xref: /openbmc/linux/fs/jffs2/build.c (revision 65e5a0e1)
1 /*
2  * JFFS2 -- Journalling Flash File System, Version 2.
3  *
4  * Copyright © 2001-2007 Red Hat, Inc.
5  * Copyright © 2004-2010 David Woodhouse <dwmw2@infradead.org>
6  *
7  * Created by David Woodhouse <dwmw2@infradead.org>
8  *
9  * For licensing information, see the file 'LICENCE' in this directory.
10  *
11  */
12 
13 #include <linux/kernel.h>
14 #include <linux/sched.h>
15 #include <linux/slab.h>
16 #include <linux/vmalloc.h>
17 #include <linux/mtd/mtd.h>
18 #include "nodelist.h"
19 
20 static void jffs2_build_remove_unlinked_inode(struct jffs2_sb_info *,
21 		struct jffs2_inode_cache *, struct jffs2_full_dirent **);
22 
23 static inline struct jffs2_inode_cache *
24 first_inode_chain(int *i, struct jffs2_sb_info *c)
25 {
26 	for (; *i < c->inocache_hashsize; (*i)++) {
27 		if (c->inocache_list[*i])
28 			return c->inocache_list[*i];
29 	}
30 	return NULL;
31 }
32 
33 static inline struct jffs2_inode_cache *
34 next_inode(int *i, struct jffs2_inode_cache *ic, struct jffs2_sb_info *c)
35 {
36 	/* More in this chain? */
37 	if (ic->next)
38 		return ic->next;
39 	(*i)++;
40 	return first_inode_chain(i, c);
41 }
42 
43 #define for_each_inode(i, c, ic)			\
44 	for (i = 0, ic = first_inode_chain(&i, (c));	\
45 	     ic;					\
46 	     ic = next_inode(&i, ic, (c)))
47 
48 
49 static void jffs2_build_inode_pass1(struct jffs2_sb_info *c,
50 				    struct jffs2_inode_cache *ic)
51 {
52 	struct jffs2_full_dirent *fd;
53 
54 	dbg_fsbuild("building directory inode #%u\n", ic->ino);
55 
56 	/* For each child, increase nlink */
57 	for(fd = ic->scan_dents; fd; fd = fd->next) {
58 		struct jffs2_inode_cache *child_ic;
59 		if (!fd->ino)
60 			continue;
61 
62 		/* we can get high latency here with huge directories */
63 
64 		child_ic = jffs2_get_ino_cache(c, fd->ino);
65 		if (!child_ic) {
66 			dbg_fsbuild("child \"%s\" (ino #%u) of dir ino #%u doesn't exist!\n",
67 				  fd->name, fd->ino, ic->ino);
68 			jffs2_mark_node_obsolete(c, fd->raw);
69 			continue;
70 		}
71 
72 		if (fd->type == DT_DIR) {
73 			if (child_ic->pino_nlink) {
74 				JFFS2_ERROR("child dir \"%s\" (ino #%u) of dir ino #%u appears to be a hard link\n",
75 					    fd->name, fd->ino, ic->ino);
76 				/* TODO: What do we do about it? */
77 			} else {
78 				child_ic->pino_nlink = ic->ino;
79 			}
80 		} else
81 			child_ic->pino_nlink++;
82 
83 		dbg_fsbuild("increased nlink for child \"%s\" (ino #%u)\n", fd->name, fd->ino);
84 		/* Can't free scan_dents so far. We might need them in pass 2 */
85 	}
86 }
87 
88 /* Scan plan:
89  - Scan physical nodes. Build map of inodes/dirents. Allocate inocaches as we go
90  - Scan directory tree from top down, setting nlink in inocaches
91  - Scan inocaches for inodes with nlink==0
92 */
93 static int jffs2_build_filesystem(struct jffs2_sb_info *c)
94 {
95 	int ret;
96 	int i;
97 	struct jffs2_inode_cache *ic;
98 	struct jffs2_full_dirent *fd;
99 	struct jffs2_full_dirent *dead_fds = NULL;
100 
101 	dbg_fsbuild("build FS data structures\n");
102 
103 	/* First, scan the medium and build all the inode caches with
104 	   lists of physical nodes */
105 
106 	c->flags |= JFFS2_SB_FLAG_SCANNING;
107 	ret = jffs2_scan_medium(c);
108 	c->flags &= ~JFFS2_SB_FLAG_SCANNING;
109 	if (ret)
110 		goto exit;
111 
112 	dbg_fsbuild("scanned flash completely\n");
113 	jffs2_dbg_dump_block_lists_nolock(c);
114 
115 	dbg_fsbuild("pass 1 starting\n");
116 	c->flags |= JFFS2_SB_FLAG_BUILDING;
117 	/* Now scan the directory tree, increasing nlink according to every dirent found. */
118 	for_each_inode(i, c, ic) {
119 		if (ic->scan_dents) {
120 			jffs2_build_inode_pass1(c, ic);
121 			cond_resched();
122 		}
123 	}
124 
125 	dbg_fsbuild("pass 1 complete\n");
126 
127 	/* Next, scan for inodes with nlink == 0 and remove them. If
128 	   they were directories, then decrement the nlink of their
129 	   children too, and repeat the scan. As that's going to be
130 	   a fairly uncommon occurrence, it's not so evil to do it this
131 	   way. Recursion bad. */
132 	dbg_fsbuild("pass 2 starting\n");
133 
134 	for_each_inode(i, c, ic) {
135 		if (ic->pino_nlink)
136 			continue;
137 
138 		jffs2_build_remove_unlinked_inode(c, ic, &dead_fds);
139 		cond_resched();
140 	}
141 
142 	dbg_fsbuild("pass 2a starting\n");
143 
144 	while (dead_fds) {
145 		fd = dead_fds;
146 		dead_fds = fd->next;
147 
148 		ic = jffs2_get_ino_cache(c, fd->ino);
149 
150 		if (ic)
151 			jffs2_build_remove_unlinked_inode(c, ic, &dead_fds);
152 		jffs2_free_full_dirent(fd);
153 	}
154 
155 	dbg_fsbuild("pass 2a complete\n");
156 	dbg_fsbuild("freeing temporary data structures\n");
157 
158 	/* Finally, we can scan again and free the dirent structs */
159 	for_each_inode(i, c, ic) {
160 		while(ic->scan_dents) {
161 			fd = ic->scan_dents;
162 			ic->scan_dents = fd->next;
163 			jffs2_free_full_dirent(fd);
164 		}
165 		ic->scan_dents = NULL;
166 		cond_resched();
167 	}
168 	jffs2_build_xattr_subsystem(c);
169 	c->flags &= ~JFFS2_SB_FLAG_BUILDING;
170 
171 	dbg_fsbuild("FS build complete\n");
172 
173 	/* Rotate the lists by some number to ensure wear levelling */
174 	jffs2_rotate_lists(c);
175 
176 	ret = 0;
177 
178 exit:
179 	if (ret) {
180 		for_each_inode(i, c, ic) {
181 			while(ic->scan_dents) {
182 				fd = ic->scan_dents;
183 				ic->scan_dents = fd->next;
184 				jffs2_free_full_dirent(fd);
185 			}
186 		}
187 		jffs2_clear_xattr_subsystem(c);
188 	}
189 
190 	return ret;
191 }
192 
193 static void jffs2_build_remove_unlinked_inode(struct jffs2_sb_info *c,
194 					struct jffs2_inode_cache *ic,
195 					struct jffs2_full_dirent **dead_fds)
196 {
197 	struct jffs2_raw_node_ref *raw;
198 	struct jffs2_full_dirent *fd;
199 
200 	dbg_fsbuild("removing ino #%u with nlink == zero.\n", ic->ino);
201 
202 	raw = ic->nodes;
203 	while (raw != (void *)ic) {
204 		struct jffs2_raw_node_ref *next = raw->next_in_ino;
205 		dbg_fsbuild("obsoleting node at 0x%08x\n", ref_offset(raw));
206 		jffs2_mark_node_obsolete(c, raw);
207 		raw = next;
208 	}
209 
210 	if (ic->scan_dents) {
211 		int whinged = 0;
212 		dbg_fsbuild("inode #%u was a directory which may have children...\n", ic->ino);
213 
214 		while(ic->scan_dents) {
215 			struct jffs2_inode_cache *child_ic;
216 
217 			fd = ic->scan_dents;
218 			ic->scan_dents = fd->next;
219 
220 			if (!fd->ino) {
221 				/* It's a deletion dirent. Ignore it */
222 				dbg_fsbuild("child \"%s\" is a deletion dirent, skipping...\n", fd->name);
223 				jffs2_free_full_dirent(fd);
224 				continue;
225 			}
226 			if (!whinged)
227 				whinged = 1;
228 
229 			dbg_fsbuild("removing child \"%s\", ino #%u\n", fd->name, fd->ino);
230 
231 			child_ic = jffs2_get_ino_cache(c, fd->ino);
232 			if (!child_ic) {
233 				dbg_fsbuild("cannot remove child \"%s\", ino #%u, because it doesn't exist\n",
234 						fd->name, fd->ino);
235 				jffs2_free_full_dirent(fd);
236 				continue;
237 			}
238 
239 			/* Reduce nlink of the child. If it's now zero, stick it on the
240 			   dead_fds list to be cleaned up later. Else just free the fd */
241 
242 			if (fd->type == DT_DIR)
243 				child_ic->pino_nlink = 0;
244 			else
245 				child_ic->pino_nlink--;
246 
247 			if (!child_ic->pino_nlink) {
248 				dbg_fsbuild("inode #%u (\"%s\") now has no links; adding to dead_fds list.\n",
249 					  fd->ino, fd->name);
250 				fd->next = *dead_fds;
251 				*dead_fds = fd;
252 			} else {
253 				dbg_fsbuild("inode #%u (\"%s\") has now got nlink %d. Ignoring.\n",
254 					  fd->ino, fd->name, child_ic->pino_nlink);
255 				jffs2_free_full_dirent(fd);
256 			}
257 		}
258 	}
259 
260 	/*
261 	   We don't delete the inocache from the hash list and free it yet.
262 	   The erase code will do that, when all the nodes are completely gone.
263 	*/
264 }
265 
266 static void jffs2_calc_trigger_levels(struct jffs2_sb_info *c)
267 {
268 	uint32_t size;
269 
270 	/* Deletion should almost _always_ be allowed. We're fairly
271 	   buggered once we stop allowing people to delete stuff
272 	   because there's not enough free space... */
273 	c->resv_blocks_deletion = 2;
274 
275 	/* Be conservative about how much space we need before we allow writes.
276 	   On top of that which is required for deletia, require an extra 2%
277 	   of the medium to be available, for overhead caused by nodes being
278 	   split across blocks, etc. */
279 
280 	size = c->flash_size / 50; /* 2% of flash size */
281 	size += c->nr_blocks * 100; /* And 100 bytes per eraseblock */
282 	size += c->sector_size - 1; /* ... and round up */
283 
284 	c->resv_blocks_write = c->resv_blocks_deletion + (size / c->sector_size);
285 
286 	/* When do we let the GC thread run in the background */
287 
288 	c->resv_blocks_gctrigger = c->resv_blocks_write + 1;
289 
290 	/* When do we allow garbage collection to merge nodes to make
291 	   long-term progress at the expense of short-term space exhaustion? */
292 	c->resv_blocks_gcmerge = c->resv_blocks_deletion + 1;
293 
294 	/* When do we allow garbage collection to eat from bad blocks rather
295 	   than actually making progress? */
296 	c->resv_blocks_gcbad = 0;//c->resv_blocks_deletion + 2;
297 
298 	/* What number of 'very dirty' eraseblocks do we allow before we
299 	   trigger the GC thread even if we don't _need_ the space. When we
300 	   can't mark nodes obsolete on the medium, the old dirty nodes cause
301 	   performance problems because we have to inspect and discard them. */
302 	c->vdirty_blocks_gctrigger = c->resv_blocks_gctrigger;
303 	if (jffs2_can_mark_obsolete(c))
304 		c->vdirty_blocks_gctrigger *= 10;
305 
306 	/* If there's less than this amount of dirty space, don't bother
307 	   trying to GC to make more space. It'll be a fruitless task */
308 	c->nospc_dirty_size = c->sector_size + (c->flash_size / 100);
309 
310 	dbg_fsbuild("JFFS2 trigger levels (size %d KiB, block size %d KiB, %d blocks)\n",
311 		  c->flash_size / 1024, c->sector_size / 1024, c->nr_blocks);
312 	dbg_fsbuild("Blocks required to allow deletion:    %d (%d KiB)\n",
313 		  c->resv_blocks_deletion, c->resv_blocks_deletion*c->sector_size/1024);
314 	dbg_fsbuild("Blocks required to allow writes:      %d (%d KiB)\n",
315 		  c->resv_blocks_write, c->resv_blocks_write*c->sector_size/1024);
316 	dbg_fsbuild("Blocks required to quiesce GC thread: %d (%d KiB)\n",
317 		  c->resv_blocks_gctrigger, c->resv_blocks_gctrigger*c->sector_size/1024);
318 	dbg_fsbuild("Blocks required to allow GC merges:   %d (%d KiB)\n",
319 		  c->resv_blocks_gcmerge, c->resv_blocks_gcmerge*c->sector_size/1024);
320 	dbg_fsbuild("Blocks required to GC bad blocks:     %d (%d KiB)\n",
321 		  c->resv_blocks_gcbad, c->resv_blocks_gcbad*c->sector_size/1024);
322 	dbg_fsbuild("Amount of dirty space required to GC: %d bytes\n",
323 		  c->nospc_dirty_size);
324 	dbg_fsbuild("Very dirty blocks before GC triggered: %d\n",
325 		  c->vdirty_blocks_gctrigger);
326 }
327 
328 int jffs2_do_mount_fs(struct jffs2_sb_info *c)
329 {
330 	int ret;
331 	int i;
332 	int size;
333 
334 	c->free_size = c->flash_size;
335 	c->nr_blocks = c->flash_size / c->sector_size;
336 	size = sizeof(struct jffs2_eraseblock) * c->nr_blocks;
337 #ifndef __ECOS
338 	if (jffs2_blocks_use_vmalloc(c))
339 		c->blocks = vmalloc(size);
340 	else
341 #endif
342 		c->blocks = kmalloc(size, GFP_KERNEL);
343 	if (!c->blocks)
344 		return -ENOMEM;
345 
346 	memset(c->blocks, 0, size);
347 	for (i=0; i<c->nr_blocks; i++) {
348 		INIT_LIST_HEAD(&c->blocks[i].list);
349 		c->blocks[i].offset = i * c->sector_size;
350 		c->blocks[i].free_size = c->sector_size;
351 	}
352 
353 	INIT_LIST_HEAD(&c->clean_list);
354 	INIT_LIST_HEAD(&c->very_dirty_list);
355 	INIT_LIST_HEAD(&c->dirty_list);
356 	INIT_LIST_HEAD(&c->erasable_list);
357 	INIT_LIST_HEAD(&c->erasing_list);
358 	INIT_LIST_HEAD(&c->erase_checking_list);
359 	INIT_LIST_HEAD(&c->erase_pending_list);
360 	INIT_LIST_HEAD(&c->erasable_pending_wbuf_list);
361 	INIT_LIST_HEAD(&c->erase_complete_list);
362 	INIT_LIST_HEAD(&c->free_list);
363 	INIT_LIST_HEAD(&c->bad_list);
364 	INIT_LIST_HEAD(&c->bad_used_list);
365 	c->highest_ino = 1;
366 	c->summary = NULL;
367 
368 	ret = jffs2_sum_init(c);
369 	if (ret)
370 		goto out_free;
371 
372 	if (jffs2_build_filesystem(c)) {
373 		dbg_fsbuild("build_fs failed\n");
374 		jffs2_free_ino_caches(c);
375 		jffs2_free_raw_node_refs(c);
376 		ret = -EIO;
377 		goto out_free;
378 	}
379 
380 	jffs2_calc_trigger_levels(c);
381 
382 	return 0;
383 
384  out_free:
385 #ifndef __ECOS
386 	if (jffs2_blocks_use_vmalloc(c))
387 		vfree(c->blocks);
388 	else
389 #endif
390 		kfree(c->blocks);
391 
392 	return ret;
393 }
394