xref: /linux/fs/squashfs/file.c (revision bcefe12eff5dca6fdfa94ed85e5bee66380d5cd9)
1 /*
2  * Squashfs - a compressed read only filesystem for Linux
3  *
4  * Copyright (c) 2002, 2003, 2004, 2005, 2006, 2007, 2008
5  * Phillip Lougher <phillip@lougher.demon.co.uk>
6  *
7  * This program is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU General Public License
9  * as published by the Free Software Foundation; either version 2,
10  * or (at your option) any later version.
11  *
12  * This program is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15  * GNU General Public License for more details.
16  *
17  * You should have received a copy of the GNU General Public License
18  * along with this program; if not, write to the Free Software
19  * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
20  *
21  * file.c
22  */
23 
24 /*
25  * This file contains code for handling regular files.  A regular file
26  * consists of a sequence of contiguous compressed blocks, and/or a
27  * compressed fragment block (tail-end packed block).   The compressed size
28  * of each datablock is stored in a block list contained within the
29  * file inode (itself stored in one or more compressed metadata blocks).
30  *
31  * To speed up access to datablocks when reading 'large' files (256 Mbytes or
32  * larger), the code implements an index cache that caches the mapping from
33  * block index to datablock location on disk.
34  *
35  * The index cache allows Squashfs to handle large files (up to 1.75 TiB) while
36  * retaining a simple and space-efficient block list on disk.  The cache
37  * is split into slots, caching up to eight 224 GiB files (128 KiB blocks).
38  * Larger files use multiple slots, with 1.75 TiB files using all 8 slots.
39  * The index cache is designed to be memory efficient, and by default uses
40  * 16 KiB.
41  */
42 
43 #include <linux/fs.h>
44 #include <linux/vfs.h>
45 #include <linux/kernel.h>
46 #include <linux/slab.h>
47 #include <linux/string.h>
48 #include <linux/pagemap.h>
49 #include <linux/mutex.h>
50 #include <linux/zlib.h>
51 
52 #include "squashfs_fs.h"
53 #include "squashfs_fs_sb.h"
54 #include "squashfs_fs_i.h"
55 #include "squashfs.h"
56 
57 /*
58  * Locate cache slot in range [offset, index] for specified inode.  If
59  * there's more than one return the slot closest to index.
60  */
61 static struct meta_index *locate_meta_index(struct inode *inode, int offset,
62 				int index)
63 {
64 	struct meta_index *meta = NULL;
65 	struct squashfs_sb_info *msblk = inode->i_sb->s_fs_info;
66 	int i;
67 
68 	mutex_lock(&msblk->meta_index_mutex);
69 
70 	TRACE("locate_meta_index: index %d, offset %d\n", index, offset);
71 
72 	if (msblk->meta_index == NULL)
73 		goto not_allocated;
74 
75 	for (i = 0; i < SQUASHFS_META_SLOTS; i++) {
76 		if (msblk->meta_index[i].inode_number == inode->i_ino &&
77 				msblk->meta_index[i].offset >= offset &&
78 				msblk->meta_index[i].offset <= index &&
79 				msblk->meta_index[i].locked == 0) {
80 			TRACE("locate_meta_index: entry %d, offset %d\n", i,
81 					msblk->meta_index[i].offset);
82 			meta = &msblk->meta_index[i];
83 			offset = meta->offset;
84 		}
85 	}
86 
87 	if (meta)
88 		meta->locked = 1;
89 
90 not_allocated:
91 	mutex_unlock(&msblk->meta_index_mutex);
92 
93 	return meta;
94 }
95 
96 
97 /*
98  * Find and initialise an empty cache slot for index offset.
99  */
100 static struct meta_index *empty_meta_index(struct inode *inode, int offset,
101 				int skip)
102 {
103 	struct squashfs_sb_info *msblk = inode->i_sb->s_fs_info;
104 	struct meta_index *meta = NULL;
105 	int i;
106 
107 	mutex_lock(&msblk->meta_index_mutex);
108 
109 	TRACE("empty_meta_index: offset %d, skip %d\n", offset, skip);
110 
111 	if (msblk->meta_index == NULL) {
112 		/*
113 		 * First time cache index has been used, allocate and
114 		 * initialise.  The cache index could be allocated at
115 		 * mount time but doing it here means it is allocated only
116 		 * if a 'large' file is read.
117 		 */
118 		msblk->meta_index = kcalloc(SQUASHFS_META_SLOTS,
119 			sizeof(*(msblk->meta_index)), GFP_KERNEL);
120 		if (msblk->meta_index == NULL) {
121 			ERROR("Failed to allocate meta_index\n");
122 			goto failed;
123 		}
124 		for (i = 0; i < SQUASHFS_META_SLOTS; i++) {
125 			msblk->meta_index[i].inode_number = 0;
126 			msblk->meta_index[i].locked = 0;
127 		}
128 		msblk->next_meta_index = 0;
129 	}
130 
131 	for (i = SQUASHFS_META_SLOTS; i &&
132 			msblk->meta_index[msblk->next_meta_index].locked; i--)
133 		msblk->next_meta_index = (msblk->next_meta_index + 1) %
134 			SQUASHFS_META_SLOTS;
135 
136 	if (i == 0) {
137 		TRACE("empty_meta_index: failed!\n");
138 		goto failed;
139 	}
140 
141 	TRACE("empty_meta_index: returned meta entry %d, %p\n",
142 			msblk->next_meta_index,
143 			&msblk->meta_index[msblk->next_meta_index]);
144 
145 	meta = &msblk->meta_index[msblk->next_meta_index];
146 	msblk->next_meta_index = (msblk->next_meta_index + 1) %
147 			SQUASHFS_META_SLOTS;
148 
149 	meta->inode_number = inode->i_ino;
150 	meta->offset = offset;
151 	meta->skip = skip;
152 	meta->entries = 0;
153 	meta->locked = 1;
154 
155 failed:
156 	mutex_unlock(&msblk->meta_index_mutex);
157 	return meta;
158 }
159 
160 
161 static void release_meta_index(struct inode *inode, struct meta_index *meta)
162 {
163 	struct squashfs_sb_info *msblk = inode->i_sb->s_fs_info;
164 	mutex_lock(&msblk->meta_index_mutex);
165 	meta->locked = 0;
166 	mutex_unlock(&msblk->meta_index_mutex);
167 }
168 
169 
170 /*
171  * Read the next n blocks from the block list, starting from
172  * metadata block <start_block, offset>.
173  */
174 static long long read_indexes(struct super_block *sb, int n,
175 				u64 *start_block, int *offset)
176 {
177 	int err, i;
178 	long long block = 0;
179 	__le32 *blist = kmalloc(PAGE_CACHE_SIZE, GFP_KERNEL);
180 
181 	if (blist == NULL) {
182 		ERROR("read_indexes: Failed to allocate block_list\n");
183 		return -ENOMEM;
184 	}
185 
186 	while (n) {
187 		int blocks = min_t(int, n, PAGE_CACHE_SIZE >> 2);
188 
189 		err = squashfs_read_metadata(sb, blist, start_block,
190 				offset, blocks << 2);
191 		if (err < 0) {
192 			ERROR("read_indexes: reading block [%llx:%x]\n",
193 				*start_block, *offset);
194 			goto failure;
195 		}
196 
197 		for (i = 0; i < blocks; i++) {
198 			int size = le32_to_cpu(blist[i]);
199 			block += SQUASHFS_COMPRESSED_SIZE_BLOCK(size);
200 		}
201 		n -= blocks;
202 	}
203 
204 	kfree(blist);
205 	return block;
206 
207 failure:
208 	kfree(blist);
209 	return err;
210 }
211 
212 
213 /*
214  * Each cache index slot has SQUASHFS_META_ENTRIES, each of which
215  * can cache one index -> datablock/blocklist-block mapping.  We wish
216  * to distribute these over the length of the file, entry[0] maps index x,
217  * entry[1] maps index x + skip, entry[2] maps index x + 2 * skip, and so on.
218  * The larger the file, the greater the skip factor.  The skip factor is
219  * limited to the size of the metadata cache (SQUASHFS_CACHED_BLKS) to ensure
220  * the number of metadata blocks that need to be read fits into the cache.
221  * If the skip factor is limited in this way then the file will use multiple
222  * slots.
223  */
224 static inline int calculate_skip(int blocks)
225 {
226 	int skip = blocks / ((SQUASHFS_META_ENTRIES + 1)
227 		 * SQUASHFS_META_INDEXES);
228 	return min(SQUASHFS_CACHED_BLKS - 1, skip + 1);
229 }
230 
231 
232 /*
233  * Search and grow the index cache for the specified inode, returning the
234  * on-disk locations of the datablock and block list metadata block
235  * <index_block, index_offset> for index (scaled to nearest cache index).
236  */
237 static int fill_meta_index(struct inode *inode, int index,
238 		u64 *index_block, int *index_offset, u64 *data_block)
239 {
240 	struct squashfs_sb_info *msblk = inode->i_sb->s_fs_info;
241 	int skip = calculate_skip(i_size_read(inode) >> msblk->block_log);
242 	int offset = 0;
243 	struct meta_index *meta;
244 	struct meta_entry *meta_entry;
245 	u64 cur_index_block = squashfs_i(inode)->block_list_start;
246 	int cur_offset = squashfs_i(inode)->offset;
247 	u64 cur_data_block = squashfs_i(inode)->start;
248 	int err, i;
249 
250 	/*
251 	 * Scale index to cache index (cache slot entry)
252 	 */
253 	index /= SQUASHFS_META_INDEXES * skip;
254 
255 	while (offset < index) {
256 		meta = locate_meta_index(inode, offset + 1, index);
257 
258 		if (meta == NULL) {
259 			meta = empty_meta_index(inode, offset + 1, skip);
260 			if (meta == NULL)
261 				goto all_done;
262 		} else {
263 			offset = index < meta->offset + meta->entries ? index :
264 				meta->offset + meta->entries - 1;
265 			meta_entry = &meta->meta_entry[offset - meta->offset];
266 			cur_index_block = meta_entry->index_block +
267 				msblk->inode_table;
268 			cur_offset = meta_entry->offset;
269 			cur_data_block = meta_entry->data_block;
270 			TRACE("get_meta_index: offset %d, meta->offset %d, "
271 				"meta->entries %d\n", offset, meta->offset,
272 				meta->entries);
273 			TRACE("get_meta_index: index_block 0x%llx, offset 0x%x"
274 				" data_block 0x%llx\n", cur_index_block,
275 				cur_offset, cur_data_block);
276 		}
277 
278 		/*
279 		 * If necessary grow cache slot by reading block list.  Cache
280 		 * slot is extended up to index or to the end of the slot, in
281 		 * which case further slots will be used.
282 		 */
283 		for (i = meta->offset + meta->entries; i <= index &&
284 				i < meta->offset + SQUASHFS_META_ENTRIES; i++) {
285 			int blocks = skip * SQUASHFS_META_INDEXES;
286 			long long res = read_indexes(inode->i_sb, blocks,
287 					&cur_index_block, &cur_offset);
288 
289 			if (res < 0) {
290 				if (meta->entries == 0)
291 					/*
292 					 * Don't leave an empty slot on read
293 					 * error allocated to this inode...
294 					 */
295 					meta->inode_number = 0;
296 				err = res;
297 				goto failed;
298 			}
299 
300 			cur_data_block += res;
301 			meta_entry = &meta->meta_entry[i - meta->offset];
302 			meta_entry->index_block = cur_index_block -
303 				msblk->inode_table;
304 			meta_entry->offset = cur_offset;
305 			meta_entry->data_block = cur_data_block;
306 			meta->entries++;
307 			offset++;
308 		}
309 
310 		TRACE("get_meta_index: meta->offset %d, meta->entries %d\n",
311 				meta->offset, meta->entries);
312 
313 		release_meta_index(inode, meta);
314 	}
315 
316 all_done:
317 	*index_block = cur_index_block;
318 	*index_offset = cur_offset;
319 	*data_block = cur_data_block;
320 
321 	/*
322 	 * Scale cache index (cache slot entry) to index
323 	 */
324 	return offset * SQUASHFS_META_INDEXES * skip;
325 
326 failed:
327 	release_meta_index(inode, meta);
328 	return err;
329 }
330 
331 
332 /*
333  * Get the on-disk location and compressed size of the datablock
334  * specified by index.  Fill_meta_index() does most of the work.
335  */
336 static int read_blocklist(struct inode *inode, int index, u64 *block)
337 {
338 	u64 start;
339 	long long blks;
340 	int offset;
341 	__le32 size;
342 	int res = fill_meta_index(inode, index, &start, &offset, block);
343 
344 	TRACE("read_blocklist: res %d, index %d, start 0x%llx, offset"
345 		       " 0x%x, block 0x%llx\n", res, index, start, offset,
346 			*block);
347 
348 	if (res < 0)
349 		return res;
350 
351 	/*
352 	 * res contains the index of the mapping returned by fill_meta_index(),
353 	 * this will likely be less than the desired index (because the
354 	 * meta_index cache works at a higher granularity).  Read any
355 	 * extra block indexes needed.
356 	 */
357 	if (res < index) {
358 		blks = read_indexes(inode->i_sb, index - res, &start, &offset);
359 		if (blks < 0)
360 			return (int) blks;
361 		*block += blks;
362 	}
363 
364 	/*
365 	 * Read length of block specified by index.
366 	 */
367 	res = squashfs_read_metadata(inode->i_sb, &size, &start, &offset,
368 			sizeof(size));
369 	if (res < 0)
370 		return res;
371 	return le32_to_cpu(size);
372 }
373 
374 
375 static int squashfs_readpage(struct file *file, struct page *page)
376 {
377 	struct inode *inode = page->mapping->host;
378 	struct squashfs_sb_info *msblk = inode->i_sb->s_fs_info;
379 	int bytes, i, offset = 0, sparse = 0;
380 	struct squashfs_cache_entry *buffer = NULL;
381 	void *pageaddr;
382 
383 	int mask = (1 << (msblk->block_log - PAGE_CACHE_SHIFT)) - 1;
384 	int index = page->index >> (msblk->block_log - PAGE_CACHE_SHIFT);
385 	int start_index = page->index & ~mask;
386 	int end_index = start_index | mask;
387 	int file_end = i_size_read(inode) >> msblk->block_log;
388 
389 	TRACE("Entered squashfs_readpage, page index %lx, start block %llx\n",
390 				page->index, squashfs_i(inode)->start);
391 
392 	if (page->index >= ((i_size_read(inode) + PAGE_CACHE_SIZE - 1) >>
393 					PAGE_CACHE_SHIFT))
394 		goto out;
395 
396 	if (index < file_end || squashfs_i(inode)->fragment_block ==
397 					SQUASHFS_INVALID_BLK) {
398 		/*
399 		 * Reading a datablock from disk.  Need to read block list
400 		 * to get location and block size.
401 		 */
402 		u64 block = 0;
403 		int bsize = read_blocklist(inode, index, &block);
404 		if (bsize < 0)
405 			goto error_out;
406 
407 		if (bsize == 0) { /* hole */
408 			bytes = index == file_end ?
409 				(i_size_read(inode) & (msblk->block_size - 1)) :
410 				 msblk->block_size;
411 			sparse = 1;
412 		} else {
413 			/*
414 			 * Read and decompress datablock.
415 			 */
416 			buffer = squashfs_get_datablock(inode->i_sb,
417 								block, bsize);
418 			if (buffer->error) {
419 				ERROR("Unable to read page, block %llx, size %x"
420 					"\n", block, bsize);
421 				squashfs_cache_put(buffer);
422 				goto error_out;
423 			}
424 			bytes = buffer->length;
425 		}
426 	} else {
427 		/*
428 		 * Datablock is stored inside a fragment (tail-end packed
429 		 * block).
430 		 */
431 		buffer = squashfs_get_fragment(inode->i_sb,
432 				squashfs_i(inode)->fragment_block,
433 				squashfs_i(inode)->fragment_size);
434 
435 		if (buffer->error) {
436 			ERROR("Unable to read page, block %llx, size %x\n",
437 				squashfs_i(inode)->fragment_block,
438 				squashfs_i(inode)->fragment_size);
439 			squashfs_cache_put(buffer);
440 			goto error_out;
441 		}
442 		bytes = i_size_read(inode) & (msblk->block_size - 1);
443 		offset = squashfs_i(inode)->fragment_offset;
444 	}
445 
446 	/*
447 	 * Loop copying datablock into pages.  As the datablock likely covers
448 	 * many PAGE_CACHE_SIZE pages (default block size is 128 KiB) explicitly
449 	 * grab the pages from the page cache, except for the page that we've
450 	 * been called to fill.
451 	 */
452 	for (i = start_index; i <= end_index && bytes > 0; i++,
453 			bytes -= PAGE_CACHE_SIZE, offset += PAGE_CACHE_SIZE) {
454 		struct page *push_page;
455 		int avail = sparse ? 0 : min_t(int, bytes, PAGE_CACHE_SIZE);
456 
457 		TRACE("bytes %d, i %d, available_bytes %d\n", bytes, i, avail);
458 
459 		push_page = (i == page->index) ? page :
460 			grab_cache_page_nowait(page->mapping, i);
461 
462 		if (!push_page)
463 			continue;
464 
465 		if (PageUptodate(push_page))
466 			goto skip_page;
467 
468 		pageaddr = kmap_atomic(push_page, KM_USER0);
469 		squashfs_copy_data(pageaddr, buffer, offset, avail);
470 		memset(pageaddr + avail, 0, PAGE_CACHE_SIZE - avail);
471 		kunmap_atomic(pageaddr, KM_USER0);
472 		flush_dcache_page(push_page);
473 		SetPageUptodate(push_page);
474 skip_page:
475 		unlock_page(push_page);
476 		if (i != page->index)
477 			page_cache_release(push_page);
478 	}
479 
480 	if (!sparse)
481 		squashfs_cache_put(buffer);
482 
483 	return 0;
484 
485 error_out:
486 	SetPageError(page);
487 out:
488 	pageaddr = kmap_atomic(page, KM_USER0);
489 	memset(pageaddr, 0, PAGE_CACHE_SIZE);
490 	kunmap_atomic(pageaddr, KM_USER0);
491 	flush_dcache_page(page);
492 	if (!PageError(page))
493 		SetPageUptodate(page);
494 	unlock_page(page);
495 
496 	return 0;
497 }
498 
499 
500 const struct address_space_operations squashfs_aops = {
501 	.readpage = squashfs_readpage
502 };
503