookup(struct inode *dir, struct dentry *dentry,
A
Al Viro 已提交
825
				 unsigned int flags)
826
{
827
	struct afs_vnode *dvnode = AFS_FS_I(dir);
828
	struct inode *inode;
829
	struct dentry *d;
D
David Howells 已提交
830
	struct key *key;
831 832
	int ret;

833
	_enter("{%llx:%llu},%p{%pd},",
834
	       dvnode->fid.vid, dvnode->fid.vnode, dentry, dentry);
835

836
	ASSERTCMP(d_inode(dentry), ==, NULL);
837

D
David Howells 已提交
838
	if (dentry->d_name.len >= AFSNAMEMAX) {
839 840 841 842
		_leave(" = -ENAMETOOLONG");
		return ERR_PTR(-ENAMETOOLONG);
	}

843
	if (test_bit(AFS_VNODE_DELETED, &dvnode->flags)) {
844 845 846 847
		_leave(" = -ESTALE");
		return ERR_PTR(-ESTALE);
	}

848
	key = afs_request_key(dvnode->volume->cell);
D
David Howells 已提交
849 850
	if (IS_ERR(key)) {
		_leave(" = %ld [key]", PTR_ERR(key));
851
		return ERR_CAST(key);
D
David Howells 已提交
852 853
	}

854
	ret = afs_validate(dvnode, key);
855 856 857 858 859 860
	if (ret < 0) {
		key_put(key);
		_leave(" = %d [val]", ret);
		return ERR_PTR(ret);
	}

861 862 863 864 865 866 867
	if (dentry->d_name.len >= 4 &&
	    dentry->d_name.name[dentry->d_name.len - 4] == '@' &&
	    dentry->d_name.name[dentry->d_name.len - 3] == 's' &&
	    dentry->d_name.name[dentry->d_name.len - 2] == 'y' &&
	    dentry->d_name.name[dentry->d_name.len - 1] == 's')
		return afs_lookup_atsys(dir, dentry, key);

868
	afs_stat_v(dvnode, n_lookup);
869
	inode = afs_do_lookup(dir, dentry, key);
D
David Howells 已提交
870
	key_put(key);
871 872 873 874 875
	if (inode == ERR_PTR(-ENOENT)) {
		inode = afs_try_auto_mntpt(dentry, dir);
	} else {
		dentry->d_fsdata =
			(void *)(unsigned long)dvnode->status.data_version;
876
	}
877 878 879 880
	d = d_splice_alias(inode, dentry);
	if (!IS_ERR_OR_NULL(d))
		d->d_fsdata = dentry->d_fsdata;
	return d;
D
David Howells 已提交
881
}
L
Linus Torvalds 已提交
882 883 884 885 886 887

/*
 * check that a dentry lookup hit has found a valid entry
 * - NOTE! the hit can be a negative hit too, so we can't assume we have an
 *   inode
 */
888
static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
L
Linus Torvalds 已提交
889
{
890
	struct afs_vnode *vnode, *dir;
A
Artem Bityutskiy 已提交
891
	struct afs_fid uninitialized_var(fid);
L
Linus Torvalds 已提交
892
	struct dentry *parent;
893
	struct inode *inode;
D
David Howells 已提交
894
	struct key *key;
895
	long dir_version, de_version;
L
Linus Torvalds 已提交
896 897
	int ret;

898
	if (flags & LOOKUP_RCU)
899 900
		return -ECHILD;

901 902
	if (d_really_is_positive(dentry)) {
		vnode = AFS_FS_I(d_inode(dentry));
903
		_enter("{v={%llx:%llu} n=%pd fl=%lx},",
A
Al Viro 已提交
904
		       vnode->fid.vid, vnode->fid.vnode, dentry,
905
		       vnode->flags);
906
	} else {
A
Al Viro 已提交
907
		_enter("{neg n=%pd}", dentry);
908
	}
L
Linus Torvalds 已提交
909

910
	key = afs_request_key(AFS_FS_S(dentry->d_sb)->volume->cell);
D
David Howells 已提交
911 912 913
	if (IS_ERR(key))
		key = NULL;

914 915 916 917 918 919 920 921 922 923
	if (d_really_is_positive(dentry)) {
		inode = d_inode(dentry);
		if (inode) {
			vnode = AFS_FS_I(inode);
			afs_validate(vnode, key);
			if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
				goto out_bad;
		}
	}

L
Linus Torvalds 已提交
924
	/* lock down the parent dentry so we can peer at it */
925
	parent = dget_parent(dentry);
926
	dir = AFS_FS_I(d_inode(parent));
L
Linus Torvalds 已提交
927

928
	/* validate the parent directory */
929
	afs_validate(dir, key);
930 931

	if (test_bit(AFS_VNODE_DELETED, &dir->flags)) {
A
Al Viro 已提交
932
		_debug("%pd: parent dir deleted", dentry);
933
		goto out_bad_parent;
L
Linus Torvalds 已提交
934 935
	}

936 937 938 939 940 941 942 943 944 945 946 947 948
	/* We only need to invalidate a dentry if the server's copy changed
	 * behind our back.  If we made the change, it's no problem.  Note that
	 * on a 32-bit system, we only have 32 bits in the dentry to store the
	 * version.
	 */
	dir_version = (long)dir->status.data_version;
	de_version = (long)dentry->d_fsdata;
	if (de_version == dir_version)
		goto out_valid;

	dir_version = (long)dir->invalid_before;
	if (de_version - dir_version >= 0)
		goto out_valid;
L
Linus Torvalds 已提交
949

950
	_debug("dir modified");
951
	afs_stat_v(dir, n_reval);
952 953

	/* search the directory for this vnode */
954
	ret = afs_do_lookup_one(&dir->vfs_inode, dentry, &fid, key);
955 956 957
	switch (ret) {
	case 0:
		/* the filename maps to something */
958
		if (d_really_is_negative(dentry))
959 960 961
			goto out_bad_parent;
		inode = d_inode(dentry);
		if (is_bad_inode(inode)) {
A
Al Viro 已提交
962 963
			printk("kAFS: afs_d_revalidate: %pd2 has bad inode\n",
			       dentry);
964
			goto out_bad_parent;
L
Linus Torvalds 已提交
965 966
		}

967 968
		vnode = AFS_FS_I(inode);

L
Linus Torvalds 已提交
969 970
		/* if the vnode ID has changed, then the dirent points to a
		 * different file */
971
		if (fid.vnode != vnode->fid.vnode) {
972
			_debug("%pd: dirent changed [%llu != %llu]",
A
Al Viro 已提交
973
			       dentry, fid.vnode,
974
			       vnode->fid.vnode);
L
Linus Torvalds 已提交
975 976 977 978
			goto not_found;
		}

		/* if the vnode ID uniqifier has changed, then the file has
979 980
		 * been deleted and replaced, and the original vnode ID has
		 * been reused */
981
		if (fid.unique != vnode->fid.unique) {
A
Al Viro 已提交
982 983
			_debug("%pd: file deleted (uq %u -> %u I:%u)",
			       dentry, fid.unique,
984
			       vnode->fid.unique,
985 986
			       vnode->vfs_inode.i_generation);
			write_seqlock(&vnode->cb_lock);
987
			set_bit(AFS_VNODE_DELETED, &vnode->flags);
988
			write_sequnlock(&vnode->cb_lock);
989
			goto not_found;
L
Linus Torvalds 已提交
990
		}
991
		goto out_valid;
992

993 994
	case -ENOENT:
		/* the filename is unknown */
A
Al Viro 已提交
995
		_debug("%pd: dirent not found", dentry);
996
		if (d_really_is_positive(dentry))
997 998
			goto not_found;
		goto out_valid;
L
Linus Torvalds 已提交
999

1000
	default:
A
Al Viro 已提交
1001 1002
		_debug("failed to iterate dir %pd: %d",
		       parent, ret);
1003
		goto out_bad_parent;
1004 1005
	}

D
David Howells 已提交
1006
out_valid:
1007
	dentry->d_fsdata = (void *)dir_version;
L
Linus Torvalds 已提交
1008
	dput(parent);
D
David Howells 已提交
1009
	key_put(key);
L
Linus Torvalds 已提交
1010 1011 1012 1013
	_leave(" = 1 [valid]");
	return 1;

	/* the dirent, if it exists, now points to a different vnode */
D
David Howells 已提交
1014
not_found:
L
Linus Torvalds 已提交
1015 1016 1017 1018
	spin_lock(&dentry->d_lock);
	dentry->d_flags |= DCACHE_NFSFS_RENAMED;
	spin_unlock(&dentry->d_lock);

1019
out_bad_parent:
A
Al Viro 已提交
1020
	_debug("dropping dentry %pd2", dentry);
L
Linus Torvalds 已提交
1021
	dput(parent);
1022
out_bad:
D
David Howells 已提交
1023
	key_put(key);
L
Linus Torvalds 已提交
1024 1025 1026

	_leave(" = 0 [bad]");
	return 0;
D
David Howells 已提交
1027
}
L
Linus Torvalds 已提交
1028 1029 1030 1031 1032 1033 1034

/*
 * allow the VFS to enquire as to whether a dentry should be unhashed (mustn't
 * sleep)
 * - called from dput() when d_count is going to 0.
 * - return 1 to request dentry be unhashed, 0 otherwise
 */
N
Nick Piggin 已提交
1035
static int afs_d_delete(const struct dentry *dentry)
L
Linus Torvalds 已提交
1036
{
A
Al Viro 已提交
1037
	_enter("%pd", dentry);
L
Linus Torvalds 已提交
1038 1039 1040 1041

	if (dentry->d_flags & DCACHE_NFSFS_RENAMED)
		goto zap;

1042 1043 1044
	if (d_really_is_positive(dentry) &&
	    (test_bit(AFS_VNODE_DELETED,   &AFS_FS_I(d_inode(dentry))->flags) ||
	     test_bit(AFS_VNODE_PSEUDODIR, &AFS_FS_I(d_inode(dentry))->flags)))
1045
		goto zap;
L
Linus Torvalds 已提交
1046 1047 1048 1049

	_leave(" = 0 [keep]");
	return 0;

D
David Howells 已提交
1050
zap:
L
Linus Torvalds 已提交
1051 1052
	_leave(" = 1 [zap]");
	return 1;
D
David Howells 已提交
1053
}
1054 1055 1056 1057

/*
 * handle dentry release
 */
1058
void afs_d_release(struct dentry *dentry)
1059
{
A
Al Viro 已提交
1060
	_enter("%pd", dentry);
1061 1062
}

1063 1064 1065 1066 1067 1068 1069 1070 1071
/*
 * Create a new inode for create/mkdir/symlink
 */
static void afs_vnode_new_inode(struct afs_fs_cursor *fc,
				struct dentry *new_dentry,
				struct afs_fid *newfid,
				struct afs_file_status *newstatus,
				struct afs_callback *newcb)
{
1072
	struct afs_vnode *vnode;
1073 1074 1075 1076 1077
	struct inode *inode;

	if (fc->ac.error < 0)
		return;

1078 1079
	d_drop(new_dentry);

1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
	inode = afs_iget(fc->vnode->vfs_inode.i_sb, fc->key,
			 newfid, newstatus, newcb, fc->cbi);
	if (IS_ERR(inode)) {
		/* ENOMEM or EINTR at a really inconvenient time - just abandon
		 * the new directory on the server.
		 */
		fc->ac.error = PTR_ERR(inode);
		return;
	}

1090 1091
	vnode = AFS_FS_I(inode);
	set_bit(AFS_VNODE_NEW_CONTENT, &vnode->flags);
1092
	afs_vnode_commit_status(fc, vnode, 0);
1093
	d_add(new_dentry, inode);
1094 1095
}

1096 1097 1098
/*
 * create a directory on an AFS filesystem
 */
1099
static int afs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
1100
{
1101 1102 1103 1104 1105
	struct afs_file_status newstatus;
	struct afs_fs_cursor fc;
	struct afs_callback newcb;
	struct afs_vnode *dvnode = AFS_FS_I(dir);
	struct afs_fid newfid;
1106
	struct key *key;
1107
	u64 data_version = dvnode->status.data_version;
1108 1109
	int ret;

1110
	mode |= S_IFDIR;
1111

1112
	_enter("{%llx:%llu},{%pd},%ho",
A
Al Viro 已提交
1113
	       dvnode->fid.vid, dvnode->fid.vnode, dentry, mode);
1114 1115 1116 1117 1118 1119 1120

	key = afs_request_key(dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1121 1122 1123
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
1124
			fc.cb_break = afs_calc_vnode_cb_break(dvnode);
1125
			afs_fs_create(&fc, dentry->d_name.name, mode, data_version,
1126 1127
				      &newfid, &newstatus, &newcb);
		}
1128

1129 1130 1131 1132 1133 1134
		afs_check_for_remote_deletion(&fc, fc.vnode);
		afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
		afs_vnode_new_inode(&fc, dentry, &newfid, &newstatus, &newcb);
		ret = afs_end_vnode_operation(&fc);
		if (ret < 0)
			goto error_key;
1135 1136
	} else {
		goto error_key;
1137 1138
	}

1139 1140 1141 1142 1143
	if (ret == 0 &&
	    test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		afs_edit_dir_add(dvnode, &dentry->d_name, &newfid,
				 afs_edit_dir_for_create);

1144 1145 1146 1147
	key_put(key);
	_leave(" = 0");
	return 0;

1148
error_key:
1149 1150 1151 1152 1153 1154 1155
	key_put(key);
error:
	d_drop(dentry);
	_leave(" = %d", ret);
	return ret;
}

1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
/*
 * Remove a subdir from a directory.
 */
static void afs_dir_remove_subdir(struct dentry *dentry)
{
	if (d_really_is_positive(dentry)) {
		struct afs_vnode *vnode = AFS_FS_I(d_inode(dentry));

		clear_nlink(&vnode->vfs_inode);
		set_bit(AFS_VNODE_DELETED, &vnode->flags);
		clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
1167
		clear_bit(AFS_VNODE_DIR_VALID, &vnode->flags);
1168 1169 1170
	}
}

1171 1172 1173 1174 1175
/*
 * remove a directory from an AFS filesystem
 */
static int afs_rmdir(struct inode *dir, struct dentry *dentry)
{
1176
	struct afs_fs_cursor fc;
1177
	struct afs_vnode *dvnode = AFS_FS_I(dir), *vnode = NULL;
1178
	struct key *key;
1179
	u64 data_version = dvnode->status.data_version;
1180 1181
	int ret;

1182
	_enter("{%llx:%llu},{%pd}",
A
Al Viro 已提交
1183
	       dvnode->fid.vid, dvnode->fid.vnode, dentry);
1184 1185 1186 1187 1188 1189 1190

	key = afs_request_key(dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1191 1192 1193 1194 1195 1196 1197 1198
	/* Try to make sure we have a callback promise on the victim. */
	if (d_really_is_positive(dentry)) {
		vnode = AFS_FS_I(d_inode(dentry));
		ret = afs_validate(vnode, key);
		if (ret < 0)
			goto error_key;
	}

1199 1200 1201
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
1202
			fc.cb_break = afs_calc_vnode_cb_break(dvnode);
1203
			afs_fs_remove(&fc, vnode, dentry->d_name.name, true,
1204
				      data_version);
1205
		}
1206

1207 1208
		afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
		ret = afs_end_vnode_operation(&fc);
1209
		if (ret == 0) {
1210
			afs_dir_remove_subdir(dentry);
1211 1212 1213 1214
			if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
				afs_edit_dir_remove(dvnode, &dentry->d_name,
						    afs_edit_dir_for_rmdir);
		}
1215 1216
	}

1217
error_key:
1218 1219 1220 1221 1222 1223
	key_put(key);
error:
	return ret;
}

/*
1224 1225 1226 1227 1228 1229 1230 1231 1232
 * Remove a link to a file or symlink from a directory.
 *
 * If the file was not deleted due to excess hard links, the fileserver will
 * break the callback promise on the file - if it had one - before it returns
 * to us, and if it was deleted, it won't
 *
 * However, if we didn't have a callback promise outstanding, or it was
 * outstanding on a different server, then it won't break it either...
 */
D
David Howells 已提交
1233 1234 1235
static int afs_dir_remove_link(struct dentry *dentry, struct key *key,
			       unsigned long d_version_before,
			       unsigned long d_version_after)
1236
{
D
David Howells 已提交
1237
	bool dir_valid;
1238 1239
	int ret = 0;

D
David Howells 已提交
1240 1241 1242 1243 1244
	/* There were no intervening changes on the server if the version
	 * number we got back was incremented by exactly 1.
	 */
	dir_valid = (d_version_after == d_version_before + 1);

1245 1246 1247
	if (d_really_is_positive(dentry)) {
		struct afs_vnode *vnode = AFS_FS_I(d_inode(dentry));

1248 1249 1250
		if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
			/* Already done */
		} else if (dir_valid) {
D
David Howells 已提交
1251 1252 1253 1254 1255
			drop_nlink(&vnode->vfs_inode);
			if (vnode->vfs_inode.i_nlink == 0) {
				set_bit(AFS_VNODE_DELETED, &vnode->flags);
				clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
			}
1256
			ret = 0;
D
David Howells 已提交
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
		} else {
			clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);

			if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
				kdebug("AFS_VNODE_DELETED");

			ret = afs_validate(vnode, key);
			if (ret == -ESTALE)
				ret = 0;
		}
1267 1268 1269 1270 1271 1272 1273 1274
		_debug("nlink %d [val %d]", vnode->vfs_inode.i_nlink, ret);
	}

	return ret;
}

/*
 * Remove a file or symlink from an AFS filesystem.
1275 1276 1277
 */
static int afs_unlink(struct inode *dir, struct dentry *dentry)
{
1278
	struct afs_fs_cursor fc;
1279
	struct afs_vnode *dvnode = AFS_FS_I(dir), *vnode = NULL;
1280
	struct key *key;
D
David Howells 已提交
1281
	unsigned long d_version = (unsigned long)dentry->d_fsdata;
1282
	u64 data_version = dvnode->status.data_version;
1283 1284
	int ret;

1285
	_enter("{%llx:%llu},{%pd}",
A
Al Viro 已提交
1286
	       dvnode->fid.vid, dvnode->fid.vnode, dentry);
1287

D
David Howells 已提交
1288
	if (dentry->d_name.len >= AFSNAMEMAX)
1289
		return -ENAMETOOLONG;
1290 1291 1292 1293 1294 1295 1296

	key = afs_request_key(dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1297
	/* Try to make sure we have a callback promise on the victim. */
1298 1299
	if (d_really_is_positive(dentry)) {
		vnode = AFS_FS_I(d_inode(dentry));
1300 1301
		ret = afs_validate(vnode, key);
		if (ret < 0)
1302
			goto error_key;
1303 1304
	}

1305 1306 1307
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
1308
			fc.cb_break = afs_calc_vnode_cb_break(dvnode);
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320

			if (test_bit(AFS_SERVER_FL_IS_YFS, &fc.cbi->server->flags) &&
			    !test_bit(AFS_SERVER_FL_NO_RM2, &fc.cbi->server->flags)) {
				yfs_fs_remove_file2(&fc, vnode, dentry->d_name.name,
						    data_version);
				if (fc.ac.error != -ECONNABORTED ||
				    fc.ac.abort_code != RXGEN_OPCODE)
					continue;
				set_bit(AFS_SERVER_FL_NO_RM2, &fc.cbi->server->flags);
			}

			afs_fs_remove(&fc, vnode, dentry->d_name.name, false,
1321
				      data_version);
1322
		}
1323

1324 1325 1326
		afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
		ret = afs_end_vnode_operation(&fc);
		if (ret == 0)
D
David Howells 已提交
1327 1328 1329
			ret = afs_dir_remove_link(
				dentry, key, d_version,
				(unsigned long)dvnode->status.data_version);
1330 1331 1332 1333
		if (ret == 0 &&
		    test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
			afs_edit_dir_remove(dvnode, &dentry->d_name,
					    afs_edit_dir_for_unlink);
1334 1335
	}

1336
error_key:
1337 1338 1339 1340 1341 1342 1343 1344 1345
	key_put(key);
error:
	_leave(" = %d", ret);
	return ret;
}

/*
 * create a regular file on an AFS filesystem
 */
A
Al Viro 已提交
1346
static int afs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
A
Al Viro 已提交
1347
		      bool excl)
1348
{
1349 1350 1351
	struct afs_fs_cursor fc;
	struct afs_file_status newstatus;
	struct afs_callback newcb;
1352
	struct afs_vnode *dvnode = AFS_FS_I(dir);
1353
	struct afs_fid newfid;
1354
	struct key *key;
1355
	u64 data_version = dvnode->status.data_version;
1356 1357
	int ret;

1358
	mode |= S_IFREG;
1359

1360
	_enter("{%llx:%llu},{%pd},%ho,",
A
Al Viro 已提交
1361
	       dvnode->fid.vid, dvnode->fid.vnode, dentry, mode);
1362

1363 1364 1365 1366
	ret = -ENAMETOOLONG;
	if (dentry->d_name.len >= AFSNAMEMAX)
		goto error;

1367 1368 1369 1370 1371 1372
	key = afs_request_key(dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1373 1374 1375
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
1376
			fc.cb_break = afs_calc_vnode_cb_break(dvnode);
1377
			afs_fs_create(&fc, dentry->d_name.name, mode, data_version,
1378 1379
				      &newfid, &newstatus, &newcb);
		}
1380

1381 1382 1383 1384 1385 1386
		afs_check_for_remote_deletion(&fc, fc.vnode);
		afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
		afs_vnode_new_inode(&fc, dentry, &newfid, &newstatus, &newcb);
		ret = afs_end_vnode_operation(&fc);
		if (ret < 0)
			goto error_key;
1387 1388
	} else {
		goto error_key;
1389 1390
	}

1391 1392 1393 1394
	if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		afs_edit_dir_add(dvnode, &dentry->d_name, &newfid,
				 afs_edit_dir_for_create);

1395 1396 1397 1398
	key_put(key);
	_leave(" = 0");
	return 0;

1399
error_key:
1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412
	key_put(key);
error:
	d_drop(dentry);
	_leave(" = %d", ret);
	return ret;
}

/*
 * create a hard link between files in an AFS filesystem
 */
static int afs_link(struct dentry *from, struct inode *dir,
		    struct dentry *dentry)
{
1413
	struct afs_fs_cursor fc;
1414 1415
	struct afs_vnode *dvnode, *vnode;
	struct key *key;
1416
	u64 data_version;
1417 1418
	int ret;

1419
	vnode = AFS_FS_I(d_inode(from));
1420
	dvnode = AFS_FS_I(dir);
1421
	data_version = dvnode->status.data_version;
1422

1423
	_enter("{%llx:%llu},{%llx:%llu},{%pd}",
1424 1425
	       vnode->fid.vid, vnode->fid.vnode,
	       dvnode->fid.vid, dvnode->fid.vnode,
A
Al Viro 已提交
1426
	       dentry);
1427

1428 1429 1430 1431
	ret = -ENAMETOOLONG;
	if (dentry->d_name.len >= AFSNAMEMAX)
		goto error;

1432 1433 1434 1435 1436 1437
	key = afs_request_key(dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1438 1439 1440 1441
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		if (mutex_lock_interruptible_nested(&vnode->io_lock, 1) < 0) {
			afs_end_vnode_operation(&fc);
1442
			goto error_key;
1443 1444 1445
		}

		while (afs_select_fileserver(&fc)) {
1446 1447
			fc.cb_break = afs_calc_vnode_cb_break(dvnode);
			fc.cb_break_2 = afs_calc_vnode_cb_break(vnode);
1448
			afs_fs_link(&fc, vnode, dentry->d_name.name, data_version);
1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
		}

		afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
		afs_vnode_commit_status(&fc, vnode, fc.cb_break_2);
		ihold(&vnode->vfs_inode);
		d_instantiate(dentry, &vnode->vfs_inode);

		mutex_unlock(&vnode->io_lock);
		ret = afs_end_vnode_operation(&fc);
		if (ret < 0)
			goto error_key;
1460 1461
	} else {
		goto error_key;
1462
	}
1463

1464 1465 1466 1467
	if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		afs_edit_dir_add(dvnode, &dentry->d_name, &vnode->fid,
				 afs_edit_dir_for_link);

1468 1469 1470 1471
	key_put(key);
	_leave(" = 0");
	return 0;

1472
error_key:
1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
	key_put(key);
error:
	d_drop(dentry);
	_leave(" = %d", ret);
	return ret;
}

/*
 * create a symlink in an AFS filesystem
 */
static int afs_symlink(struct inode *dir, struct dentry *dentry,
		       const char *content)
{
1486 1487 1488 1489
	struct afs_fs_cursor fc;
	struct afs_file_status newstatus;
	struct afs_vnode *dvnode = AFS_FS_I(dir);
	struct afs_fid newfid;
1490
	struct key *key;
1491
	u64 data_version = dvnode->status.data_version;
1492 1493
	int ret;

1494
	_enter("{%llx:%llu},{%pd},%s",
A
Al Viro 已提交
1495
	       dvnode->fid.vid, dvnode->fid.vnode, dentry,
1496 1497
	       content);

1498 1499 1500 1501
	ret = -ENAMETOOLONG;
	if (dentry->d_name.len >= AFSNAMEMAX)
		goto error;

1502
	ret = -EINVAL;
D
David Howells 已提交
1503
	if (strlen(content) >= AFSPATHMAX)
1504 1505 1506 1507 1508 1509 1510 1511
		goto error;

	key = afs_request_key(dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1512 1513 1514
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, dvnode, key)) {
		while (afs_select_fileserver(&fc)) {
1515
			fc.cb_break = afs_calc_vnode_cb_break(dvnode);
1516 1517
			afs_fs_symlink(&fc, dentry->d_name.name,
				       content, data_version,
1518 1519
				       &newfid, &newstatus);
		}
1520

1521 1522 1523 1524 1525 1526
		afs_check_for_remote_deletion(&fc, fc.vnode);
		afs_vnode_commit_status(&fc, dvnode, fc.cb_break);
		afs_vnode_new_inode(&fc, dentry, &newfid, &newstatus, NULL);
		ret = afs_end_vnode_operation(&fc);
		if (ret < 0)
			goto error_key;
1527 1528
	} else {
		goto error_key;
1529 1530
	}

1531 1532 1533 1534
	if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		afs_edit_dir_add(dvnode, &dentry->d_name, &newfid,
				 afs_edit_dir_for_symlink);

1535 1536 1537 1538
	key_put(key);
	_leave(" = 0");
	return 0;

1539
error_key:
1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
	key_put(key);
error:
	d_drop(dentry);
	_leave(" = %d", ret);
	return ret;
}

/*
 * rename a file in an AFS filesystem and/or move it between directories
 */
static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
1551 1552
		      struct inode *new_dir, struct dentry *new_dentry,
		      unsigned int flags)
1553
{
1554
	struct afs_fs_cursor fc;
1555 1556
	struct afs_vnode *orig_dvnode, *new_dvnode, *vnode;
	struct key *key;
1557 1558
	u64 orig_data_version, new_data_version;
	bool new_negative = d_is_negative(new_dentry);
1559 1560
	int ret;

1561 1562 1563
	if (flags)
		return -EINVAL;

1564
	vnode = AFS_FS_I(d_inode(old_dentry));
1565 1566
	orig_dvnode = AFS_FS_I(old_dir);
	new_dvnode = AFS_FS_I(new_dir);
1567 1568
	orig_data_version = orig_dvnode->status.data_version;
	new_data_version = new_dvnode->status.data_version;
1569

1570
	_enter("{%llx:%llu},{%llx:%llu},{%llx:%llu},{%pd}",
1571 1572 1573
	       orig_dvnode->fid.vid, orig_dvnode->fid.vnode,
	       vnode->fid.vid, vnode->fid.vnode,
	       new_dvnode->fid.vid, new_dvnode->fid.vnode,
A
Al Viro 已提交
1574
	       new_dentry);
1575 1576 1577 1578 1579 1580 1581

	key = afs_request_key(orig_dvnode->volume->cell);
	if (IS_ERR(key)) {
		ret = PTR_ERR(key);
		goto error;
	}

1582 1583 1584 1585 1586
	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, orig_dvnode, key)) {
		if (orig_dvnode != new_dvnode) {
			if (mutex_lock_interruptible_nested(&new_dvnode->io_lock, 1) < 0) {
				afs_end_vnode_operation(&fc);
1587
				goto error_key;
1588 1589 1590
			}
		}
		while (afs_select_fileserver(&fc)) {
1591 1592
			fc.cb_break = afs_calc_vnode_cb_break(orig_dvnode);
			fc.cb_break_2 = afs_calc_vnode_cb_break(new_dvnode);
1593
			afs_fs_rename(&fc, old_dentry->d_name.name,
1594 1595
				      new_dvnode, new_dentry->d_name.name,
				      orig_data_version, new_data_version);
1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606
		}

		afs_vnode_commit_status(&fc, orig_dvnode, fc.cb_break);
		afs_vnode_commit_status(&fc, new_dvnode, fc.cb_break_2);
		if (orig_dvnode != new_dvnode)
			mutex_unlock(&new_dvnode->io_lock);
		ret = afs_end_vnode_operation(&fc);
		if (ret < 0)
			goto error_key;
	}

1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
	if (ret == 0) {
		if (test_bit(AFS_VNODE_DIR_VALID, &orig_dvnode->flags))
		    afs_edit_dir_remove(orig_dvnode, &old_dentry->d_name,
					afs_edit_dir_for_rename);

		if (!new_negative &&
		    test_bit(AFS_VNODE_DIR_VALID, &new_dvnode->flags))
			afs_edit_dir_remove(new_dvnode, &new_dentry->d_name,
					    afs_edit_dir_for_rename);

		if (test_bit(AFS_VNODE_DIR_VALID, &new_dvnode->flags))
			afs_edit_dir_add(new_dvnode, &new_dentry->d_name,
					 &vnode->fid,  afs_edit_dir_for_rename);
	}

1622
error_key:
1623 1624 1625 1626 1627
	key_put(key);
error:
	_leave(" = %d", ret);
	return ret;
}
D
David Howells 已提交
1628 1629 1630 1631 1632 1633 1634 1635 1636

/*
 * Release a directory page and clean up its private state if it's not busy
 * - return true if the page can now be released, false if not
 */
static int afs_dir_releasepage(struct page *page, gfp_t gfp_flags)
{
	struct afs_vnode *dvnode = AFS_FS_I(page->mapping->host);

1637
	_enter("{{%llx:%llu}[%lu]}", dvnode->fid.vid, dvnode->fid.vnode, page->index);
D
David Howells 已提交
1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671

	set_page_private(page, 0);
	ClearPagePrivate(page);

	/* The directory will need reloading. */
	if (test_and_clear_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		afs_stat_v(dvnode, n_relpg);
	return 1;
}

/*
 * invalidate part or all of a page
 * - release a page and clean up its private data if offset is 0 (indicating
 *   the entire page)
 */
static void afs_dir_invalidatepage(struct page *page, unsigned int offset,
				   unsigned int length)
{
	struct afs_vnode *dvnode = AFS_FS_I(page->mapping->host);

	_enter("{%lu},%u,%u", page->index, offset, length);

	BUG_ON(!PageLocked(page));

	/* The directory will need reloading. */
	if (test_and_clear_bit(AFS_VNODE_DIR_VALID, &dvnode->flags))
		afs_stat_v(dvnode, n_inval);

	/* we clean up only if the entire page is being invalidated */
	if (offset == 0 && length == PAGE_SIZE) {
		set_page_private(page, 0);
		ClearPagePrivate(page);
	}
}