Related to https://lkml.org/lkml/2018/4/8/661

Sometimes, we need to write meta data to new allocated block address,
then we will allocate a zeroed page in inner inode's address space, and
fill partial data in it, and leave other place with zero value which means
some fields are initial status.

There are two inner inodes (meta inode and node inode) setting __GFP_ZERO,
I have just checked them, for both of them, we can avoid using __GFP_ZERO,
and do initialization by ourselves to avoid unneeded/redundant zeroing
from mm.

Cc: <stable@vger.kernel.org>
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously, we use generic FS_*_FL defined by vfs to indicate inode status
for each bit of i_flags, so f2fs's flag status definition is tied to vfs'
one, it will be hard for f2fs to reuse bits f2fs never used to indicate
new status..

In order to solve this issue, we introduce private inode status mapping,
Note, for these bits have already been persisted into disk, we should
never change their definition, for other ones, we can remap them for
later new coming status.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This fixes xfstests/generic/392.

The failure was caused by different times between 1) one marked in the last
fsync(2) call and 2) the other given by roll-forward recovery after power-cut.
The reason was that we skipped updating inode block at 1), since its i_size
was recoverable along with 4KB-aligned data writes, which was fixed by:
  "f2fs: fix a wrong condition in f2fs_skip_inode_update"
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

When setting COLD_BIT_SHIFT flag in node block, we only need to call
set_cold_node() in new_node_page() and recover_inode_page() during
node page initialization. So remove unneeded set_cold_node() in other
places.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previous dentry page uses highmem, which will cause panic in platforms
using highmem (such as arm), since the address space of dentry pages
from highmem directly goes into the decryption path via the function
fscrypt_fname_disk_to_usr. But sg_init_one assumes the address is not
from highmem, and then cause panic since it doesn't call kmap_high but
kunmap_high is triggered at the end. To fix this problem in a simple
way, this patch avoids to put dentry page in pagecache into highmem.
Signed-off-by: NYunlong Song <yunlong.song@huawei.com>
Reviewed-by: NChao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: fix coding style]
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds creation time field in inode layout to support showing
kstat.btime in ->statx.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Now, we invoke f2fs_mark_inode_dirty_sync() to make an inode dirty in
advance of creating a new node page for the inode. By this, some inodes
whose node page is not created yet can be linked into the global dirty
list.

If the checkpoint is executed at this moment, the inode will be written
back by writeback_single_inode() and finally update_inode_page() will
fail to detach the inode from the global dirty list because the inode
doesn't have a node page.

The problem is that the inode's state in VFS layer will become clean
after execution of writeback_single_inode() and it's still linked in
the global dirty list of f2fs and this will cause a kernel panic.

So, we will prevent the newly created inode from being dirtied during
the FI_NEW_INODE flag of the inode is set. We will make it dirty
right after the flag is cleared.
Signed-off-by: NDaeho Jeong <daeho.jeong@samsung.com>
Signed-off-by: NYoungjin Gil <youngjin.gil@samsung.com>
Tested-by: NHobin Woo <hobin.woo@samsung.com>
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

fscrypt_put_encryption_info() is only called when evicting an inode, so
the 'struct fscrypt_info *ci' parameter is always NULL, and there cannot
be races with other threads.  This was cruft left over from the broken
key revocation code.  Remove the unused parameter and the cmpxchg().

Also remove the #ifdefs around the fscrypt_put_encryption_info() calls,
since fscrypt_notsupp.h defines a no-op stub for it.
Signed-off-by: NEric Biggers <ebiggers@google.com>
Signed-off-by: NTheodore Ts'o <tytso@mit.edu>

-Thread A                             Thread B

-write_checkpoint
 -block_operations
  -f2fs_unlock_all                    -f2fs_sync_file
                                       -f2fs_write_inode
                                        -f2fs_inode_synced
    -f2fs_sync_inode_meta
     -sync_node_pages
                                        -set_page_drity

In this case, if sudden power off without next new checkpoint,
the last inode page update will lost. wb_writeback is same with
fsync.

Yunlei also reproduced the bug by:

@@ -366,7 +366,7 @@ int update_inode(struct inode *inode, struct page *node_page)
        struct extent_tree *et = F2FS_I(inode)->extent_tree;

        f2fs_inode_synced(inode);
-
+       msleep(10000);
        f2fs_wait_on_page_writeback(node_page, NODE, true);

shell 1:                                       shell2:

dd if=/dev/zero of=./test bs=1M count=10
sync
echo "hello" >> ./test
fsync test  // sleep 10s
                                               sync //return quickly
echo c > /proc/sysrq-trigger
Signed-off-by: NYunlei He <heyunlei@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Now, in product, more and more features based on file encryption were
introduced, their demand of xattr space is increasing, however, inline
xattr has fixed-size of 200 bytes, once inline xattr space is full, new
increased xattr data would occupy additional xattr block which may bring
us more space usage and performance regression during persisting.

In order to resolve above issue, it's better to expand inline xattr size
flexibly according to user's requirement.

So this patch introduces new filesystem feature 'flexible inline xattr',
and new mount option 'inline_xattr_size=%u', once mkfs enables the
feature, we can use the option to make f2fs supporting flexible inline
xattr size.

To support this feature, we add extra attribute i_inline_xattr_size in
inode layout, indicating that how many space inline xattr borrows from
block address mapping space in inode layout, by this, we can easily
locate and store flexible-sized inline xattr data in inode.

Inode disk layout:
  +----------------------+
  | .i_mode              |
  | ...                  |
  | .i_ext               |
  +----------------------+
  | .i_extra_isize       |
  | .i_inline_xattr_size |-----------+
  | ...                  |           |
  +----------------------+           |
  | .i_addr              |           |
  |  - block address or  |           |
  |  - inline data       |           |
  +----------------------+<---+      v
  |    inline xattr      |    +---inline xattr range
  +----------------------+<---+
  | .i_nid               |
  +----------------------+
  |   node_footer        |
  | (nid, ino, offset)   |
  +----------------------+

Note that, we have to cnosider backward compatibility which reserved
inline_data space, 200 bytes, all the time, reported by Sheng Yong.

Previous inline data or directory always reserved 200 bytes in inode layout,
even if inline_xattr is disabled. In order to keep inline_dentry's structure
for backward compatibility, we get the space back only from inline_data.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Reported-by: NSheng Yong <shengyong1@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

When doing fault injection test, f2fs_evict_inode() didn't remove gdirty_list
which incurs a kernel panic due to wrong pointer access.
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Introduce a flag S_ENCRYPTED which can be set in ->i_flags to indicate
that the inode is encrypted using the fscrypt (fs/crypto/) mechanism.

Checking this flag will give the same information that
inode->i_sb->s_cop->is_encrypted(inode) currently does, but will be more
efficient.  This will be useful for adding higher-level helper functions
for filesystems to use.  For example we'll be able to replace this:

	if (ext4_encrypted_inode(inode)) {
		ret = fscrypt_get_encryption_info(inode);
		if (ret)
			return ret;
		if (!fscrypt_has_encryption_key(inode))
			return -ENOKEY;
	}

with this:

	ret = fscrypt_require_key(inode);
	if (ret)
		return ret;

... since we'll be able to retain the fast path for unencrypted files as
a single flag check, using an inline function.  This wasn't possible
before because we'd have had to frequently call through the
->i_sb->s_cop->is_encrypted function pointer, even when the encryption
support was disabled or not being used.

Note: we don't define S_ENCRYPTED to 0 if CONFIG_FS_ENCRYPTION is
disabled because we want to continue to return an error if an encrypted
file is accessed without encryption support, rather than pretending that
it is unencrypted.
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Acked-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NEric Biggers <ebiggers@google.com>
Signed-off-by: NTheodore Ts'o <tytso@mit.edu>

When multiple device feature is enabled, during ->fsync we will issue
flush in all devices to make sure node/data of the file being persisted
into storage. But some flushes of device could be unneeded as file's
data may be not writebacked into those devices. So this patch adds and
manage bitmap per inode in global cache to indicate which device is
dirty and it needs to issue flush during ->fsync, hence, we could improve
performance of fsync in scenario of multiple device.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Add a bugon in f2fs_evict_inode to detect inconsistent status between
inode cache and related node page cache.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If another thread already made the page dirtied or writebacked, we must avoid
to verify checksum. If we got an error, we need to remove its uptodate as well.
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds to support inode checksum in f2fs.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: fix verification flow]
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds to support plain project quota.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch add new flag F2FS_EXTRA_ATTR storing in inode.i_inline
to indicate that on-disk structure of current inode is extended.

In order to extend, we changed the inode structure a bit:

Original one:

struct f2fs_inode {
	...
	struct f2fs_extent i_ext;
	__le32 i_addr[DEF_ADDRS_PER_INODE];
	__le32 i_nid[DEF_NIDS_PER_INODE];
}

Extended one:

struct f2fs_inode {
        ...
        struct f2fs_extent i_ext;
	union {
		struct {
			__le16 i_extra_isize;
			__le16 i_padding;
			__le32 i_extra_end[0];
		};
		__le32 i_addr[DEF_ADDRS_PER_INODE];
	};
        __le32 i_nid[DEF_NIDS_PER_INODE];
}

Once F2FS_EXTRA_ATTR is set, we will steal four bytes in the head of
i_addr field for storing i_extra_isize and i_padding. with i_extra_isize,
we can calculate actual size of reserved space in i_addr, available
attribute fields included in total extra attribute fields for current
inode can be described as below:

  +--------------------+
  | .i_mode            |
  | ...                |
  | .i_ext             |
  +--------------------+
  | .i_extra_isize     |-----+
  | .i_padding         |     |
  | .i_prjid           |     |
  | .i_atime_extra     |     |
  | .i_ctime_extra     |     |
  | .i_mtime_extra     |<----+
  | .i_inode_cs        |<----- store blkaddr/inline from here
  | .i_xattr_cs        |
  | ...                |
  +--------------------+
  |                    |
  |    block address   |
  |                    |
  +--------------------+
  | .i_nid             |
  +--------------------+
  |   node_footer      |
  | (nid, ino, offset) |
  +--------------------+

Hence, with this patch, we would enhance scalability of f2fs inode for
storing more newly added attribute.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch tries to make below macros calculating max inline size,
inline dentry field size considerring reserving size-changeable
space:
- MAX_INLINE_DATA
- NR_INLINE_DENTRY
- INLINE_DENTRY_BITMAP_SIZE
- INLINE_RESERVED_SIZE

Then, when inline_{data,dentry} options is enabled, it allows us to
reserve inline space with different size flexibly for adding newly
introduced inode attribute.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds to support plain user/group quota.

Change Note by Jaegeuk Kim.

- Use f2fs page cache for quota files in order to consider garbage collection.
  so, quota files are not tolerable for sudden power-cuts, so user needs to do
  quotacheck.

- setattr() calls dquot_transfer which will transfer inode->i_blocks.
  We can't reclaim that during f2fs_evict_inode(). So, we need to count
  node blocks as well in order to match i_blocks with dquot's space.

  Note that, Chao wrote a patch to count inode->i_blocks without inode block.
  (f2fs: don't count inode block in in-memory inode.i_blocks)

- in f2fs_remount, we need to make RW in prior to dquot_resume.

- handle fault_injection case during f2fs_quota_off_umount

- TODO: Project quota
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously, we count all inode consumed blocks including inode block,
xattr block, index block, data block into i_blocks, for other generic
filesystems, they won't count inode block into i_blocks, so for
userspace applications or quota system, they may detect incorrect block
count according to i_blocks value in inode.

This patch changes to count all blocks into inode.i_blocks excluding
inode block, for on-disk i_blocks, we keep counting inode block for
backward compatibility.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Both in memory or on disk, generic filesystems record i_blocks with
512bytes sized sector count, also VFS sub module such as disk quota
follows this rule, but f2fs records it with 4096bytes sized block
count, this difference leads to that once we use dquota's function
which inc/dec iblocks, it will make i_blocks of f2fs being inconsistent
between in memory and on disk.

In order to resolve this issue, this patch changes to make in-memory
i_blocks of f2fs recording sector count instead of block count,
meanwhile leaving on-disk i_blocks recording block count.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

For example,

f2fs_create
 - new_node_page is failed
 - handle_failed_inode
  - skip to add it into orphan list, since ni.blk_addr == NULL_ADDR
   : set_inode_flag(inode, FI_FREE_NID)

f2fs_evict_inode
 - EIO due to fault injection
 - f2fs_bug_on() is triggered

So, we don't need to call f2fs_bug_on in this case.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch fixes missing inode flag loaded from disk, reported by Tom.

[tom@localhost ~]$ sudo mount /dev/loop0 /mnt/
[tom@localhost ~]$ sudo chown tom:tom /mnt/
[tom@localhost ~]$ touch /mnt/testfile
[tom@localhost ~]$ sudo chattr +i /mnt/testfile
[tom@localhost ~]$ echo test > /mnt/testfile
bash: /mnt/testfile: Operation not permitted
[tom@localhost ~]$ rm /mnt/testfile
rm: cannot remove '/mnt/testfile': Operation not permitted
[tom@localhost ~]$ sudo umount /mnt/
[tom@localhost ~]$ sudo mount /dev/loop0 /mnt/
[tom@localhost ~]$ lsattr /mnt/testfile
----i-------------- /mnt/testfile
[tom@localhost ~]$ echo test > /mnt/testfile
[tom@localhost ~]$ rm /mnt/testfile
[tom@localhost ~]$ sudo umount /mnt/

Cc: stable@vger.kernel.org
Reported-by: NTom Yan <tom.ty89@outlook.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch also reverts d0db7703 ("f2fs: do SSR in higher priority").

This patch fixes out of free segments caused by many small file creation by
1) mkfs -s 1 2G
2) mount
3) untar
 - preoduce 60000 small files burstly
4) sync
 - flush node pages
 - flush imeta

Here, when we do f2fs_balance_fs, we missed # of imeta blocks, resulting in
skipping to check has_not_enough_free_secs.

Another test is done by
1) mkfs -s 12 2G
2) mount
3) untar
 - preoduce 60000 small files burstly
4) sync
 - flush node pages
 - flush imeta

In this case, this patch also fixes wrong block allocation under large section
size.
Reported-by: NWilliam Brana <wbrana@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch fixes the following scenario.

- f2fs_create/f2fs_mkdir             - write_checkpoint
 - f2fs_mark_inode_dirty_sync         - block_operations
                                       - f2fs_lock_all
                                       - f2fs_sync_inode_meta
                                        - f2fs_unlock_all
                                        - sync_inode_metadata
 - f2fs_lock_op
                                         - f2fs_write_inode
                                          - update_inode_page
                                           - get_node_page
                                             return -ENOENT
 - new_inode_page
  - fill_node_footer
 - f2fs_mark_inode_dirty_sync
 - ...
 - f2fs_unlock_op
                                          - f2fs_inode_synced
                                       - f2fs_lock_all
                                       - do_checkpoint

In this checkpoint, we can get an inode page which contains zeros having valid
node footer only.

Cc: <stable@vger.kernel.org>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

The previous one was not a proper location to inject an error, since there
is no point to get errors. Instead, we can emulate EIO during truncation,
and the below logic should handle it correctly.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If f2fs_new_inode() is failed, the bad inode will invalidate 0'th node page
during f2fs_evict_inode(), which doesn't need to do.
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously kernel message can show that in which function we do the
injection, but unfortunately, most of the caller are the same, for
tracking more information of injection path, it needs to show upper
caller's name. This patch supports that ability.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

We should record updating status of inode only for living inode, for those
unlinked inode it needs to clear its ino cache, otherwise after the ino
was been reused, it will cause unneeded node page writing during ->fsync.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch tries to make more clean inodes when flushing dirty inodes in
checkpoint.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This is to avoid no free segment bug during checkpoint caused by a number of
dirty inodes.

The case was reported by Chao like this.
1. mount with lazytime option
2. fill 4k file until disk is full
3. sync filesystem
4. read all files in the image
5. umount

In this case, we actually don't need to flush dirty inode to inode page during
checkpoint.
Reviewed-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

We don't need to keep incomplete created inode in cache, so if we fail to
add link into directory during new inode creation, it's better to set
nlink of inode to zero, then we can evict inode immediately. Otherwise
release of nid belong to inode will be delayed until inode cache is being
shrunk, it may cause a seemingly endless loop while allocating free nids
in time of testing generic/269 case of fstest suit.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: add update_inode_page to fix kernel panic]
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

In order to avoid racing problem, make largest extent cache being updated
under lock.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously, we only support global fault injection configuration, so that
when we configure type/rate of fault injection through sysfs, mount
option, it will influence all f2fs partition which is being used.

It is not make sence, since it will be not convenient if developer want
to test separated partitions with different fault injection rate/type
simultaneously, also it's not possible to enable fault injection in one
partition and disable fault injection in other one.

>From now on, we move global configuration of fault injection in module
into per-superblock, hence injection testing can be more flexible.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch gives another chances during roll-forward recovery regarding to
-ENOMEM.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Let's check inode's dirtiness before calling mark_inode_dirty.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Let's store orphan inode pages right away.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

FI_DIRTY_INODE flag is not covered by inode page lock, so it can be unset
at any time like below.

Thread #1                        Thread #2
- lock_page(ipage)
- update i_fields
                                 - update i_size/i_blocks/and so on
				 - set FI_DIRTY_INODE
- reset FI_DIRTY_INODE
- set_page_dirty(ipage)

In this case, we can lose the latest i_field information.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

We don't need lock parameter, which is always true.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>