1 files changed, 178 insertions, 176 deletions

diff --git a/fs/ubifs/recovery.c b/fs/ubifs/recovery.c
index 3dbad6f..731d9e2 100644
--- a/fs/ubifs/recovery.c
+++ b/fs/ubifs/recovery.c
@@ -564,13 +564,16 @@ static int fix_unclean_leb(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
 }
 
 /**
- * drop_incomplete_group - drop nodes from an incomplete group.
+ * drop_last_node - drop the last node or group of nodes.
  * @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 function returns %1 if nodes are dropped and %0 otherwise.
+ * This is a helper function for 'ubifs_recover_leb()' which drops the last
+ * node of the scanned LEB or the last group of nodes if @grouped is not zero.
+ * This function returns %1 if a node was dropped and %0 otherwise.
  */
-static int drop_incomplete_group(struct ubifs_scan_leb *sleb, int *offs)
+static int drop_last_node(struct ubifs_scan_leb *sleb, int *offs, int grouped)
 {
 	int dropped = 0;
 
@@ -589,6 +592,8 @@ static int drop_incomplete_group(struct ubifs_scan_leb *sleb, int *offs)
 		kfree(snod);
 		sleb->nodes_cnt -= 1;
 		dropped = 1;
+		if (!grouped)
+			break;
 	}
 	return dropped;
 }
@@ -609,8 +614,7 @@ static int drop_incomplete_group(struct ubifs_scan_leb *sleb, int *offs)
 struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum,
 					 int offs, void *sbuf, int grouped)
 {
-	int err, len = c->leb_size - offs, need_clean = 0, quiet = 1;
-	int empty_chkd = 0, start = offs;
+	int ret = 0, err, len = c->leb_size - offs, start = offs, min_io_unit;
 	struct ubifs_scan_leb *sleb;
 	void *buf = sbuf + offs;
 
@@ -620,12 +624,8 @@ struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum,
 	if (IS_ERR(sleb))
 		return sleb;
 
-	if (sleb->ecc)
-		need_clean = 1;
-
+	ubifs_assert(len >= 8);
 	while (len >= 8) {
-		int ret;
-
 		dbg_scan("look at LEB %d:%d (%d bytes left)",
 			 lnum, offs, len);
 
@@ -635,8 +635,7 @@ struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum,
 		 * Scan quietly until there is an error from which we cannot
 		 * recover
 		 */
-		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, quiet);
-
+		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 0);
 		if (ret == SCANNED_A_NODE) {
 			/* A valid node, and not a padding node */
 			struct ubifs_ch *ch = buf;
@@ -649,70 +648,32 @@ struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum,
 			offs += node_len;
 			buf += node_len;
 			len -= node_len;
-			continue;
-		}
-
-		if (ret > 0) {
+		} else if (ret > 0) {
 			/* Padding bytes or a valid padding node */
 			offs += ret;
 			buf += ret;
 			len -= ret;
-			continue;
-		}
-
-		if (ret == SCANNED_EMPTY_SPACE) {
-			if (!is_empty(buf, len)) {
-				if (!is_last_write(c, buf, offs))
-					break;
-				clean_buf(c, &buf, lnum, &offs, &len);
-				need_clean = 1;
-			}
-			empty_chkd = 1;
+		} 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", ret);
 			break;
-		}
-
-		if (ret == SCANNED_GARBAGE || ret == SCANNED_A_BAD_PAD_NODE)
-			if (is_last_write(c, buf, offs)) {
-				clean_buf(c, &buf, lnum, &offs, &len);
-				need_clean = 1;
-				empty_chkd = 1;
-				break;
-			}
-
-		if (ret == SCANNED_A_CORRUPT_NODE)
-			if (no_more_nodes(c, buf, len, lnum, offs)) {
-				clean_buf(c, &buf, lnum, &offs, &len);
-				need_clean = 1;
-				empty_chkd = 1;
-				break;
-			}
-
-		if (quiet) {
-			/* Redo the last scan but noisily */
-			quiet = 0;
-			continue;
-		}
-
-		switch (ret) {
-		case SCANNED_GARBAGE:
-			dbg_err("garbage");
-			goto corrupted;
-		case SCANNED_A_CORRUPT_NODE:
-		case SCANNED_A_BAD_PAD_NODE:
-			dbg_err("bad node");
-			goto corrupted;
-		default:
-			dbg_err("unknown");
+		} else {
+			dbg_err("unexpected return value %d", ret);
 			err = -EINVAL;
 			goto error;
 		}
 	}
 
-	if (!empty_chkd && !is_empty(buf, len)) {
-		if (is_last_write(c, buf, offs)) {
-			clean_buf(c, &buf, lnum, &offs, &len);
-			need_clean = 1;
-		} else {
+	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)) {
 			int corruption = first_non_ff(buf, len);
 
 			/*
@@ -728,29 +689,82 @@ struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum,
 		}
 	}
 
-	/* Drop nodes from incomplete group */
-	if (grouped && drop_incomplete_group(sleb, &offs)) {
-		buf = sbuf + offs;
-		len = c->leb_size - offs;
-		clean_buf(c, &buf, lnum, &offs, &len);
-		need_clean = 1;
-	}
+	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_node(sleb, &offs, 1);
 
-	if (offs % c->min_io_size) {
-		clean_buf(c, &buf, lnum, &offs, &len);
-		need_clean = 1;
-	}
+	/*
+	 * 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 node 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? Whey 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. The description follows, but it is worth noting that that is
+	 * only about the GC head, so we could do this trick only if the bud
+	 * belongs to the GC head, but it does not seem to be worth an
+	 * additional "if" statement.
+	 *
+	 * So, imagine the file-system is full, we run GC which is 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 (min_io_unit == round_down(offs, c->min_io_size) &&
+	       min_io_unit != offs &&
+	       drop_last_node(sleb, &offs, grouped));
+
+	buf = sbuf + offs;
+	len = c->leb_size - offs;
 
+	clean_buf(c, &buf, lnum, &offs, &len);
 	ubifs_end_scan(c, sleb, lnum, offs);
 
-	if (need_clean) {
-		err = fix_unclean_leb(c, sleb, start);
-		if (err)
-			goto error;
-	}
+	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);
 	err = -EUCLEAN;
@@ -1070,6 +1084,53 @@ int ubifs_clean_lebs(const struct ubifs_info *c, void *sbuf)
 }
 
 /**
+ * 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
  *
@@ -1091,71 +1152,26 @@ int ubifs_rcvry_gc_commit(struct ubifs_info *c)
 {
 	struct ubifs_wbuf *wbuf = &c->jheads[GCHD].wbuf;
 	struct ubifs_lprops lp;
-	int lnum, err;
+	int err;
+
+	dbg_rcvry("GC head LEB %d, offs %d", wbuf->lnum, wbuf->offs);
 
 	c->gc_lnum = -1;
-	if (wbuf->lnum == -1) {
-		dbg_rcvry("no GC head LEB");
-		goto find_free;
-	}
-	/*
-	 * See whether the used space in the dirtiest LEB fits in the GC head
-	 * LEB.
-	 */
-	if (wbuf->offs == c->leb_size) {
-		dbg_rcvry("no room in GC head LEB");
-		goto find_free;
-	}
+	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) {
-		/*
-		 * There are no dirty or empty LEBs subject to here being
-		 * enough for the index. Try to use
-		 * 'ubifs_find_free_leb_for_idx()', which will return any empty
-		 * LEBs (ignoring index requirements). If the index then
-		 * doesn't have enough LEBs the recovery commit will fail -
-		 * which is the  same result anyway i.e. recovery fails. So
-		 * there is no problem ignoring index  requirements and just
-		 * grabbing a free LEB since we have already established there
-		 * is not a dirty LEB we could have used instead.
-		 */
-		if (err == -ENOSPC) {
-			dbg_rcvry("could not find a dirty LEB");
-			goto find_free;
-		}
-		return err;
-	}
-	ubifs_assert(!(lp.flags & LPROPS_INDEX));
-	lnum = lp.lnum;
-	if (lp.free + lp.dirty == c->leb_size) {
-		/* An empty LEB was returned */
-		if (lp.free != c->leb_size) {
-			err = ubifs_change_one_lp(c, lnum, c->leb_size,
-						  0, 0, 0, 0);
-			if (err)
-				return err;
-		}
-		err = ubifs_leb_unmap(c, lnum);
-		if (err)
+		if (err != -ENOSPC)
 			return err;
-		c->gc_lnum = lnum;
-		dbg_rcvry("allocated LEB %d for GC", lnum);
-		/* Run the commit */
-		dbg_rcvry("committing");
-		return ubifs_run_commit(c);
-	}
-	/*
-	 * There was no empty LEB so the used space in the dirtiest LEB must fit
-	 * in the GC head LEB.
-	 */
-	if (lp.free + lp.dirty < wbuf->offs) {
-		dbg_rcvry("LEB %d doesn't fit in GC head LEB %d:%d",
-			  lnum, wbuf->lnum, wbuf->offs);
-		err = ubifs_return_leb(c, lnum);
-		if (err)
-			return err;
-		goto find_free;
+
+		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.
@@ -1164,11 +1180,8 @@ int ubifs_rcvry_gc_commit(struct ubifs_info *c)
 	err = ubifs_run_commit(c);
 	if (err)
 		return err;
-	/*
-	 * The data in the dirtiest LEB fits in the GC head LEB, so do the GC
-	 * - use locking to keep 'ubifs_assert()' happy.
-	 */
-	dbg_rcvry("GC'ing LEB %d", lnum);
+
+	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) {
@@ -1184,37 +1197,17 @@ int ubifs_rcvry_gc_commit(struct ubifs_info *c)
 			err = -EINVAL;
 		return err;
 	}
-	if (err != LEB_RETAINED) {
-		dbg_err("GC returned %d", 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", lnum);
-	return 0;
 
-find_free:
-	/*
-	 * There is no GC head LEB or the free space in the GC head LEB is too
-	 * small, or there are not dirty LEBs. Allocate gc_lnum by calling
-	 * 'ubifs_find_free_leb_for_idx()' so GC is not run.
-	 */
-	lnum = ubifs_find_free_leb_for_idx(c);
-	if (lnum < 0) {
-		dbg_err("could not find an empty LEB");
-		return lnum;
-	}
-	/* And 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("allocated LEB %d for GC", lnum);
-	/* Run the commit */
-	dbg_rcvry("committing");
-	return ubifs_run_commit(c);
+	dbg_rcvry("allocated LEB %d for GC", lp.lnum);
+	return 0;
 }
 
 /**
@@ -1456,7 +1449,7 @@ static int fix_size_in_place(struct ubifs_info *c, struct size_entry *e)
 	err = ubi_leb_change(c->ubi, lnum, c->sbuf, len, UBI_UNKNOWN);
 	if (err)
 		goto out;
-	dbg_rcvry("inode %lu at %d:%d size %lld -> %lld ",
+	dbg_rcvry("inode %lu at %d:%d size %lld -> %lld",
 		  (unsigned long)e->inum, lnum, offs, i_size, e->d_size);
 	return 0;
 
@@ -1505,20 +1498,27 @@ int ubifs_recover_size(struct ubifs_info *c)
 				e->i_size = le64_to_cpu(ino->size);
 			}
 		}
+
 		if (e->exists && e->i_size < e->d_size) {
-			if (!e->inode && c->ro_mount) {
+			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,
-						  e->d_size, inode->i_size);
+						  inode->i_size, e->d_size);
 					inode->i_size = e->d_size;
-					ubifs_inode(inode)->ui_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;
@@ -1533,9 +1533,11 @@ int ubifs_recover_size(struct ubifs_info *c)
 					iput(e->inode);
 			}
 		}
+
 		this = rb_next(this);
 		rb_erase(&e->rb, &c->size_tree);
 		kfree(e);
 	}
+
 	return 0;
 }