- 11 May, 2016 3 commits
-
-
Chao Yu authored
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:
Chao Yu <yuchao0@huawei.com> [Jaegeuk Kim: fix bugs and add performance numbers measured in x86.] Signed-off-by:
Jaegeuk Kim <jaegeuk@kernel.org>
-
Chao Yu authored
atomic/volatile ioctl interfaces are exposed to user like other file operation interface, it needs to make them getting exclusion against to each other to avoid potential conflict among these operations in concurrent scenario. Signed-off-by:
-
Chao Yu authored
In interfaces of ioctl, mnt_{want,drop}_write_file should be used for: - get exclusion against file system freezing which may used by lvm snapshot. - do telling filesystem that a write is about to be performed on it, and make sure that the writes are permitted. Signed-off-by:
-
- 07 May, 2016 24 commits
-
-
Jaegeuk Kim authored
Previously f2fs_preallocate_blocks() tries to allocate unaligned blocks. In f2fs_write_begin(), however, prepare_write_begin() does not skip its allocation due to (len != 4KB). So, it needs locking node page twice unexpectedly. Signed-off-by: -
Jaegeuk Kim authored
This patch enables reading node blocks in advance when truncating large data blocks. > time rm $MNT/testfile (500GB) after drop_cachees Before : 9.422 s After : 4.821 s Reported-by:
Stephen Bates <stephen.bates@microsemi.com> Signed-off-by:
-
Jaegeuk Kim authored
This patch is to improve the expand_inode speed in fallocate by allocating data blocks as many as possible in single locked node page. In SSD, # time fallocate -l 500G $MNT/testfile Before : 1m 33.410 s After : 24.758 s Reported-by:
-
Chao Yu authored
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:
-
Jaegeuk Kim authored
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: -
Chao Yu authored
The following panic occurs when truncating inode which has inline xattr to max filesize. [<ffffffffa013d3be>] get_dnode_of_data+0x4e/0x580 [f2fs] [<ffffffffa013aca1>] ? read_node_page+0x51/0x90 [f2fs] [<ffffffffa013ad99>] ? get_node_page.part.34+0xb9/0x170 [f2fs] [<ffffffffa01235b1>] truncate_blocks+0x131/0x3f0 [f2fs] [<ffffffffa01238e3>] f2fs_truncate+0x73/0x100 [f2fs] [<ffffffffa01239d2>] f2fs_setattr+0x62/0x2a0 [f2fs] [<ffffffff811a72c8>] notify_change+0x158/0x300 [<ffffffff8118a42b>] do_truncate+0x6b/0xa0 [<ffffffff8118e539>] ? __sb_start_write+0x49/0x100 [<ffffffff8118a798>] do_sys_ftruncate.constprop.12+0x118/0x170 [<ffffffff8118a82e>] SyS_ftruncate+0xe/0x10 [<ffffffff8169efcf>] tracesys+0xe1/0xe6 [<ffffffffa0139ae0>] get_node_path+0x210/0x220 [f2fs] <ffff880206a89ce8> --[ end trace 5fea664dfbcc6625 ]--- The reason is truncate_blocks tries to truncate all node and data blocks start from specified block offset with value of (max filesize / block size), but actually, our valid max block offset is (max filesize / block size) - 1, so f2fs detects such invalid block offset with BUG_ON in truncation path. This patch lets f2fs skip truncating data which is exceeding max filesize. Signed-off-by:
-
Chao Yu authored
Currently, generic_block_bmap is used in f2fs_bmap, its semantics is when the mapping is been found, return position of target physical block, otherwise return zero. But, previously, when there is no mapping info for specified logical block, f2fs_bmap will map target physical block to a uninitialized variable, which should be wrong. Fix it. Signed-off-by:
-
Jaegeuk Kim authored
This patch removes an obsolete variable used in add_free_nid. Signed-off-by: -
Jaegeuk Kim authored
Even if an inode failed to release its blocks, it should be kept in an orphan inode list, so it will be released later. Signed-off-by: -
Chao Yu authored
Restructure struct seg_entry to eliminate holes in it, after that, in 32-bits machine, it reduces size from 32 bytes to 24 bytes; in 64-bits machine, it reduces size from 56 bytes to 40 bytes. Signed-off-by:
-
Chao Yu authored
Reuse get_extent_info for readability. Signed-off-by:
-
Chao Yu authored
Each of fields in struct f2fs_xattr_entry will be assigned later, so previously we don't need to memset the struct. Signed-off-by:
-
Chao Yu authored
In find_fsync_dnodes, get_tmp_page will read dnode page synchronously, previously, ra_meta_page did the same work, which is redundant, remove it. Signed-off-by:
-
Jaegeuk Kim authored
This patch modifies to retry truncating node blocks in -ENOMEM case. Signed-off-by:
Hou Pengyang <houpengyang@huawei.com> Signed-off-by:
-
Jaegeuk Kim authored
When unmounting filesystem, we should release all the ino entries. Signed-off-by: -
Jaegeuk Kim authored
This patch fixes a couple of bugs regarding to orphan inodes when handling errors. This tries to - call alloc_nid_done with add_orphan_inode in handle_failed_inode - let truncate blocks in f2fs_evict_inode - not make a bad inode due to i_mode change Signed-off-by: -
Jaegeuk Kim authored
This patch injects ENOSPC failures. Signed-off-by: -
Jaegeuk Kim authored
This patch adds page allocation failures. Signed-off-by: -
Jaegeuk Kim authored
This patch injects kmalloc failure given a fault injection rate. Signed-off-by: -
Jaegeuk Kim authored
This patch adds a mount option to select fault ratio. Signed-off-by: -
Jaegeuk Kim authored
This patch converts grab_cache_page to f2fs_grab_cache_page. Signed-off-by: -
Jaegeuk Kim authored
This patch adds f2fs_kmalloc. Signed-off-by: -
Jaegeuk Kim authored
This patch adds a new proc entry to show segment information in more detail. Signed-off-by: -
Jaegeuk Kim authored
This adds macros to be used multiple proc entries. Signed-off-by:
-
- 03 May, 2016 3 commits
-
-
Chao Yu authored
Factor out fsync inode entry operations into {add,del}_fsync_inode. Signed-off-by: -
Chao Yu authored
Commit 28bc106b ("f2fs: support revoking atomic written pages") forgot to clear page private flag correctly, fix it. Signed-off-by:
-
Chao Yu authored
Private data in page should be removed during ->releasepage or ->invalidatepage, otherwise garbage data would be remained in that page. Signed-off-by:
-
- 27 Apr, 2016 4 commits
-
-
Yunlong Song authored
Commit 57b62d29 ("f2fs: fix to report error in f2fs_readdir") causes f2fs_readdir to return -ENOENT when get_lock_data_page returns -ENOENT. However, the original logic is to continue when get_lock_data_page returns -ENOENT, but it forgets to reset err to 0. This will cause getdents64 incorretly return -ENOENT when lastdirent is NULL in getdents64. This will lead to a wrong return value for syscall caller. Signed-off-by:
Yunlong Song <yunlong.song@huawei.com> Reviewed-by:
-
Chao Yu authored
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:
-
Chao Yu authored
The filename length in dirent of may become zero-sized after random junk data injection, once encounter such dirent, find_target_dentry or f2fs_add_inline_entries will run into an infinite loop. So let f2fs being aware of that to avoid deadloop. Signed-off-by:
-
Chao Yu authored
I've changed employer, update my email address to the new one. Signed-off-by:
-
- 26 Apr, 2016 6 commits
-
-
Jaegeuk Kim authored
Under direct IO path with O_(D)SYNC, it needs to set proper APPEND or UPDATE flags, so taht f2fs_sync_file can make its data safe. Acked-by:
-
Jaegeuk Kim authored
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:
-
Jaegeuk Kim authored
The fsync_node_pages should return pass or failure so that user could know fsync is completed or not. Acked-by:
-
Jaegeuk Kim authored
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:
-
Jaegeuk Kim authored
The first page of volatile writes usually contains a sort of header information which will be used for recovery. (e.g., journal header of sqlite) If this is written without other journal data, user needs to handle the stale journal information. Acked-by:
-
Jaegeuk Kim authored
When fsync is called, sync_node_pages finds a proper direct node pages to flush. But, it locks unrelated direct node pages together unnecessarily. Acked-by:
-
