UBIFS updates for 2.6.27

UBIFS updates for 2.6.27

[Artem Bityutskiy]

Hi,

Christoph asked me to send UBIFS patches here before I put them
to Linux next.

The following are patches we plan to send to Linus before 2.6.27
is out. Some of the patches are not fixes, but because UBIFS is
new and 2.6.27 will be its first release, we think it is OK to
merge "non-fix" patches for UBIFS as an exception.

Christoph, I do not send NFS patch for now because I'd like to
work on it some more, taking into account your suggestions.

[PATCH] Documentation: fix typo in ubifs.txt

[Artem Bityutskiy]

From: Sebastian Siewior <bigeasy@linutronix.de>

Signed-off-by: Sebastian Siewior <bigeasy@linutronix.de>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 Documentation/filesystems/ubifs.txt |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/Documentation/filesystems/ubifs.txt b/Documentation/filesystems/ubifs.txt
index 540e9e7..6a0d70a 100644
--- a/Documentation/filesystems/ubifs.txt
+++ b/Documentation/filesystems/ubifs.txt
@@ -57,7 +57,7 @@ Similarly to JFFS2, UBIFS supports on-the-flight compression which makes
 it possible to fit quite a lot of data to the flash.
 
 Similarly to JFFS2, UBIFS is tolerant of unclean reboots and power-cuts.
-It does not need stuff like ckfs.ext2. UBIFS automatically replays its
+It does not need stuff like fsck.ext2. UBIFS automatically replays its
 journal and recovers from crashes, ensuring that the on-flash data
 structures are consistent.
 
-- 
1.5.4.1

[PATCH] UBIFS: print volume name as well

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

We encouredge people to mount using volume name, not device
numbers. So print the name of the mounted UBI volume, not just
IDs.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/super.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index ca1e2d4..43af934 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -1122,8 +1122,8 @@ static int mount_ubifs(struct ubifs_info *c)
 	if (err)
 		goto out_infos;
 
-	ubifs_msg("mounted UBI device %d, volume %d", c->vi.ubi_num,
-		  c->vi.vol_id);
+	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)
 		ubifs_msg("mounted read-only");
 	x = (long long)c->main_lebs * c->leb_size;
-- 
1.5.4.1

[PATCH] UBIFS: fix budgeting calculations

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

The 'ubifs_release_dirty_inode_budget()' was buggy and incorrectly
freed the budget, which led to not freeing all dirty data budget.
This patch fixes that.

Also, this patch fixes ubifs_mkdir() which passed 1 in dirty_ino_d,
which makes no sense. Well, it is harmless though.

Also, add few more useful assertions. And improve few debugging
messages.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/budget.c |    5 +++--
 fs/ubifs/dir.c    |    3 +--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index d81fb9e..12a1717 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -686,9 +686,10 @@ void ubifs_convert_page_budget(struct ubifs_info *c)
 void ubifs_release_dirty_inode_budget(struct ubifs_info *c,
 				      struct ubifs_inode *ui)
 {
-	struct ubifs_budget_req req = {.dd_growth = c->inode_budget,
-				       .dirtied_ino_d = ui->data_len};
+	struct ubifs_budget_req req;
 
+	memset(&req, 0, sizeof(struct ubifs_budget_req));
+	req.dd_growth = c->inode_budget + ui->data_len;
 	ubifs_release_budget(c, &req);
 }
 
diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index e90374b..a79e850 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -727,8 +727,7 @@ static int ubifs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 	struct ubifs_inode *dir_ui = ubifs_inode(dir);
 	struct ubifs_info *c = dir->i_sb->s_fs_info;
 	int err, sz_change = CALC_DENT_SIZE(dentry->d_name.len);
-	struct ubifs_budget_req req = { .new_ino = 1, .new_dent = 1,
-					.dirtied_ino_d = 1 };
+	struct ubifs_budget_req req = { .new_ino = 1, .new_dent = 1 };
 
 	/*
 	 * Budget request settings: new inode, new direntry and changing parent
-- 
1.5.4.1

[PATCH] UBIFS: improve debugging

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

1. Print inode mode in some of debugging messages
2. Add few more useful assertions

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/file.c  |    3 ++-
 fs/ubifs/super.c |    6 ++++--
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 8565e58..01598f2 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -941,7 +941,8 @@ int ubifs_setattr(struct dentry *dentry, struct iattr *attr)
 	struct inode *inode = dentry->d_inode;
 	struct ubifs_info *c = inode->i_sb->s_fs_info;
 
-	dbg_gen("ino %lu, ia_valid %#x", inode->i_ino, attr->ia_valid);
+	dbg_gen("ino %lu, mode %#x, ia_valid %#x",
+		inode->i_ino, inode->i_mode, attr->ia_valid);
 	err = inode_change_ok(inode, attr);
 	if (err)
 		return err;
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index 43af934..06e3b22 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -299,7 +299,7 @@ static int ubifs_write_inode(struct inode *inode, int wait)
 		return 0;
 	}
 
-	dbg_gen("inode %lu", inode->i_ino);
+	dbg_gen("inode %lu, mode %#x", inode->i_ino, (int)inode->i_mode);
 	err = ubifs_jnl_write_inode(c, inode, 0);
 	if (err)
 		ubifs_err("can't write inode %lu, error %d", inode->i_ino, err);
@@ -323,9 +323,10 @@ static void ubifs_delete_inode(struct inode *inode)
 		 */
 		goto out;
 
-	dbg_gen("inode %lu", inode->i_ino);
+	dbg_gen("inode %lu, mode %#x", inode->i_ino, (int)inode->i_mode);
 	ubifs_assert(!atomic_read(&inode->i_count));
 	ubifs_assert(inode->i_nlink == 0);
+	ubifs_assert(!ubifs_inode(inode)->dirty);
 
 	truncate_inode_pages(&inode->i_data, 0);
 	if (is_bad_inode(inode))
@@ -1469,6 +1470,7 @@ static void ubifs_put_super(struct super_block *sb)
 	 */
 	ubifs_assert(atomic_long_read(&c->dirty_pg_cnt) == 0);
 	ubifs_assert(c->budg_idx_growth == 0);
+	ubifs_assert(c->budg_dd_growth == 0);
 	ubifs_assert(c->budg_data_growth == 0);
 
 	/*
-- 
1.5.4.1

[PATCH] UBIFS: free budget in delete_inode as well

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Although the inode is marked as clean when it is being deleted,
it might stay and be used as orphan, and be marked as dirty.
So we have to free the budget when we delete it.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/super.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index 06e3b22..884beed 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -314,8 +314,9 @@ static void ubifs_delete_inode(struct inode *inode)
 {
 	int err;
 	struct ubifs_info *c = inode->i_sb->s_fs_info;
+	struct ubifs_inode *ui = ubifs_inode(inode);
 
-	if (ubifs_inode(inode)->xattr)
+	if (ui->xattr)
 		/*
 		 * Extended attribute inode deletions are fully handled in
 		 * 'ubifs_removexattr()'. These inodes are special and have
@@ -326,13 +327,12 @@ static void ubifs_delete_inode(struct inode *inode)
 	dbg_gen("inode %lu, mode %#x", inode->i_ino, (int)inode->i_mode);
 	ubifs_assert(!atomic_read(&inode->i_count));
 	ubifs_assert(inode->i_nlink == 0);
-	ubifs_assert(!ubifs_inode(inode)->dirty);
 
 	truncate_inode_pages(&inode->i_data, 0);
 	if (is_bad_inode(inode))
 		goto out;
 
-	ubifs_inode(inode)->ui_size = inode->i_size = 0;
+	ui->ui_size = inode->i_size = 0;
 	err = ubifs_jnl_write_inode(c, inode, 1);
 	if (err)
 		/*
@@ -341,6 +341,8 @@ static void ubifs_delete_inode(struct inode *inode)
 		 */
 		ubifs_err("can't write inode %lu, error %d", inode->i_ino, err);
 out:
+	if (ui->dirty)
+		ubifs_release_dirty_inode_budget(c, ui);
 	clear_inode(inode);
 }
 
-- 
1.5.4.1

[PATCH] UBIFS: fix error return in failure mode

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

UBIFS recovery testing debug facility simulates media failures.
When simulating an IO error, the error code returned must be
-EIO but it was not always if the user switched off the
debug recovery testing option at the same time.

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
---
 fs/ubifs/debug.c |    7 ++++---
 1 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/ubifs/debug.c b/fs/ubifs/debug.c
index 4e3aaeb..0adfb29 100644
--- a/fs/ubifs/debug.c
+++ b/fs/ubifs/debug.c
@@ -2208,16 +2208,17 @@ int dbg_leb_read(struct ubi_volume_desc *desc, int lnum, char *buf, int offset,
 int dbg_leb_write(struct ubi_volume_desc *desc, int lnum, const void *buf,
 		  int offset, int len, int dtype)
 {
-	int err;
+	int err, failing;
 
 	if (in_failure_mode(desc))
 		return -EIO;
-	if (do_fail(desc, lnum, 1))
+	failing = do_fail(desc, lnum, 1);
+	if (failing)
 		cut_data(buf, len);
 	err = ubi_leb_write(desc, lnum, buf, offset, len, dtype);
 	if (err)
 		return err;
-	if (in_failure_mode(desc))
+	if (failing)
 		return -EIO;
 	return 0;
 }
-- 
1.5.4.1

[PATCH] UBIFS: ensure UBIFS switches to read-only on error

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

UBI transparently handles write errors by automatically copying
and remapping the affected eraseblock. If UBI is unable to do
that, for example its pool of eraseblocks reserved for bad block
handling is empty, then the error is propagated to UBIFS. UBIFS
must protect the media from falling into an inconsistent state
by immediately switching to read-only mode. In the case of log
updates, this was not being done.

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
---
 fs/ubifs/log.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/fs/ubifs/log.c b/fs/ubifs/log.c
index 36857b9..e14829e 100644
--- a/fs/ubifs/log.c
+++ b/fs/ubifs/log.c
@@ -317,6 +317,8 @@ int ubifs_add_bud_to_log(struct ubifs_info *c, int jhead, int lnum, int offs)
 	return 0;
 
 out_unlock:
+	if (err != -EAGAIN)
+		ubifs_ro_mode(c, err);
 	mutex_unlock(&c->log_mutex);
 	kfree(ref);
 	kfree(bud);
-- 
1.5.4.1

[PATCH] UBIFS: make ubifs_ro_mode() not inline

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

We use ubifs_ro_mode() quite a lot, and not in fast-path, so
there is no reason to blow the code up by having it inlined.
Also, we usually want R/O mode change to be seen to other
CPUs as soon as possible, so when we make this a function
call, we will automatically have a memory barrier.

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/io.c    |   14 ++++++++++++++
 fs/ubifs/misc.h  |   14 --------------
 fs/ubifs/ubifs.h |    1 +
 3 files changed, 15 insertions(+), 14 deletions(-)

diff --git a/fs/ubifs/io.c b/fs/ubifs/io.c
index 3374f91..054363f 100644
--- a/fs/ubifs/io.c
+++ b/fs/ubifs/io.c
@@ -54,6 +54,20 @@
 #include "ubifs.h"
 
 /**
+ * ubifs_ro_mode - switch UBIFS to read read-only mode.
+ * @c: UBIFS file-system description object
+ * @err: error code which is the reason of switching to R/O mode
+ */
+void ubifs_ro_mode(struct ubifs_info *c, int err)
+{
+	if (!c->ro_media) {
+		c->ro_media = 1;
+		ubifs_warn("switched to read-only mode, error %d", err);
+		dbg_dump_stack();
+	}
+}
+
+/**
  * ubifs_check_node - check node.
  * @c: UBIFS file-system description object
  * @buf: node to check
diff --git a/fs/ubifs/misc.h b/fs/ubifs/misc.h
index 4beccfc..cd83ffc 100644
--- a/fs/ubifs/misc.h
+++ b/fs/ubifs/misc.h
@@ -80,20 +80,6 @@ static inline struct ubifs_inode *ubifs_inode(const struct inode *inode)
 }
 
 /**
- * ubifs_ro_mode - switch UBIFS to read read-only mode.
- * @c: UBIFS file-system description object
- * @err: error code which is the reason of switching to R/O mode
- */
-static inline void ubifs_ro_mode(struct ubifs_info *c, int err)
-{
-	if (!c->ro_media) {
-		c->ro_media = 1;
-		ubifs_warn("switched to read-only mode, error %d", err);
-		dbg_dump_stack();
-	}
-}
-
-/**
  * ubifs_compr_present - check if compressor was compiled in.
  * @compr_type: compressor type to check
  *
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index e4f89f2..c488d43 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -1346,6 +1346,7 @@ extern struct backing_dev_info ubifs_backing_dev_info;
 extern struct ubifs_compressor *ubifs_compressors[UBIFS_COMPR_TYPES_CNT];
 
 /* io.c */
+void ubifs_ro_mode(struct ubifs_info *c, int err);
 int ubifs_wbuf_write_nolock(struct ubifs_wbuf *wbuf, void *buf, int len);
 int ubifs_wbuf_seek_nolock(struct ubifs_wbuf *wbuf, int lnum, int offs,
 			   int dtype);
-- 
1.5.4.1

[PATCH] UBIFS: do not write orphans back

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Orphan inodes are deleted inodes which will disappear after FS
re-mount. There is not need to write orphan inodes back, because
they are not needed on the flash media.

So optimize orphans a little by not writing them back. Just mark
them as clean, free the budget, and report success to VFS.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/super.c |   18 +++++++++++++-----
 1 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index 884beed..13e90b0 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -278,7 +278,7 @@ static void ubifs_destroy_inode(struct inode *inode)
  */
 static int ubifs_write_inode(struct inode *inode, int wait)
 {
-	int err;
+	int err = 0;
 	struct ubifs_info *c = inode->i_sb->s_fs_info;
 	struct ubifs_inode *ui = ubifs_inode(inode);
 
@@ -299,10 +299,18 @@ static int ubifs_write_inode(struct inode *inode, int wait)
 		return 0;
 	}
 
-	dbg_gen("inode %lu, mode %#x", inode->i_ino, (int)inode->i_mode);
-	err = ubifs_jnl_write_inode(c, inode, 0);
-	if (err)
-		ubifs_err("can't write inode %lu, error %d", inode->i_ino, err);
+	/*
+	 * As an optimization, do not write orphan inodes to the media just
+	 * because this is not needed.
+	 */
+	dbg_gen("inode %lu, mode %#x, nlink %u",
+		inode->i_ino, (int)inode->i_mode, inode->i_nlink);
+	if (inode->i_nlink) {
+		err = ubifs_jnl_write_inode(c, inode, 0);
+		if (err)
+			ubifs_err("can't write inode %lu, error %d",
+				  inode->i_ino, err);
+	}
 
 	ui->dirty = 0;
 	mutex_unlock(&ui->ui_mutex);
-- 
1.5.4.1

[PATCH] UBIFS: remove unneeded function parameter

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Simplify 'ubifs_jnl_write_inode()' by removing the 'deletion'
parameter which is not really needed because we may test
inode->i_nlink and check whether this is a deletion or not.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/journal.c |   19 +++++++------------
 fs/ubifs/super.c   |    4 ++--
 fs/ubifs/ubifs.h   |    3 +--
 3 files changed, 10 insertions(+), 16 deletions(-)

diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 283155a..666ad82 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -750,30 +750,25 @@ out_free:
  * ubifs_jnl_write_inode - flush inode to the journal.
  * @c: UBIFS file-system description object
  * @inode: inode to flush
- * @deletion: inode has been deleted
  *
  * This function writes inode @inode to the journal. If the inode is
  * synchronous, it also synchronizes the write-buffer. Returns zero in case of
  * success and a negative error code in case of failure.
  */
-int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode,
-			  int deletion)
+int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode)
 {
-	int err, len, lnum, offs, sync = 0;
+	int err, lnum, offs;
 	struct ubifs_ino_node *ino;
 	struct ubifs_inode *ui = ubifs_inode(inode);
+	int sync = 0, len = UBIFS_INO_NODE_SZ, last_reference = !inode->i_nlink;
 
-	dbg_jnl("ino %lu%s", inode->i_ino,
-		deletion ? " (last reference)" : "");
-	if (deletion)
-		ubifs_assert(inode->i_nlink == 0);
+	dbg_jnl("ino %lu, nlink %u", inode->i_ino, inode->i_nlink);
 
-	len = UBIFS_INO_NODE_SZ;
 	/*
 	 * If the inode is being deleted, do not write the attached data. No
 	 * need to synchronize the write-buffer either.
 	 */
-	if (!deletion) {
+	if (!last_reference) {
 		len += ui->data_len;
 		sync = IS_SYNC(inode);
 	}
@@ -786,7 +781,7 @@ int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode,
 	if (err)
 		goto out_free;
 
-	pack_inode(c, ino, inode, 1, deletion);
+	pack_inode(c, ino, inode, 1, last_reference);
 	err = write_head(c, BASEHD, ino, len, &lnum, &offs, sync);
 	if (err)
 		goto out_release;
@@ -795,7 +790,7 @@ int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode,
 					  inode->i_ino);
 	release_head(c, BASEHD);
 
-	if (deletion) {
+	if (last_reference) {
 		err = ubifs_tnc_remove_ino(c, inode->i_ino);
 		if (err)
 			goto out_ro;
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index 13e90b0..cf1fb6c 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -306,7 +306,7 @@ static int ubifs_write_inode(struct inode *inode, int wait)
 	dbg_gen("inode %lu, mode %#x, nlink %u",
 		inode->i_ino, (int)inode->i_mode, inode->i_nlink);
 	if (inode->i_nlink) {
-		err = ubifs_jnl_write_inode(c, inode, 0);
+		err = ubifs_jnl_write_inode(c, inode);
 		if (err)
 			ubifs_err("can't write inode %lu, error %d",
 				  inode->i_ino, err);
@@ -341,7 +341,7 @@ static void ubifs_delete_inode(struct inode *inode)
 		goto out;
 
 	ui->ui_size = inode->i_size = 0;
-	err = ubifs_jnl_write_inode(c, inode, 1);
+	err = ubifs_jnl_write_inode(c, inode);
 	if (err)
 		/*
 		 * Worst case we have a lost orphan inode wasting space, so a
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index c488d43..6ddd1de 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -1400,8 +1400,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 		     int deletion, int xent);
 int ubifs_jnl_write_data(struct ubifs_info *c, const struct inode *inode,
 			 const union ubifs_key *key, const void *buf, int len);
-int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode,
-			  int last_reference);
+int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode);
 int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 		     const struct dentry *old_dentry,
 		     const struct inode *new_dir,
-- 
1.5.4.1

[PATCH] UBIFS: remove another unneeded function parameter

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

The 'last_reference' parameter of 'pack_inode()' is not really
needed because 'inode->i_nlink' may be tested instead. Zap it.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/journal.c |   30 ++++++++++++++----------------
 1 files changed, 14 insertions(+), 16 deletions(-)

diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 666ad82..3bc3fc9 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -447,13 +447,11 @@ static int get_dent_type(int mode)
  * @ino: buffer in which to pack inode node
  * @inode: inode to pack
  * @last: indicates the last node of the group
- * @last_reference: non-zero if this is a deletion inode
  */
 static void pack_inode(struct ubifs_info *c, struct ubifs_ino_node *ino,
-		       const struct inode *inode, int last,
-		       int last_reference)
+		       const struct inode *inode, int last)
 {
-	int data_len = 0;
+	int data_len = 0, last_reference = !inode->i_nlink;
 	struct ubifs_inode *ui = ubifs_inode(inode);
 
 	ino->ch.node_type = UBIFS_INO_NODE;
@@ -596,9 +594,9 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 	ubifs_prep_grp_node(c, dent, dlen, 0);
 
 	ino = (void *)dent + aligned_dlen;
-	pack_inode(c, ino, inode, 0, last_reference);
+	pack_inode(c, ino, inode, 0);
 	ino = (void *)ino + aligned_ilen;
-	pack_inode(c, ino, dir, 1, 0);
+	pack_inode(c, ino, dir, 1);
 
 	if (last_reference) {
 		err = ubifs_add_orphan(c, inode->i_ino);
@@ -781,7 +779,7 @@ int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode)
 	if (err)
 		goto out_free;
 
-	pack_inode(c, ino, inode, 1, last_reference);
+	pack_inode(c, ino, inode, 1);
 	err = write_head(c, BASEHD, ino, len, &lnum, &offs, sync);
 	if (err)
 		goto out_release;
@@ -912,16 +910,16 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 
 	p = (void *)dent2 + aligned_dlen2;
 	if (new_inode) {
-		pack_inode(c, p, new_inode, 0, last_reference);
+		pack_inode(c, p, new_inode, 0);
 		p += ALIGN(ilen, 8);
 	}
 
 	if (!move)
-		pack_inode(c, p, old_dir, 1, 0);
+		pack_inode(c, p, old_dir, 1);
 	else {
-		pack_inode(c, p, old_dir, 0, 0);
+		pack_inode(c, p, old_dir, 0);
 		p += ALIGN(plen, 8);
-		pack_inode(c, p, new_dir, 1, 0);
+		pack_inode(c, p, new_dir, 1);
 	}
 
 	if (last_reference) {
@@ -1126,7 +1124,7 @@ int ubifs_jnl_truncate(struct ubifs_info *c, const struct inode *inode,
 	if (err)
 		goto out_free;
 
-	pack_inode(c, ino, inode, 0, 0);
+	pack_inode(c, ino, inode, 0);
 	ubifs_prep_grp_node(c, trun, UBIFS_TRUN_NODE_SZ, dlen ? 0 : 1);
 	if (dlen)
 		ubifs_prep_grp_node(c, dn, dlen, 1);
@@ -1246,9 +1244,9 @@ int ubifs_jnl_delete_xattr(struct ubifs_info *c, const struct inode *host,
 	ubifs_prep_grp_node(c, xent, xlen, 0);
 
 	ino = (void *)xent + aligned_xlen;
-	pack_inode(c, ino, inode, 0, 1);
+	pack_inode(c, ino, inode, 0);
 	ino = (void *)ino + UBIFS_INO_NODE_SZ;
-	pack_inode(c, ino, host, 1, 0);
+	pack_inode(c, ino, host, 1);
 
 	err = write_head(c, BASEHD, xent, len, &lnum, &xent_offs, sync);
 	if (!sync && !err)
@@ -1339,8 +1337,8 @@ int ubifs_jnl_change_xattr(struct ubifs_info *c, const struct inode *inode,
 	if (err)
 		goto out_free;
 
-	pack_inode(c, ino, host, 0, 0);
-	pack_inode(c, (void *)ino + aligned_len1, inode, 1, 0);
+	pack_inode(c, ino, host, 0);
+	pack_inode(c, (void *)ino + aligned_len1, inode, 1);
 
 	err = write_head(c, BASEHD, ino, aligned_len, &lnum, &offs, 0);
 	if (!sync && !err) {
-- 
1.5.4.1

[PATCH] UBIFS: increment commit number earlier

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Increment the commit number at the beginnig of the commit, instead
of doing this after the commit. This is needed for further
optimizations.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/commit.c |    3 ++-
 fs/ubifs/log.c    |    2 +-
 fs/ubifs/orphan.c |    4 ++--
 fs/ubifs/ubifs.h  |    3 ++-
 4 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/fs/ubifs/commit.c b/fs/ubifs/commit.c
index 3b51631..0a6aa2c 100644
--- a/fs/ubifs/commit.c
+++ b/fs/ubifs/commit.c
@@ -74,6 +74,7 @@ static int do_commit(struct ubifs_info *c)
 			goto out_up;
 	}
 
+	c->cmt_no += 1;
 	err = ubifs_gc_start_commit(c);
 	if (err)
 		goto out_up;
@@ -115,7 +116,7 @@ static int do_commit(struct ubifs_info *c)
 		goto out;
 
 	mutex_lock(&c->mst_mutex);
-	c->mst_node->cmt_no      = cpu_to_le64(++c->cmt_no);
+	c->mst_node->cmt_no      = cpu_to_le64(c->cmt_no);
 	c->mst_node->log_lnum    = cpu_to_le32(new_ltail_lnum);
 	c->mst_node->root_lnum   = cpu_to_le32(zroot.lnum);
 	c->mst_node->root_offs   = cpu_to_le32(zroot.offs);
diff --git a/fs/ubifs/log.c b/fs/ubifs/log.c
index e14829e..3e0aa73 100644
--- a/fs/ubifs/log.c
+++ b/fs/ubifs/log.c
@@ -412,7 +412,7 @@ int ubifs_log_start_commit(struct ubifs_info *c, int *ltail_lnum)
 		return -ENOMEM;
 
 	cs->ch.node_type = UBIFS_CS_NODE;
-	cs->cmt_no = cpu_to_le64(c->cmt_no + 1);
+	cs->cmt_no = cpu_to_le64(c->cmt_no);
 	ubifs_prepare_node(c, cs, UBIFS_CS_NODE_SZ, 0);
 
 	/*
diff --git a/fs/ubifs/orphan.c b/fs/ubifs/orphan.c
index 3afeb92..02d3462 100644
--- a/fs/ubifs/orphan.c
+++ b/fs/ubifs/orphan.c
@@ -310,10 +310,10 @@ static int write_orph_node(struct ubifs_info *c, int atomic)
 	c->cmt_orphans -= cnt;
 	spin_unlock(&c->orphan_lock);
 	if (c->cmt_orphans)
-		orph->cmt_no = cpu_to_le64(c->cmt_no + 1);
+		orph->cmt_no = cpu_to_le64(c->cmt_no);
 	else
 		/* Mark the last node of the commit */
-		orph->cmt_no = cpu_to_le64((c->cmt_no + 1) | (1ULL << 63));
+		orph->cmt_no = cpu_to_le64((c->cmt_no) | (1ULL << 63));
 	ubifs_assert(c->ohead_offs + len <= c->leb_size);
 	ubifs_assert(c->ohead_lnum >= c->orph_first);
 	ubifs_assert(c->ohead_lnum <= c->orph_last);
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index 6ddd1de..21502b6 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -865,7 +865,8 @@ struct ubifs_mount_opts {
  * @highest_inum: highest used inode number
  * @vfs_gen: VFS inode generation counter
  * @max_sqnum: current global sequence number
- * @cmt_no: commit number (last successfully completed commit)
+ * @cmt_no: commit number of the last successfully completed commit, protected
+ *          by @commit_sem
  * @cnt_lock: protects @highest_inum, @vfs_gen, and @max_sqnum counters
  * @fmt_version: UBIFS on-flash format version
  * @uuid: UUID from super block
-- 
1.5.4.1

[PATCH] UBIFS: optimize deletions

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Every time anything is deleted, UBIFS writes the deletion inode
node twice - once in 'ubifs_jnl_update()' and the second time in
'ubifs_jnl_write_inode()'. However, the second write is not needed
if no commit happened after 'ubifs_jnl_update()'. This patch
checks that condition and avoids writing the deletion inode for
the second time.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/journal.c |   60 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ubifs/super.c   |    6 +++-
 fs/ubifs/ubifs.h   |   12 +++++++--
 3 files changed, 73 insertions(+), 5 deletions(-)

diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 3bc3fc9..0bcee7d 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -604,6 +604,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 			release_head(c, BASEHD);
 			goto out_finish;
 		}
+		ui->del_cmtno = c->cmt_no;
 	}
 
 	err = write_head(c, BASEHD, dent, len, &lnum, &dent_offs, sync);
@@ -821,6 +822,64 @@ out_free:
 }
 
 /**
+ * ubifs_jnl_write_inode - delete an inode.
+ * @c: UBIFS file-system description object
+ * @inode: inode to delete
+ *
+ * This function deletes inode @inode which includes removing it from orphans,
+ * deleting it from TNC and, in some cases, writing a deletion inode to the
+ * journal.
+ *
+ * When regular file inodes are unlinked or a directory inode is removed, the
+ * 'ubifs_jnl_update()' function write corresponding deletion inode and
+ * direntry to the media, and adds the inode to orphans. After this, when the
+ * last reference to this inode has been dropped, this function is called. In
+ * general, it has to write one more deletion inode to the media, because if
+ * a commit happened between 'ubifs_jnl_update()' and
+ * 'ubifs_jnl_delete_inode()', the deletion inode is not in the journal
+ * anymore, and in fact it might be not on the flash anymore, becouse it might
+ * have been garbage-collected already. And for optimization reasond UBIFS does
+ * not read the orphan area if it has been unmounted cleanly, so it would have
+ * no indication in the journal that there is a deleted inode which has to be
+ * removed from TNC.
+ *
+ * However, if there was no commit between 'ubifs_jnl_update()' and
+ * 'ubifs_jnl_delete_inode()', then there is no need to write the deletion
+ * inode to the media for the second time. And this is quite typical case.
+ *
+ * This function returns zero in case of success and a negative error code in
+ * case of failure.
+ */
+int ubifs_jnl_delete_inode(struct ubifs_info *c, const struct inode *inode)
+{
+	int err;
+	struct ubifs_inode *ui = ubifs_inode(inode);
+
+	ubifs_assert(inode->i_nlink == 0);
+
+	if (ui->del_cmtno != c->cmt_no)
+		/* A commit happened for sure */
+		return ubifs_jnl_write_inode(c, inode);
+
+	down_read(&c->commit_sem);
+	/*
+	 * Check commit number again, because the first test has been done
+	 * without @c->commit_sem, so a commit might have happened.
+	 */
+	if (ui->del_cmtno != c->cmt_no) {
+		up_read(&c->commit_sem);
+		return ubifs_jnl_write_inode(c, inode);
+	}
+
+	ubifs_delete_orphan(c, inode->i_ino);
+	err = ubifs_tnc_remove_ino(c, inode->i_ino);
+	if (err)
+		ubifs_ro_mode(c, err);
+	up_read(&c->commit_sem);
+	return err;
+}
+
+/**
  * ubifs_jnl_rename - rename a directory entry.
  * @c: UBIFS file-system description object
  * @old_dir: parent inode of directory entry to rename
@@ -928,6 +987,7 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 			release_head(c, BASEHD);
 			goto out_finish;
 		}
+		new_ui->del_cmtno = c->cmt_no;
 	}
 
 	err = write_head(c, BASEHD, dent, len, &lnum, &offs, sync);
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index cf1fb6c..6cc4175 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -341,13 +341,15 @@ static void ubifs_delete_inode(struct inode *inode)
 		goto out;
 
 	ui->ui_size = inode->i_size = 0;
-	err = ubifs_jnl_write_inode(c, inode);
+	err = ubifs_jnl_delete_inode(c, inode);
 	if (err)
 		/*
 		 * Worst case we have a lost orphan inode wasting space, so a
 		 * simple error message is ok here.
 		 */
-		ubifs_err("can't write inode %lu, error %d", inode->i_ino, err);
+		ubifs_err("can't delete inode %lu, error %d",
+			  inode->i_ino, err);
+
 out:
 	if (ui->dirty)
 		ubifs_release_dirty_inode_budget(c, ui);
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index 21502b6..dfb4b93 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -322,6 +322,8 @@ struct ubifs_gced_idx_leb {
  * struct ubifs_inode - UBIFS in-memory inode description.
  * @vfs_inode: VFS inode description object
  * @creat_sqnum: sequence number at time of creation
+ * @del_cmtno: commit number corresponding to the time the inode was deleted,
+ *             protected by @c->commit_sem;
  * @xattr_size: summarized size of all extended attributes in bytes
  * @xattr_cnt: count of extended attributes this inode has
  * @xattr_names: sum of lengths of all extended attribute names belonging to
@@ -372,7 +374,10 @@ struct ubifs_gced_idx_leb {
  */
 struct ubifs_inode {
 	struct inode vfs_inode;
-	unsigned long long creat_sqnum;
+	union {
+		unsigned long long creat_sqnum;
+		unsigned long long del_cmtno;
+	};
 	unsigned int xattr_size;
 	unsigned int xattr_cnt;
 	unsigned int xattr_names;
@@ -779,7 +784,7 @@ struct ubifs_compressor {
 /**
  * struct ubifs_budget_req - budget requirements of an operation.
  *
- * @fast: non-zero if the budgeting should try to aquire budget quickly and
+ * @fast: non-zero if the budgeting should try to acquire budget quickly and
  *        should not try to call write-back
  * @recalculate: non-zero if @idx_growth, @data_growth, and @dd_growth fields
  *               have to be re-calculated
@@ -860,7 +865,7 @@ struct ubifs_mount_opts {
  * struct ubifs_info - UBIFS file-system description data structure
  * (per-superblock).
  * @vfs_sb: VFS @struct super_block object
- * @bdi: backing device info object to make VFS happy and disable readahead
+ * @bdi: backing device info object to make VFS happy and disable read-ahead
  *
  * @highest_inum: highest used inode number
  * @vfs_gen: VFS inode generation counter
@@ -1402,6 +1407,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 int ubifs_jnl_write_data(struct ubifs_info *c, const struct inode *inode,
 			 const union ubifs_key *key, const void *buf, int len);
 int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode);
+int ubifs_jnl_delete_inode(struct ubifs_info *c, const struct inode *inode);
 int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 		     const struct dentry *old_dentry,
 		     const struct inode *new_dir,
-- 
1.5.4.1

[PATCH] UBIFS: do not union creat_sqnum and del_cmtno

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

The values in these two fields need to be preserved independently
and so a union cannot be used.

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
---
 fs/ubifs/ubifs.h |    6 ++----
 1 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index dfb4b93..d342c69 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -374,10 +374,8 @@ struct ubifs_gced_idx_leb {
  */
 struct ubifs_inode {
 	struct inode vfs_inode;
-	union {
-		unsigned long long creat_sqnum;
-		unsigned long long del_cmtno;
-	};
+	unsigned long long creat_sqnum;
+	unsigned long long del_cmtno;
 	unsigned int xattr_size;
 	unsigned int xattr_cnt;
 	unsigned int xattr_names;
-- 
1.5.4.1

[PATCH] UBIFS: fix typos in comments

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
---
 fs/ubifs/journal.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 0bcee7d..25de6fd 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -822,7 +822,7 @@ out_free:
 }
 
 /**
- * ubifs_jnl_write_inode - delete an inode.
+ * ubifs_jnl_delete_inode - delete an inode.
  * @c: UBIFS file-system description object
  * @inode: inode to delete
  *
@@ -831,21 +831,21 @@ out_free:
  * journal.
  *
  * When regular file inodes are unlinked or a directory inode is removed, the
- * 'ubifs_jnl_update()' function write corresponding deletion inode and
+ * 'ubifs_jnl_update()' function writes a corresponding deletion inode and
  * direntry to the media, and adds the inode to orphans. After this, when the
  * last reference to this inode has been dropped, this function is called. In
  * general, it has to write one more deletion inode to the media, because if
  * a commit happened between 'ubifs_jnl_update()' and
  * 'ubifs_jnl_delete_inode()', the deletion inode is not in the journal
- * anymore, and in fact it might be not on the flash anymore, becouse it might
- * have been garbage-collected already. And for optimization reasond UBIFS does
+ * anymore, and in fact it might not be on the flash anymore, because it might
+ * have been garbage-collected already. And for optimization reasons UBIFS does
  * not read the orphan area if it has been unmounted cleanly, so it would have
  * no indication in the journal that there is a deleted inode which has to be
  * removed from TNC.
  *
  * However, if there was no commit between 'ubifs_jnl_update()' and
  * 'ubifs_jnl_delete_inode()', then there is no need to write the deletion
- * inode to the media for the second time. And this is quite typical case.
+ * inode to the media for the second time. And this is quite a typical case.
  *
  * This function returns zero in case of success and a negative error code in
  * case of failure.
-- 
1.5.4.1

[PATCH] UBIFS: correct orphan deletion order

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

The debug function that checks orphans, does so using the
TNC mutex. That means it will not see a correct picture
if the inode is removed from the orphan tree before it is
removed from TNC.

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
---
 fs/ubifs/journal.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 25de6fd..acdae00 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -871,10 +871,11 @@ int ubifs_jnl_delete_inode(struct ubifs_info *c, const struct inode *inode)
 		return ubifs_jnl_write_inode(c, inode);
 	}
 
-	ubifs_delete_orphan(c, inode->i_ino);
 	err = ubifs_tnc_remove_ino(c, inode->i_ino);
 	if (err)
 		ubifs_ro_mode(c, err);
+	else
+		ubifs_delete_orphan(c, inode->i_ino);
 	up_read(&c->commit_sem);
 	return err;
 }
-- 
1.5.4.1

[PATCH] UBIFS: improve budgeting checks

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Budgeting is a crucial UBIFS subsystem - add more assertions
to improve requests checking. This is not compiled in when
UBIFS debugging is disabled.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/budget.c |   12 ++++++++++++
 fs/ubifs/ubifs.h  |    8 +++++++-
 2 files changed, 19 insertions(+), 1 deletions(-)

diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index 12a1717..f5afce5 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -543,6 +543,12 @@ int ubifs_budget_space(struct ubifs_info *c, struct ubifs_budget_req *req)
 	int err, idx_growth, data_growth, dd_growth;
 	struct retries_info ri;
 
+	ubifs_assert(req->new_page <= 1);
+	ubifs_assert(req->dirtied_page <= 1);
+	ubifs_assert(req->new_dent <= 1);
+	ubifs_assert(req->mod_dent <= 1);
+	ubifs_assert(req->new_ino <= 1);
+	ubifs_assert(req->new_ino_d <= UBIFS_MAX_INO_DATA);
 	ubifs_assert(req->dirtied_ino <= 4);
 	ubifs_assert(req->dirtied_ino_d <= UBIFS_MAX_INO_DATA * 4);
 
@@ -618,6 +624,12 @@ again:
  */
 void ubifs_release_budget(struct ubifs_info *c, struct ubifs_budget_req *req)
 {
+	ubifs_assert(req->new_page <= 1);
+	ubifs_assert(req->dirtied_page <= 1);
+	ubifs_assert(req->new_dent <= 1);
+	ubifs_assert(req->mod_dent <= 1);
+	ubifs_assert(req->new_ino <= 1);
+	ubifs_assert(req->new_ino_d <= UBIFS_MAX_INO_DATA);
 	ubifs_assert(req->dirtied_ino <= 4);
 	ubifs_assert(req->dirtied_ino_d <= UBIFS_MAX_INO_DATA * 4);
 	if (!req->recalculate) {
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index d342c69..565dca2 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -812,17 +812,23 @@ struct ubifs_compressor {
 struct ubifs_budget_req {
 	unsigned int fast:1;
 	unsigned int recalculate:1;
+#ifndef UBIFS_DEBUG
 	unsigned int new_page:1;
 	unsigned int dirtied_page:1;
 	unsigned int new_dent:1;
 	unsigned int mod_dent:1;
 	unsigned int new_ino:1;
 	unsigned int new_ino_d:13;
-#ifndef UBIFS_DEBUG
 	unsigned int dirtied_ino:4;
 	unsigned int dirtied_ino_d:15;
 #else
 	/* Not bit-fields to check for overflows */
+	unsigned int new_page;
+	unsigned int dirtied_page;
+	unsigned int new_dent;
+	unsigned int mod_dent;
+	unsigned int new_ino;
+	unsigned int new_ino_d;
 	unsigned int dirtied_ino;
 	unsigned int dirtied_ino_d;
 #endif
-- 
1.5.4.1

[PATCH] UBIFS: align inode data to eight

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

UBIFS aligns node lengths to 8, so budgeting has to do the
same. Well, direntry, inode, and page budgets are already
aligned, but not inode data budget (e.g., data in special
devices or symlinks). Do this for inode data as well.
Also, add corresponding debugging checks.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/budget.c |   10 +++++++++-
 fs/ubifs/dir.c    |   10 ++++++----
 fs/ubifs/file.c   |    4 ++--
 fs/ubifs/ubifs.h  |    4 ++++
 fs/ubifs/xattr.c  |    4 ++--
 5 files changed, 23 insertions(+), 9 deletions(-)

diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index f5afce5..a3978ba 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -551,6 +551,8 @@ int ubifs_budget_space(struct ubifs_info *c, struct ubifs_budget_req *req)
 	ubifs_assert(req->new_ino_d <= UBIFS_MAX_INO_DATA);
 	ubifs_assert(req->dirtied_ino <= 4);
 	ubifs_assert(req->dirtied_ino_d <= UBIFS_MAX_INO_DATA * 4);
+	ubifs_assert(!(req->new_ino_d & 7));
+	ubifs_assert(!(req->dirtied_ino_d & 7));
 
 	data_growth = calc_data_growth(c, req);
 	dd_growth = calc_dd_growth(c, req);
@@ -632,6 +634,8 @@ void ubifs_release_budget(struct ubifs_info *c, struct ubifs_budget_req *req)
 	ubifs_assert(req->new_ino_d <= UBIFS_MAX_INO_DATA);
 	ubifs_assert(req->dirtied_ino <= 4);
 	ubifs_assert(req->dirtied_ino_d <= UBIFS_MAX_INO_DATA * 4);
+	ubifs_assert(!(req->new_ino_d & 7));
+	ubifs_assert(!(req->dirtied_ino_d & 7));
 	if (!req->recalculate) {
 		ubifs_assert(req->idx_growth >= 0);
 		ubifs_assert(req->data_growth >= 0);
@@ -659,7 +663,11 @@ void ubifs_release_budget(struct ubifs_info *c, struct ubifs_budget_req *req)
 
 	ubifs_assert(c->budg_idx_growth >= 0);
 	ubifs_assert(c->budg_data_growth >= 0);
+	ubifs_assert(c->budg_dd_growth >= 0);
 	ubifs_assert(c->min_idx_lebs < c->main_lebs);
+	ubifs_assert(!(c->budg_idx_growth & 7));
+	ubifs_assert(!(c->budg_data_growth & 7));
+	ubifs_assert(!(c->budg_dd_growth & 7));
 	spin_unlock(&c->space_lock);
 }
 
@@ -701,7 +709,7 @@ void ubifs_release_dirty_inode_budget(struct ubifs_info *c,
 	struct ubifs_budget_req req;
 
 	memset(&req, 0, sizeof(struct ubifs_budget_req));
-	req.dd_growth = c->inode_budget + ui->data_len;
+	req.dd_growth = c->inode_budget + ALIGN(ui->data_len, 8);
 	ubifs_release_budget(c, &req);
 }
 
diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index a79e850..eba3a8a 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -525,7 +525,7 @@ static int ubifs_link(struct dentry *old_dentry, struct inode *dir,
 	struct ubifs_inode *dir_ui = ubifs_inode(dir);
 	int err, sz_change = CALC_DENT_SIZE(dentry->d_name.len);
 	struct ubifs_budget_req req = { .new_dent = 1, .dirtied_ino = 2,
-					.dirtied_ino_d = ui->data_len };
+				.dirtied_ino_d = ALIGN(ui->data_len, 8) };
 
 	/*
 	 * Budget request settings: new direntry, changing the target inode,
@@ -788,7 +788,8 @@ static int ubifs_mknod(struct inode *dir, struct dentry *dentry,
 	int sz_change = CALC_DENT_SIZE(dentry->d_name.len);
 	int err, devlen = 0;
 	struct ubifs_budget_req req = { .new_ino = 1, .new_dent = 1,
-					.new_ino_d = devlen, .dirtied_ino = 1 };
+					.new_ino_d = ALIGN(devlen, 8),
+					.dirtied_ino = 1 };
 
 	/*
 	 * Budget request settings: new inode, new direntry and changing parent
@@ -862,7 +863,8 @@ static int ubifs_symlink(struct inode *dir, struct dentry *dentry,
 	int err, len = strlen(symname);
 	int sz_change = CALC_DENT_SIZE(dentry->d_name.len);
 	struct ubifs_budget_req req = { .new_ino = 1, .new_dent = 1,
-					.new_ino_d = len, .dirtied_ino = 1 };
+					.new_ino_d = ALIGN(len, 8),
+					.dirtied_ino = 1 };
 
 	/*
 	 * Budget request settings: new inode, new direntry and changing parent
@@ -1011,7 +1013,7 @@ static int ubifs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	struct ubifs_budget_req req = { .new_dent = 1, .mod_dent = 1,
 					.dirtied_ino = 3 };
 	struct ubifs_budget_req ino_req = { .dirtied_ino = 1,
-				.dirtied_ino_d = old_inode_ui->data_len };
+			.dirtied_ino_d = ALIGN(old_inode_ui->data_len, 8) };
 	struct timespec time;
 
 	/*
diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 01598f2..9fecab2 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -890,7 +890,7 @@ static int do_setattr(struct ubifs_info *c, struct inode *inode,
 	loff_t new_size = attr->ia_size;
 	struct ubifs_inode *ui = ubifs_inode(inode);
 	struct ubifs_budget_req req = { .dirtied_ino = 1,
-					.dirtied_ino_d = ui->data_len };
+				.dirtied_ino_d = ALIGN(ui->data_len, 8) };
 
 	err = ubifs_budget_space(c, &req);
 	if (err)
@@ -1052,7 +1052,7 @@ static int update_mctime(struct ubifs_info *c, struct inode *inode)
 	if (mctime_update_needed(inode, &now)) {
 		int err, release;
 		struct ubifs_budget_req req = { .dirtied_ino = 1,
-						.dirtied_ino_d = ui->data_len };
+				.dirtied_ino_d = ALIGN(ui->data_len, 8) };
 
 		err = ubifs_budget_space(c, &req);
 		if (err)
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index 565dca2..73ca8a0 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -808,6 +808,10 @@ struct ubifs_compressor {
  * An inode may contain 4KiB of data at max., thus the widths of @new_ino_d
  * is 13 bits, and @dirtied_ino_d - 15, because up to 4 inodes may be made
  * dirty by the re-name operation.
+ *
+ * Note, UBIFS aligns node lengths to 8-bytes boundary, so the requester has to
+ * make sure the amount of inode data which contribute to @new_ino_d and
+ * @dirtied_ino_d fields are aligned.
  */
 struct ubifs_budget_req {
 	unsigned int fast:1;
diff --git a/fs/ubifs/xattr.c b/fs/ubifs/xattr.c
index 1388a07..39e831d 100644
--- a/fs/ubifs/xattr.c
+++ b/fs/ubifs/xattr.c
@@ -103,8 +103,8 @@ static int create_xattr(struct ubifs_info *c, struct inode *host,
 	struct inode *inode;
 	struct ubifs_inode *ui, *host_ui = ubifs_inode(host);
 	struct ubifs_budget_req req = { .new_ino = 1, .new_dent = 1,
-					.new_ino_d = size, .dirtied_ino = 1,
-					.dirtied_ino_d = host_ui->data_len};
+				.new_ino_d = size, .dirtied_ino = 1,
+				.dirtied_ino_d = ALIGN(host_ui->data_len, 8)};
 
 	if (host_ui->xattr_cnt >= MAX_XATTRS_PER_INODE)
 		return -ENOSPC;
-- 
1.5.4.1

[PATCH] UBIFS: print pid in dump function

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Useful when something fails and there are many processes
racing.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/debug.c |   20 ++++++++++----------
 1 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/fs/ubifs/debug.c b/fs/ubifs/debug.c
index 0adfb29..b9cb774 100644
--- a/fs/ubifs/debug.c
+++ b/fs/ubifs/debug.c
@@ -568,8 +568,8 @@ void dbg_dump_budget_req(const struct ubifs_budget_req *req)
 void dbg_dump_lstats(const struct ubifs_lp_stats *lst)
 {
 	spin_lock(&dbg_lock);
-	printk(KERN_DEBUG "Lprops statistics: empty_lebs %d, idx_lebs  %d\n",
-	       lst->empty_lebs, lst->idx_lebs);
+	printk(KERN_DEBUG "(pid %d) Lprops statistics: empty_lebs %d, "
+	       "idx_lebs  %d\n", current->pid, lst->empty_lebs, lst->idx_lebs);
 	printk(KERN_DEBUG "\ttaken_empty_lebs %d, total_free %lld, "
 	       "total_dirty %lld\n", lst->taken_empty_lebs, lst->total_free,
 	       lst->total_dirty);
@@ -587,8 +587,8 @@ void dbg_dump_budg(struct ubifs_info *c)
 	struct ubifs_gced_idx_leb *idx_gc;
 
 	spin_lock(&dbg_lock);
-	printk(KERN_DEBUG "Budgeting info: budg_data_growth %lld, "
-	       "budg_dd_growth %lld, budg_idx_growth %lld\n",
+	printk(KERN_DEBUG "(pid %d) Budgeting info: budg_data_growth %lld, "
+	       "budg_dd_growth %lld, budg_idx_growth %lld\n", current->pid,
 	       c->budg_data_growth, c->budg_dd_growth, c->budg_idx_growth);
 	printk(KERN_DEBUG "\tdata budget sum %lld, total budget sum %lld, "
 	       "freeable_cnt %d\n", c->budg_data_growth + c->budg_dd_growth,
@@ -634,7 +634,7 @@ void dbg_dump_lprops(struct ubifs_info *c)
 	struct ubifs_lprops lp;
 	struct ubifs_lp_stats lst;
 
-	printk(KERN_DEBUG "Dumping LEB properties\n");
+	printk(KERN_DEBUG "(pid %d) Dumping LEB properties\n", current->pid);
 	ubifs_get_lp_stats(c, &lst);
 	dbg_dump_lstats(&lst);
 
@@ -655,7 +655,7 @@ void dbg_dump_leb(const struct ubifs_info *c, int lnum)
 	if (dbg_failure_mode)
 		return;
 
-	printk(KERN_DEBUG "Dumping LEB %d\n", lnum);
+	printk(KERN_DEBUG "(pid %d) Dumping LEB %d\n", current->pid, lnum);
 
 	sleb = ubifs_scan(c, lnum, 0, c->dbg_buf);
 	if (IS_ERR(sleb)) {
@@ -720,8 +720,8 @@ void dbg_dump_heap(struct ubifs_info *c, struct ubifs_lpt_heap *heap, int cat)
 {
 	int i;
 
-	printk(KERN_DEBUG "Dumping heap cat %d (%d elements)\n",
-	       cat, heap->cnt);
+	printk(KERN_DEBUG "(pid %d) Dumping heap cat %d (%d elements)\n",
+	       current->pid, cat, heap->cnt);
 	for (i = 0; i < heap->cnt; i++) {
 		struct ubifs_lprops *lprops = heap->arr[i];
 
@@ -736,7 +736,7 @@ void dbg_dump_pnode(struct ubifs_info *c, struct ubifs_pnode *pnode,
 {
 	int i;
 
-	printk(KERN_DEBUG "Dumping pnode:\n");
+	printk(KERN_DEBUG "(pid %d) Dumping pnode:\n", current->pid);
 	printk(KERN_DEBUG "\taddress %zx parent %zx cnext %zx\n",
 	       (size_t)pnode, (size_t)parent, (size_t)pnode->cnext);
 	printk(KERN_DEBUG "\tflags %lu iip %d level %d num %d\n",
@@ -755,7 +755,7 @@ void dbg_dump_tnc(struct ubifs_info *c)
 	int level;
 
 	printk(KERN_DEBUG "\n");
-	printk(KERN_DEBUG "Dumping the TNC tree\n");
+	printk(KERN_DEBUG "(pid %d) Dumping the TNC tree\n", current->pid);
 	znode = ubifs_tnc_levelorder_next(c->zroot.znode, NULL);
 	level = znode->level;
 	printk(KERN_DEBUG "== Level %d ==\n", level);
-- 
1.5.4.1

[PATCH] UBIFS: reserve more space for index

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

At the moment UBIFS reserves twice old index size space for the
index. But this is not enough in some cases, because if the indexing
node are very fragmented and there are many small gaps, while the
dirty index has big znodes - in-the-gaps method would fail.

Thus, reserve trise as more, in which case we are guaranteed that
we can commit in any case.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/budget.c      |    8 ++++----
 fs/ubifs/find.c        |    9 +++++++--
 fs/ubifs/misc.h        |    2 +-
 fs/ubifs/ubifs-media.h |    4 ++--
 4 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index a3978ba..323d83a 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -263,8 +263,8 @@ int ubifs_calc_min_idx_lebs(struct ubifs_info *c)
 
 	idx_size = c->old_idx_sz + c->budg_idx_growth + c->budg_uncommitted_idx;
 
-	/* And make sure we have twice the index size of space reserved */
-	idx_size <<= 1;
+	/* And make sure we have trice the index size of space reserved */
+	idx_size = idx_size + (idx_size << 1);
 
 	/*
 	 * We do not maintain 'old_idx_size' as 'old_idx_lebs'/'old_idx_bytes'
@@ -388,11 +388,11 @@ static int can_use_rp(struct ubifs_info *c)
  * This function makes sure UBIFS has enough free eraseblocks for index growth
  * and data.
  *
- * When budgeting index space, UBIFS reserves twice as more LEBs as the index
+ * When budgeting index space, UBIFS reserves trice as more LEBs as the index
  * would take if it was consolidated and written to the flash. This guarantees
  * that the "in-the-gaps" commit method always succeeds and UBIFS will always
  * be able to commit dirty index. So this function basically adds amount of
- * budgeted index space to the size of the current index, multiplies this by 2,
+ * budgeted index space to the size of the current index, multiplies this by 3,
  * and makes sure this does not exceed the amount of free eraseblocks.
  *
  * Notes about @c->min_idx_lebs and @c->lst.idx_lebs variables:
diff --git a/fs/ubifs/find.c b/fs/ubifs/find.c
index 10394c5..c70c767 100644
--- a/fs/ubifs/find.c
+++ b/fs/ubifs/find.c
@@ -290,9 +290,14 @@ int ubifs_find_dirty_leb(struct ubifs_info *c, struct ubifs_lprops *ret_lp,
 		idx_lp = idx_heap->arr[0];
 		sum = idx_lp->free + idx_lp->dirty;
 		/*
-		 * Since we reserve twice as more space for the index than it
+		 * Since we reserve trice as more space for the index than it
 		 * actually takes, it does not make sense to pick indexing LEBs
-		 * with less than half LEB of dirty space.
+		 * with less than, say, half LEB of dirty space. May be half is
+		 * not the optimal boundary - this should be tested and
+		 * checked. This boundary should determine how much we use
+		 * in-the-gaps to consolidate the index comparing to how much
+		 * we use garbage collector to consolidate it. The "half"
+		 * criteria just feels to be fine.
 		 */
 		if (sum < min_space || sum < c->half_leb_size)
 			idx_lp = NULL;
diff --git a/fs/ubifs/misc.h b/fs/ubifs/misc.h
index cd83ffc..87dabf9 100644
--- a/fs/ubifs/misc.h
+++ b/fs/ubifs/misc.h
@@ -308,7 +308,7 @@ static inline long long ubifs_reported_space(const struct ubifs_info *c,
 {
 	int divisor, factor;
 
-	divisor = UBIFS_MAX_DATA_NODE_SZ + (c->max_idx_node_sz << 1);
+	divisor = UBIFS_MAX_DATA_NODE_SZ + (c->max_idx_node_sz * 3);
 	factor = UBIFS_MAX_DATA_NODE_SZ - UBIFS_DATA_NODE_SZ;
 	do_div(free, divisor);
 
diff --git a/fs/ubifs/ubifs-media.h b/fs/ubifs/ubifs-media.h
index 0cc7da9..bd2121f 100644
--- a/fs/ubifs/ubifs-media.h
+++ b/fs/ubifs/ubifs-media.h
@@ -228,10 +228,10 @@ enum {
 /* Minimum number of orphan area logical eraseblocks */
 #define UBIFS_MIN_ORPH_LEBS 1
 /*
- * Minimum number of main area logical eraseblocks (buds, 2 for the index, 1
+ * Minimum number of main area logical eraseblocks (buds, 3 for the index, 1
  * for GC, 1 for deletions, and at least 1 for committed data).
  */
-#define UBIFS_MIN_MAIN_LEBS (UBIFS_MIN_BUD_LEBS + 5)
+#define UBIFS_MIN_MAIN_LEBS (UBIFS_MIN_BUD_LEBS + 6)
 
 /* Minimum number of logical eraseblocks */
 #define UBIFS_MIN_LEB_CNT (UBIFS_SB_LEBS + UBIFS_MST_LEBS + \
-- 
1.5.4.1

[PATCH] UBIFS: minor tweaks in commit

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

No functional changes, just lessen the amount of indentations.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/tnc_commit.c |   37 ++++++++++++++++++-------------------
 1 files changed, 18 insertions(+), 19 deletions(-)

diff --git a/fs/ubifs/tnc_commit.c b/fs/ubifs/tnc_commit.c
index 8117e65..8ac76b1 100644
--- a/fs/ubifs/tnc_commit.c
+++ b/fs/ubifs/tnc_commit.c
@@ -372,26 +372,25 @@ static int layout_in_gaps(struct ubifs_info *c, int cnt)
 		written = layout_leb_in_gaps(c, p);
 		if (written < 0) {
 			err = written;
-			if (err == -ENOSPC) {
-				if (!dbg_force_in_the_gaps_enabled) {
-					/*
-					 * Do not print scary warnings if the
-					 * debugging option which forces
-					 * in-the-gaps is enabled.
-					 */
-					ubifs_err("out of space");
-					spin_lock(&c->space_lock);
-					dbg_dump_budg(c);
-					spin_unlock(&c->space_lock);
-					dbg_dump_lprops(c);
-				}
-				/* Try to commit anyway */
-				err = 0;
-				break;
+			if (err != -ENOSPC) {
+				kfree(c->gap_lebs);
+				c->gap_lebs = NULL;
+				return err;
 			}
-			kfree(c->gap_lebs);
-			c->gap_lebs = NULL;
-			return err;
+			if (!dbg_force_in_the_gaps_enabled) {
+				/*
+				 * Do not print scary warnings if the debugging
+				 * option which forces in-the-gaps is enabled.
+				 */
+				ubifs_err("out of space");
+				spin_lock(&c->space_lock);
+				dbg_dump_budg(c);
+				spin_unlock(&c->space_lock);
+				dbg_dump_lprops(c);
+			}
+			/* Try to commit anyway */
+			err = 0;
+			break;
 		}
 		p++;
 		cnt -= written;
-- 
1.5.4.1

[PATCH] UBIFS: support splice_write

[Artem Bityutskiy]

From: Zoltan Sogor <weth@inf.u-szeged.hu>

Signed-off-by: Zoltan Sogor <weth@inf.u-szeged.hu>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/file.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 9fecab2..4071d1c 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -1271,6 +1271,7 @@ struct file_operations ubifs_file_operations = {
 	.fsync          = ubifs_fsync,
 	.unlocked_ioctl = ubifs_ioctl,
 	.splice_read	= generic_file_splice_read,
+	.splice_write	= generic_file_splice_write,
 #ifdef CONFIG_COMPAT
 	.compat_ioctl   = ubifs_compat_ioctl,
 #endif
-- 
1.5.4.1

[PATCH] UBIFS: correct spelling of "thrice".

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
---
 fs/ubifs/budget.c |    4 ++--
 fs/ubifs/find.c   |    2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/fs/ubifs/budget.c b/fs/ubifs/budget.c
index 323d83a..1540981 100644
--- a/fs/ubifs/budget.c
+++ b/fs/ubifs/budget.c
@@ -263,7 +263,7 @@ int ubifs_calc_min_idx_lebs(struct ubifs_info *c)
 
 	idx_size = c->old_idx_sz + c->budg_idx_growth + c->budg_uncommitted_idx;
 
-	/* And make sure we have trice the index size of space reserved */
+	/* And make sure we have thrice the index size of space reserved */
 	idx_size = idx_size + (idx_size << 1);
 
 	/*
@@ -388,7 +388,7 @@ static int can_use_rp(struct ubifs_info *c)
  * This function makes sure UBIFS has enough free eraseblocks for index growth
  * and data.
  *
- * When budgeting index space, UBIFS reserves trice as more LEBs as the index
+ * When budgeting index space, UBIFS reserves thrice as many LEBs as the index
  * would take if it was consolidated and written to the flash. This guarantees
  * that the "in-the-gaps" commit method always succeeds and UBIFS will always
  * be able to commit dirty index. So this function basically adds amount of
diff --git a/fs/ubifs/find.c b/fs/ubifs/find.c
index c70c767..adee7b5 100644
--- a/fs/ubifs/find.c
+++ b/fs/ubifs/find.c
@@ -290,7 +290,7 @@ int ubifs_find_dirty_leb(struct ubifs_info *c, struct ubifs_lprops *ret_lp,
 		idx_lp = idx_heap->arr[0];
 		sum = idx_lp->free + idx_lp->dirty;
 		/*
-		 * Since we reserve trice as more space for the index than it
+		 * Since we reserve thrice as much space for the index than it
 		 * actually takes, it does not make sense to pick indexing LEBs
 		 * with less than, say, half LEB of dirty space. May be half is
 		 * not the optimal boundary - this should be tested and
-- 
1.5.4.1

[PATCH] UBIFS: always set i_generation to 0

[Artem Bityutskiy]

From: Adrian Hunter <ext-adrian.hunter@nokia.com>

UBIFS does not presently re-use inode numbers, so leaving
i_generation zero is most appropriate for now.

Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/dir.c   |    1 -
 fs/ubifs/super.c |    2 --
 fs/ubifs/ubifs.h |    4 +---
 3 files changed, 1 insertions(+), 6 deletions(-)

diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index eba3a8a..0d1ab89 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -165,7 +165,6 @@ struct inode *ubifs_new_inode(struct ubifs_info *c, const struct inode *dir,
 	}
 
 	inode->i_ino = ++c->highest_inum;
-	inode->i_generation = ++c->vfs_gen;
 	/*
 	 * The creation sequence number remains with this inode for its
 	 * lifetime. All nodes for this inode have a greater sequence number,
diff --git a/fs/ubifs/super.c b/fs/ubifs/super.c
index 6cc4175..2c268a4 100644
--- a/fs/ubifs/super.c
+++ b/fs/ubifs/super.c
@@ -30,7 +30,6 @@
 #include <linux/slab.h>
 #include <linux/module.h>
 #include <linux/ctype.h>
-#include <linux/random.h>
 #include <linux/kthread.h>
 #include <linux/parser.h>
 #include <linux/seq_file.h>
@@ -1671,7 +1670,6 @@ static int ubifs_fill_super(struct super_block *sb, void *data, int silent)
 	INIT_LIST_HEAD(&c->orph_new);
 
 	c->highest_inum = UBIFS_FIRST_INO;
-	get_random_bytes(&c->vfs_gen, sizeof(int));
 	c->lhead_lnum = c->ltail_lnum = UBIFS_LOG_LNUM;
 
 	ubi_get_volume_info(ubi, &c->vi);
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index 73ca8a0..f2dd749 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -876,11 +876,10 @@ struct ubifs_mount_opts {
  * @bdi: backing device info object to make VFS happy and disable read-ahead
  *
  * @highest_inum: highest used inode number
- * @vfs_gen: VFS inode generation counter
  * @max_sqnum: current global sequence number
  * @cmt_no: commit number of the last successfully completed commit, protected
  *          by @commit_sem
- * @cnt_lock: protects @highest_inum, @vfs_gen, and @max_sqnum counters
+ * @cnt_lock: protects @highest_inum and @max_sqnum counters
  * @fmt_version: UBIFS on-flash format version
  * @uuid: UUID from super block
  *
@@ -1117,7 +1116,6 @@ struct ubifs_info {
 	struct backing_dev_info bdi;
 
 	ino_t highest_inum;
-	unsigned int vfs_gen;
 	unsigned long long max_sqnum;
 	unsigned long long cmt_no;
 	spinlock_t cnt_lock;
-- 
1.5.4.1

[PATCH] UBIFS: improve arguments checking in debugging messages

[Artem Bityutskiy]

From: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

Use "if (0) printk()" construct in debugging print macros to
make the debugging messages be checked even if debugging is
off.

This patch also removes some unneeded spaces and blank lines.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/debug.h |  143 ++++++++++++++++++++++++++----------------------------
 1 files changed, 69 insertions(+), 74 deletions(-)

diff --git a/fs/ubifs/debug.h b/fs/ubifs/debug.h
index 3c4f1e9..50315fc 100644
--- a/fs/ubifs/debug.h
+++ b/fs/ubifs/debug.h
@@ -27,7 +27,7 @@
 
 #define UBIFS_DBG(op) op
 
-#define ubifs_assert(expr)  do {                                               \
+#define ubifs_assert(expr) do {                                                \
 	if (unlikely(!(expr))) {                                               \
 		printk(KERN_CRIT "UBIFS assert failed in %s at %u (pid %d)\n", \
 		       __func__, __LINE__, current->pid);                      \
@@ -73,50 +73,50 @@ const char *dbg_key_str1(const struct ubifs_info *c,
 			 const union ubifs_key *key);
 
 /*
- * DBGKEY macros require dbg_lock to be held, which it is in the dbg message
+ * DBGKEY macros require @dbg_lock to be held, which it is in the dbg message
  * macros.
  */
 #define DBGKEY(key) dbg_key_str0(c, (key))
 #define DBGKEY1(key) dbg_key_str1(c, (key))
 
 /* General messages */
-#define dbg_gen(fmt, ...)        dbg_do_msg(UBIFS_MSG_GEN, fmt, ##__VA_ARGS__)
+#define dbg_gen(fmt, ...)   dbg_do_msg(UBIFS_MSG_GEN, fmt, ##__VA_ARGS__)
 
 /* Additional journal messages */
-#define dbg_jnl(fmt, ...)        dbg_do_msg(UBIFS_MSG_JNL, fmt, ##__VA_ARGS__)
+#define dbg_jnl(fmt, ...)   dbg_do_msg(UBIFS_MSG_JNL, fmt, ##__VA_ARGS__)
 
 /* Additional TNC messages */
-#define dbg_tnc(fmt, ...)        dbg_do_msg(UBIFS_MSG_TNC, fmt, ##__VA_ARGS__)
+#define dbg_tnc(fmt, ...)   dbg_do_msg(UBIFS_MSG_TNC, fmt, ##__VA_ARGS__)
 
 /* Additional lprops messages */
-#define dbg_lp(fmt, ...)         dbg_do_msg(UBIFS_MSG_LP, fmt, ##__VA_ARGS__)
+#define dbg_lp(fmt, ...)    dbg_do_msg(UBIFS_MSG_LP, fmt, ##__VA_ARGS__)
 
 /* Additional LEB find messages */
-#define dbg_find(fmt, ...)       dbg_do_msg(UBIFS_MSG_FIND, fmt, ##__VA_ARGS__)
+#define dbg_find(fmt, ...)  dbg_do_msg(UBIFS_MSG_FIND, fmt, ##__VA_ARGS__)
 
 /* Additional mount messages */
-#define dbg_mnt(fmt, ...)        dbg_do_msg(UBIFS_MSG_MNT, fmt, ##__VA_ARGS__)
+#define dbg_mnt(fmt, ...)   dbg_do_msg(UBIFS_MSG_MNT, fmt, ##__VA_ARGS__)
 
 /* Additional I/O messages */
-#define dbg_io(fmt, ...)         dbg_do_msg(UBIFS_MSG_IO, fmt, ##__VA_ARGS__)
+#define dbg_io(fmt, ...)    dbg_do_msg(UBIFS_MSG_IO, fmt, ##__VA_ARGS__)
 
 /* Additional commit messages */
-#define dbg_cmt(fmt, ...)        dbg_do_msg(UBIFS_MSG_CMT, fmt, ##__VA_ARGS__)
+#define dbg_cmt(fmt, ...)   dbg_do_msg(UBIFS_MSG_CMT, fmt, ##__VA_ARGS__)
 
 /* Additional budgeting messages */
-#define dbg_budg(fmt, ...)       dbg_do_msg(UBIFS_MSG_BUDG, fmt, ##__VA_ARGS__)
+#define dbg_budg(fmt, ...)  dbg_do_msg(UBIFS_MSG_BUDG, fmt, ##__VA_ARGS__)
 
 /* Additional log messages */
-#define dbg_log(fmt, ...)        dbg_do_msg(UBIFS_MSG_LOG, fmt, ##__VA_ARGS__)
+#define dbg_log(fmt, ...)   dbg_do_msg(UBIFS_MSG_LOG, fmt, ##__VA_ARGS__)
 
 /* Additional gc messages */
-#define dbg_gc(fmt, ...)         dbg_do_msg(UBIFS_MSG_GC, fmt, ##__VA_ARGS__)
+#define dbg_gc(fmt, ...)    dbg_do_msg(UBIFS_MSG_GC, fmt, ##__VA_ARGS__)
 
 /* Additional scan messages */
-#define dbg_scan(fmt, ...)       dbg_do_msg(UBIFS_MSG_SCAN, fmt, ##__VA_ARGS__)
+#define dbg_scan(fmt, ...)  dbg_do_msg(UBIFS_MSG_SCAN, fmt, ##__VA_ARGS__)
 
 /* Additional recovery messages */
-#define dbg_rcvry(fmt, ...)      dbg_do_msg(UBIFS_MSG_RCVRY, fmt, ##__VA_ARGS__)
+#define dbg_rcvry(fmt, ...) dbg_do_msg(UBIFS_MSG_RCVRY, fmt, ##__VA_ARGS__)
 
 /*
  * Debugging message type flags (must match msg_type_names in debug.c).
@@ -239,34 +239,23 @@ typedef int (*dbg_leaf_callback)(struct ubifs_info *c,
 				 struct ubifs_zbranch *zbr, void *priv);
 typedef int (*dbg_znode_callback)(struct ubifs_info *c,
 				  struct ubifs_znode *znode, void *priv);
-
 int dbg_walk_index(struct ubifs_info *c, dbg_leaf_callback leaf_cb,
 		   dbg_znode_callback znode_cb, void *priv);
 
 /* Checking functions */
 
 int dbg_check_lprops(struct ubifs_info *c);
-
 int dbg_old_index_check_init(struct ubifs_info *c, struct ubifs_zbranch *zroot);
 int dbg_check_old_index(struct ubifs_info *c, struct ubifs_zbranch *zroot);
-
 int dbg_check_cats(struct ubifs_info *c);
-
 int dbg_check_ltab(struct ubifs_info *c);
-
 int dbg_check_synced_i_size(struct inode *inode);
-
 int dbg_check_dir_size(struct ubifs_info *c, const struct inode *dir);
-
 int dbg_check_tnc(struct ubifs_info *c, int extra);
-
 int dbg_check_idx_size(struct ubifs_info *c, long long idx_size);
-
 int dbg_check_filesystem(struct ubifs_info *c);
-
 void dbg_check_heap(struct ubifs_info *c, struct ubifs_lpt_heap *heap, int cat,
 		    int add_pos);
-
 int dbg_check_lprops(struct ubifs_info *c);
 int dbg_check_lpt_nodes(struct ubifs_info *c, struct ubifs_cnode *cnode,
 			int row, int col);
@@ -329,71 +318,77 @@ static inline int dbg_change(struct ubi_volume_desc *desc, int lnum,
 #else /* !CONFIG_UBIFS_FS_DEBUG */
 
 #define UBIFS_DBG(op)
-#define ubifs_assert(expr)                         ({})
-#define ubifs_assert_cmt_locked(c)
+
+/* Use "if (0)" to make compiler check arguments even if debugging is off */
+#define ubifs_assert(expr)  do {                                               \
+	if (0 && (expr))                                                       \
+		printk(KERN_CRIT "UBIFS assert failed in %s at %u (pid %d)\n", \
+		       __func__, __LINE__, current->pid);                      \
+} while (0)
+
+#define dbg_err(fmt, ...)   do {                                               \
+	if (0)                                                                 \
+		ubifs_err(fmt, ##__VA_ARGS__);                                 \
+} while (0)
+
+#define dbg_msg(fmt, ...) do {                                                 \
+	if (0)                                                                 \
+		printk(KERN_DEBUG "UBIFS DBG (pid %d): %s: " fmt "\n",         \
+		       current->pid, __func__, ##__VA_ARGS__);                 \
+} while (0)
+
 #define dbg_dump_stack()
-#define dbg_err(fmt, ...)                          ({})
-#define dbg_msg(fmt, ...)                          ({})
-#define dbg_key(c, key, fmt, ...)                  ({})
-
-#define dbg_gen(fmt, ...)                          ({})
-#define dbg_jnl(fmt, ...)                          ({})
-#define dbg_tnc(fmt, ...)                          ({})
-#define dbg_lp(fmt, ...)                           ({})
-#define dbg_find(fmt, ...)                         ({})
-#define dbg_mnt(fmt, ...)                          ({})
-#define dbg_io(fmt, ...)                           ({})
-#define dbg_cmt(fmt, ...)                          ({})
-#define dbg_budg(fmt, ...)                         ({})
-#define dbg_log(fmt, ...)                          ({})
-#define dbg_gc(fmt, ...)                           ({})
-#define dbg_scan(fmt, ...)                         ({})
-#define dbg_rcvry(fmt, ...)                        ({})
-
-#define dbg_ntype(type)                            ""
-#define dbg_cstate(cmt_state)                      ""
-#define dbg_get_key_dump(c, key)                   ({})
-#define dbg_dump_inode(c, inode)                   ({})
-#define dbg_dump_node(c, node)                     ({})
-#define dbg_dump_budget_req(req)                   ({})
-#define dbg_dump_lstats(lst)                       ({})
-#define dbg_dump_budg(c)                           ({})
-#define dbg_dump_lprop(c, lp)                      ({})
-#define dbg_dump_lprops(c)                         ({})
-#define dbg_dump_leb(c, lnum)                      ({})
-#define dbg_dump_znode(c, znode)                   ({})
-#define dbg_dump_heap(c, heap, cat)                ({})
-#define dbg_dump_pnode(c, pnode, parent, iip)      ({})
-#define dbg_dump_tnc(c)                            ({})
-#define dbg_dump_index(c)                          ({})
+#define ubifs_assert_cmt_locked(c)
 
-#define dbg_walk_index(c, leaf_cb, znode_cb, priv) 0
+#define dbg_gen(fmt, ...)   dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_jnl(fmt, ...)   dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_tnc(fmt, ...)   dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_lp(fmt, ...)    dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_find(fmt, ...)  dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_mnt(fmt, ...)   dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_io(fmt, ...)    dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_cmt(fmt, ...)   dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_budg(fmt, ...)  dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_log(fmt, ...)   dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_gc(fmt, ...)    dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_scan(fmt, ...)  dbg_msg(fmt, ##__VA_ARGS__)
+#define dbg_rcvry(fmt, ...) dbg_msg(fmt, ##__VA_ARGS__)
+
+#define DBGKEY(key)  ((char *)(key))
+#define DBGKEY1(key) ((char *)(key))
+
+#define dbg_ntype(type)                       ""
+#define dbg_cstate(cmt_state)                 ""
+#define dbg_get_key_dump(c, key)              ({})
+#define dbg_dump_inode(c, inode)              ({})
+#define dbg_dump_node(c, node)                ({})
+#define dbg_dump_budget_req(req)              ({})
+#define dbg_dump_lstats(lst)                  ({})
+#define dbg_dump_budg(c)                      ({})
+#define dbg_dump_lprop(c, lp)                 ({})
+#define dbg_dump_lprops(c)                    ({})
+#define dbg_dump_leb(c, lnum)                 ({})
+#define dbg_dump_znode(c, znode)              ({})
+#define dbg_dump_heap(c, heap, cat)           ({})
+#define dbg_dump_pnode(c, pnode, parent, iip) ({})
+#define dbg_dump_tnc(c)                       ({})
+#define dbg_dump_index(c)                     ({})
 
+#define dbg_walk_index(c, leaf_cb, znode_cb, priv) 0
 #define dbg_old_index_check_init(c, zroot)         0
 #define dbg_check_old_index(c, zroot)              0
-
 #define dbg_check_cats(c)                          0
-
 #define dbg_check_ltab(c)                          0
-
 #define dbg_check_synced_i_size(inode)             0
-
 #define dbg_check_dir_size(c, dir)                 0
-
 #define dbg_check_tnc(c, x)                        0
-
 #define dbg_check_idx_size(c, idx_size)            0
-
 #define dbg_check_filesystem(c)                    0
-
 #define dbg_check_heap(c, heap, cat, add_pos)      ({})
-
 #define dbg_check_lprops(c)                        0
 #define dbg_check_lpt_nodes(c, cnode, row, col)    0
-
 #define dbg_force_in_the_gaps_enabled              0
 #define dbg_force_in_the_gaps()                    0
-
 #define dbg_failure_mode                           0
 #define dbg_failure_mode_registration(c)           ({})
 #define dbg_failure_mode_deregistration(c)         ({})
-- 
1.5.4.1

[PATCH] UBIFS: fix budgeting request alignment in xattr code

[Artem Bityutskiy]

From: Zoltan Sogor <weth@inf.u-szeged.hu>

Data length has to be aligned in the budgeting request. Code
in xattr.c did not do this.

Signed-off-by: Zoltan Sogor <weth@inf.u-szeged.hu>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
---
 fs/ubifs/xattr.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/fs/ubifs/xattr.c b/fs/ubifs/xattr.c
index 39e831d..6f493de 100644
--- a/fs/ubifs/xattr.c
+++ b/fs/ubifs/xattr.c
@@ -103,8 +103,8 @@ static int create_xattr(struct ubifs_info *c, struct inode *host,
 	struct inode *inode;
 	struct ubifs_inode *ui, *host_ui = ubifs_inode(host);
 	struct ubifs_budget_req req = { .new_ino = 1, .new_dent = 1,
-				.new_ino_d = size, .dirtied_ino = 1,
-				.dirtied_ino_d = ALIGN(host_ui->data_len, 8)};
+				.new_ino_d = ALIGN(size, 8), .dirtied_ino = 1,
+				.dirtied_ino_d = ALIGN(host_ui->data_len, 8) };
 
 	if (host_ui->xattr_cnt >= MAX_XATTRS_PER_INODE)
 		return -ENOSPC;
@@ -200,7 +200,7 @@ static int change_xattr(struct ubifs_info *c, struct inode *host,
 	struct ubifs_inode *host_ui = ubifs_inode(host);
 	struct ubifs_inode *ui = ubifs_inode(inode);
 	struct ubifs_budget_req req = { .dirtied_ino = 2,
-				.dirtied_ino_d = size + host_ui->data_len };
+		.dirtied_ino_d = ALIGN(size, 8) + ALIGN(host_ui->data_len, 8) };
 
 	ubifs_assert(ui->data_len == inode->i_size);
 	err = ubifs_budget_space(c, &req);
@@ -497,8 +497,8 @@ static int remove_xattr(struct ubifs_info *c, struct inode *host,
 	int err;
 	struct ubifs_inode *host_ui = ubifs_inode(host);
 	struct ubifs_inode *ui = ubifs_inode(inode);
-	struct ubifs_budget_req req = { .dirtied_ino = 1, .mod_dent = 1,
-					.dirtied_ino_d = host_ui->data_len };
+	struct ubifs_budget_req req = { .dirtied_ino = 2, .mod_dent = 1,
+				.dirtied_ino_d = ALIGN(host_ui->data_len, 8) };
 
 	ubifs_assert(ui->data_len == inode->i_size);
 
-- 
1.5.4.1

Re: UBIFS updates for 2.6.27

[Christoph Hellwig]

On Wed, Aug 13, 2008 at 01:17:29PM +0300, Artem Bityutskiy wrote:
> Hi,
> 
> Christoph asked me to send UBIFS patches here before I put them
> to Linux next.
> 
> The following are patches we plan to send to Linus before 2.6.27
> is out. Some of the patches are not fixes, but because UBIFS is
> new and 2.6.27 will be its first release, we think it is OK to
> merge "non-fix" patches for UBIFS as an exception.
> 
> Christoph, I do not send NFS patch for now because I'd like to
> work on it some more, taking into account your suggestions.

Ok.  The other patch looks good to me from a quick glance.  BTW,
-fsdevel might be a better list than lkml for fs patches review.

UBIFS updates for 2.6.27

[Artem Bityutskiy]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1935 bytes --]

Hi,

here is the second bunch of UBIFS updates we would like to send to
Linus for 2.6.27 inclusion. The updates contain the following:

Adrian Hunter (2):
      UBIFS: always read hashed-key nodes under TNC mutex
      UBIFS: allow for racing between GC and TNC

Artem Bityutskiy (10):
      UBIFS: fix zero-length truncations
      UBIFS: do not update min_idx_lebs in stafs
      UBIFS: push empty flash hack down
      UBIFS: remove incorrect index space check
      UBIFS: improve statfs reporting
      UBIFS: fix assertion
      UBIFS: add forgotten gc_idx_lebs component
      UBIFS: introduce LEB overhead
      UBIFS: improve statfs reporting even more
      UBIFS: fill f_fsid

 fs/ubifs/budget.c |  113 +++++++++++++++++++++++++++++++++++++++------------
 fs/ubifs/dir.c    |    1 -
 fs/ubifs/file.c   |   20 +++++++--
 fs/ubifs/find.c   |   18 ++++-----
 fs/ubifs/gc.c     |    6 +++
 fs/ubifs/misc.h   |   49 ++++++++---------------
 fs/ubifs/super.c  |   19 +++++---
 fs/ubifs/tnc.c    |  116 ++++++++++++++++++++++++++++-------------------------
 fs/ubifs/ubifs.h  |   14 +++++-
 9 files changed, 217 insertions(+), 139 deletions(-)

These patches are pure fixes:
UBIFS: fix zero-length truncations
UBIFS: always read hashed-key nodes under TNC mutex
UBIFS: allow for racing between GC and TNC
UBIFS: fix assertion
UBIFS: add forgotten gc_idx_lebs component
UBIFS: remove incorrect index space check

These patches partially fix the problem reported in the MTD ML:
http://lists.infradead.org/pipermail/linux-mtd/2008-August/022579.html

UBIFS: do not update min_idx_lebs in stafs
UBIFS: push empty flash hack down
UBIFS: improve statfs reporting
UBIFS: introduce LEB overhead
UBIFS: improve statfs reporting even more

Thus, it should be fine to send them mainline at this late stage.
Unless there are complaints, I'll send a pull request soon.

Thanks.

-- 
Best regards,
Artem Bityutskiy (Битюцкий Артём)

Re: UBIFS updates for 2.6.27

[Artem Bityutskiy]

Artem Bityutskiy wrote:
> Hi,
> 
> here is the second bunch of UBIFS updates we would like to send to
> Linus for 2.6.27 inclusion. The updates contain the following:

Oh, meant to send this to linux-fsdevel@vger.kernel.org, not to
linux-fsdevel-owner@vger.kernel.org. Apologies.

-- 
Best Regards,
Artem Bityutskiy (Артём Битюцкий)