diff options
Diffstat (limited to 'fs/ubifs/recovery.c')
| -rw-r--r-- | fs/ubifs/recovery.c | 1572 |
1 files changed, 1572 insertions, 0 deletions
diff --git a/fs/ubifs/recovery.c b/fs/ubifs/recovery.c new file mode 100644 index 00000000..0398d717 --- /dev/null +++ b/fs/ubifs/recovery.c @@ -0,0 +1,1572 @@ +/* + * This file is part of UBIFS. + * + * Copyright (C) 2006-2008 Nokia Corporation + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License version 2 as published by + * the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 51 + * Franklin St, Fifth Floor, Boston, MA 02110-1301 USA + * + * Authors: Adrian Hunter + * Artem Bityutskiy (Битюцкий Артём) + */ + +/* + * This file implements functions needed to recover from unclean un-mounts. + * When UBIFS is mounted, it checks a flag on the master node to determine if + * an un-mount was completed successfully. If not, the process of mounting + * incorporates additional checking and fixing of on-flash data structures. + * UBIFS always cleans away all remnants of an unclean un-mount, so that + * errors do not accumulate. However UBIFS defers recovery if it is mounted + * read-only, and the flash is not modified in that case. + * + * The general UBIFS approach to the recovery is that it recovers from + * corruptions which could be caused by power cuts, but it refuses to recover + * from corruption caused by other reasons. And UBIFS tries to distinguish + * between these 2 reasons of corruptions and silently recover in the former + * case and loudly complain in the latter case. + * + * UBIFS writes only to erased LEBs, so it writes only to the flash space + * containing only 0xFFs. UBIFS also always writes strictly from the beginning + * of the LEB to the end. And UBIFS assumes that the underlying flash media + * writes in @c->max_write_size bytes at a time. + * + * Hence, if UBIFS finds a corrupted node at offset X, it expects only the min. + * I/O unit corresponding to offset X to contain corrupted data, all the + * following min. I/O units have to contain empty space (all 0xFFs). If this is + * not true, the corruption cannot be the result of a power cut, and UBIFS + * refuses to mount. + */ + +#include <linux/crc32.h> +#include <linux/slab.h> +#include "ubifs.h" + +/** + * is_empty - determine whether a buffer is empty (contains all 0xff). + * @buf: buffer to clean + * @len: length of buffer + * + * This function returns %1 if the buffer is empty (contains all 0xff) otherwise + * %0 is returned. + */ +static int is_empty(void *buf, int len) +{ + uint8_t *p = buf; + int i; + + for (i = 0; i < len; i++) + if (*p++ != 0xff) + return 0; + return 1; +} + +/** + * first_non_ff - find offset of the first non-0xff byte. + * @buf: buffer to search in + * @len: length of buffer + * + * This function returns offset of the first non-0xff byte in @buf or %-1 if + * the buffer contains only 0xff bytes. + */ +static int first_non_ff(void *buf, int len) +{ + uint8_t *p = buf; + int i; + + for (i = 0; i < len; i++) + if (*p++ != 0xff) + return i; + return -1; +} + +/** + * get_master_node - get the last valid master node allowing for corruption. + * @c: UBIFS file-system description object + * @lnum: LEB number + * @pbuf: buffer containing the LEB read, is returned here + * @mst: master node, if found, is returned here + * @cor: corruption, if found, is returned here + * + * This function allocates a buffer, reads the LEB into it, and finds and + * returns the last valid master node allowing for one area of corruption. + * The corrupt area, if there is one, must be consistent with the assumption + * that it is the result of an unclean unmount while the master node was being + * written. Under those circumstances, it is valid to use the previously written + * master node. + * + * This function returns %0 on success and a negative error code on failure. + */ +static int get_master_node(const struct ubifs_info *c, int lnum, void **pbuf, + struct ubifs_mst_node **mst, void **cor) +{ + const int sz = c->mst_node_alsz; + int err, offs, len; + void *sbuf, *buf; + + sbuf = vmalloc(c->leb_size); + if (!sbuf) + return -ENOMEM; + + err = ubifs_leb_read(c, lnum, sbuf, 0, c->leb_size, 0); + if (err && err != -EBADMSG) + goto out_free; + + /* Find the first position that is definitely not a node */ + offs = 0; + buf = sbuf; + len = c->leb_size; + while (offs + UBIFS_MST_NODE_SZ <= c->leb_size) { + struct ubifs_ch *ch = buf; + + if (le32_to_cpu(ch->magic) != UBIFS_NODE_MAGIC) + break; + offs += sz; + buf += sz; + len -= sz; + } + /* See if there was a valid master node before that */ + if (offs) { + int ret; + + offs -= sz; + buf -= sz; + len += sz; + ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 1); + if (ret != SCANNED_A_NODE && offs) { + /* Could have been corruption so check one place back */ + offs -= sz; + buf -= sz; + len += sz; + ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 1); + if (ret != SCANNED_A_NODE) + /* + * We accept only one area of corruption because + * we are assuming that it was caused while + * trying to write a master node. + */ + goto out_err; + } + if (ret == SCANNED_A_NODE) { + struct ubifs_ch *ch = buf; + + if (ch->node_type != UBIFS_MST_NODE) + goto out_err; + dbg_rcvry("found a master node at %d:%d", lnum, offs); + *mst = buf; + offs += sz; + buf += sz; + len -= sz; + } + } + /* Check for corruption */ + if (offs < c->leb_size) { + if (!is_empty(buf, min_t(int, len, sz))) { + *cor = buf; + dbg_rcvry("found corruption at %d:%d", lnum, offs); + } + offs += sz; + buf += sz; + len -= sz; + } + /* Check remaining empty space */ + if (offs < c->leb_size) + if (!is_empty(buf, len)) + goto out_err; + *pbuf = sbuf; + return 0; + +out_err: + err = -EINVAL; +out_free: + vfree(sbuf); + *mst = NULL; + *cor = NULL; + return err; +} + +/** + * write_rcvrd_mst_node - write recovered master node. + * @c: UBIFS file-system description object + * @mst: master node + * + * This function returns %0 on success and a negative error code on failure. + */ +static int write_rcvrd_mst_node(struct ubifs_info *c, + struct ubifs_mst_node *mst) +{ + int err = 0, lnum = UBIFS_MST_LNUM, sz = c->mst_node_alsz; + __le32 save_flags; + + dbg_rcvry("recovery"); + + save_flags = mst->flags; + mst->flags |= cpu_to_le32(UBIFS_MST_RCVRY); + + ubifs_prepare_node(c, mst, UBIFS_MST_NODE_SZ, 1); + err = ubifs_leb_change(c, lnum, mst, sz, UBI_SHORTTERM); + if (err) + goto out; + err = ubifs_leb_change(c, lnum + 1, mst, sz, UBI_SHORTTERM); + if (err) + goto out; +out: + mst->flags = save_flags; + return err; +} + +/** + * ubifs_recover_master_node - recover the master node. + * @c: UBIFS file-system description object + * + * This function recovers the master node from corruption that may occur due to + * an unclean unmount. + * + * This function returns %0 on success and a negative error code on failure. + */ +int ubifs_recover_master_node(struct ubifs_info *c) +{ + void *buf1 = NULL, *buf2 = NULL, *cor1 = NULL, *cor2 = NULL; + struct ubifs_mst_node *mst1 = NULL, *mst2 = NULL, *mst; + const int sz = c->mst_node_alsz; + int err, offs1, offs2; + + dbg_rcvry("recovery"); + + err = get_master_node(c, UBIFS_MST_LNUM, &buf1, &mst1, &cor1); + if (err) + goto out_free; + + err = get_master_node(c, UBIFS_MST_LNUM + 1, &buf2, &mst2, &cor2); + if (err) + goto out_free; + + if (mst1) { + offs1 = (void *)mst1 - buf1; + if ((le32_to_cpu(mst1->flags) & UBIFS_MST_RCVRY) && + (offs1 == 0 && !cor1)) { + /* + * mst1 was written by recovery at offset 0 with no + * corruption. + */ + dbg_rcvry("recovery recovery"); + mst = mst1; + } else if (mst2) { + offs2 = (void *)mst2 - buf2; + if (offs1 == offs2) { + /* Same offset, so must be the same */ + if (memcmp((void *)mst1 + UBIFS_CH_SZ, + (void *)mst2 + UBIFS_CH_SZ, + UBIFS_MST_NODE_SZ - UBIFS_CH_SZ)) + goto out_err; + mst = mst1; + } else if (offs2 + sz == offs1) { + /* 1st LEB was written, 2nd was not */ + if (cor1) + goto out_err; + mst = mst1; + } else if (offs1 == 0 && + c->leb_size - offs2 - sz < sz) { + /* 1st LEB was unmapped and written, 2nd not */ + if (cor1) + goto out_err; + mst = mst1; + } else + goto out_err; + } else { + /* + * 2nd LEB was unmapped and about to be written, so + * there must be only one master node in the first LEB + * and no corruption. + */ + if (offs1 != 0 || cor1) + goto out_err; + mst = mst1; + } + } else { + if (!mst2) + goto out_err; + /* + * 1st LEB was unmapped and about to be written, so there must + * be no room left in 2nd LEB. + */ + offs2 = (void *)mst2 - buf2; + if (offs2 + sz + sz <= c->leb_size) + goto out_err; + mst = mst2; + } + + ubifs_msg("recovered master node from LEB %d", + (mst == mst1 ? UBIFS_MST_LNUM : UBIFS_MST_LNUM + 1)); + + memcpy(c->mst_node, mst, UBIFS_MST_NODE_SZ); + + 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) { + err = -ENOMEM; + goto out_free; + } + memcpy(c->rcvrd_mst_node, c->mst_node, UBIFS_MST_NODE_SZ); + + /* + * We had to recover the master node, which means there was an + * unclean reboot. However, it is possible that the master node + * is clean at this point, i.e., %UBIFS_MST_DIRTY is not set. + * E.g., consider the following chain of events: + * + * 1. UBIFS was cleanly unmounted, so the master node is clean + * 2. UBIFS is being mounted R/W and starts changing the master + * node in the first (%UBIFS_MST_LNUM). A power cut happens, + * so this LEB ends up with some amount of garbage at the + * end. + * 3. UBIFS is being mounted R/O. We reach this place and + * recover the master node from the second LEB + * (%UBIFS_MST_LNUM + 1). But we cannot update the media + * because we are being mounted R/O. We have to defer the + * operation. + * 4. However, this master node (@c->mst_node) is marked as + * clean (since the step 1). And if we just return, the + * mount code will be confused and won't recover the master + * node when it is re-mounter R/W later. + * + * Thus, to force the recovery by marking the master node as + * dirty. + */ + c->mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY); + } else { + /* Write the recovered master node */ + c->max_sqnum = le64_to_cpu(mst->ch.sqnum) - 1; + err = write_rcvrd_mst_node(c, c->mst_node); + if (err) + goto out_free; + } + + vfree(buf2); + vfree(buf1); + + return 0; + +out_err: + err = -EINVAL; +out_free: + ubifs_err("failed to recover master node"); + if (mst1) { + dbg_err("dumping first master node"); + dbg_dump_node(c, mst1); + } + if (mst2) { + dbg_err("dumping second master node"); + dbg_dump_node(c, mst2); + } + vfree(buf2); + vfree(buf1); + return err; +} + +/** + * ubifs_write_rcvrd_mst_node - write the recovered master node. + * @c: UBIFS file-system description object + * + * This function writes the master node that was recovered during mounting in + * read-only mode and must now be written because we are remounting rw. + * + * This function returns %0 on success and a negative error code on failure. + */ +int ubifs_write_rcvrd_mst_node(struct ubifs_info *c) +{ + int err; + + if (!c->rcvrd_mst_node) + return 0; + c->rcvrd_mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY); + c->mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY); + err = write_rcvrd_mst_node(c, c->rcvrd_mst_node); + if (err) + return err; + kfree(c->rcvrd_mst_node); + c->rcvrd_mst_node = NULL; + return 0; +} + +/** + * is_last_write - determine if an offset was in the last write to a LEB. + * @c: UBIFS file-system description object + * @buf: buffer to check + * @offs: offset to check + * + * This function returns %1 if @offs was in the last write to the LEB whose data + * is in @buf, otherwise %0 is returned. The determination is made by checking + * for subsequent empty space starting from the next @c->max_write_size + * boundary. + */ +static int is_last_write(const struct ubifs_info *c, void *buf, int offs) +{ + int empty_offs, check_len; + uint8_t *p; + + /* + * Round up to the next @c->max_write_size boundary i.e. @offs is in + * the last wbuf written. After that should be empty space. + */ + empty_offs = ALIGN(offs + 1, c->max_write_size); + check_len = c->leb_size - empty_offs; + p = buf + empty_offs - offs; + return is_empty(p, check_len); +} + +/** + * clean_buf - clean the data from an LEB sitting in a buffer. + * @c: UBIFS file-system description object + * @buf: buffer to clean + * @lnum: LEB number to clean + * @offs: offset from which to clean + * @len: length of buffer + * + * This function pads up to the next min_io_size boundary (if there is one) and + * sets empty space to all 0xff. @buf, @offs and @len are updated to the next + * @c->min_io_size boundary. + */ +static void clean_buf(const struct ubifs_info *c, void **buf, int lnum, + int *offs, int *len) +{ + int empty_offs, pad_len; + + lnum = lnum; + dbg_rcvry("cleaning corruption at %d:%d", lnum, *offs); + + ubifs_assert(!(*offs & 7)); + empty_offs = ALIGN(*offs, c->min_io_size); + pad_len = empty_offs - *offs; + ubifs_pad(c, *buf, pad_len); + *offs += pad_len; + *buf += pad_len; + *len -= pad_len; + memset(*buf, 0xff, c->leb_size - empty_offs); +} + +/** + * no_more_nodes - determine if there are no more nodes in a buffer. + * @c: UBIFS file-system description object + * @buf: buffer to check + * @len: length of buffer + * @lnum: LEB number of the LEB from which @buf was read + * @offs: offset from which @buf was read + * + * This function ensures that the corrupted node at @offs is the last thing + * written to a LEB. This function returns %1 if more data is not found and + * %0 if more data is found. + */ +static int no_more_nodes(const struct ubifs_info *c, void *buf, int len, + int lnum, int offs) +{ + struct ubifs_ch *ch = buf; + int skip, dlen = le32_to_cpu(ch->len); + + /* Check for empty space after the corrupt node's common header */ + skip = ALIGN(offs + UBIFS_CH_SZ, c->max_write_size) - offs; + if (is_empty(buf + skip, len - skip)) + return 1; + /* + * The area after the common header size is not empty, so the common + * header must be intact. Check it. + */ + if (ubifs_check_node(c, buf, lnum, offs, 1, 0) != -EUCLEAN) { + dbg_rcvry("unexpected bad common header at %d:%d", lnum, offs); + return 0; + } + /* Now we know the corrupt node's length we can skip over it */ + skip = ALIGN(offs + dlen, c->max_write_size) - offs; + /* After which there should be empty space */ + if (is_empty(buf + skip, len - skip)) + return 1; + dbg_rcvry("unexpected data at %d:%d", lnum, offs + skip); + return 0; +} + +/** + * fix_unclean_leb - fix an unclean LEB. + * @c: UBIFS file-system description object + * @sleb: scanned LEB information + * @start: offset where scan started + */ +static int fix_unclean_leb(struct ubifs_info *c, struct ubifs_scan_leb *sleb, + int start) +{ + int lnum = sleb->lnum, endpt = start; + + /* Get the end offset of the last node we are keeping */ + if (!list_empty(&sleb->nodes)) { + struct ubifs_scan_node *snod; + + snod = list_entry(sleb->nodes.prev, + struct ubifs_scan_node, list); + endpt = snod->offs + snod->len; + } + + if (c->ro_mount && !c->remounting_rw) { + /* Add to recovery list */ + struct ubifs_unclean_leb *ucleb; + + dbg_rcvry("need to fix LEB %d start %d endpt %d", + lnum, start, sleb->endpt); + ucleb = kzalloc(sizeof(struct ubifs_unclean_leb), GFP_NOFS); + if (!ucleb) + return -ENOMEM; + ucleb->lnum = lnum; + ucleb->endpt = endpt; + list_add_tail(&ucleb->list, &c->unclean_leb_list); + } else { + /* Write the fixed LEB back to flash */ + int err; + + dbg_rcvry("fixing LEB %d start %d endpt %d", + lnum, start, sleb->endpt); + if (endpt == 0) { + err = ubifs_leb_unmap(c, lnum); + if (err) + return err; + } else { + int len = ALIGN(endpt, c->min_io_size); + + if (start) { + err = ubifs_leb_read(c, lnum, sleb->buf, 0, + start, 1); + if (err) + return err; + } + /* Pad to min_io_size */ + if (len > endpt) { + int pad_len = len - ALIGN(endpt, 8); + + if (pad_len > 0) { + void *buf = sleb->buf + len - pad_len; + + ubifs_pad(c, buf, pad_len); + } + } + err = ubifs_leb_change(c, lnum, sleb->buf, len, + UBI_UNKNOWN); + if (err) + return err; + } + } + return 0; +} + +/** + * drop_last_group - drop the last group of nodes. + * @sleb: scanned LEB information + * @offs: offset of dropped nodes is returned here + * + * This is a helper function for 'ubifs_recover_leb()' which drops the last + * group of nodes of the scanned LEB. + */ +static void drop_last_group(struct ubifs_scan_leb *sleb, int *offs) +{ + while (!list_empty(&sleb->nodes)) { + struct ubifs_scan_node *snod; + struct ubifs_ch *ch; + + snod = list_entry(sleb->nodes.prev, struct ubifs_scan_node, + list); + ch = snod->node; + if (ch->group_type != UBIFS_IN_NODE_GROUP) + break; + + dbg_rcvry("dropping grouped node at %d:%d", + sleb->lnum, snod->offs); + *offs = snod->offs; + list_del(&snod->list); + kfree(snod); + sleb->nodes_cnt -= 1; + } +} + +/** + * drop_last_node - drop the last node. + * @sleb: scanned LEB information + * @offs: offset of dropped nodes is returned here + * @grouped: non-zero if whole group of nodes have to be dropped + * + * This is a helper function for 'ubifs_recover_leb()' which drops the last + * node of the scanned LEB. + */ +static void drop_last_node(struct ubifs_scan_leb *sleb, int *offs) +{ + struct ubifs_scan_node *snod; + + if (!list_empty(&sleb->nodes)) { + snod = list_entry(sleb->nodes.prev, struct ubifs_scan_node, + list); + + dbg_rcvry("dropping last node at %d:%d", sleb->lnum, snod->offs); + *offs = snod->offs; + list_del(&snod->list); + kfree(snod); + sleb->nodes_cnt -= 1; + } +} + +/** + * ubifs_recover_leb - scan and recover a LEB. + * @c: UBIFS file-system description object + * @lnum: LEB number + * @offs: offset + * @sbuf: LEB-sized buffer to use + * @jhead: journal head number this LEB belongs to (%-1 if the LEB does not + * belong to any journal head) + * + * 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. + * Returns %0 in case of success, %-EUCLEAN if an unrecoverable corruption is + * found, and a negative error code in case of failure. + */ +struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum, + int offs, void *sbuf, int jhead) +{ + int ret = 0, err, len = c->leb_size - offs, start = offs, min_io_unit; + int grouped = jhead == -1 ? 0 : c->jheads[jhead].grouped; + struct ubifs_scan_leb *sleb; + void *buf = sbuf + offs; + int corruption; + dbg_rcvry("%d:%d, jhead %d, grouped %d", lnum, offs, jhead, grouped); + + sleb = ubifs_start_scan(c, lnum, offs, sbuf); + if (IS_ERR(sleb)) + return sleb; + + ubifs_assert(len >= 8); + while (len >= 8) { + dbg_scan("look at LEB %d:%d (%d bytes left)", + lnum, offs, len); + + cond_resched(); + + /* + * Scan quietly until there is an error from which we cannot + * recover + */ + ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 1); + if (ret == SCANNED_A_NODE) { + /* A valid node, and not a padding node */ + struct ubifs_ch *ch = buf; + int node_len; + + err = ubifs_add_snod(c, sleb, buf, offs); + if (err) + goto error; + node_len = ALIGN(le32_to_cpu(ch->len), 8); + offs += node_len; + buf += node_len; + len -= node_len; + } else if (ret > 0) { + /* Padding bytes or a valid padding node */ + offs += ret; + buf += ret; + len -= ret; + } else if (ret == SCANNED_EMPTY_SPACE || + ret == SCANNED_GARBAGE || + ret == SCANNED_A_BAD_PAD_NODE || + ret == SCANNED_A_CORRUPT_NODE) { + dbg_rcvry("found corruption (%d) at %d:%d", + ret, lnum, offs); + break; + } else { + dbg_err("unexpected return value %d", ret); + err = -EINVAL; + goto error; + } + } + + if (ret == SCANNED_GARBAGE || ret == SCANNED_A_BAD_PAD_NODE) { + if (!is_last_write(c, buf, offs)) + goto corrupted_rescan; + } else if (ret == SCANNED_A_CORRUPT_NODE) { + if (!no_more_nodes(c, buf, len, lnum, offs)) + goto corrupted_rescan; + } else if (!is_empty(buf, len)) { + if (!is_last_write(c, buf, offs)) { + corruption = first_non_ff(buf, len); + + /* + * See header comment for this file for more + * explanations about the reasons we have this check. + */ + ubifs_err("corrupt empty space LEB %d:%d, corruption " + "starts at %d", lnum, offs, corruption); + /* Make sure we dump interesting non-0xFF data */ + offs += corruption; + buf += corruption; + goto corrupted; + } + } +test: + min_io_unit = round_down(offs, c->min_io_size); + if (grouped) + /* + * If nodes are grouped, always drop the incomplete group at + * the end. + */ + drop_last_group(sleb, &offs); + + if (jhead == GCHD) { + /* + * If this LEB belongs to the GC head then while we are in the + * middle of the same min. I/O unit keep dropping nodes. So + * basically, what we want is to make sure that the last min. + * I/O unit where we saw the corruption is dropped completely + * with all the uncorrupted nodes which may possibly sit there. + * + * In other words, let's name the min. I/O unit where the + * corruption starts B, and the previous min. I/O unit A. The + * below code tries to deal with a situation when half of B + * contains valid nodes or the end of a valid node, and the + * second half of B contains corrupted data or garbage. This + * means that UBIFS had been writing to B just before the power + * cut happened. I do not know how realistic is this scenario + * that half of the min. I/O unit had been written successfully + * and the other half not, but this is possible in our 'failure + * mode emulation' infrastructure at least. + * + * So what is the problem, why we need to drop those nodes? Why + * can't we just clean-up the second half of B by putting a + * padding node there? We can, and this works fine with one + * exception which was reproduced with power cut emulation + * testing and happens extremely rarely. + * + * Imagine the file-system is full, we run GC which starts + * moving valid nodes from LEB X to LEB Y (obviously, LEB Y is + * the current GC head LEB). The @c->gc_lnum is -1, which means + * that GC will retain LEB X and will try to continue. Imagine + * that LEB X is currently the dirtiest LEB, and the amount of + * used space in LEB Y is exactly the same as amount of free + * space in LEB X. + * + * And a power cut happens when nodes are moved from LEB X to + * LEB Y. We are here trying to recover LEB Y which is the GC + * head LEB. We find the min. I/O unit B as described above. + * Then we clean-up LEB Y by padding min. I/O unit. And later + * 'ubifs_rcvry_gc_commit()' function fails, because it cannot + * find a dirty LEB which could be GC'd into LEB Y! Even LEB X + * does not match because the amount of valid nodes there does + * not fit the free space in LEB Y any more! And this is + * because of the padding node which we added to LEB Y. The + * user-visible effect of this which I once observed and + * analysed is that we cannot mount the file-system with + * -ENOSPC error. + * + * So obviously, to make sure that situation does not happen we + * should free min. I/O unit B in LEB Y completely and the last + * used min. I/O unit in LEB Y should be A. This is basically + * what the below code tries to do. + */ + while (offs > min_io_unit) + drop_last_node(sleb, &offs); + } + + buf = sbuf + offs; + len = c->leb_size - offs; + + clean_buf(c, &buf, lnum, &offs, &len); + ubifs_end_scan(c, sleb, lnum, offs); + + err = fix_unclean_leb(c, sleb, start); + if (err) + goto error; + + return sleb; + +corrupted_rescan: + /* Re-scan the corrupted data with verbose messages */ + dbg_err("corruptio %d", ret); + ubifs_scan_a_node(c, buf, len, lnum, offs, 1); +corrupted: + ubifs_scanned_corruption(c, lnum, offs, buf); + offs -= corruption; + buf -= corruption; + goto test; + err = -EUCLEAN; +error: + ubifs_err("LEB %d scanning failed", lnum); + ubifs_scan_destroy(sleb); + return ERR_PTR(err); +} + +/** + * get_cs_sqnum - get commit start sequence number. + * @c: UBIFS file-system description object + * @lnum: LEB number of commit start node + * @offs: offset of commit start node + * @cs_sqnum: commit start sequence number is returned here + * + * This function returns %0 on success and a negative error code on failure. + */ +static int get_cs_sqnum(struct ubifs_info *c, int lnum, int offs, + unsigned long long *cs_sqnum) +{ + struct ubifs_cs_node *cs_node = NULL; + int err, ret; + + dbg_rcvry("at %d:%d", lnum, offs); + cs_node = kmalloc(UBIFS_CS_NODE_SZ, GFP_KERNEL); + if (!cs_node) + return -ENOMEM; + if (c->leb_size - offs < UBIFS_CS_NODE_SZ) + goto out_err; + err = ubifs_leb_read(c, lnum, (void *)cs_node, offs, + UBIFS_CS_NODE_SZ, 0); + if (err && err != -EBADMSG) + goto out_free; + ret = ubifs_scan_a_node(c, cs_node, UBIFS_CS_NODE_SZ, lnum, offs, 0); + if (ret != SCANNED_A_NODE) { + dbg_err("Not a valid node"); + goto out_err; + } + if (cs_node->ch.node_type != UBIFS_CS_NODE) { + dbg_err("Node a CS node, type is %d", cs_node->ch.node_type); + goto out_err; + } + if (le64_to_cpu(cs_node->cmt_no) != c->cmt_no) { + dbg_err("CS node cmt_no %llu != current cmt_no %llu", + (unsigned long long)le64_to_cpu(cs_node->cmt_no), + c->cmt_no); + goto out_err; + } + *cs_sqnum = le64_to_cpu(cs_node->ch.sqnum); + dbg_rcvry("commit start sqnum %llu", *cs_sqnum); + kfree(cs_node); + return 0; + +out_err: + err = -EINVAL; +out_free: + ubifs_err("failed to get CS sqnum"); + kfree(cs_node); + return err; +} + +/** + * ubifs_recover_log_leb - scan and recover a log LEB. + * @c: UBIFS file-system description object + * @lnum: LEB number + * @offs: offset + * @sbuf: LEB-sized buffer to use + * + * This function does a scan of a LEB, but caters for errors that might have + * 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. + */ +struct ubifs_scan_leb *ubifs_recover_log_leb(struct ubifs_info *c, int lnum, + int offs, void *sbuf) +{ + struct ubifs_scan_leb *sleb; + int next_lnum; + + dbg_rcvry("LEB %d", lnum); + next_lnum = lnum + 1; + if (next_lnum >= UBIFS_LOG_LNUM + c->log_lebs) + next_lnum = UBIFS_LOG_LNUM; + if (next_lnum != c->ltail_lnum) { + /* + * We can only recover at the end of the log, so check that the + * next log LEB is empty or out of date. + */ + sleb = ubifs_scan(c, next_lnum, 0, sbuf, 0); + if (IS_ERR(sleb)) + return sleb; + if (sleb->nodes_cnt) { + struct ubifs_scan_node *snod; + unsigned long long cs_sqnum = c->cs_sqnum; + + snod = list_entry(sleb->nodes.next, + struct ubifs_scan_node, list); + if (cs_sqnum == 0) { + int err; + + err = get_cs_sqnum(c, lnum, offs, &cs_sqnum); + if (err) { + ubifs_scan_destroy(sleb); + return ERR_PTR(err); + } + } + if (snod->sqnum > cs_sqnum) { + ubifs_err("unrecoverable log corruption " + "in LEB %d", lnum); + ubifs_scan_destroy(sleb); + return ERR_PTR(-EUCLEAN); + } + } + ubifs_scan_destroy(sleb); + } + return ubifs_recover_leb(c, lnum, offs, sbuf, -1); +} + +/** + * recover_head - recover a head. + * @c: UBIFS file-system description object + * @lnum: LEB number of head to recover + * @offs: offset of head to recover + * @sbuf: LEB-sized buffer to use + * + * This function ensures that there is no data on the flash at a head location. + * + * This function returns %0 on success and a negative error code on failure. + */ +static int recover_head(struct ubifs_info *c, int lnum, int offs, void *sbuf) +{ + int len = c->max_write_size, err; + + if (offs + len > c->leb_size) + len = c->leb_size - offs; + + if (!len) + return 0; + + /* Read at the head location and check it is empty flash */ + err = ubifs_leb_read(c, lnum, sbuf, offs, len, 1); + if (err || !is_empty(sbuf, len)) { + dbg_rcvry("cleaning head at %d:%d", lnum, offs); + if (offs == 0) + return ubifs_leb_unmap(c, lnum); + err = ubifs_leb_read(c, lnum, sbuf, 0, offs, 1); + if (err) + return err; + return ubifs_leb_change(c, lnum, sbuf, offs, UBI_UNKNOWN); + } + + return 0; +} + +/** + * ubifs_recover_inl_heads - recover index and LPT heads. + * @c: UBIFS file-system description object + * @sbuf: LEB-sized buffer to use + * + * This function ensures that there is no data on the flash at the index and + * LPT head locations. + * + * This deals with the recovery of a half-completed journal commit. UBIFS is + * careful never to overwrite the last version of the index or the LPT. Because + * the index and LPT are wandering trees, data from a half-completed commit will + * not be referenced anywhere in UBIFS. The data will be either in LEBs that are + * assumed to be empty and will be unmapped anyway before use, or in the index + * and LPT heads. + * + * This function returns %0 on success and a negative error code on failure. + */ +int ubifs_recover_inl_heads(struct ubifs_info *c, void *sbuf) +{ + int err; + + 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); + if (err) + return err; + + dbg_rcvry("checking LPT head at %d:%d", c->nhead_lnum, c->nhead_offs); + err = recover_head(c, c->nhead_lnum, c->nhead_offs, sbuf); + if (err) + return err; + + return 0; +} + +/** + * clean_an_unclean_leb - read and write a LEB to remove corruption. + * @c: UBIFS file-system description object + * @ucleb: unclean LEB information + * @sbuf: LEB-sized buffer to use + * + * This function reads a LEB up to a point pre-determined by the mount recovery, + * checks the nodes, and writes the result back to the flash, thereby cleaning + * off any following corruption, or non-fatal ECC errors. + * + * This function returns %0 on success and a negative error code on failure. + */ +static int clean_an_unclean_leb(struct ubifs_info *c, + struct ubifs_unclean_leb *ucleb, void *sbuf) +{ + int err, lnum = ucleb->lnum, offs = 0, len = ucleb->endpt, quiet = 1; + void *buf = sbuf; + + dbg_rcvry("LEB %d len %d", lnum, len); + + if (len == 0) { + /* Nothing to read, just unmap it */ + err = ubifs_leb_unmap(c, lnum); + if (err) + return err; + return 0; + } + + err = ubifs_leb_read(c, lnum, buf, offs, len, 0); + if (err && err != -EBADMSG) + return err; + + while (len >= 8) { + int ret; + + cond_resched(); + + /* Scan quietly until there is an error */ + ret = ubifs_scan_a_node(c, buf, len, lnum, offs, quiet); + + if (ret == SCANNED_A_NODE) { + /* A valid node, and not a padding node */ + struct ubifs_ch *ch = buf; + int node_len; + + node_len = ALIGN(le32_to_cpu(ch->len), 8); + offs += node_len; + buf += node_len; + len -= node_len; + continue; + } + + if (ret > 0) { + /* Padding bytes or a valid padding node */ + offs += ret; + buf += ret; + len -= ret; + continue; + } + + if (ret == SCANNED_EMPTY_SPACE) { + ubifs_err("unexpected empty space at %d:%d", + lnum, offs); + return -EUCLEAN; + } + + if (quiet) { + /* Redo the last scan but noisily */ + quiet = 0; + continue; + } + + ubifs_scanned_corruption(c, lnum, offs, buf); + return -EUCLEAN; + } + + /* Pad to min_io_size */ + len = ALIGN(ucleb->endpt, c->min_io_size); + if (len > ucleb->endpt) { + int pad_len = len - ALIGN(ucleb->endpt, 8); + + if (pad_len > 0) { + buf = c->sbuf + len - pad_len; + ubifs_pad(c, buf, pad_len); + } + } + + /* Write back the LEB atomically */ + err = ubifs_leb_change(c, lnum, sbuf, len, UBI_UNKNOWN); + if (err) + return err; + + dbg_rcvry("cleaned LEB %d", lnum); + + return 0; +} + +/** + * ubifs_clean_lebs - clean LEBs recovered during read-only mount. + * @c: UBIFS file-system description object + * @sbuf: LEB-sized buffer to use + * + * This function cleans a LEB identified during recovery that needs to be + * written but was not because UBIFS was mounted read-only. This happens when + * remounting to read-write mode. + * + * This function returns %0 on success and a negative error code on failure. + */ +int ubifs_clean_lebs(struct ubifs_info *c, void *sbuf) +{ + dbg_rcvry("recovery"); + while (!list_empty(&c->unclean_leb_list)) { + struct ubifs_unclean_leb *ucleb; + int err; + + ucleb = list_entry(c->unclean_leb_list.next, + struct ubifs_unclean_leb, list); + err = clean_an_unclean_leb(c, ucleb, sbuf); + if (err) + return err; + list_del(&ucleb->list); + kfree(ucleb); + } + return 0; +} + +/** + * grab_empty_leb - grab an empty LEB to use as GC LEB and run commit. + * @c: UBIFS file-system description object + * + * This is a helper function for 'ubifs_rcvry_gc_commit()' which grabs an empty + * LEB to be used as GC LEB (@c->gc_lnum), and then runs the commit. Returns + * zero in case of success and a negative error code in case of failure. + */ +static int grab_empty_leb(struct ubifs_info *c) +{ + int lnum, err; + + /* + * Note, it is very important to first search for an empty LEB and then + * run the commit, not vice-versa. The reason is that there might be + * only one empty LEB at the moment, the one which has been the + * @c->gc_lnum just before the power cut happened. During the regular + * UBIFS operation (not now) @c->gc_lnum is marked as "taken", so no + * one but GC can grab it. But at this moment this single empty LEB is + * not marked as taken, so if we run commit - what happens? Right, the + * commit will grab it and write the index there. Remember that the + * index always expands as long as there is free space, and it only + * starts consolidating when we run out of space. + * + * IOW, if we run commit now, we might not be able to find a free LEB + * after this. + */ + lnum = ubifs_find_free_leb_for_idx(c); + if (lnum < 0) { + dbg_err("could not find an empty LEB"); + dbg_dump_lprops(c); + dbg_dump_budg(c, &c->bi); + return lnum; + } + + /* Reset the index flag */ + err = ubifs_change_one_lp(c, lnum, LPROPS_NC, LPROPS_NC, 0, + LPROPS_INDEX, 0); + if (err) + return err; + + c->gc_lnum = lnum; + dbg_rcvry("found empty LEB %d, run commit", lnum); + + return ubifs_run_commit(c); +} + +/** + * ubifs_rcvry_gc_commit - recover the GC LEB number and run the commit. + * @c: UBIFS file-system description object + * + * Out-of-place garbage collection requires always one empty LEB with which to + * start garbage collection. The LEB number is recorded in c->gc_lnum and is + * written to the master node on unmounting. In the case of an unclean unmount + * the value of gc_lnum recorded in the master node is out of date and cannot + * be used. Instead, recovery must allocate an empty LEB for this purpose. + * However, there may not be enough empty space, in which case it must be + * possible to GC the dirtiest LEB into the GC head LEB. + * + * This function also runs the commit which causes the TNC updates from + * size-recovery and orphans to be written to the flash. That is important to + * ensure correct replay order for subsequent mounts. + * + * This function returns %0 on success and a negative error code on failure. + */ +int ubifs_rcvry_gc_commit(struct ubifs_info *c) +{ + struct ubifs_wbuf *wbuf = &c->jheads[GCHD].wbuf; + struct ubifs_lprops lp; + int err; + + dbg_rcvry("GC head LEB %d, offs %d", wbuf->lnum, wbuf->offs); + + c->gc_lnum = -1; + if (wbuf->lnum == -1 || wbuf->offs == c->leb_size) + return grab_empty_leb(c); + + err = ubifs_find_dirty_leb(c, &lp, wbuf->offs, 2); + if (err) { + if (err != -ENOSPC) + return err; + + dbg_rcvry("could not find a dirty LEB"); + return grab_empty_leb(c); + } + + ubifs_assert(!(lp.flags & LPROPS_INDEX)); + ubifs_assert(lp.free + lp.dirty >= wbuf->offs); + + /* + * We run the commit before garbage collection otherwise subsequent + * mounts will see the GC and orphan deletion in a different order. + */ + dbg_rcvry("committing"); + err = ubifs_run_commit(c); + if (err) + return err; + + dbg_rcvry("GC'ing LEB %d", lp.lnum); + mutex_lock_nested(&wbuf->io_mutex, wbuf->jhead); + err = ubifs_garbage_collect_leb(c, &lp); + if (err >= 0) { + int err2 = ubifs_wbuf_sync_nolock(wbuf); + + if (err2) + err = err2; + } + mutex_unlock(&wbuf->io_mutex); + if (err < 0) { + dbg_err("GC failed, error %d", err); + if (err == -EAGAIN) + err = -EINVAL; + return err; + } + + ubifs_assert(err == LEB_RETAINED); + if (err != LEB_RETAINED) + return -EINVAL; + + err = ubifs_leb_unmap(c, c->gc_lnum); + if (err) + return err; + + dbg_rcvry("allocated LEB %d for GC", lp.lnum); + return 0; +} + +/** + * struct size_entry - inode size information for recovery. + * @rb: link in the RB-tree of sizes + * @inum: inode number + * @i_size: size on inode + * @d_size: maximum size based on data nodes + * @exists: indicates whether the inode exists + * @inode: inode if pinned in memory awaiting rw mode to fix it + */ +struct size_entry { + struct rb_node rb; + ino_t inum; + loff_t i_size; + loff_t d_size; + int exists; + struct inode *inode; +}; + +/** + * add_ino - add an entry to the size tree. + * @c: UBIFS file-system description object + * @inum: inode number + * @i_size: size on inode + * @d_size: maximum size based on data nodes + * @exists: indicates whether the inode exists + */ +static int add_ino(struct ubifs_info *c, ino_t inum, loff_t i_size, + loff_t d_size, int exists) +{ + struct rb_node **p = &c->size_tree.rb_node, *parent = NULL; + struct size_entry *e; + + while (*p) { + parent = *p; + e = rb_entry(parent, struct size_entry, rb); + if (inum < e->inum) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + e = kzalloc(sizeof(struct size_entry), GFP_KERNEL); + if (!e) + return -ENOMEM; + + e->inum = inum; + e->i_size = i_size; + e->d_size = d_size; + e->exists = exists; + + rb_link_node(&e->rb, parent, p); + rb_insert_color(&e->rb, &c->size_tree); + + return 0; +} + +/** + * find_ino - find an entry on the size tree. + * @c: UBIFS file-system description object + * @inum: inode number + */ +static struct size_entry *find_ino(struct ubifs_info *c, ino_t inum) +{ + struct rb_node *p = c->size_tree.rb_node; + struct size_entry *e; + + while (p) { + e = rb_entry(p, struct size_entry, rb); + if (inum < e->inum) + p = p->rb_left; + else if (inum > e->inum) + p = p->rb_right; + else + return e; + } + return NULL; +} + +/** + * remove_ino - remove an entry from the size tree. + * @c: UBIFS file-system description object + * @inum: inode number + */ +static void remove_ino(struct ubifs_info *c, ino_t inum) +{ + struct size_entry *e = find_ino(c, inum); + + if (!e) + return; + rb_erase(&e->rb, &c->size_tree); + kfree(e); +} + +/** + * ubifs_destroy_size_tree - free resources related to the size tree. + * @c: UBIFS file-system description object + */ +void ubifs_destroy_size_tree(struct ubifs_info *c) +{ + struct rb_node *this = c->size_tree.rb_node; + struct size_entry *e; + + while (this) { + if (this->rb_left) { + this = this->rb_left; + continue; + } else if (this->rb_right) { + this = this->rb_right; + continue; + } + e = rb_entry(this, struct size_entry, rb); + if (e->inode) + iput(e->inode); + this = rb_parent(this); + if (this) { + if (this->rb_left == &e->rb) + this->rb_left = NULL; + else + this->rb_right = NULL; + } + kfree(e); + } + c->size_tree = RB_ROOT; +} + +/** + * ubifs_recover_size_accum - accumulate inode sizes for recovery. + * @c: UBIFS file-system description object + * @key: node key + * @deletion: node is for a deletion + * @new_size: inode size + * + * This function has two purposes: + * 1) to ensure there are no data nodes that fall outside the inode size + * 2) to ensure there are no data nodes for inodes that do not exist + * To accomplish those purposes, a rb-tree is constructed containing an entry + * for each inode number in the journal that has not been deleted, and recording + * the size from the inode node, the maximum size of any data node (also altered + * by truncations) and a flag indicating a inode number for which no inode node + * was present in the journal. + * + * Note that there is still the possibility that there are data nodes that have + * been committed that are beyond the inode size, however the only way to find + * them would be to scan the entire index. Alternatively, some provision could + * be made to record the size of inodes at the start of commit, which would seem + * very cumbersome for a scenario that is quite unlikely and the only negative + * consequence of which is wasted space. + * + * This functions returns %0 on success and a negative error code on failure. + */ +int ubifs_recover_size_accum(struct ubifs_info *c, union ubifs_key *key, + int deletion, loff_t new_size) +{ + ino_t inum = key_inum(c, key); + struct size_entry *e; + int err; + + switch (key_type(c, key)) { + case UBIFS_INO_KEY: + if (deletion) + remove_ino(c, inum); + else { + e = find_ino(c, inum); + if (e) { + e->i_size = new_size; + e->exists = 1; + } else { + err = add_ino(c, inum, new_size, 0, 1); + if (err) + return err; + } + } + break; + case UBIFS_DATA_KEY: + e = find_ino(c, inum); + if (e) { + if (new_size > e->d_size) + e->d_size = new_size; + } else { + err = add_ino(c, inum, 0, new_size, 0); + if (err) + return err; + } + break; + case UBIFS_TRUN_KEY: + e = find_ino(c, inum); + if (e) + e->d_size = new_size; + break; + } + return 0; +} + +/** + * fix_size_in_place - fix inode size in place on flash. + * @c: UBIFS file-system description object + * @e: inode size information for recovery + */ +static int fix_size_in_place(struct ubifs_info *c, struct size_entry *e) +{ + struct ubifs_ino_node *ino = c->sbuf; + unsigned char *p; + union ubifs_key key; + int err, lnum, offs, len; + loff_t i_size; + uint32_t crc; + + /* Locate the inode node LEB number and offset */ + ino_key_init(c, &key, e->inum); + err = ubifs_tnc_locate(c, &key, ino, &lnum, &offs); + if (err) + goto out; + /* + * If the size recorded on the inode node is greater than the size that + * was calculated from nodes in the journal then don't change the inode. + */ + i_size = le64_to_cpu(ino->size); + if (i_size >= e->d_size) + return 0; + /* Read the LEB */ + err = ubifs_leb_read(c, lnum, c->sbuf, 0, c->leb_size, 1); + if (err) + goto out; + /* Change the size field and recalculate the CRC */ + ino = c->sbuf + offs; + ino->size = cpu_to_le64(e->d_size); + len = le32_to_cpu(ino->ch.len); + crc = crc32(UBIFS_CRC32_INIT, (void *)ino + 8, len - 8); + ino->ch.crc = cpu_to_le32(crc); + /* Work out where data in the LEB ends and free space begins */ + p = c->sbuf; + len = c->leb_size - 1; + while (p[len] == 0xff) + len -= 1; + len = ALIGN(len + 1, c->min_io_size); + /* Atomically write the fixed LEB back again */ + err = ubifs_leb_change(c, lnum, c->sbuf, len, UBI_UNKNOWN); + if (err) + goto out; + dbg_rcvry("inode %lu at %d:%d size %lld -> %lld", + (unsigned long)e->inum, lnum, offs, i_size, e->d_size); + return 0; + +out: + ubifs_warn("inode %lu failed to fix size %lld -> %lld error %d", + (unsigned long)e->inum, e->i_size, e->d_size, err); + return err; +} + +/** + * ubifs_recover_size - recover inode size. + * @c: UBIFS file-system description object + * + * This function attempts to fix inode size discrepancies identified by the + * 'ubifs_recover_size_accum()' function. + * + * This functions returns %0 on success and a negative error code on failure. + */ +int ubifs_recover_size(struct ubifs_info *c) +{ + struct rb_node *this = rb_first(&c->size_tree); + + while (this) { + struct size_entry *e; + int err; + + e = rb_entry(this, struct size_entry, rb); + if (!e->exists) { + union ubifs_key key; + + ino_key_init(c, &key, e->inum); + err = ubifs_tnc_lookup(c, &key, c->sbuf); + if (err && err != -ENOENT) + return err; + if (err == -ENOENT) { + /* Remove data nodes that have no inode */ + dbg_rcvry("removing ino %lu", + (unsigned long)e->inum); + err = ubifs_tnc_remove_ino(c, e->inum); + if (err) + return err; + } else { + struct ubifs_ino_node *ino = c->sbuf; + + e->exists = 1; + e->i_size = le64_to_cpu(ino->size); + } + } + + if (e->exists && e->i_size < e->d_size) { + if (c->ro_mount) { + /* Fix the inode size and pin it in memory */ + struct inode *inode; + struct ubifs_inode *ui; + + ubifs_assert(!e->inode); + + inode = ubifs_iget(c->vfs_sb, e->inum); + if (IS_ERR(inode)) + return PTR_ERR(inode); + + ui = ubifs_inode(inode); + if (inode->i_size < e->d_size) { + dbg_rcvry("ino %lu size %lld -> %lld", + (unsigned long)e->inum, + inode->i_size, e->d_size); + inode->i_size = e->d_size; + ui->ui_size = e->d_size; + ui->synced_i_size = e->d_size; + e->inode = inode; + this = rb_next(this); + continue; + } + iput(inode); + } else { + /* Fix the size in place */ + err = fix_size_in_place(c, e); + if (err) + return err; + if (e->inode) + iput(e->inode); + } + } + + this = rb_next(this); + rb_erase(&e->rb, &c->size_tree); + kfree(e); + } + + return 0; +} |