xref: /openbmc/linux/fs/jffs2/build.c (revision 27c72b04)
1 /*
2  * JFFS2 -- Journalling Flash File System, Version 2.
3  *
4  * Copyright © 2001-2007 Red Hat, Inc.
5  *
6  * Created by David Woodhouse <dwmw2@infradead.org>
7  *
8  * For licensing information, see the file 'LICENCE' in this directory.
9  *
10  */
11 
12 #include <linux/kernel.h>
13 #include <linux/sched.h>
14 #include <linux/slab.h>
15 #include <linux/vmalloc.h>
16 #include <linux/mtd/mtd.h>
17 #include "nodelist.h"
18 
19 static void jffs2_build_remove_unlinked_inode(struct jffs2_sb_info *,
20 		struct jffs2_inode_cache *, struct jffs2_full_dirent **);
21 
22 static inline struct jffs2_inode_cache *
23 first_inode_chain(int *i, struct jffs2_sb_info *c)
24 {
25 	for (; *i < INOCACHE_HASHSIZE; (*i)++) {
26 		if (c->inocache_list[*i])
27 			return c->inocache_list[*i];
28 	}
29 	return NULL;
30 }
31 
32 static inline struct jffs2_inode_cache *
33 next_inode(int *i, struct jffs2_inode_cache *ic, struct jffs2_sb_info *c)
34 {
35 	/* More in this chain? */
36 	if (ic->next)
37 		return ic->next;
38 	(*i)++;
39 	return first_inode_chain(i, c);
40 }
41 
42 #define for_each_inode(i, c, ic)			\
43 	for (i = 0, ic = first_inode_chain(&i, (c));	\
44 	     ic;					\
45 	     ic = next_inode(&i, ic, (c)))
46 
47 
48 static void jffs2_build_inode_pass1(struct jffs2_sb_info *c,
49 				    struct jffs2_inode_cache *ic)
50 {
51 	struct jffs2_full_dirent *fd;
52 
53 	dbg_fsbuild("building directory inode #%u\n", ic->ino);
54 
55 	/* For each child, increase nlink */
56 	for(fd = ic->scan_dents; fd; fd = fd->next) {
57 		struct jffs2_inode_cache *child_ic;
58 		if (!fd->ino)
59 			continue;
60 
61 		/* we can get high latency here with huge directories */
62 
63 		child_ic = jffs2_get_ino_cache(c, fd->ino);
64 		if (!child_ic) {
65 			dbg_fsbuild("child \"%s\" (ino #%u) of dir ino #%u doesn't exist!\n",
66 				  fd->name, fd->ino, ic->ino);
67 			jffs2_mark_node_obsolete(c, fd->raw);
68 			continue;
69 		}
70 
71 		if (fd->type == DT_DIR) {
72 			if (child_ic->pino_nlink) {
73 				JFFS2_ERROR("child dir \"%s\" (ino #%u) of dir ino #%u appears to be a hard link\n",
74 					    fd->name, fd->ino, ic->ino);
75 				/* TODO: What do we do about it? */
76 			} else {
77 				child_ic->pino_nlink = ic->ino;
78 			}
79 		} else
80 			child_ic->pino_nlink++;
81 
82 		dbg_fsbuild("increased nlink for child \"%s\" (ino #%u)\n", fd->name, fd->ino);
83 		/* Can't free scan_dents so far. We might need them in pass 2 */
84 	}
85 }
86 
87 /* Scan plan:
88  - Scan physical nodes. Build map of inodes/dirents. Allocate inocaches as we go
89  - Scan directory tree from top down, setting nlink in inocaches
90  - Scan inocaches for inodes with nlink==0
91 */
92 static int jffs2_build_filesystem(struct jffs2_sb_info *c)
93 {
94 	int ret;
95 	int i;
96 	struct jffs2_inode_cache *ic;
97 	struct jffs2_full_dirent *fd;
98 	struct jffs2_full_dirent *dead_fds = NULL;
99 
100 	dbg_fsbuild("build FS data structures\n");
101 
102 	/* First, scan the medium and build all the inode caches with
103 	   lists of physical nodes */
104 
105 	c->flags |= JFFS2_SB_FLAG_SCANNING;
106 	ret = jffs2_scan_medium(c);
107 	c->flags &= ~JFFS2_SB_FLAG_SCANNING;
108 	if (ret)
109 		goto exit;
110 
111 	dbg_fsbuild("scanned flash completely\n");
112 	jffs2_dbg_dump_block_lists_nolock(c);
113 
114 	dbg_fsbuild("pass 1 starting\n");
115 	c->flags |= JFFS2_SB_FLAG_BUILDING;
116 	/* Now scan the directory tree, increasing nlink according to every dirent found. */
117 	for_each_inode(i, c, ic) {
118 		if (ic->scan_dents) {
119 			jffs2_build_inode_pass1(c, ic);
120 			cond_resched();
121 		}
122 	}
123 
124 	dbg_fsbuild("pass 1 complete\n");
125 
126 	/* Next, scan for inodes with nlink == 0 and remove them. If
127 	   they were directories, then decrement the nlink of their
128 	   children too, and repeat the scan. As that's going to be
129 	   a fairly uncommon occurrence, it's not so evil to do it this
130 	   way. Recursion bad. */
131 	dbg_fsbuild("pass 2 starting\n");
132 
133 	for_each_inode(i, c, ic) {
134 		if (ic->pino_nlink)
135 			continue;
136 
137 		jffs2_build_remove_unlinked_inode(c, ic, &dead_fds);
138 		cond_resched();
139 	}
140 
141 	dbg_fsbuild("pass 2a starting\n");
142 
143 	while (dead_fds) {
144 		fd = dead_fds;
145 		dead_fds = fd->next;
146 
147 		ic = jffs2_get_ino_cache(c, fd->ino);
148 
149 		if (ic)
150 			jffs2_build_remove_unlinked_inode(c, ic, &dead_fds);
151 		jffs2_free_full_dirent(fd);
152 	}
153 
154 	dbg_fsbuild("pass 2a complete\n");
155 	dbg_fsbuild("freeing temporary data structures\n");
156 
157 	/* Finally, we can scan again and free the dirent structs */
158 	for_each_inode(i, c, ic) {
159 		while(ic->scan_dents) {
160 			fd = ic->scan_dents;
161 			ic->scan_dents = fd->next;
162 			jffs2_free_full_dirent(fd);
163 		}
164 		ic->scan_dents = NULL;
165 		cond_resched();
166 	}
167 	jffs2_build_xattr_subsystem(c);
168 	c->flags &= ~JFFS2_SB_FLAG_BUILDING;
169 
170 	dbg_fsbuild("FS build complete\n");
171 
172 	/* Rotate the lists by some number to ensure wear levelling */
173 	jffs2_rotate_lists(c);
174 
175 	ret = 0;
176 
177 exit:
178 	if (ret) {
179 		for_each_inode(i, c, ic) {
180 			while(ic->scan_dents) {
181 				fd = ic->scan_dents;
182 				ic->scan_dents = fd->next;
183 				jffs2_free_full_dirent(fd);
184 			}
185 		}
186 		jffs2_clear_xattr_subsystem(c);
187 	}
188 
189 	return ret;
190 }
191 
192 static void jffs2_build_remove_unlinked_inode(struct jffs2_sb_info *c,
193 					struct jffs2_inode_cache *ic,
194 					struct jffs2_full_dirent **dead_fds)
195 {
196 	struct jffs2_raw_node_ref *raw;
197 	struct jffs2_full_dirent *fd;
198 
199 	dbg_fsbuild("removing ino #%u with nlink == zero.\n", ic->ino);
200 
201 	raw = ic->nodes;
202 	while (raw != (void *)ic) {
203 		struct jffs2_raw_node_ref *next = raw->next_in_ino;
204 		dbg_fsbuild("obsoleting node at 0x%08x\n", ref_offset(raw));
205 		jffs2_mark_node_obsolete(c, raw);
206 		raw = next;
207 	}
208 
209 	if (ic->scan_dents) {
210 		int whinged = 0;
211 		dbg_fsbuild("inode #%u was a directory which may have children...\n", ic->ino);
212 
213 		while(ic->scan_dents) {
214 			struct jffs2_inode_cache *child_ic;
215 
216 			fd = ic->scan_dents;
217 			ic->scan_dents = fd->next;
218 
219 			if (!fd->ino) {
220 				/* It's a deletion dirent. Ignore it */
221 				dbg_fsbuild("child \"%s\" is a deletion dirent, skipping...\n", fd->name);
222 				jffs2_free_full_dirent(fd);
223 				continue;
224 			}
225 			if (!whinged)
226 				whinged = 1;
227 
228 			dbg_fsbuild("removing child \"%s\", ino #%u\n", fd->name, fd->ino);
229 
230 			child_ic = jffs2_get_ino_cache(c, fd->ino);
231 			if (!child_ic) {
232 				dbg_fsbuild("cannot remove child \"%s\", ino #%u, because it doesn't exist\n",
233 						fd->name, fd->ino);
234 				jffs2_free_full_dirent(fd);
235 				continue;
236 			}
237 
238 			/* Reduce nlink of the child. If it's now zero, stick it on the
239 			   dead_fds list to be cleaned up later. Else just free the fd */
240 
241 			if (fd->type == DT_DIR)
242 				child_ic->pino_nlink = 0;
243 			else
244 				child_ic->pino_nlink--;
245 
246 			if (!child_ic->pino_nlink) {
247 				dbg_fsbuild("inode #%u (\"%s\") now has no links; adding to dead_fds list.\n",
248 					  fd->ino, fd->name);
249 				fd->next = *dead_fds;
250 				*dead_fds = fd;
251 			} else {
252 				dbg_fsbuild("inode #%u (\"%s\") has now got nlink %d. Ignoring.\n",
253 					  fd->ino, fd->name, child_ic->pino_nlink);
254 				jffs2_free_full_dirent(fd);
255 			}
256 		}
257 	}
258 
259 	/*
260 	   We don't delete the inocache from the hash list and free it yet.
261 	   The erase code will do that, when all the nodes are completely gone.
262 	*/
263 }
264 
265 static void jffs2_calc_trigger_levels(struct jffs2_sb_info *c)
266 {
267 	uint32_t size;
268 
269 	/* Deletion should almost _always_ be allowed. We're fairly
270 	   buggered once we stop allowing people to delete stuff
271 	   because there's not enough free space... */
272 	c->resv_blocks_deletion = 2;
273 
274 	/* Be conservative about how much space we need before we allow writes.
275 	   On top of that which is required for deletia, require an extra 2%
276 	   of the medium to be available, for overhead caused by nodes being
277 	   split across blocks, etc. */
278 
279 	size = c->flash_size / 50; /* 2% of flash size */
280 	size += c->nr_blocks * 100; /* And 100 bytes per eraseblock */
281 	size += c->sector_size - 1; /* ... and round up */
282 
283 	c->resv_blocks_write = c->resv_blocks_deletion + (size / c->sector_size);
284 
285 	/* When do we let the GC thread run in the background */
286 
287 	c->resv_blocks_gctrigger = c->resv_blocks_write + 1;
288 
289 	/* When do we allow garbage collection to merge nodes to make
290 	   long-term progress at the expense of short-term space exhaustion? */
291 	c->resv_blocks_gcmerge = c->resv_blocks_deletion + 1;
292 
293 	/* When do we allow garbage collection to eat from bad blocks rather
294 	   than actually making progress? */
295 	c->resv_blocks_gcbad = 0;//c->resv_blocks_deletion + 2;
296 
297 	/* What number of 'very dirty' eraseblocks do we allow before we
298 	   trigger the GC thread even if we don't _need_ the space. When we
299 	   can't mark nodes obsolete on the medium, the old dirty nodes cause
300 	   performance problems because we have to inspect and discard them. */
301 	c->vdirty_blocks_gctrigger = c->resv_blocks_gctrigger;
302 	if (jffs2_can_mark_obsolete(c))
303 		c->vdirty_blocks_gctrigger *= 10;
304 
305 	/* If there's less than this amount of dirty space, don't bother
306 	   trying to GC to make more space. It'll be a fruitless task */
307 	c->nospc_dirty_size = c->sector_size + (c->flash_size / 100);
308 
309 	dbg_fsbuild("JFFS2 trigger levels (size %d KiB, block size %d KiB, %d blocks)\n",
310 		  c->flash_size / 1024, c->sector_size / 1024, c->nr_blocks);
311 	dbg_fsbuild("Blocks required to allow deletion:    %d (%d KiB)\n",
312 		  c->resv_blocks_deletion, c->resv_blocks_deletion*c->sector_size/1024);
313 	dbg_fsbuild("Blocks required to allow writes:      %d (%d KiB)\n",
314 		  c->resv_blocks_write, c->resv_blocks_write*c->sector_size/1024);
315 	dbg_fsbuild("Blocks required to quiesce GC thread: %d (%d KiB)\n",
316 		  c->resv_blocks_gctrigger, c->resv_blocks_gctrigger*c->sector_size/1024);
317 	dbg_fsbuild("Blocks required to allow GC merges:   %d (%d KiB)\n",
318 		  c->resv_blocks_gcmerge, c->resv_blocks_gcmerge*c->sector_size/1024);
319 	dbg_fsbuild("Blocks required to GC bad blocks:     %d (%d KiB)\n",
320 		  c->resv_blocks_gcbad, c->resv_blocks_gcbad*c->sector_size/1024);
321 	dbg_fsbuild("Amount of dirty space required to GC: %d bytes\n",
322 		  c->nospc_dirty_size);
323 	dbg_fsbuild("Very dirty blocks before GC triggered: %d\n",
324 		  c->vdirty_blocks_gctrigger);
325 }
326 
327 int jffs2_do_mount_fs(struct jffs2_sb_info *c)
328 {
329 	int ret;
330 	int i;
331 	int size;
332 
333 	c->free_size = c->flash_size;
334 	c->nr_blocks = c->flash_size / c->sector_size;
335 	size = sizeof(struct jffs2_eraseblock) * c->nr_blocks;
336 #ifndef __ECOS
337 	if (jffs2_blocks_use_vmalloc(c))
338 		c->blocks = vmalloc(size);
339 	else
340 #endif
341 		c->blocks = kmalloc(size, GFP_KERNEL);
342 	if (!c->blocks)
343 		return -ENOMEM;
344 
345 	memset(c->blocks, 0, size);
346 	for (i=0; i<c->nr_blocks; i++) {
347 		INIT_LIST_HEAD(&c->blocks[i].list);
348 		c->blocks[i].offset = i * c->sector_size;
349 		c->blocks[i].free_size = c->sector_size;
350 	}
351 
352 	INIT_LIST_HEAD(&c->clean_list);
353 	INIT_LIST_HEAD(&c->very_dirty_list);
354 	INIT_LIST_HEAD(&c->dirty_list);
355 	INIT_LIST_HEAD(&c->erasable_list);
356 	INIT_LIST_HEAD(&c->erasing_list);
357 	INIT_LIST_HEAD(&c->erase_checking_list);
358 	INIT_LIST_HEAD(&c->erase_pending_list);
359 	INIT_LIST_HEAD(&c->erasable_pending_wbuf_list);
360 	INIT_LIST_HEAD(&c->erase_complete_list);
361 	INIT_LIST_HEAD(&c->free_list);
362 	INIT_LIST_HEAD(&c->bad_list);
363 	INIT_LIST_HEAD(&c->bad_used_list);
364 	c->highest_ino = 1;
365 	c->summary = NULL;
366 
367 	ret = jffs2_sum_init(c);
368 	if (ret)
369 		goto out_free;
370 
371 	if (jffs2_build_filesystem(c)) {
372 		dbg_fsbuild("build_fs failed\n");
373 		jffs2_free_ino_caches(c);
374 		jffs2_free_raw_node_refs(c);
375 		ret = -EIO;
376 		goto out_free;
377 	}
378 
379 	jffs2_calc_trigger_levels(c);
380 
381 	return 0;
382 
383  out_free:
384 #ifndef __ECOS
385 	if (jffs2_blocks_use_vmalloc(c))
386 		vfree(c->blocks);
387 	else
388 #endif
389 		kfree(c->blocks);
390 
391 	return ret;
392 }
393