kernel/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 "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 }