xref: /openbmc/linux/fs/jffs2/build.c (revision 5a528957)
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 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
14 
15 #include <linux/kernel.h>
16 #include <linux/sched.h>
17 #include <linux/slab.h>
18 #include <linux/vmalloc.h>
19 #include <linux/mtd/mtd.h>
20 #include "nodelist.h"
21 
22 static void jffs2_build_remove_unlinked_inode(struct jffs2_sb_info *,
23 		struct jffs2_inode_cache *, struct jffs2_full_dirent **);
24 
25 static inline struct jffs2_inode_cache *
26 first_inode_chain(int *i, struct jffs2_sb_info *c)
27 {
28 	for (; *i < c->inocache_hashsize; (*i)++) {
29 		if (c->inocache_list[*i])
30 			return c->inocache_list[*i];
31 	}
32 	return NULL;
33 }
34 
35 static inline struct jffs2_inode_cache *
36 next_inode(int *i, struct jffs2_inode_cache *ic, struct jffs2_sb_info *c)
37 {
38 	/* More in this chain? */
39 	if (ic->next)
40 		return ic->next;
41 	(*i)++;
42 	return first_inode_chain(i, c);
43 }
44 
45 #define for_each_inode(i, c, ic)			\
46 	for (i = 0, ic = first_inode_chain(&i, (c));	\
47 	     ic;					\
48 	     ic = next_inode(&i, ic, (c)))
49 
50 
51 static void jffs2_build_inode_pass1(struct jffs2_sb_info *c,
52 				    struct jffs2_inode_cache *ic)
53 {
54 	struct jffs2_full_dirent *fd;
55 
56 	dbg_fsbuild("building directory inode #%u\n", ic->ino);
57 
58 	/* For each child, increase nlink */
59 	for(fd = ic->scan_dents; fd; fd = fd->next) {
60 		struct jffs2_inode_cache *child_ic;
61 		if (!fd->ino)
62 			continue;
63 
64 		/* we can get high latency here with huge directories */
65 
66 		child_ic = jffs2_get_ino_cache(c, fd->ino);
67 		if (!child_ic) {
68 			dbg_fsbuild("child \"%s\" (ino #%u) of dir ino #%u doesn't exist!\n",
69 				  fd->name, fd->ino, ic->ino);
70 			jffs2_mark_node_obsolete(c, fd->raw);
71 			continue;
72 		}
73 
74 		if (fd->type == DT_DIR) {
75 			if (child_ic->pino_nlink) {
76 				JFFS2_ERROR("child dir \"%s\" (ino #%u) of dir ino #%u appears to be a hard link\n",
77 					    fd->name, fd->ino, ic->ino);
78 				/* TODO: What do we do about it? */
79 			} else {
80 				child_ic->pino_nlink = ic->ino;
81 			}
82 		} else
83 			child_ic->pino_nlink++;
84 
85 		dbg_fsbuild("increased nlink for child \"%s\" (ino #%u)\n", fd->name, fd->ino);
86 		/* Can't free scan_dents so far. We might need them in pass 2 */
87 	}
88 }
89 
90 /* Scan plan:
91  - Scan physical nodes. Build map of inodes/dirents. Allocate inocaches as we go
92  - Scan directory tree from top down, setting nlink in inocaches
93  - Scan inocaches for inodes with nlink==0
94 */
95 static int jffs2_build_filesystem(struct jffs2_sb_info *c)
96 {
97 	int ret;
98 	int i;
99 	struct jffs2_inode_cache *ic;
100 	struct jffs2_full_dirent *fd;
101 	struct jffs2_full_dirent *dead_fds = NULL;
102 
103 	dbg_fsbuild("build FS data structures\n");
104 
105 	/* First, scan the medium and build all the inode caches with
106 	   lists of physical nodes */
107 
108 	c->flags |= JFFS2_SB_FLAG_SCANNING;
109 	ret = jffs2_scan_medium(c);
110 	c->flags &= ~JFFS2_SB_FLAG_SCANNING;
111 	if (ret)
112 		goto exit;
113 
114 	dbg_fsbuild("scanned flash completely\n");
115 	jffs2_dbg_dump_block_lists_nolock(c);
116 
117 	dbg_fsbuild("pass 1 starting\n");
118 	c->flags |= JFFS2_SB_FLAG_BUILDING;
119 	/* Now scan the directory tree, increasing nlink according to every dirent found. */
120 	for_each_inode(i, c, ic) {
121 		if (ic->scan_dents) {
122 			jffs2_build_inode_pass1(c, ic);
123 			cond_resched();
124 		}
125 	}
126 
127 	dbg_fsbuild("pass 1 complete\n");
128 
129 	/* Next, scan for inodes with nlink == 0 and remove them. If
130 	   they were directories, then decrement the nlink of their
131 	   children too, and repeat the scan. As that's going to be
132 	   a fairly uncommon occurrence, it's not so evil to do it this
133 	   way. Recursion bad. */
134 	dbg_fsbuild("pass 2 starting\n");
135 
136 	for_each_inode(i, c, ic) {
137 		if (ic->pino_nlink)
138 			continue;
139 
140 		jffs2_build_remove_unlinked_inode(c, ic, &dead_fds);
141 		cond_resched();
142 	}
143 
144 	dbg_fsbuild("pass 2a starting\n");
145 
146 	while (dead_fds) {
147 		fd = dead_fds;
148 		dead_fds = fd->next;
149 
150 		ic = jffs2_get_ino_cache(c, fd->ino);
151 
152 		if (ic)
153 			jffs2_build_remove_unlinked_inode(c, ic, &dead_fds);
154 		jffs2_free_full_dirent(fd);
155 	}
156 
157 	dbg_fsbuild("pass 2a complete\n");
158 	dbg_fsbuild("freeing temporary data structures\n");
159 
160 	/* Finally, we can scan again and free the dirent structs */
161 	for_each_inode(i, c, ic) {
162 		while(ic->scan_dents) {
163 			fd = ic->scan_dents;
164 			ic->scan_dents = fd->next;
165 			jffs2_free_full_dirent(fd);
166 		}
167 		ic->scan_dents = NULL;
168 		cond_resched();
169 	}
170 	jffs2_build_xattr_subsystem(c);
171 	c->flags &= ~JFFS2_SB_FLAG_BUILDING;
172 
173 	dbg_fsbuild("FS build complete\n");
174 
175 	/* Rotate the lists by some number to ensure wear levelling */
176 	jffs2_rotate_lists(c);
177 
178 	ret = 0;
179 
180 exit:
181 	if (ret) {
182 		for_each_inode(i, c, ic) {
183 			while(ic->scan_dents) {
184 				fd = ic->scan_dents;
185 				ic->scan_dents = fd->next;
186 				jffs2_free_full_dirent(fd);
187 			}
188 		}
189 		jffs2_clear_xattr_subsystem(c);
190 	}
191 
192 	return ret;
193 }
194 
195 static void jffs2_build_remove_unlinked_inode(struct jffs2_sb_info *c,
196 					struct jffs2_inode_cache *ic,
197 					struct jffs2_full_dirent **dead_fds)
198 {
199 	struct jffs2_raw_node_ref *raw;
200 	struct jffs2_full_dirent *fd;
201 
202 	dbg_fsbuild("removing ino #%u with nlink == zero.\n", ic->ino);
203 
204 	raw = ic->nodes;
205 	while (raw != (void *)ic) {
206 		struct jffs2_raw_node_ref *next = raw->next_in_ino;
207 		dbg_fsbuild("obsoleting node at 0x%08x\n", ref_offset(raw));
208 		jffs2_mark_node_obsolete(c, raw);
209 		raw = next;
210 	}
211 
212 	if (ic->scan_dents) {
213 		int whinged = 0;
214 		dbg_fsbuild("inode #%u was a directory which may have children...\n", ic->ino);
215 
216 		while(ic->scan_dents) {
217 			struct jffs2_inode_cache *child_ic;
218 
219 			fd = ic->scan_dents;
220 			ic->scan_dents = fd->next;
221 
222 			if (!fd->ino) {
223 				/* It's a deletion dirent. Ignore it */
224 				dbg_fsbuild("child \"%s\" is a deletion dirent, skipping...\n", fd->name);
225 				jffs2_free_full_dirent(fd);
226 				continue;
227 			}
228 			if (!whinged)
229 				whinged = 1;
230 
231 			dbg_fsbuild("removing child \"%s\", ino #%u\n", fd->name, fd->ino);
232 
233 			child_ic = jffs2_get_ino_cache(c, fd->ino);
234 			if (!child_ic) {
235 				dbg_fsbuild("cannot remove child \"%s\", ino #%u, because it doesn't exist\n",
236 						fd->name, fd->ino);
237 				jffs2_free_full_dirent(fd);
238 				continue;
239 			}
240 
241 			/* Reduce nlink of the child. If it's now zero, stick it on the
242 			   dead_fds list to be cleaned up later. Else just free the fd */
243 
244 			if (fd->type == DT_DIR)
245 				child_ic->pino_nlink = 0;
246 			else
247 				child_ic->pino_nlink--;
248 
249 			if (!child_ic->pino_nlink) {
250 				dbg_fsbuild("inode #%u (\"%s\") now has no links; adding to dead_fds list.\n",
251 					  fd->ino, fd->name);
252 				fd->next = *dead_fds;
253 				*dead_fds = fd;
254 			} else {
255 				dbg_fsbuild("inode #%u (\"%s\") has now got nlink %d. Ignoring.\n",
256 					  fd->ino, fd->name, child_ic->pino_nlink);
257 				jffs2_free_full_dirent(fd);
258 			}
259 		}
260 	}
261 
262 	/*
263 	   We don't delete the inocache from the hash list and free it yet.
264 	   The erase code will do that, when all the nodes are completely gone.
265 	*/
266 }
267 
268 static void jffs2_calc_trigger_levels(struct jffs2_sb_info *c)
269 {
270 	uint32_t size;
271 
272 	/* Deletion should almost _always_ be allowed. We're fairly
273 	   buggered once we stop allowing people to delete stuff
274 	   because there's not enough free space... */
275 	c->resv_blocks_deletion = 2;
276 
277 	/* Be conservative about how much space we need before we allow writes.
278 	   On top of that which is required for deletia, require an extra 2%
279 	   of the medium to be available, for overhead caused by nodes being
280 	   split across blocks, etc. */
281 
282 	size = c->flash_size / 50; /* 2% of flash size */
283 	size += c->nr_blocks * 100; /* And 100 bytes per eraseblock */
284 	size += c->sector_size - 1; /* ... and round up */
285 
286 	c->resv_blocks_write = c->resv_blocks_deletion + (size / c->sector_size);
287 
288 	/* When do we let the GC thread run in the background */
289 
290 	c->resv_blocks_gctrigger = c->resv_blocks_write + 1;
291 
292 	/* When do we allow garbage collection to merge nodes to make
293 	   long-term progress at the expense of short-term space exhaustion? */
294 	c->resv_blocks_gcmerge = c->resv_blocks_deletion + 1;
295 
296 	/* When do we allow garbage collection to eat from bad blocks rather
297 	   than actually making progress? */
298 	c->resv_blocks_gcbad = 0;//c->resv_blocks_deletion + 2;
299 
300 	/* What number of 'very dirty' eraseblocks do we allow before we
301 	   trigger the GC thread even if we don't _need_ the space. When we
302 	   can't mark nodes obsolete on the medium, the old dirty nodes cause
303 	   performance problems because we have to inspect and discard them. */
304 	c->vdirty_blocks_gctrigger = c->resv_blocks_gctrigger;
305 	if (jffs2_can_mark_obsolete(c))
306 		c->vdirty_blocks_gctrigger *= 10;
307 
308 	/* If there's less than this amount of dirty space, don't bother
309 	   trying to GC to make more space. It'll be a fruitless task */
310 	c->nospc_dirty_size = c->sector_size + (c->flash_size / 100);
311 
312 	dbg_fsbuild("trigger levels (size %d KiB, block size %d KiB, %d blocks)\n",
313 		    c->flash_size / 1024, c->sector_size / 1024, c->nr_blocks);
314 	dbg_fsbuild("Blocks required to allow deletion:    %d (%d KiB)\n",
315 		  c->resv_blocks_deletion, c->resv_blocks_deletion*c->sector_size/1024);
316 	dbg_fsbuild("Blocks required to allow writes:      %d (%d KiB)\n",
317 		  c->resv_blocks_write, c->resv_blocks_write*c->sector_size/1024);
318 	dbg_fsbuild("Blocks required to quiesce GC thread: %d (%d KiB)\n",
319 		  c->resv_blocks_gctrigger, c->resv_blocks_gctrigger*c->sector_size/1024);
320 	dbg_fsbuild("Blocks required to allow GC merges:   %d (%d KiB)\n",
321 		  c->resv_blocks_gcmerge, c->resv_blocks_gcmerge*c->sector_size/1024);
322 	dbg_fsbuild("Blocks required to GC bad blocks:     %d (%d KiB)\n",
323 		  c->resv_blocks_gcbad, c->resv_blocks_gcbad*c->sector_size/1024);
324 	dbg_fsbuild("Amount of dirty space required to GC: %d bytes\n",
325 		  c->nospc_dirty_size);
326 	dbg_fsbuild("Very dirty blocks before GC triggered: %d\n",
327 		  c->vdirty_blocks_gctrigger);
328 }
329 
330 int jffs2_do_mount_fs(struct jffs2_sb_info *c)
331 {
332 	int ret;
333 	int i;
334 	int size;
335 
336 	c->free_size = c->flash_size;
337 	c->nr_blocks = c->flash_size / c->sector_size;
338 	size = sizeof(struct jffs2_eraseblock) * c->nr_blocks;
339 #ifndef __ECOS
340 	if (jffs2_blocks_use_vmalloc(c))
341 		c->blocks = vzalloc(size);
342 	else
343 #endif
344 		c->blocks = kzalloc(size, GFP_KERNEL);
345 	if (!c->blocks)
346 		return -ENOMEM;
347 
348 	for (i=0; i<c->nr_blocks; i++) {
349 		INIT_LIST_HEAD(&c->blocks[i].list);
350 		c->blocks[i].offset = i * c->sector_size;
351 		c->blocks[i].free_size = c->sector_size;
352 	}
353 
354 	INIT_LIST_HEAD(&c->clean_list);
355 	INIT_LIST_HEAD(&c->very_dirty_list);
356 	INIT_LIST_HEAD(&c->dirty_list);
357 	INIT_LIST_HEAD(&c->erasable_list);
358 	INIT_LIST_HEAD(&c->erasing_list);
359 	INIT_LIST_HEAD(&c->erase_checking_list);
360 	INIT_LIST_HEAD(&c->erase_pending_list);
361 	INIT_LIST_HEAD(&c->erasable_pending_wbuf_list);
362 	INIT_LIST_HEAD(&c->erase_complete_list);
363 	INIT_LIST_HEAD(&c->free_list);
364 	INIT_LIST_HEAD(&c->bad_list);
365 	INIT_LIST_HEAD(&c->bad_used_list);
366 	c->highest_ino = 1;
367 	c->summary = NULL;
368 
369 	ret = jffs2_sum_init(c);
370 	if (ret)
371 		goto out_free;
372 
373 	if (jffs2_build_filesystem(c)) {
374 		dbg_fsbuild("build_fs failed\n");
375 		jffs2_free_ino_caches(c);
376 		jffs2_free_raw_node_refs(c);
377 		ret = -EIO;
378 		goto out_free;
379 	}
380 
381 	jffs2_calc_trigger_levels(c);
382 
383 	return 0;
384 
385  out_free:
386 #ifndef __ECOS
387 	if (jffs2_blocks_use_vmalloc(c))
388 		vfree(c->blocks);
389 	else
390 #endif
391 		kfree(c->blocks);
392 
393 	return ret;
394 }
395