The scenario is like this.

One trhead triggers:
  f2fs_write_data_pages
    lock_page
    f2fs_write_data_page
      f2fs_lock_op  <- wait

The other thread triggers:
  f2fs_truncate
    truncate_blocks
      f2fs_lock_op
        truncate_partial_data_page
          lock_page  <- wait for locking the page

This patch resolves this bug by relocating truncate_partial_data_page.
This function is just to truncate user data page and not related to FS
consistency as well.
And, we don't need to call truncate_inline_data. Rather than that,
f2fs_write_data_page will finally update inline_data later.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

The # of inline_data inode is decreased only when it has inline_data.
After clearing the flag, we can't decreased the number.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If a mount option has dirsync, we should call checkpoint for all the directory
operations.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Under memory pressure, let's avoid skipping data writes.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

It needs to write node pages if checkpoint is not doing in order to avoid
memory pressure.
Reviewed-by: NChangman Lee <cm224.lee@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds to control the memory footprint used by ino entries.
This will conduct best effort, not strictly.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds to monitor the number of ino entries.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

The roll-forward mechanism should be activated when the number of active
logs is not 2.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If a system wants to reduce the booting time as a top priority, now we can
use a mount option, -o fastboot.
With this option, f2fs conducts a little bit slow write_checkpoint, but
it can avoid the node page reads during the next mount time.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Let's remove unused macro.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

__submit_merged_bio    f2fs_write_end_io        f2fs_write_end_io
                       wait_io = X              wait_io = x
                       complete(X)              complete(X)
                       wait_io = NULL
wait_for_completion()
free(X)
                                                 spin_lock(X)
                                                 kernel panic

In order to avoid this, this patch removes the wait_io facility.
Instead, we can use wait_on_all_pages_writeback(sbi) to wait for end_ios.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If there is a chance to make a huge sized discard command, we don't need
to split it out, since each blkdev_issue_discard should wait one at a
time.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch simplifies the inline_data usage with the following rule.
1. inline_data is set during the file creation.
2. If new data is requested to be written ranges out of inline_data,
 f2fs converts that inode permanently.
3. There is no cases which converts non-inline_data inode to inline_data.
4. The inline_data flag should be changed under inode page lock.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

There's no point in using test_and_clear_bit_le() when we don't use the
return value of the function. Just use clear_bit_le() instead.

Coverity-id: 1016434
Signed-off-by: NJan Kara <jack@suse.cz>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

We should not discard any data protected by the previous checkpoint all
the time.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

When reading inline data, we should call flush_dcache_page.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

During the write_checkpoint, we should avoid f2fs_gc trigger to avoid any
filesystem consistency.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

f2fs_write_begin() doesn't initialize the 'dn' variable if the inode has
inline data. However it uses its contents to decide whether it should
just zero out the page or load data to it. Thus if we are unlucky we can
zero out page contents instead of loading inline data into a page.

CC: stable@vger.kernel.org
CC: Changman Lee <cm224.lee@samsung.com>
Signed-off-by: NJan Kara <jack@suse.cz>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Signed-off-by: NGu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Rename f2fs_set/clear_bit to f2fs_test_and_set/clear_bit, which mean
set/clear bit and return the old value, for better readability.
Signed-off-by: NGu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Set raw_super default to NULL to avoid the possibly used
uninitialized warning, though we may never hit it in fact.
Signed-off-by: NGu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Introduce f2fs_change_bit to simplify the change bit logic in
function set_to_next_nat{sit}.
Signed-off-by: NGu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Use clear_inode_flag to replace the redundant cond_clear_inode_flag.
Signed-off-by: NGu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Remove the unneeded argument 'type' from __get_victim, use
NO_CHECK_TYPE directly when calling v_ops->get_victim().
Signed-off-by: NGu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If user specifies too low end sector for trimming, f2fs_trim_fs() will
use uninitialized value as a number of trimmed blocks and returns it to
userspace. Initialize number of trimmed blocks early to avoid the
problem.

Coverity-id: 1248809
CC: stable@vger.kernel.org
Signed-off-by: NJan Kara <jack@suse.cz>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch declares f2fs_convert_inline_dir as a static function, which was
reported by kbuild test robot.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

For better performance, we need to use kmap_atomic instead of kmap.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces do_make_empty_dir to mitigate code redundancy
for inline_dentry.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces f2fs_dentry_ptr structure for the use of a function
parameter in inline_dentry operations.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If the inode has inline_dentry, we should not truncate any block indices.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces a core function, f2fs_fill_dentries, to remove
redundant code in f2fs_readdir and f2fs_read_inline_dir.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch fixes wrongly counting inline_data inode numbers.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds status information for inline_dentry inodes.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously, init_inode_metadata does not hold any parent directory's inode
page. So, f2fs_init_acl can grab its parent inode page without any problem.
But, when we use inline_dentry, that page is grabbed during f2fs_add_link,
so that we can fall into deadlock condition like below.

INFO: task mknod:11006 blocked for more than 120 seconds.
      Tainted: G           OE  3.17.0-rc1+ #13
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
mknod           D ffff88003fc94580     0 11006  11004 0x00000000
 ffff880007717b10 0000000000000002 ffff88003c323220 ffff880007717fd8
 0000000000014580 0000000000014580 ffff88003daecb30 ffff88003c323220
 ffff88003fc94e80 ffff88003ffbb4e8 ffff880007717ba0 0000000000000002
Call Trace:
 [<ffffffff8173dc40>] ? bit_wait+0x50/0x50
 [<ffffffff8173d4cd>] io_schedule+0x9d/0x130
 [<ffffffff8173dc6c>] bit_wait_io+0x2c/0x50
 [<ffffffff8173da3b>] __wait_on_bit_lock+0x4b/0xb0
 [<ffffffff811640a7>] __lock_page+0x67/0x70
 [<ffffffff810acf50>] ? autoremove_wake_function+0x40/0x40
 [<ffffffff811652cc>] pagecache_get_page+0x14c/0x1e0
 [<ffffffffa029afa9>] get_node_page+0x59/0x130 [f2fs]
 [<ffffffffa02a63ad>] read_all_xattrs+0x24d/0x430 [f2fs]
 [<ffffffffa02a6ca2>] f2fs_getxattr+0x52/0xe0 [f2fs]
 [<ffffffffa02a7481>] f2fs_get_acl+0x41/0x2d0 [f2fs]
 [<ffffffff8122d847>] get_acl+0x47/0x70
 [<ffffffff8122db5a>] posix_acl_create+0x5a/0x150
 [<ffffffffa02a7759>] f2fs_init_acl+0x29/0xcb [f2fs]
 [<ffffffffa0286a8d>] init_inode_metadata+0x5d/0x340 [f2fs]
 [<ffffffffa029253a>] f2fs_add_inline_entry+0x12a/0x2e0 [f2fs]
 [<ffffffffa0286ea5>] __f2fs_add_link+0x45/0x4a0 [f2fs]
 [<ffffffffa028b5b6>] ? f2fs_new_inode+0x146/0x220 [f2fs]
 [<ffffffffa028b816>] f2fs_mknod+0x86/0xf0 [f2fs]
 [<ffffffff811e3ec1>] vfs_mknod+0xe1/0x160
 [<ffffffff811e4b26>] SyS_mknod+0x1f6/0x200
 [<ffffffff81741d7f>] tracesys+0xe1/0xe6
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

The inode page needs to wait NODE block io.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch removes redundant copied code in find_in_inline_dir.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces to reuse the existing room_for_filename for inline dentry
operation.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds descriptions for the inline dir support in f2fs document.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Add inline dir functions into normal dir ops' function to handle inline ops.
Besides, we enable inline dir mode when a new dir inode is created if
inline_data option is on.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Adds Functions to implement inline dir init/lookup/insert/delete/convert ops.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
[Jaegeuk Kim: remove needless reserved area copy, pointed by Dan Carpenter]
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>