When testing f2fs on an SSD, I found some 128 page IOs followed by 1 page IO
were issued by f2fs_write_node_pages.
This means that there were some mishandling flows which degrades performance.

Previous f2fs_write_node_pages determines the number of pages to be written,
nr_to_write, as follows.

1. The bio_get_nr_vecs returns 129 pages.
2. The bio_alloc makes a room for 128 pages.
3. The initial 128 pages go into one bio.
4. The existing bio is submitted, and a new bio is prepared for the last 1 page.
5. Finally, sync_node_pages submits the last 1 page bio.

The problem is from the use of bio_get_nr_vecs, so this patch replace it
with max_hw_blocks using queue_max_sectors.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

try_to_free_nats() is usually called with parameter nr_shrink as
	"nm_i->nat_cnt - NM_WOUT_THRESHOLD"
by flush_nat_entries() during checkpointing process.

However, this is inconsistent with the actual threshold check as
	"if (nm_i->nat_cnt < 2 * NM_WOUT_THRESHOLD)"
, which will ignore the free_nats requests when
	NM_WOUT_THRESHOLD < nm_i->nat_cnt < 2 * NM_WOUT_THRESHOLD

So fix the threshold check condition.
Signed-off-by: NHaicheng Li <haicheng.li@linux.intel.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

We call lock_page when we need to update a page after readpage.
Between grab and lock page, the page can be truncated by other thread.
So, we should check the page after lock_page whether it was truncated or not.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

In order to avoid build_free_nid lock contention, let's change the order of
function calls as follows.

At first, check whether there is enough free nids.
 - If available, just get a free nid with spin_lock without any overhead.
 - Otherwise, conduct build_free_nids.
  : scan nat pages, journal nat entries, and nat cache entries.

We should consider carefullly not to serve free nids intermediately made by
build_free_nids.
We can get stable free nids only after build_free_nids is done.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

It is more obvious that add_free_nid checks whether the free nid is zero or not.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Previously, background GC submits many 4KB read requests to load victim blocks
and/or its (i)node blocks.

...
f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb61, blkaddr = 0x3b964ed
f2fs_gc : block_rq_complete: 8,16 R () 499854968 + 8 [0]
f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb6f, blkaddr = 0x3b964ee
f2fs_gc : block_rq_complete: 8,16 R () 499854976 + 8 [0]
f2fs_gc : f2fs_readpage: ino = 1, page_index = 0xb79, blkaddr = 0x3b964ef
f2fs_gc : block_rq_complete: 8,16 R () 499854984 + 8 [0]
...

However, by the fact that many IOs are sequential, we can give a chance to merge
the IOs by IO scheduler.
In order to do that, let's use blk_plug.

...
f2fs_gc : f2fs_iget: ino = 143
f2fs_gc : f2fs_readpage: ino = 143, page_index = 0x1c6, blkaddr = 0x2e6ee
f2fs_gc : f2fs_iget: ino = 143
f2fs_gc : f2fs_readpage: ino = 143, page_index = 0x1c7, blkaddr = 0x2e6ef
<idle> : block_rq_complete: 8,16 R () 1519616 + 8 [0]
<idle> : block_rq_complete: 8,16 R () 1519848 + 8 [0]
<idle> : block_rq_complete: 8,16 R () 1520432 + 96 [0]
<idle> : block_rq_complete: 8,16 R () 1520536 + 104 [0]
<idle> : block_rq_complete: 8,16 R () 1521008 + 112 [0]
<idle> : block_rq_complete: 8,16 R () 1521440 + 152 [0]
<idle> : block_rq_complete: 8,16 R () 1521688 + 144 [0]
<idle> : block_rq_complete: 8,16 R () 1522128 + 192 [0]
<idle> : block_rq_complete: 8,16 R () 1523256 + 328 [0]
...

Note that this issue should be addressed in checkpoint, and some readahead
flows too.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

add tracepoints for tracing the truncate operations
like truncate node/data blocks, f2fs_truncate etc.

Tracepoints are added at entry and exit of operation
to trace the success & failure of operation.
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NPankaj Kumar <pankaj.km@samsung.com>
Acked-by: NSteven Rostedt <rostedt@goodmis.org>
[Jaegeuk: combine and modify the tracepoint structures]
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

In the previous version, f2fs uses global locks according to the usage types,
such as directory operations, block allocation, block write, and so on.

Reference the following lock types in f2fs.h.
enum lock_type {
	RENAME,		/* for renaming operations */
	DENTRY_OPS,	/* for directory operations */
	DATA_WRITE,	/* for data write */
	DATA_NEW,	/* for data allocation */
	DATA_TRUNC,	/* for data truncate */
	NODE_NEW,	/* for node allocation */
	NODE_TRUNC,	/* for node truncate */
	NODE_WRITE,	/* for node write */
	NR_LOCK_TYPE,
};

In that case, we lose the performance under the multi-threading environment,
since every types of operations must be conducted one at a time.

In order to address the problem, let's share the locks globally with a mutex
array regardless of any types.
So, let users grab a mutex and perform their jobs in parallel as much as
possbile.

For this, I propose a new global lock scheme as follows.

0. Data structure
 - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS]
 - f2fs_sb_info -> node_write

1. mutex_lock_op(sbi)
 - try to get an avaiable lock from the array.
 - returns the index of the gottern lock variable.

2. mutex_unlock_op(sbi, index of the lock)
 - unlock the given index of the lock.

3. mutex_lock_all(sbi)
 - grab all the locks in the array before the checkpoint.

4. mutex_unlock_all(sbi)
 - release all the locks in the array after checkpoint.

5. block_operations()
 - call mutex_lock_all()
 - sync_dirty_dir_inodes()
 - grab node_write
 - sync_node_pages()

Note that,
 the pairs of mutex_lock_op()/mutex_unlock_op() and
 mutex_lock_all()/mutex_unlock_all() should be used together.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This patch reduces redundant spin_lock operations in alloc_nid_failed().
The alloc_nid_failed() does not need to delete entry and add one again
by triggering spin_lock and spin_unlock redundantly.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

In order to do GC more reliably, I'd like to lock the vicitm summary page
until its GC is completed, and also prevent any checkpoint process.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

In get_node_page, we do not need to call lock_page all the time.

If the node page is cached as uptodate,

1. grab_cache_page locks the page,
2. read_node_page unlocks the page, and
3. lock_page is called for further process.

Let's avoid this.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Use kmemdup instead of kzalloc and memcpy.
Signed-off-by: NAlexandru Gheorghiu <gheorghiuandru@gmail.com>
Acked-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

In the checkpoint flow, the f2fs investigates the total nat cache entries.
Previously, if an entry has NULL_ADDR, f2fs drops the entry and adds the
obsolete nid to the free nid list.
However, this free nid will be reused sooner, resulting in its nat entry miss.
In order to avoid this, we don't need to drop the nat cache entry at this moment.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

The build_free_nid should not add free nids over nm_i->max_nid.
But, there was a hole that invalid free nid was added by the following scenario.

Let's suppose nm_i->max_nid = 150 and the last NAT page has 100 ~ 200 nids.

build_free_nids
  - get_current_nat_page loads the last NAT page
  - scan_nat_page can add 100 ~ 200 nids
    -> Bug here!
So, when scanning an NAT page, we should check each candidate whether it is
over max_nid or not.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

If the return value of releasepage is equal to zero, the page cannot be reclaimed.
Instead, we should return 1 in order to reclaim clean pages.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

When we build new free nids, let's scan the just next NAT page instead of
skipping a couple of previously scanned pages in order to reuse free nids in
there.
Otherwise, we can use too much wide range of nids even though several nids were
deallocated, and also their node pages can be cached in the node_inode's address
space.
This means that we can retain lots of clean pages in the main memory, which
induces mm's reclaiming overhead.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Currently, f2fs doesn't reclaim any node pages.
However, if we found that a node page was truncated by checking its block
address with zero during f2fs_write_node_page, we should not skip that node
page and return zero to reclaim it.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This patch reduces redundant locking and unlocking pages during read operations.
In f2fs_readpage, let's use wait_on_page_locked() instead of lock_page.
And then, when we need to modify any data finally, let's lock the page so that
we can avoid lock contention.

[readpage rule]
- The f2fs_readpage returns unlocked page, or released page too in error cases.
- Its caller should handle read error, -EIO, after locking the page, which
  indicates read completion.
- Its caller should check PageUptodate after grab_cache_page.
Signed-off-by: NChangman Lee <cm224.lee@samsung.com>
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

In all the breaking conditions in get_node_path, 'n' is used to
track index in offset[] array, but while breaking out also, in all
paths n++ is done.
So, remove the ++ from breaking paths. Also, avoid
reset of 'level=0' in first case.
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NAmit Sahrawat <a.sahrawat@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Optimize and change return path in lookup_free_nid_list
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NAmit Sahrawat <a.sahrawat@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

We can remove the call to find_get_page to get a page from the cache
and check for up-to-date, instead we can make use of grab_cache_page
part itself to fetch the page from the cache.
So, removing the call and moving the PageUptodate at proper place, also
taken care of moving the lock_page condition in the page_hit part.
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NAmit Sahrawat <a.sahrawat@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

The caller of get_nid should be careful not to put lower value than
NODE_DIR1_BLOCK in case of level is zero.
Signed-off-by: NChangman Lee <cm224.lee@samsung.com>
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Previously, f2fs reads several node pages ahead when get_dnode_of_data is called
with RDONLY_NODE flag.
And, this flag is set by the following functions.
- get_data_block_ro
- get_lock_data_page
- do_write_data_page
- truncate_blocks
- truncate_hole

However, this readahead mechanism is initially introduced for the use of
get_data_block_ro to enhance the sequential read performance.

So, let's clarify all the cases with the additional modes as follows.

enum {
	ALLOC_NODE,	/* allocate a new node page if needed */
	LOOKUP_NODE,	/* look up a node without readahead */
	LOOKUP_NODE_RA,	/*
			 * look up a node with readahead called
			 * by get_datablock_ro.
			 */
}
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>

The get_node_page_ra tries to:
1. grab or read a target node page for the given nid,
2. then, call ra_node_page to read other adjacent node pages in advance.

So, when we try to read a target node page by #1, we should submit bio with
READ_SYNC instead of READA.
And, in #2, READA should be used.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>

If the node page was truncated, its block address became zero.
This means that we don't need to write the node page, but have to unlock
NODE_WRITE, decrease the number of dirty node pages, and then unlock_page
before returning the f2fs_write_node_page with zero.
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This patch removes the following build warning:
fs/f2fs/node.c: warning: 'nofs' may be used uninitialized in this function
[-Wuninitialized]:  => 738:8

Note that this is a false alarm.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This patch makes clearer the ambiguous f2fs_gc flow as follows.

1. Remove intermediate checkpoint condition during f2fs_gc
 (i.e., should_do_checkpoint() and GC_BLOCKED)

2. Remove unnecessary return values of f2fs_gc because of #1.
 (i.e., GC_NODE, GC_OK, etc)

3. Simplify write_checkpoint() because of #2.

4. Clarify the main f2fs_gc flow.
 o monitor how many freed sections during one iteration of do_garbage_collect().
 o do GC more without checkpoints if we can't get enough free sections.
 o do checkpoint once we've got enough free sections through forground GCs.

5. Adopt thread-logging (Slack-Space-Recycle) scheme more aggressively on data
  log types. See. get_ssr_segement()
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Let's remove the use of page_cache_release() in f2fs, and instead, use
f2fs_put_page(page, 0) which is exactly same but for code readability.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

We can remove unneeded label unlock_out, avoid unnecessary jump
and reorganize the returning conditions in this function.
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NAmit Sahrawat <a.sahrawat@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

for one thing, it doesn't (and shouldn't) use anything else from dentry;
for another, on some call chains the dentry is fake and should
be eliminated completely.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

If some small bios of dirty node pages are supposed to be issued during the
sequential data writes, there-in well-produced consecutive data bios are able
to be split by the small node bios, resulting in performance degradation.
So, let's collect a number of dirty node pages until reaching a threshold.
And, by default, I set the threshold as 2MB, a segment size.

This improves sequential write performance on i5, 512GB SSD (830 w/ SATA2) as
follows.
Before: 231 MB/s -> After: 255 MB/s
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>
Reviewed-by: NNamjae Jeon <namjae.jeon@samsung.com>

Add __init to functions in init_f2fs_fs for code consistency.
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NAmit Sahrawat <a.sahrawat@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This patch cleans up a couple of unnecessary codes related to unused variables
and return values.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

The new_node_page() is processed as the following procedure.

1. A new node page is allocated.
2. Set PageUptodate with proper footer information.
3. Check if there is a free space for allocation
 4.a. If there is no space, f2fs returns with -ENOSPC.
 4.b. Otherwise, go next.

In the case of step #4.a, f2fs remains a wrong node page in the page cache
with the uptodate flag.

Also, even though a new node page is allocated successfully, an error can be
occurred afterwards due to allocation failure of the other data structures.
In such a case, remove_inode_page() would be triggered, so that we have to
clear uptodate flag in truncate_node() too.

So, we should remove the uptodate flag, if allocation is failed.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

Ruslan reported that f2fs hangs with an infinite loop in f2fs_sync_file():

	while (sync_node_pages(sbi, inode->i_ino, &wbc) == 0)
		f2fs_write_inode(inode, NULL);

The reason was revealed that the cold flag is not set even thought this inode is
a normal file. Therefore, sync_node_pages() skips to write node blocks since it
only writes cold node blocks.

The cold flag is stored to the node_footer in node block, and whenever a new
node page is allocated, it is set according to its file type, file or directory.

But, after sudden-power-off, when recovering the inode page, f2fs doesn't recover
its cold flag.

So, let's assign the cold flag in more right places.

One more thing:
If f2fs_write_inode() returns an error due to whatever situations, there would
be no dirty node pages so that sync_node_pages() returns zero.
(i.e., zero means nothing was written.)
Reported-by: NRuslan N. Marchenko <me@ruff.mobi>
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

When CONFIG_CC_OPTIMIZE_FOR_SIZE is enabled in the kernel, -Os optimisation
flag is passed to gcc for compilation, and somehow while trying to optimize
the code, compiler is might not able to see the initialisation of variable
ne struct variable inside the get_node_info() function and results into
following warning:

fs/f2fs/node.c: In function 'get_node_info':
fs/f2fs/node.c:175:3: warning: 'ne.block_addr' may be used uninitialized in
this function [-Wuninitialized]
fs/f2fs/node.c:265:24: note: 'ne.block_addr' was declared here
fs/f2fs/node.c:176:3: warning: 'ne.ino' may be used uninitialized in this
function [-Wuninitialized]
fs/f2fs/node.c:265:24: note: 'ne.ino' was declared here
fs/f2fs/node.c:177:3: warning: 'ne.version' may be used uninitialized in
this function [-Wuninitialized]
fs/f2fs/node.c:265:24: note: 'ne.version' was declared here

Hence, lets initialise the ne struct variable to zero, which will remove
this warning and also doing this does not seems to making any impact on the
code behavior.
Signed-off-by: NNamjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: NPankaj Kumar <pankaj.km@samsung.com>

As pointed out by Randy Dunlap, this patch removes all usage of "/**" for comment
blocks. Instead, just use "/*".
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This patch should resolve the bugs reported by the sparse tool.
Initial reports were written by "kbuild test robot" managed by fengguang.wu.

In my local machines, I've tested also by running:
> make C=2 CF="-D__CHECK_ENDIAN__"

Accordingly, I've found lots of warnings and bugs related to the endian
conversion. And I've fixed all at this moment.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>

This adds specific functions to manage NAT pages, a cache for NAT entries, free
nids, direct/indirect node blocks for indexing data, and address space for node
pages.

- The key information of an NAT entry consists of a node id and a block address.

- An NAT page is composed of block addresses covered by a certain range of NAT
  entries, which is maintained by the address space of meta_inode.

- A radix tree structure is used to cache NAT entries. The index for the tree
  is a node id.

- When there is no free nid, F2FS should scan NAT entries to find new one. In
  order to avoid scanning frequently, F2FS manages a list containing a number of
  free nids in memory. Only when free nids in the list are exhausted, scanning
  process, build_free_nids(), is triggered.

- F2FS has direct and indirect node blocks for indexing data. This patch adds
  fuctions related to the node block management such as getting, allocating, and
  truncating node blocks to index data.

- In order to cache node blocks in memory, F2FS has a node_inode with an address
  space for node pages. This patch also adds the address space operations for
  node_inode.
Signed-off-by: NJaegeuk Kim <jaegeuk.kim@samsung.com>