From: Linus Torvalds Date: Fri, 22 Oct 2010 23:34:03 +0000 (-0700) Subject: Merge branch 'linux-next' of git://git.infradead.org/ubifs-2.6 X-Git-Tag: v2.6.37-rc1~170 X-Git-Url: https://git.openpandora.org/cgi-bin/gitweb.cgi?p=pandora-kernel.git;a=commitdiff_plain;h=06d362931a530e0d48c1a9554a752da4ed240f0b;hp=f5d9d249b9a6884daff513ef08afa43d3f7e085f Merge branch 'linux-next' of git://git.infradead.org/ubifs-2.6 * 'linux-next' of git://git.infradead.org/ubifs-2.6: UBIFS: do not allocate unneeded scan buffer UBIFS: do not forget to cancel timers UBIFS: remove a bit of unneeded code UBIFS: add a commentary about log recovery UBIFS: avoid kernel error if ubifs superblock read fails UBIFS: introduce new flags for RO mounts UBIFS: introduce new flag for RO due to errors UBIFS: check return code of pnode_lookup UBIFS: check return code of ubifs_lpt_lookup UBIFS: improve error reporting when reading bad node UBIFS: introduce list sorting debugging checks UBIFS: fix assertion warnings in comparison function UBIFS: mark unused key objects as invalid UBIFS: do not write rubbish into truncation scanning node UBIFS: improve assertion in node comparison functions UBIFS: do not use key type in list_sort UBIFS: do not look up truncation nodes UBIFS: fix assertion warning UBIFS: do not treat ENOSPC specially UBIFS: switch to RO mode after synchronizing --- diff --git a/fs/ubifs/commit.c b/fs/ubifs/commit.c index 37fa7ed062d8..02429d81ca33 100644 --- a/fs/ubifs/commit.c +++ b/fs/ubifs/commit.c @@ -63,7 +63,9 @@ static int do_commit(struct ubifs_info *c) struct ubifs_lp_stats lst; dbg_cmt("start"); - if (c->ro_media) { + ubifs_assert(!c->ro_media && !c->ro_mount); + + if (c->ro_error) { err = -EROFS; goto out_up; } diff --git a/fs/ubifs/debug.c b/fs/ubifs/debug.c index c6c553fd0b3d..0bee4dbffc31 100644 --- a/fs/ubifs/debug.c +++ b/fs/ubifs/debug.c @@ -2239,6 +2239,162 @@ out_free: return err; } +/** + * dbg_check_data_nodes_order - check that list of data nodes is sorted. + * @c: UBIFS file-system description object + * @head: the list of nodes ('struct ubifs_scan_node' objects) + * + * This function returns zero if the list of data nodes is sorted correctly, + * and %-EINVAL if not. + */ +int dbg_check_data_nodes_order(struct ubifs_info *c, struct list_head *head) +{ + struct list_head *cur; + struct ubifs_scan_node *sa, *sb; + + if (!(ubifs_chk_flags & UBIFS_CHK_GEN)) + return 0; + + for (cur = head->next; cur->next != head; cur = cur->next) { + ino_t inuma, inumb; + uint32_t blka, blkb; + + cond_resched(); + sa = container_of(cur, struct ubifs_scan_node, list); + sb = container_of(cur->next, struct ubifs_scan_node, list); + + if (sa->type != UBIFS_DATA_NODE) { + ubifs_err("bad node type %d", sa->type); + dbg_dump_node(c, sa->node); + return -EINVAL; + } + if (sb->type != UBIFS_DATA_NODE) { + ubifs_err("bad node type %d", sb->type); + dbg_dump_node(c, sb->node); + return -EINVAL; + } + + inuma = key_inum(c, &sa->key); + inumb = key_inum(c, &sb->key); + + if (inuma < inumb) + continue; + if (inuma > inumb) { + ubifs_err("larger inum %lu goes before inum %lu", + (unsigned long)inuma, (unsigned long)inumb); + goto error_dump; + } + + blka = key_block(c, &sa->key); + blkb = key_block(c, &sb->key); + + if (blka > blkb) { + ubifs_err("larger block %u goes before %u", blka, blkb); + goto error_dump; + } + if (blka == blkb) { + ubifs_err("two data nodes for the same block"); + goto error_dump; + } + } + + return 0; + +error_dump: + dbg_dump_node(c, sa->node); + dbg_dump_node(c, sb->node); + return -EINVAL; +} + +/** + * dbg_check_nondata_nodes_order - check that list of data nodes is sorted. + * @c: UBIFS file-system description object + * @head: the list of nodes ('struct ubifs_scan_node' objects) + * + * This function returns zero if the list of non-data nodes is sorted correctly, + * and %-EINVAL if not. + */ +int dbg_check_nondata_nodes_order(struct ubifs_info *c, struct list_head *head) +{ + struct list_head *cur; + struct ubifs_scan_node *sa, *sb; + + if (!(ubifs_chk_flags & UBIFS_CHK_GEN)) + return 0; + + for (cur = head->next; cur->next != head; cur = cur->next) { + ino_t inuma, inumb; + uint32_t hasha, hashb; + + cond_resched(); + sa = container_of(cur, struct ubifs_scan_node, list); + sb = container_of(cur->next, struct ubifs_scan_node, list); + + if (sa->type != UBIFS_INO_NODE && sa->type != UBIFS_DENT_NODE && + sa->type != UBIFS_XENT_NODE) { + ubifs_err("bad node type %d", sa->type); + dbg_dump_node(c, sa->node); + return -EINVAL; + } + if (sa->type != UBIFS_INO_NODE && sa->type != UBIFS_DENT_NODE && + sa->type != UBIFS_XENT_NODE) { + ubifs_err("bad node type %d", sb->type); + dbg_dump_node(c, sb->node); + return -EINVAL; + } + + if (sa->type != UBIFS_INO_NODE && sb->type == UBIFS_INO_NODE) { + ubifs_err("non-inode node goes before inode node"); + goto error_dump; + } + + if (sa->type == UBIFS_INO_NODE && sb->type != UBIFS_INO_NODE) + continue; + + if (sa->type == UBIFS_INO_NODE && sb->type == UBIFS_INO_NODE) { + /* Inode nodes are sorted in descending size order */ + if (sa->len < sb->len) { + ubifs_err("smaller inode node goes first"); + goto error_dump; + } + continue; + } + + /* + * This is either a dentry or xentry, which should be sorted in + * ascending (parent ino, hash) order. + */ + inuma = key_inum(c, &sa->key); + inumb = key_inum(c, &sb->key); + + if (inuma < inumb) + continue; + if (inuma > inumb) { + ubifs_err("larger inum %lu goes before inum %lu", + (unsigned long)inuma, (unsigned long)inumb); + goto error_dump; + } + + hasha = key_block(c, &sa->key); + hashb = key_block(c, &sb->key); + + if (hasha > hashb) { + ubifs_err("larger hash %u goes before %u", hasha, hashb); + goto error_dump; + } + } + + return 0; + +error_dump: + ubifs_msg("dumping first node"); + dbg_dump_node(c, sa->node); + ubifs_msg("dumping second node"); + dbg_dump_node(c, sb->node); + return -EINVAL; + return 0; +} + static int invocation_cnt; int dbg_force_in_the_gaps(void) diff --git a/fs/ubifs/debug.h b/fs/ubifs/debug.h index 29d960101ea6..69ebe4729151 100644 --- a/fs/ubifs/debug.h +++ b/fs/ubifs/debug.h @@ -324,6 +324,8 @@ int dbg_check_lpt_nodes(struct ubifs_info *c, struct ubifs_cnode *cnode, int row, int col); int dbg_check_inode_size(struct ubifs_info *c, const struct inode *inode, loff_t size); +int dbg_check_data_nodes_order(struct ubifs_info *c, struct list_head *head); +int dbg_check_nondata_nodes_order(struct ubifs_info *c, struct list_head *head); /* Force the use of in-the-gaps method for testing */ @@ -465,6 +467,8 @@ void dbg_debugfs_exit_fs(struct ubifs_info *c); #define dbg_check_lprops(c) 0 #define dbg_check_lpt_nodes(c, cnode, row, col) 0 #define dbg_check_inode_size(c, inode, size) 0 +#define dbg_check_data_nodes_order(c, head) 0 +#define dbg_check_nondata_nodes_order(c, head) 0 #define dbg_force_in_the_gaps_enabled 0 #define dbg_force_in_the_gaps() 0 #define dbg_failure_mode 0 diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c index 03ae894c45de..d77db7e36484 100644 --- a/fs/ubifs/file.c +++ b/fs/ubifs/file.c @@ -433,8 +433,9 @@ static int ubifs_write_begin(struct file *file, struct address_space *mapping, struct page *page; ubifs_assert(ubifs_inode(inode)->ui_size == inode->i_size); + ubifs_assert(!c->ro_media && !c->ro_mount); - if (unlikely(c->ro_media)) + if (unlikely(c->ro_error)) return -EROFS; /* Try out the fast-path part first */ @@ -1439,9 +1440,9 @@ static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vm dbg_gen("ino %lu, pg %lu, i_size %lld", inode->i_ino, page->index, i_size_read(inode)); - ubifs_assert(!(inode->i_sb->s_flags & MS_RDONLY)); + ubifs_assert(!c->ro_media && !c->ro_mount); - if (unlikely(c->ro_media)) + if (unlikely(c->ro_error)) return VM_FAULT_SIGBUS; /* -EROFS */ /* diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c index 918d1582ca05..151f10882820 100644 --- a/fs/ubifs/gc.c +++ b/fs/ubifs/gc.c @@ -125,10 +125,16 @@ int data_nodes_cmp(void *priv, struct list_head *a, struct list_head *b) struct ubifs_scan_node *sa, *sb; cond_resched(); + if (a == b) + return 0; + sa = list_entry(a, struct ubifs_scan_node, list); sb = list_entry(b, struct ubifs_scan_node, list); + ubifs_assert(key_type(c, &sa->key) == UBIFS_DATA_KEY); ubifs_assert(key_type(c, &sb->key) == UBIFS_DATA_KEY); + ubifs_assert(sa->type == UBIFS_DATA_NODE); + ubifs_assert(sb->type == UBIFS_DATA_NODE); inuma = key_inum(c, &sa->key); inumb = key_inum(c, &sb->key); @@ -157,28 +163,40 @@ int data_nodes_cmp(void *priv, struct list_head *a, struct list_head *b) */ int nondata_nodes_cmp(void *priv, struct list_head *a, struct list_head *b) { - int typea, typeb; ino_t inuma, inumb; struct ubifs_info *c = priv; struct ubifs_scan_node *sa, *sb; cond_resched(); + if (a == b) + return 0; + sa = list_entry(a, struct ubifs_scan_node, list); sb = list_entry(b, struct ubifs_scan_node, list); - typea = key_type(c, &sa->key); - typeb = key_type(c, &sb->key); - ubifs_assert(typea != UBIFS_DATA_KEY && typeb != UBIFS_DATA_KEY); + + ubifs_assert(key_type(c, &sa->key) != UBIFS_DATA_KEY && + key_type(c, &sb->key) != UBIFS_DATA_KEY); + ubifs_assert(sa->type != UBIFS_DATA_NODE && + sb->type != UBIFS_DATA_NODE); /* Inodes go before directory entries */ - if (typea == UBIFS_INO_KEY) { - if (typeb == UBIFS_INO_KEY) + if (sa->type == UBIFS_INO_NODE) { + if (sb->type == UBIFS_INO_NODE) return sb->len - sa->len; return -1; } - if (typeb == UBIFS_INO_KEY) + if (sb->type == UBIFS_INO_NODE) return 1; - ubifs_assert(typea == UBIFS_DENT_KEY && typeb == UBIFS_DENT_KEY); + ubifs_assert(key_type(c, &sa->key) == UBIFS_DENT_KEY || + key_type(c, &sa->key) == UBIFS_XENT_KEY); + ubifs_assert(key_type(c, &sb->key) == UBIFS_DENT_KEY || + key_type(c, &sb->key) == UBIFS_XENT_KEY); + ubifs_assert(sa->type == UBIFS_DENT_NODE || + sa->type == UBIFS_XENT_NODE); + ubifs_assert(sb->type == UBIFS_DENT_NODE || + sb->type == UBIFS_XENT_NODE); + inuma = key_inum(c, &sa->key); inumb = key_inum(c, &sb->key); @@ -224,17 +242,33 @@ int nondata_nodes_cmp(void *priv, struct list_head *a, struct list_head *b) static int sort_nodes(struct ubifs_info *c, struct ubifs_scan_leb *sleb, struct list_head *nondata, int *min) { + int err; struct ubifs_scan_node *snod, *tmp; *min = INT_MAX; /* Separate data nodes and non-data nodes */ list_for_each_entry_safe(snod, tmp, &sleb->nodes, list) { - int err; + ubifs_assert(snod->type == UBIFS_INO_NODE || + snod->type == UBIFS_DATA_NODE || + snod->type == UBIFS_DENT_NODE || + snod->type == UBIFS_XENT_NODE || + snod->type == UBIFS_TRUN_NODE); + + if (snod->type != UBIFS_INO_NODE && + snod->type != UBIFS_DATA_NODE && + snod->type != UBIFS_DENT_NODE && + snod->type != UBIFS_XENT_NODE) { + /* Probably truncation node, zap it */ + list_del(&snod->list); + kfree(snod); + continue; + } - ubifs_assert(snod->type != UBIFS_IDX_NODE); - ubifs_assert(snod->type != UBIFS_REF_NODE); - ubifs_assert(snod->type != UBIFS_CS_NODE); + ubifs_assert(key_type(c, &snod->key) == UBIFS_DATA_KEY || + key_type(c, &snod->key) == UBIFS_INO_KEY || + key_type(c, &snod->key) == UBIFS_DENT_KEY || + key_type(c, &snod->key) == UBIFS_XENT_KEY); err = ubifs_tnc_has_node(c, &snod->key, 0, sleb->lnum, snod->offs, 0); @@ -258,6 +292,13 @@ static int sort_nodes(struct ubifs_info *c, struct ubifs_scan_leb *sleb, /* Sort data and non-data nodes */ list_sort(c, &sleb->nodes, &data_nodes_cmp); list_sort(c, nondata, &nondata_nodes_cmp); + + err = dbg_check_data_nodes_order(c, &sleb->nodes); + if (err) + return err; + err = dbg_check_nondata_nodes_order(c, nondata); + if (err) + return err; return 0; } @@ -575,13 +616,14 @@ int ubifs_garbage_collect(struct ubifs_info *c, int anyway) struct ubifs_wbuf *wbuf = &c->jheads[GCHD].wbuf; ubifs_assert_cmt_locked(c); + ubifs_assert(!c->ro_media && !c->ro_mount); if (ubifs_gc_should_commit(c)) return -EAGAIN; mutex_lock_nested(&wbuf->io_mutex, wbuf->jhead); - if (c->ro_media) { + if (c->ro_error) { ret = -EROFS; goto out_unlock; } @@ -677,14 +719,12 @@ int ubifs_garbage_collect(struct ubifs_info *c, int anyway) ret = ubifs_garbage_collect_leb(c, &lp); if (ret < 0) { - if (ret == -EAGAIN || ret == -ENOSPC) { + if (ret == -EAGAIN) { /* - * These codes are not errors, so we have to - * return the LEB to lprops. But if the - * 'ubifs_return_leb()' function fails, its - * failure code is propagated to the caller - * instead of the original '-EAGAIN' or - * '-ENOSPC'. + * This is not error, so we have to return the + * LEB to lprops. But if 'ubifs_return_leb()' + * fails, its failure code is propagated to the + * caller instead of the original '-EAGAIN'. */ err = ubifs_return_leb(c, lp.lnum); if (err) @@ -774,8 +814,8 @@ out_unlock: out: ubifs_assert(ret < 0); ubifs_assert(ret != -ENOSPC && ret != -EAGAIN); - ubifs_ro_mode(c, ret); ubifs_wbuf_sync_nolock(wbuf); + ubifs_ro_mode(c, ret); mutex_unlock(&wbuf->io_mutex); ubifs_return_leb(c, lp.lnum); return ret; diff --git a/fs/ubifs/io.c b/fs/ubifs/io.c index bcf5a16f30bb..d82173182eeb 100644 --- a/fs/ubifs/io.c +++ b/fs/ubifs/io.c @@ -61,8 +61,8 @@ */ void ubifs_ro_mode(struct ubifs_info *c, int err) { - if (!c->ro_media) { - c->ro_media = 1; + if (!c->ro_error) { + c->ro_error = 1; c->no_chk_data_crc = 0; c->vfs_sb->s_flags |= MS_RDONLY; ubifs_warn("switched to read-only mode, error %d", err); @@ -356,11 +356,11 @@ int ubifs_wbuf_sync_nolock(struct ubifs_wbuf *wbuf) dbg_io("LEB %d:%d, %d bytes, jhead %s", wbuf->lnum, wbuf->offs, wbuf->used, dbg_jhead(wbuf->jhead)); - ubifs_assert(!(c->vfs_sb->s_flags & MS_RDONLY)); ubifs_assert(!(wbuf->avail & 7)); ubifs_assert(wbuf->offs + c->min_io_size <= c->leb_size); + ubifs_assert(!c->ro_media && !c->ro_mount); - if (c->ro_media) + if (c->ro_error) return -EROFS; ubifs_pad(c, wbuf->buf + wbuf->used, wbuf->avail); @@ -440,11 +440,12 @@ int ubifs_bg_wbufs_sync(struct ubifs_info *c) { int err, i; + ubifs_assert(!c->ro_media && !c->ro_mount); if (!c->need_wbuf_sync) return 0; c->need_wbuf_sync = 0; - if (c->ro_media) { + if (c->ro_error) { err = -EROFS; goto out_timers; } @@ -519,6 +520,7 @@ int ubifs_wbuf_write_nolock(struct ubifs_wbuf *wbuf, void *buf, int len) ubifs_assert(!(wbuf->offs & 7) && wbuf->offs <= c->leb_size); ubifs_assert(wbuf->avail > 0 && wbuf->avail <= c->min_io_size); ubifs_assert(mutex_is_locked(&wbuf->io_mutex)); + ubifs_assert(!c->ro_media && !c->ro_mount); if (c->leb_size - wbuf->offs - wbuf->used < aligned_len) { err = -ENOSPC; @@ -527,7 +529,7 @@ int ubifs_wbuf_write_nolock(struct ubifs_wbuf *wbuf, void *buf, int len) cancel_wbuf_timer_nolock(wbuf); - if (c->ro_media) + if (c->ro_error) return -EROFS; if (aligned_len <= wbuf->avail) { @@ -663,8 +665,9 @@ int ubifs_write_node(struct ubifs_info *c, void *buf, int len, int lnum, buf_len); ubifs_assert(lnum >= 0 && lnum < c->leb_cnt && offs >= 0); ubifs_assert(offs % c->min_io_size == 0 && offs < c->leb_size); + ubifs_assert(!c->ro_media && !c->ro_mount); - if (c->ro_media) + if (c->ro_error) return -EROFS; ubifs_prepare_node(c, buf, len, 1); @@ -815,7 +818,8 @@ int ubifs_read_node(const struct ubifs_info *c, void *buf, int type, int len, return 0; out: - ubifs_err("bad node at LEB %d:%d", lnum, offs); + ubifs_err("bad node at LEB %d:%d, LEB mapping status %d", lnum, offs, + ubi_is_mapped(c->ubi, lnum)); dbg_dump_node(c, buf); dbg_dump_stack(); return -EINVAL; diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c index d321baeca68d..914f1bd89e57 100644 --- a/fs/ubifs/journal.c +++ b/fs/ubifs/journal.c @@ -122,11 +122,12 @@ static int reserve_space(struct ubifs_info *c, int jhead, int len) * better to try to allocate space at the ends of eraseblocks. This is * what the squeeze parameter does. */ + ubifs_assert(!c->ro_media && !c->ro_mount); squeeze = (jhead == BASEHD); again: mutex_lock_nested(&wbuf->io_mutex, wbuf->jhead); - if (c->ro_media) { + if (c->ro_error) { err = -EROFS; goto out_unlock; } diff --git a/fs/ubifs/key.h b/fs/ubifs/key.h index 0f530c684f0b..92a8491a8f8c 100644 --- a/fs/ubifs/key.h +++ b/fs/ubifs/key.h @@ -305,6 +305,20 @@ static inline void trun_key_init(const struct ubifs_info *c, key->u32[1] = UBIFS_TRUN_KEY << UBIFS_S_KEY_BLOCK_BITS; } +/** + * invalid_key_init - initialize invalid node key. + * @c: UBIFS file-system description object + * @key: key to initialize + * + * This is a helper function which marks a @key object as invalid. + */ +static inline void invalid_key_init(const struct ubifs_info *c, + union ubifs_key *key) +{ + key->u32[0] = 0xDEADBEAF; + key->u32[1] = UBIFS_INVALID_KEY; +} + /** * key_type - get key type. * @c: UBIFS file-system description object diff --git a/fs/ubifs/log.c b/fs/ubifs/log.c index c345e125f42c..4d0cb1241460 100644 --- a/fs/ubifs/log.c +++ b/fs/ubifs/log.c @@ -159,7 +159,7 @@ void ubifs_add_bud(struct ubifs_info *c, struct ubifs_bud *bud) jhead = &c->jheads[bud->jhead]; list_add_tail(&bud->list, &jhead->buds_list); } else - ubifs_assert(c->replaying && (c->vfs_sb->s_flags & MS_RDONLY)); + ubifs_assert(c->replaying && c->ro_mount); /* * Note, although this is a new bud, we anyway account this space now, @@ -223,8 +223,8 @@ int ubifs_add_bud_to_log(struct ubifs_info *c, int jhead, int lnum, int offs) } mutex_lock(&c->log_mutex); - - if (c->ro_media) { + ubifs_assert(!c->ro_media && !c->ro_mount); + if (c->ro_error) { err = -EROFS; goto out_unlock; } diff --git a/fs/ubifs/lpt.c b/fs/ubifs/lpt.c index 0084a33c4c69..72775d35b99e 100644 --- a/fs/ubifs/lpt.c +++ b/fs/ubifs/lpt.c @@ -1363,6 +1363,7 @@ static int read_lsave(struct ubifs_info *c) goto out; for (i = 0; i < c->lsave_cnt; i++) { int lnum = c->lsave[i]; + struct ubifs_lprops *lprops; /* * Due to automatic resizing, the values in the lsave table @@ -1370,7 +1371,11 @@ static int read_lsave(struct ubifs_info *c) */ if (lnum >= c->leb_cnt) continue; - ubifs_lpt_lookup(c, lnum); + lprops = ubifs_lpt_lookup(c, lnum); + if (IS_ERR(lprops)) { + err = PTR_ERR(lprops); + goto out; + } } out: vfree(buf); diff --git a/fs/ubifs/lpt_commit.c b/fs/ubifs/lpt_commit.c index d12535b7fc78..5c90dec5db0b 100644 --- a/fs/ubifs/lpt_commit.c +++ b/fs/ubifs/lpt_commit.c @@ -705,6 +705,9 @@ static int make_tree_dirty(struct ubifs_info *c) struct ubifs_pnode *pnode; pnode = pnode_lookup(c, 0); + if (IS_ERR(pnode)) + return PTR_ERR(pnode); + while (pnode) { do_make_pnode_dirty(c, pnode); pnode = next_pnode_to_dirty(c, pnode); diff --git a/fs/ubifs/master.c b/fs/ubifs/master.c index 28beaeedadc0..21f47afdacff 100644 --- a/fs/ubifs/master.c +++ b/fs/ubifs/master.c @@ -361,7 +361,8 @@ int ubifs_write_master(struct ubifs_info *c) { int err, lnum, offs, len; - if (c->ro_media) + ubifs_assert(!c->ro_media && !c->ro_mount); + if (c->ro_error) return -EROFS; lnum = UBIFS_MST_LNUM; diff --git a/fs/ubifs/misc.h b/fs/ubifs/misc.h index 4fa81d867e41..c3de04dc952a 100644 --- a/fs/ubifs/misc.h +++ b/fs/ubifs/misc.h @@ -132,7 +132,8 @@ static inline int ubifs_leb_unmap(const struct ubifs_info *c, int lnum) { int err; - if (c->ro_media) + ubifs_assert(!c->ro_media && !c->ro_mount); + if (c->ro_error) return -EROFS; err = ubi_leb_unmap(c->ubi, lnum); if (err) { @@ -159,7 +160,8 @@ static inline int ubifs_leb_write(const struct ubifs_info *c, int lnum, { int err; - if (c->ro_media) + ubifs_assert(!c->ro_media && !c->ro_mount); + if (c->ro_error) return -EROFS; err = ubi_leb_write(c->ubi, lnum, buf, offs, len, dtype); if (err) { @@ -186,7 +188,8 @@ static inline int ubifs_leb_change(const struct ubifs_info *c, int lnum, { int err; - if (c->ro_media) + ubifs_assert(!c->ro_media && !c->ro_mount); + if (c->ro_error) return -EROFS; err = ubi_leb_change(c->ubi, lnum, buf, len, dtype); if (err) { diff --git a/fs/ubifs/recovery.c b/fs/ubifs/recovery.c index daae9e1f5382..77e9b874b6c2 100644 --- a/fs/ubifs/recovery.c +++ b/fs/ubifs/recovery.c @@ -292,7 +292,7 @@ int ubifs_recover_master_node(struct ubifs_info *c) memcpy(c->mst_node, mst, UBIFS_MST_NODE_SZ); - if ((c->vfs_sb->s_flags & MS_RDONLY)) { + if (c->ro_mount) { /* Read-only mode. Keep a copy for switching to rw mode */ c->rcvrd_mst_node = kmalloc(sz, GFP_KERNEL); if (!c->rcvrd_mst_node) { @@ -469,7 +469,7 @@ static int fix_unclean_leb(struct ubifs_info *c, struct ubifs_scan_leb *sleb, endpt = snod->offs + snod->len; } - if ((c->vfs_sb->s_flags & MS_RDONLY) && !c->remounting_rw) { + if (c->ro_mount && !c->remounting_rw) { /* Add to recovery list */ struct ubifs_unclean_leb *ucleb; @@ -772,7 +772,8 @@ out_free: * @sbuf: LEB-sized buffer to use * * This function does a scan of a LEB, but caters for errors that might have - * been caused by the unclean unmount from which we are attempting to recover. + * been caused by unclean reboots from which we are attempting to recover + * (assume that only the last log LEB can be corrupted by an unclean reboot). * * This function returns %0 on success and a negative error code on failure. */ @@ -883,7 +884,7 @@ int ubifs_recover_inl_heads(const struct ubifs_info *c, void *sbuf) { int err; - ubifs_assert(!(c->vfs_sb->s_flags & MS_RDONLY) || c->remounting_rw); + ubifs_assert(!c->ro_mount || c->remounting_rw); dbg_rcvry("checking index head at %d:%d", c->ihead_lnum, c->ihead_offs); err = recover_head(c, c->ihead_lnum, c->ihead_offs, sbuf); @@ -1461,7 +1462,7 @@ int ubifs_recover_size(struct ubifs_info *c) } } if (e->exists && e->i_size < e->d_size) { - if (!e->inode && (c->vfs_sb->s_flags & MS_RDONLY)) { + if (!e->inode && c->ro_mount) { /* Fix the inode size and pin it in memory */ struct inode *inode; diff --git a/fs/ubifs/replay.c b/fs/ubifs/replay.c index 5c2d6d759a3e..eed0fcff8d73 100644 --- a/fs/ubifs/replay.c +++ b/fs/ubifs/replay.c @@ -627,8 +627,7 @@ static int replay_bud(struct ubifs_info *c, int lnum, int offs, int jhead, ubifs_assert(sleb->endpt - offs >= used); ubifs_assert(sleb->endpt % c->min_io_size == 0); - if (sleb->endpt + c->min_io_size <= c->leb_size && - !(c->vfs_sb->s_flags & MS_RDONLY)) + if (sleb->endpt + c->min_io_size <= c->leb_size && !c->ro_mount) err = ubifs_wbuf_seek_nolock(&c->jheads[jhead].wbuf, lnum, sleb->endpt, UBI_SHORTTERM); @@ -840,6 +839,11 @@ static int replay_log_leb(struct ubifs_info *c, int lnum, int offs, void *sbuf) if (IS_ERR(sleb)) { if (PTR_ERR(sleb) != -EUCLEAN || !c->need_recovery) return PTR_ERR(sleb); + /* + * Note, the below function will recover this log LEB only if + * it is the last, because unclean reboots can possibly corrupt + * only the tail of the log. + */ sleb = ubifs_recover_log_leb(c, lnum, offs, sbuf); if (IS_ERR(sleb)) return PTR_ERR(sleb); @@ -851,7 +855,6 @@ static int replay_log_leb(struct ubifs_info *c, int lnum, int offs, void *sbuf) } node = sleb->buf; - snod = list_entry(sleb->nodes.next, struct ubifs_scan_node, list); if (c->cs_sqnum == 0) { /* @@ -898,7 +901,6 @@ static int replay_log_leb(struct ubifs_info *c, int lnum, int offs, void *sbuf) } list_for_each_entry(snod, &sleb->nodes, list) { - cond_resched(); if (snod->sqnum >= SQNUM_WATERMARK) { @@ -1011,7 +1013,6 @@ out: int ubifs_replay_journal(struct ubifs_info *c) { int err, i, lnum, offs, free; - void *sbuf = NULL; BUILD_BUG_ON(UBIFS_TRUN_KEY > 5); @@ -1026,14 +1027,8 @@ int ubifs_replay_journal(struct ubifs_info *c) return -EINVAL; } - sbuf = vmalloc(c->leb_size); - if (!sbuf) - return -ENOMEM; - dbg_mnt("start replaying the journal"); - c->replaying = 1; - lnum = c->ltail_lnum = c->lhead_lnum; offs = c->lhead_offs; @@ -1046,7 +1041,7 @@ int ubifs_replay_journal(struct ubifs_info *c) lnum = UBIFS_LOG_LNUM; offs = 0; } - err = replay_log_leb(c, lnum, offs, sbuf); + err = replay_log_leb(c, lnum, offs, c->sbuf); if (err == 1) /* We hit the end of the log */ break; @@ -1079,7 +1074,6 @@ int ubifs_replay_journal(struct ubifs_info *c) out: destroy_replay_tree(c); destroy_bud_list(c); - vfree(sbuf); c->replaying = 0; return err; } diff --git a/fs/ubifs/sb.c b/fs/ubifs/sb.c index 96cb62c8a9dd..bf31b4729e51 100644 --- a/fs/ubifs/sb.c +++ b/fs/ubifs/sb.c @@ -542,11 +542,8 @@ int ubifs_read_superblock(struct ubifs_info *c) * due to the unavailability of time-travelling equipment. */ if (c->fmt_version > UBIFS_FORMAT_VERSION) { - struct super_block *sb = c->vfs_sb; - int mounting_ro = sb->s_flags & MS_RDONLY; - - ubifs_assert(!c->ro_media || mounting_ro); - if (!mounting_ro || + ubifs_assert(!c->ro_media || c->ro_mount); + if (!c->ro_mount || c->ro_compat_version > UBIFS_RO_COMPAT_VERSION) { ubifs_err("on-flash format version is w%d/r%d, but " "software only supports up to version " @@ -624,7 +621,7 @@ int ubifs_read_superblock(struct ubifs_info *c) c->old_leb_cnt = c->leb_cnt; if (c->leb_cnt < c->vi.size && c->leb_cnt < c->max_leb_cnt) { c->leb_cnt = min_t(int, c->max_leb_cnt, c->vi.size); - if (c->vfs_sb->s_flags & MS_RDONLY) + if (c->ro_mount) dbg_mnt("Auto resizing (ro) from %d LEBs to %d LEBs", c->old_leb_cnt, c->leb_cnt); else { diff --git a/fs/ubifs/scan.c b/fs/ubifs/scan.c index 96c525384191..3e1ee57dbeaa 100644 --- a/fs/ubifs/scan.c +++ b/fs/ubifs/scan.c @@ -197,7 +197,7 @@ int ubifs_add_snod(const struct ubifs_info *c, struct ubifs_scan_leb *sleb, struct ubifs_ino_node *ino = buf; struct ubifs_scan_node *snod; - snod = kzalloc(sizeof(struct ubifs_scan_node), GFP_NOFS); + snod = kmalloc(sizeof(struct ubifs_scan_node), GFP_NOFS); if (!snod) return -ENOMEM; @@ -212,13 +212,15 @@ int ubifs_add_snod(const struct ubifs_info *c, struct ubifs_scan_leb *sleb, case UBIFS_DENT_NODE: case UBIFS_XENT_NODE: case UBIFS_DATA_NODE: - case UBIFS_TRUN_NODE: /* * The key is in the same place in all keyed * nodes. */ key_read(c, &ino->key, &snod->key); break; + default: + invalid_key_init(c, &snod->key); + break; } list_add_tail(&snod->list, &sleb->nodes); sleb->nodes_cnt += 1; diff --git a/fs/ubifs/shrinker.c b/fs/ubifs/shrinker.c index 0b201114a5ad..46961c003236 100644 --- a/fs/ubifs/shrinker.c +++ b/fs/ubifs/shrinker.c @@ -250,7 +250,7 @@ static int kick_a_thread(void) dirty_zn_cnt = atomic_long_read(&c->dirty_zn_cnt); if (!dirty_zn_cnt || c->cmt_state == COMMIT_BROKEN || - c->ro_media) { + c->ro_mount || c->ro_error) { mutex_unlock(&c->umount_mutex); continue; } diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c index cd5900b85d38..9a47c9f0ad07 100644 --- a/fs/ubifs/super.c +++ b/fs/ubifs/super.c @@ -1137,11 +1137,11 @@ static int check_free_space(struct ubifs_info *c) */ static int mount_ubifs(struct ubifs_info *c) { - struct super_block *sb = c->vfs_sb; - int err, mounted_read_only = (sb->s_flags & MS_RDONLY); + int err; long long x; size_t sz; + c->ro_mount = !!(c->vfs_sb->s_flags & MS_RDONLY); err = init_constants_early(c); if (err) return err; @@ -1154,7 +1154,7 @@ static int mount_ubifs(struct ubifs_info *c) if (err) goto out_free; - if (c->empty && (mounted_read_only || c->ro_media)) { + if (c->empty && (c->ro_mount || c->ro_media)) { /* * This UBI volume is empty, and read-only, or the file system * is mounted read-only - we cannot format it. @@ -1165,7 +1165,7 @@ static int mount_ubifs(struct ubifs_info *c) goto out_free; } - if (c->ro_media && !mounted_read_only) { + if (c->ro_media && !c->ro_mount) { ubifs_err("cannot mount read-write - read-only media"); err = -EROFS; goto out_free; @@ -1185,7 +1185,7 @@ static int mount_ubifs(struct ubifs_info *c) if (!c->sbuf) goto out_free; - if (!mounted_read_only) { + if (!c->ro_mount) { c->ileb_buf = vmalloc(c->leb_size); if (!c->ileb_buf) goto out_free; @@ -1228,7 +1228,7 @@ static int mount_ubifs(struct ubifs_info *c) } sprintf(c->bgt_name, BGT_NAME_PATTERN, c->vi.ubi_num, c->vi.vol_id); - if (!mounted_read_only) { + if (!c->ro_mount) { err = alloc_wbufs(c); if (err) goto out_cbuf; @@ -1254,12 +1254,12 @@ static int mount_ubifs(struct ubifs_info *c) if ((c->mst_node->flags & cpu_to_le32(UBIFS_MST_DIRTY)) != 0) { ubifs_msg("recovery needed"); c->need_recovery = 1; - if (!mounted_read_only) { + if (!c->ro_mount) { err = ubifs_recover_inl_heads(c, c->sbuf); if (err) goto out_master; } - } else if (!mounted_read_only) { + } else if (!c->ro_mount) { /* * Set the "dirty" flag so that if we reboot uncleanly we * will notice this immediately on the next mount. @@ -1270,7 +1270,7 @@ static int mount_ubifs(struct ubifs_info *c) goto out_master; } - err = ubifs_lpt_init(c, 1, !mounted_read_only); + err = ubifs_lpt_init(c, 1, !c->ro_mount); if (err) goto out_lpt; @@ -1285,11 +1285,11 @@ static int mount_ubifs(struct ubifs_info *c) /* Calculate 'min_idx_lebs' after journal replay */ c->min_idx_lebs = ubifs_calc_min_idx_lebs(c); - err = ubifs_mount_orphans(c, c->need_recovery, mounted_read_only); + err = ubifs_mount_orphans(c, c->need_recovery, c->ro_mount); if (err) goto out_orphans; - if (!mounted_read_only) { + if (!c->ro_mount) { int lnum; err = check_free_space(c); @@ -1351,7 +1351,7 @@ static int mount_ubifs(struct ubifs_info *c) spin_unlock(&ubifs_infos_lock); if (c->need_recovery) { - if (mounted_read_only) + if (c->ro_mount) ubifs_msg("recovery deferred"); else { c->need_recovery = 0; @@ -1378,7 +1378,7 @@ static int mount_ubifs(struct ubifs_info *c) ubifs_msg("mounted UBI device %d, volume %d, name \"%s\"", c->vi.ubi_num, c->vi.vol_id, c->vi.name); - if (mounted_read_only) + if (c->ro_mount) ubifs_msg("mounted read-only"); x = (long long)c->main_lebs * c->leb_size; ubifs_msg("file system size: %lld bytes (%lld KiB, %lld MiB, %d " @@ -1640,7 +1640,7 @@ static int ubifs_remount_rw(struct ubifs_info *c) } dbg_gen("re-mounted read-write"); - c->vfs_sb->s_flags &= ~MS_RDONLY; + c->ro_mount = 0; c->remounting_rw = 0; c->always_chk_crc = 0; err = dbg_check_space_info(c); @@ -1676,7 +1676,7 @@ static void ubifs_remount_ro(struct ubifs_info *c) int i, err; ubifs_assert(!c->need_recovery); - ubifs_assert(!(c->vfs_sb->s_flags & MS_RDONLY)); + ubifs_assert(!c->ro_mount); mutex_lock(&c->umount_mutex); if (c->bgt) { @@ -1686,10 +1686,8 @@ static void ubifs_remount_ro(struct ubifs_info *c) dbg_save_space_info(c); - for (i = 0; i < c->jhead_cnt; i++) { + for (i = 0; i < c->jhead_cnt; i++) ubifs_wbuf_sync(&c->jheads[i].wbuf); - hrtimer_cancel(&c->jheads[i].wbuf.timer); - } c->mst_node->flags &= ~cpu_to_le32(UBIFS_MST_DIRTY); c->mst_node->flags |= cpu_to_le32(UBIFS_MST_NO_ORPHS); @@ -1704,6 +1702,7 @@ static void ubifs_remount_ro(struct ubifs_info *c) vfree(c->ileb_buf); c->ileb_buf = NULL; ubifs_lpt_free(c, 1); + c->ro_mount = 1; err = dbg_check_space_info(c); if (err) ubifs_ro_mode(c, err); @@ -1735,7 +1734,7 @@ static void ubifs_put_super(struct super_block *sb) * the mutex is locked. */ mutex_lock(&c->umount_mutex); - if (!(c->vfs_sb->s_flags & MS_RDONLY)) { + if (!c->ro_mount) { /* * First of all kill the background thread to make sure it does * not interfere with un-mounting and freeing resources. @@ -1745,23 +1744,22 @@ static void ubifs_put_super(struct super_block *sb) c->bgt = NULL; } - /* Synchronize write-buffers */ - if (c->jheads) - for (i = 0; i < c->jhead_cnt; i++) - ubifs_wbuf_sync(&c->jheads[i].wbuf); - /* - * On fatal errors c->ro_media is set to 1, in which case we do + * On fatal errors c->ro_error is set to 1, in which case we do * not write the master node. */ - if (!c->ro_media) { + if (!c->ro_error) { + int err; + + /* Synchronize write-buffers */ + for (i = 0; i < c->jhead_cnt; i++) + ubifs_wbuf_sync(&c->jheads[i].wbuf); + /* * We are being cleanly unmounted which means the * orphans were killed - indicate this in the master * node. Also save the reserved GC LEB number. */ - int err; - c->mst_node->flags &= ~cpu_to_le32(UBIFS_MST_DIRTY); c->mst_node->flags |= cpu_to_le32(UBIFS_MST_NO_ORPHS); c->mst_node->gc_lnum = cpu_to_le32(c->gc_lnum); @@ -1774,6 +1772,10 @@ static void ubifs_put_super(struct super_block *sb) */ ubifs_err("failed to write master node, " "error %d", err); + } else { + for (i = 0; i < c->jhead_cnt; i++) + /* Make sure write-buffer timers are canceled */ + hrtimer_cancel(&c->jheads[i].wbuf.timer); } } @@ -1797,17 +1799,21 @@ static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data) return err; } - if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) { + if (c->ro_mount && !(*flags & MS_RDONLY)) { + if (c->ro_error) { + ubifs_msg("cannot re-mount R/W due to prior errors"); + return -EROFS; + } if (c->ro_media) { - ubifs_msg("cannot re-mount due to prior errors"); + ubifs_msg("cannot re-mount R/W - UBI volume is R/O"); return -EROFS; } err = ubifs_remount_rw(c); if (err) return err; - } else if (!(sb->s_flags & MS_RDONLY) && (*flags & MS_RDONLY)) { - if (c->ro_media) { - ubifs_msg("cannot re-mount due to prior errors"); + } else if (!c->ro_mount && (*flags & MS_RDONLY)) { + if (c->ro_error) { + ubifs_msg("cannot re-mount R/O due to prior errors"); return -EROFS; } ubifs_remount_ro(c); @@ -2049,8 +2055,8 @@ static int ubifs_get_sb(struct file_system_type *fs_type, int flags, */ ubi = open_ubi(name, UBI_READONLY); if (IS_ERR(ubi)) { - ubifs_err("cannot open \"%s\", error %d", - name, (int)PTR_ERR(ubi)); + dbg_err("cannot open \"%s\", error %d", + name, (int)PTR_ERR(ubi)); return PTR_ERR(ubi); } ubi_get_volume_info(ubi, &vi); @@ -2064,9 +2070,11 @@ static int ubifs_get_sb(struct file_system_type *fs_type, int flags, } if (sb->s_root) { + struct ubifs_info *c1 = sb->s_fs_info; + /* A new mount point for already mounted UBIFS */ dbg_gen("this ubi volume is already mounted"); - if ((flags ^ sb->s_flags) & MS_RDONLY) { + if (!!(flags & MS_RDONLY) != c1->ro_mount) { err = -EBUSY; goto out_deact; } diff --git a/fs/ubifs/tnc.c b/fs/ubifs/tnc.c index 2194915220e5..ad9cf0133622 100644 --- a/fs/ubifs/tnc.c +++ b/fs/ubifs/tnc.c @@ -1177,6 +1177,7 @@ int ubifs_lookup_level0(struct ubifs_info *c, const union ubifs_key *key, unsigned long time = get_seconds(); dbg_tnc("search key %s", DBGKEY(key)); + ubifs_assert(key_type(c, key) < UBIFS_INVALID_KEY); znode = c->zroot.znode; if (unlikely(!znode)) { @@ -2966,7 +2967,7 @@ static struct ubifs_znode *right_znode(struct ubifs_info *c, * * This function searches an indexing node by its first key @key and its * address @lnum:@offs. It looks up the indexing tree by pulling all indexing - * nodes it traverses to TNC. This function is called fro indexing nodes which + * nodes it traverses to TNC. This function is called for indexing nodes which * were found on the media by scanning, for example when garbage-collecting or * when doing in-the-gaps commit. This means that the indexing node which is * looked for does not have to have exactly the same leftmost key @key, because @@ -2988,6 +2989,8 @@ static struct ubifs_znode *lookup_znode(struct ubifs_info *c, struct ubifs_znode *znode, *zn; int n, nn; + ubifs_assert(key_type(c, key) < UBIFS_INVALID_KEY); + /* * The arguments have probably been read off flash, so don't assume * they are valid. diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h index 0c9876b396dd..381d6b207a52 100644 --- a/fs/ubifs/ubifs.h +++ b/fs/ubifs/ubifs.h @@ -119,8 +119,12 @@ * in TNC. However, when replaying, it is handy to introduce fake "truncation" * keys for truncation nodes because the code becomes simpler. So we define * %UBIFS_TRUN_KEY type. + * + * But otherwise, out of the journal reply scope, the truncation keys are + * invalid. */ -#define UBIFS_TRUN_KEY UBIFS_KEY_TYPES_CNT +#define UBIFS_TRUN_KEY UBIFS_KEY_TYPES_CNT +#define UBIFS_INVALID_KEY UBIFS_KEY_TYPES_CNT /* * How much a directory entry/extended attribute entry adds to the parent/host @@ -1028,6 +1032,8 @@ struct ubifs_debug_info; * @max_leb_cnt: maximum count of logical eraseblocks * @old_leb_cnt: count of logical eraseblocks before re-size * @ro_media: the underlying UBI volume is read-only + * @ro_mount: the file-system was mounted as read-only + * @ro_error: UBIFS switched to R/O mode because an error happened * * @dirty_pg_cnt: number of dirty pages (not used) * @dirty_zn_cnt: number of dirty znodes @@ -1168,11 +1174,14 @@ struct ubifs_debug_info; * @replay_sqnum: sequence number of node currently being replayed * @need_recovery: file-system needs recovery * @replaying: set to %1 during journal replay - * @unclean_leb_list: LEBs to recover when mounting ro to rw - * @rcvrd_mst_node: recovered master node to write when mounting ro to rw + * @unclean_leb_list: LEBs to recover when re-mounting R/O mounted FS to R/W + * mode + * @rcvrd_mst_node: recovered master node to write when re-mounting R/O mounted + * FS to R/W mode * @size_tree: inode size information for recovery - * @remounting_rw: set while remounting from ro to rw (sb flags have MS_RDONLY) - * @always_chk_crc: always check CRCs (while mounting and remounting rw) + * @remounting_rw: set while re-mounting from R/O mode to R/W mode + * @always_chk_crc: always check CRCs (while mounting and remounting to R/W + * mode) * @mount_opts: UBIFS-specific mount options * * @dbg: debugging-related information @@ -1268,7 +1277,9 @@ struct ubifs_info { int leb_cnt; int max_leb_cnt; int old_leb_cnt; - int ro_media; + unsigned int ro_media:1; + unsigned int ro_mount:1; + unsigned int ro_error:1; atomic_long_t dirty_pg_cnt; atomic_long_t dirty_zn_cnt;