This reports performance regression by Yuanhan Liu.
The basic idea was to reduce one-point mutex, but it turns out this causes
another contention like context swithes.

https://lkml.org/lkml/2015/4/21/11

Until finishing the analysis on this issue, I'd like to revert this for a while.

This reverts commit 78373b73.

in the f2fs_fill_super function, variable "retry" is bool type
i think that it should be set as false.
Signed-off-by: NTaehee Yoo <ap420073@gmail.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Enable inline_data feature by default since it brings us better
performance and space utilization and now has already stable.
Add another option noinline_data to disable it during mount.
Suggested-by: NJaegeuk Kim <jaegeuk@kernel.org>
Suggested-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NWanpeng Li <wanpeng.li@linux.intel.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously, f2fs_write_data_pages has a mutex, sbi->writepages, to serialize
data writes to maximize write bandwidth, while sacrificing multi-threads
performance.
Practically, however, multi-threads environment is much more important for
users. So this patch tries to remove the mutex.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch tries to set SBI_NEED_FSCK flag into sbi only when we fail to recover
in fill_super, so we could skip fscking image when we fail to fill super for
other reason.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

extent tree/node slab cache is created during f2fs insmod,
how, it isn't destroyed during f2fs rmmod, this patch fix
it by destroy extent tree/node slab cache once rmmod f2fs.
Signed-off-by: NWanpeng Li <wanpeng.li@linux.intel.com>
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch enables rb-tree based extent cache in f2fs.

When we mount with "-o extent_cache", f2fs will try to add recently accessed
page-block mappings into rb-tree based extent cache as much as possible, instead
of original one extent info cache.

By this way, f2fs can support more effective cache between dnode page cache and
disk. It will supply high hit ratio in the cache with fewer memory when dnode
page cache are reclaimed in environment of low memory.

Storage: Sandisk sd card 64g
1.append write file (offset: 0, size: 128M);
2.override write file (offset: 2M, size: 1M);
3.override write file (offset: 4M, size: 1M);
...
4.override write file (offset: 48M, size: 1M);
...
5.override write file (offset: 112M, size: 1M);
6.sync
7.echo 3 > /proc/sys/vm/drop_caches
8.read file (size:128M, unit: 4k, count: 32768)
(time dd if=/mnt/f2fs/128m bs=4k count=32768)

Extent Hit Ratio:
		before		patched
Hit Ratio	121 / 1071	1071 / 1071

Performance:
		before		patched
real    	0m37.051s	0m35.556s
user    	0m0.040s	0m0.026s
sys     	0m2.990s	0m2.251s

Memory Cost:
		before		patched
Tree Count:	0		1 (size: 24 bytes)
Node Count:	0		45 (size: 1440 bytes)

v3:
 o retest and given more details of test result.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds a mount option 'extent_cache' in f2fs.

It is try to use a rb-tree based extent cache to cache more mapping information
with less memory if this option is set, otherwise we will use the original one
extent info cache.
Suggested-by: NChangman Lee <cm224.lee@samsung.com>
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Move ext_lock out of struct extent_info, then in the following patches we can
use variables with struct extent_info type as a parameter to pass pure data.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces a batched trimming feature, which submits split discard
commands.

This is to avoid long latency due to huge trim commands.
If fstrim was triggered ranging from 0 to the end of device, we should lock
all the checkpoint-related mutexes, resulting in very long latency.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If device is read-only, we should not proceed data recovery.
But, if the previous checkpoint was done by normal clean shutdown, it's safe to
proceed the recovery, since there will be no data to be recovered.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds FASTBOOT flag into checkpoint as follows.

 - CP_UMOUNT_FLAG is set when system is umounted.
 - CP_FASTBOOT_FLAG is set when intermediate checkpoint having node summaries
   was done.

So, if you get CP_UMOUNT_FLAG from checkpoint, the system was umounted cleanly.
Instead, if there was sudden-power-off, you can get CP_FASTBOOT_FLAG or nothing.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds a mount option, norecovery, which is mostly same as
disable_roll_forward. The only difference is that norecovery should be activated
with read-only mount option.

This can be used when user wants to check whether f2fs is mountable or not
without any recovery process. (e.g., xfstests/200)
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If wrong mount option was requested, f2fs tries to fill_super again.
But, during the next trial, f2fs has no valid mount options, since
parse_options deleted all the separators in the original string.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Currently, there are several variables with Boolean type as below:

struct f2fs_sb_info {
...
	int s_dirty;
	bool need_fsck;
	bool s_closing;
...
	bool por_doing;
...
}

For this there are some issues:
1. there are some space of f2fs_sb_info is wasted due to aligning after Boolean
   type variables by compiler.
2. if we continuously add new flag into f2fs_sb_info, structure will be messed
   up.

So in this patch, we try to:
1. switch s_dirty to Boolean type variable since it has two status 0/1.
2. merge s_dirty/need_fsck/s_closing/por_doing variables into s_flag.
3. introduce an enum type which can indicate different states of sbi.
4. use new introduced universal interfaces is_sbi_flag_set/{set,clear}_sbi_flag
   to operate flags for sbi.

After that, above issues will be fixed.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If the previous checkpoint was done without CP_UMOUNT flag, it needs to do
checkpoint with CP_UMOUNT for the next fast boot.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch fixes to trigger checkpoint with umount flag when kill_sb was called.
In kill_sb, f2fs_sync_fs was finally called, but at this time, f2fs can't do
checkpoint with CP_UMOUNT.
After then, f2fs_put_super is not doing checkpoint, since it is not dirty.

So, this patch adds a flag to indicate f2fs_sync_fs is called during umount.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch removes radix tree after finishing tracing IOs.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds spin_lock to cover radix tree operations in IO tracer.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

There are two slab cache inode_entry_slab and winode_slab using the same
structure as below:

struct dir_inode_entry {
	struct list_head list;	/* list head */
	struct inode *inode;	/* vfs inode pointer */
};

struct inode_entry {
	struct list_head list;
	struct inode *inode;
};

It's a little waste that the two cache can not share their memory space for each
other.
So in this patch we remove one redundant winode_slab slab cache, then use more
universal name struct inode_entry as remaining data structure name of slab,
finally we reuse the inode_entry_slab to store dirty dir item and gc item for
more effective.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch activates f2fs_trace_ios.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Because we have checked the contrary condition in case of "if" judgment, we do
not need to check the condition again in case of "else" judgment. Let's remove
it.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

In f2fs_remount, we will stop gc thread and set need_restart_gc as true when new
option is set without BG_GC, then if any error occurred in the following
procedure, we can restore to start the gc thread.
But after that, We will fail to restore gc thread in start_gc_thread as BG_GC is
not set in new option, so we'd better move this condition judgment out of
start_gc_thread to fix this issue.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
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>

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>

Adds a new mount option 'inline_dentry' for inline dir.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch let inmemory pages be clean all the time.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces a very limited functionality for atomic write support.
In order to support atomic write, this patch adds two ioctls:
 o F2FS_IOC_START_ATOMIC_WRITE
 o F2FS_IOC_COMMIT_ATOMIC_WRITE

The database engine should be aware of the following sequence.
1. open
 -> ioctl(F2FS_IOC_START_ATOMIC_WRITE);
2. writes
  : all the written data will be treated as atomic pages.
3. commit
 -> ioctl(F2FS_IOC_COMMIT_ATOMIC_WRITE);
  : this flushes all the data blocks to the disk, which will be shown all or
  nothing by f2fs recovery procedure.
4. repeat to #2.

The IO pattens should be:

  ,- START_ATOMIC_WRITE                  ,- COMMIT_ATOMIC_WRITE
 CP | D D D D D D | FSYNC | D D D D | FSYNC ...
                      `- COMMIT_ATOMIC_WRITE
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces FITRIM in f2fs_ioctl.
In this case, f2fs will issue small discards and prefree discards as many as
possible for the given area.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch add a new data structure to control checkpoint parameters.
Currently, it presents the reason of checkpoint such as is_umount and normal
sync.
Reviewed-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

In manual of mount, we descript remount as below:

"mount -o remount,rw /dev/foo /dir
After  this call all old mount options are replaced and arbitrary stuff from
fstab is ignored, except the loop= option which is internally generated and
maintained by the mount command."

Previously f2fs do not clear up old mount options when remount_fs, so we have no
chance of disabling previous option (e.g. flush_merge). Fix it.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Block size in f2fs is 4096 bytes, so theoretically, f2fs can support 4096 bytes
sector device at maximum. But now f2fs only support 512 bytes size sector, so
block device such as zRAM which uses page cache as its block storage space will
not be mounted successfully as mismatch between sector size of zRAM and sector
size of f2fs supported.

In this patch we support large sector size in f2fs, so block device with sector
size of 512/1024/2048/4096 bytes can be supported in f2fs.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If user wrote F2FS_IPU_FSYNC:4 in /sys/fs/f2fs/ipu_policy, f2fs_sync_file
only starts to try in-place-updates.
And, if the number of dirty pages is over /sys/fs/f2fs/min_fsync_blocks, it
keeps out-of-order manner. Otherwise, it triggers in-place-updates.

This may be used by storage showing very high random write performance.

For example, it can be used when,

Seq. writes (Data) + wait + Seq. writes (Node)

is pretty much slower than,

Rand. writes (Data)
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Previously f2fs only counts dirty dentry pages, but there is no reason not to
expand the scope.

This patch changes the names on the management of dirty pages and to count
dirty pages in each inode info as well.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If the roll-forward recovery was failed, we'd better conduct fsck.f2fs.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds sbi->need_fsck to conduct fsck.f2fs later.
This flag can only be removed by fsck.f2fs.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

We can make the code a bit simpler because we know that "!retry" is
zero.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Theoretically, our total inodes number is the same as total node number, but
there are three node ids are reserved in f2fs, they are 0, 1 (node nid), and 2
(meta nid), and they should never be used by user, so our total/free inode
number calculated in ->statfs is wrong.

This patch indroduces F2FS_RESERVED_NODE_NUM and then fixes this issue by
recalculating total/free inode number with the macro.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

There are two rules when EIO is occurred.
1. don't write any checkpoint data to preserve the previous checkpoint
2. don't lose the cached dentry/node/meta pages

So, at first, this patch adds set_page_dirty in f2fs_write_end_io's failure.
Then, writing checkpoint/dentry/node blocks is not allowed.

Note that, for the data pages, we can't just throw away by redirtying them.
Otherwise, kworker can fall into infinite loop to flush them.
(Ref. xfstests/019)
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

It needs to check s_dirty under cp_mutex, since s_dirty is reset under that
mutex.
And previous condition was not correct, since we can omit doing checkpoint
when checkpoint was done followed by all the node pages were written back.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>