There is a data race between allocate_data_block() and f2fs_sbumit_page_mbio(),
which incur unnecessary reversed bio submission.
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>

Remove deprecated paramter.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

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

This patch removes writepages lock.
We can improve multi-threading performance.

tiobench, 32 threads, 4KB write per fsync on SSD
Before: 25.88 MB/s
After: 28.03 MB/s
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If flush commands do not incur any congestion, we don't need to throw that to
dispatching queue which causes unnecessary latency.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

If roll-forward recovery can recover i_size, we don't need to update inode's
metadata during fsync.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch reduces to call them across the whole tree.
- sync_inode_page()
- update_inode_page()
- update_inode()
- f2fs_write_inode()

Instead, checkpoint will flush all the dirty inode metadata before syncing
node pages.
Note that, this is doable, since we call mark_inode_dirty_sync() for all
inode's field change which needs to update on-disk inode as well.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch registers all the inodes which have dirty metadata to sync when
checkpoint is doing.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch calls mark_inode_dirty_sync() for the following on-disk inode
changes.

 -> largest
 -> ctime/mtime/atime
 -> i_current_depth
 -> i_xattr_nid
 -> i_pino
 -> i_advise
 -> i_flags
 -> i_mode
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces f2fs_i_links_write() to call mark_inode_dirty_sync() when
changing inode->i_links.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces f2fs_i_blocks_write() to call mark_inode_dirty_sync() when
changing inode->i_blocks.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces f2fs_i_size_write() to call mark_inode_dirty_sync() with
i_size_write().
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch refactors to use inode pointer for set_inode_flag and
clear_inode_flag.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Given errors, this patch flushes pending bios as soon as possible.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch uses percpu_counter to avoid stat_lock.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch uses percpu_count for sbi->alloc_valid_block_count.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds percpu_counter for # of dirty pages in inode.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch substitutes percpu_counter for atomic_counter when counting
various types of pages.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This can reduce page counting overhead.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces a new struct f2fs_fault_info and a global f2fs_fault
to save fault injection status. Fault injection entries are created in
/sys/fs/f2fs/fault_injection/ during initializing f2fs module.
Signed-off-by: NSheng Yong <shengyong1@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This adds debug information for # of orphan inodes.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch introduces reserve_new_blocks to make preallocation of multi
blocks as in batch operation, so it can avoid lots of redundant
operation, result in better performance.

In virtual machine, with rotational device:

time fallocate -l 32G /mnt/f2fs/file

Before:
real	0m4.584s
user	0m0.000s
sys	0m4.580s

After:
real	0m0.292s
user	0m0.000s
sys	0m0.272s

In x86, with SSD:

time fallocate -l 500G $MNT/testfile

Before : 24.758 s
After  :  1.604 s
Signed-off-by: NChao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: fix bugs and add performance numbers measured in x86.]
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

When testing f2fs with inline_dentry option, generic/342 reports:
VFS: Busy inodes after unmount of dm-0. Self-destruct in 5 seconds.  Have a nice day...

After rmmod f2fs module, kenrel shows following dmesg:
 =============================================================================
 BUG f2fs_inode_cache (Tainted: G           O   ): Objects remaining in f2fs_inode_cache on __kmem_cache_shutdown()
 -----------------------------------------------------------------------------

 Disabling lock debugging due to kernel taint
 INFO: Slab 0xf51ca0e0 objects=22 used=1 fp=0xd1e6fc60 flags=0x40004080
 CPU: 3 PID: 7455 Comm: rmmod Tainted: G    B      O    4.6.0-rc4+ #16
 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
  00000086 00000086 d062fe18 c13a83a0 f51ca0e0 d062fe38 d062fea4 c11c7276
  c1981040 f51ca0e0 00000016 00000001 d1e6fc60 40004080 656a624f 20737463
  616d6572 6e696e69 6e692067 66326620 6e695f73 5f65646f 68636163 6e6f2065
 Call Trace:
  [<c13a83a0>] dump_stack+0x5f/0x8f
  [<c11c7276>] slab_err+0x76/0x80
  [<c11cbfc0>] ? __kmem_cache_shutdown+0x100/0x2f0
  [<c11cbfc0>] ? __kmem_cache_shutdown+0x100/0x2f0
  [<c11cbfe5>] __kmem_cache_shutdown+0x125/0x2f0
  [<c1198a38>] kmem_cache_destroy+0x158/0x1f0
  [<c176b43d>] ? mutex_unlock+0xd/0x10
  [<f8f15aa3>] exit_f2fs_fs+0x4b/0x5a8 [f2fs]
  [<c10f596c>] SyS_delete_module+0x16c/0x1d0
  [<c1001b10>] ? do_fast_syscall_32+0x30/0x1c0
  [<c13c59bf>] ? __this_cpu_preempt_check+0xf/0x20
  [<c10afa7d>] ? trace_hardirqs_on_caller+0xdd/0x210
  [<c10ad50b>] ? trace_hardirqs_off+0xb/0x10
  [<c1001b81>] do_fast_syscall_32+0xa1/0x1c0
  [<c176d888>] sysenter_past_esp+0x45/0x74
 INFO: Object 0xd1e6d9e0 @offset=6624
 kmem_cache_destroy f2fs_inode_cache: Slab cache still has objects
 CPU: 3 PID: 7455 Comm: rmmod Tainted: G    B      O    4.6.0-rc4+ #16
 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
  00000286 00000286 d062fef4 c13a83a0 f174b000 d062ff14 d062ff28 c1198ac7
  c197fe18 f3c5b980 d062ff20 000d04f2 d062ff0c d062ff0c d062ff14 d062ff14
  f8f20dc0 fffffff5 d062e000 d062ff30 f8f15aa3 d062ff7c c10f596c 73663266
 Call Trace:
  [<c13a83a0>] dump_stack+0x5f/0x8f
  [<c1198ac7>] kmem_cache_destroy+0x1e7/0x1f0
  [<f8f15aa3>] exit_f2fs_fs+0x4b/0x5a8 [f2fs]
  [<c10f596c>] SyS_delete_module+0x16c/0x1d0
  [<c1001b10>] ? do_fast_syscall_32+0x30/0x1c0
  [<c13c59bf>] ? __this_cpu_preempt_check+0xf/0x20
  [<c10afa7d>] ? trace_hardirqs_on_caller+0xdd/0x210
  [<c10ad50b>] ? trace_hardirqs_off+0xb/0x10
  [<c1001b81>] do_fast_syscall_32+0xa1/0x1c0
  [<c176d888>] sysenter_past_esp+0x45/0x74

The reason is: in recovery flow, we use delayed iput mechanism for directory
which has recovered dentry block. It means the reference of inode will be
held until last dirty dentry page being writebacked.

But when we mount f2fs with inline_dentry option, during recovery, dirent
may only be recovered into dir inode page rather than dentry page, so there
are no chance for us to release inode reference in ->writepage when
writebacking last dentry page.

We can call paired iget/iput explicityly for inline_dentry case, but for
non-inline_dentry case, iput will call writeback_single_inode to write all
data pages synchronously, but during recovery, ->writepages of f2fs skips
writing all pages, result in losing dirent.

This patch fixes this issue by obsoleting old mechanism, and introduce a
new dir_list to hold all directory inodes which has recovered datas until
finishing recovery.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch allows fscrypto to handle a second key prefix given by filesystem.
The main reason is to provide backward compatibility, since previously f2fs
used "f2fs:" as a crypto prefix instead of "fscrypt:".
Later, ext4 should also provide key_prefix() to give "ext4:".

One concern decribed by Ted would be kinda double check overhead of prefixes.
In x86, for example, validate_user_key consumes 8 ms after boot-up, which turns
out derive_key_aes() consumed most of the time to load specific crypto module.
After such the cold miss, it shows almost zero latencies, which treats as a
negligible overhead.
Note that request_key() detects wrong prefix in prior to derive_key_aes() even.

Cc: Ted Tso <tytso@mit.edu>
Cc: stable@vger.kernel.org # v4.6
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Reuse get_extent_info for readability.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

When unmounting filesystem, we should release all the ino entries.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch injects ENOSPC failures.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds page allocation failures.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch injects kmalloc failure given a fault injection rate.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds a mount option to select fault ratio.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds f2fs_kmalloc.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

For foreground GC, we cache node blocks in victim section and set them
dirty, then we call sync_node_pages to flush these node pages, but
meanwhile, those node pages which does not locate in victim section
will be flushed together, so more bandwidth and continuous free space
would be occupied.

So for this condition, it's better to leave those unrelated node page
in cache for further write hit, and let CP or VM to flush them afterward.
Signed-off-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

In order to give atomic writes, we should consider power failure during
sync_node_pages in fsync.
So, this patch marks fsync flag only in the last dnode block.
Acked-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch splits the existing sync_node_pages into (f)sync_node_pages.
The fsync_node_pages is used for f2fs_sync_file only.
Acked-by: NChao Yu <yuchao0@huawei.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

With below serials, we will lose parts of dirents:

1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [1-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir

ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root    0 Feb 19 15:12 1
-rw-r--r-- 1 root root    0 Feb 19 15:12 10
-rw-r--r-- 1 root root    0 Feb 19 15:12 100
-????????? ? ?    ?       ?            ? 101
-????????? ? ?    ?       ?            ? 102
-????????? ? ?    ?       ?            ? 103
...

The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.

By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.

However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.

This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.
Signed-off-by: NChao Yu <chao2.yu@samsung.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

Once detecting something to recover, f2fs should stop mounting, given norecovery
and rw mount options.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

This patch adds a sbi flag, SBI_NEED_SB_WRITE, which indicates it needs to
recover superblock when (re)mounting as RW. This is set only when f2fs is
mounted as RO.
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>

PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.

This promise never materialized.  And unlikely will.

We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE.  And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.

Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.

Let's stop pretending that pages in page cache are special.  They are
not.

The changes are pretty straight-forward:

 - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;

 - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;

 - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};

 - page_cache_get() -> get_page();

 - page_cache_release() -> put_page();

This patch contains automated changes generated with coccinelle using
script below.  For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.

The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.

There are few places in the code where coccinelle didn't reach.  I'll
fix them manually in a separate patch.  Comments and documentation also
will be addressed with the separate patch.

virtual patch

@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E

@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E

@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT

@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE

@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK

@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)

@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)

@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: NKirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: NMichal Hocko <mhocko@suse.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The crc function is done bit by bit.
Optimize this by use cryptoapi
crc32 function which is backed by h/w acceleration.
Signed-off-by: NKeith Mok <ek9852@gmail.com>
Signed-off-by: NJaegeuk Kim <jaegeuk@kernel.org>