fs/jffs2/build.c

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