From d323d005ac4a2d413128267af76bb9d71f7303da Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 27 Oct 2015 09:53:45 +0800 Subject: f2fs: support file defragment This patch introduces a new ioctl F2FS_IOC_DEFRAGMENT to support file defragment in a specified range of regular file. This ioctl can be used in very limited workload: if user expects high sequential read performance in randomly written file, this interface can be used for defragmentation, after that file can be written as continuous as possible in the device. Meanwhile, it has side-effect, it will make holes in segments where blocks located originally, so it's better to trigger GC to eliminate fragment in segments. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 195 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 195 insertions(+) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index a197215ad52b..2f392982c597 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1646,6 +1646,199 @@ static int f2fs_ioc_write_checkpoint(struct file *filp, unsigned long arg) return 0; } +static int f2fs_defragment_range(struct f2fs_sb_info *sbi, + struct file *filp, + struct f2fs_defragment *range) +{ + struct inode *inode = file_inode(filp); + struct f2fs_map_blocks map; + struct extent_info ei; + pgoff_t pg_start, pg_end; + unsigned int blk_per_seg = 1 << sbi->log_blocks_per_seg; + unsigned int total = 0, sec_num; + unsigned int pages_per_sec = sbi->segs_per_sec * + (1 << sbi->log_blocks_per_seg); + block_t blk_end = 0; + bool fragmented = false; + int err; + + /* if in-place-update policy is enabled, don't waste time here */ + if (need_inplace_update(inode)) + return -EINVAL; + + pg_start = range->start >> PAGE_CACHE_SHIFT; + pg_end = (range->start + range->len) >> PAGE_CACHE_SHIFT; + + f2fs_balance_fs(sbi); + + mutex_lock(&inode->i_mutex); + + /* writeback all dirty pages in the range */ + err = filemap_write_and_wait_range(inode->i_mapping, range->start, + range->start + range->len); + if (err) + goto out; + + /* + * lookup mapping info in extent cache, skip defragmenting if physical + * block addresses are continuous. + */ + if (f2fs_lookup_extent_cache(inode, pg_start, &ei)) { + if (ei.fofs + ei.len >= pg_end) + goto out; + } + + map.m_lblk = pg_start; + map.m_len = pg_end - pg_start; + + /* + * lookup mapping info in dnode page cache, skip defragmenting if all + * physical block addresses are continuous even if there are hole(s) + * in logical blocks. + */ + while (map.m_lblk < pg_end) { + map.m_flags = 0; + err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_READ); + if (err) + goto out; + + if (!(map.m_flags & F2FS_MAP_FLAGS)) { + map.m_lblk++; + map.m_len--; + continue; + } + + if (blk_end && blk_end != map.m_pblk) { + fragmented = true; + break; + } + blk_end = map.m_pblk + map.m_len; + + map.m_lblk += map.m_len; + map.m_len = pg_end - map.m_lblk; + } + + if (!fragmented) + goto out; + + map.m_lblk = pg_start; + map.m_len = pg_end - pg_start; + + sec_num = (map.m_len + pages_per_sec - 1) / pages_per_sec; + + /* + * make sure there are enough free section for LFS allocation, this can + * avoid defragment running in SSR mode when free section are allocated + * intensively + */ + if (has_not_enough_free_secs(sbi, sec_num)) { + err = -EAGAIN; + goto out; + } + + while (map.m_lblk < pg_end) { + pgoff_t idx; + int cnt = 0; + +do_map: + map.m_flags = 0; + err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_READ); + if (err) + goto clear_out; + + if (!(map.m_flags & F2FS_MAP_FLAGS)) { + map.m_lblk++; + continue; + } + + set_inode_flag(F2FS_I(inode), FI_DO_DEFRAG); + + idx = map.m_lblk; + while (idx < map.m_lblk + map.m_len && cnt < blk_per_seg) { + struct page *page; + + page = get_lock_data_page(inode, idx, true); + if (IS_ERR(page)) { + err = PTR_ERR(page); + goto clear_out; + } + + set_page_dirty(page); + f2fs_put_page(page, 1); + + idx++; + cnt++; + total++; + } + + map.m_lblk = idx; + map.m_len = pg_end - idx; + + if (idx < pg_end && cnt < blk_per_seg) + goto do_map; + + clear_inode_flag(F2FS_I(inode), FI_DO_DEFRAG); + + err = filemap_fdatawrite(inode->i_mapping); + if (err) + goto out; + } +clear_out: + clear_inode_flag(F2FS_I(inode), FI_DO_DEFRAG); +out: + mutex_unlock(&inode->i_mutex); + if (!err) + range->len = (u64)total << PAGE_CACHE_SHIFT; + return err; +} + +static int f2fs_ioc_defragment(struct file *filp, unsigned long arg) +{ + struct inode *inode = file_inode(filp); + struct f2fs_sb_info *sbi = F2FS_I_SB(inode); + struct f2fs_defragment range; + int err; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + if (!S_ISREG(inode->i_mode)) + return -EINVAL; + + err = mnt_want_write_file(filp); + if (err) + return err; + + if (f2fs_readonly(sbi->sb)) { + err = -EROFS; + goto out; + } + + if (copy_from_user(&range, (struct f2fs_defragment __user *)arg, + sizeof(range))) { + err = -EFAULT; + goto out; + } + + /* verify alignment of offset & size */ + if (range.start & (F2FS_BLKSIZE - 1) || + range.len & (F2FS_BLKSIZE - 1)) { + err = -EINVAL; + goto out; + } + + err = f2fs_defragment_range(sbi, filp, &range); + if (err < 0) + goto out; + + if (copy_to_user((struct f2fs_defragment __user *)arg, &range, + sizeof(range))) + err = -EFAULT; +out: + mnt_drop_write_file(filp); + return err; +} + long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) { switch (cmd) { @@ -1679,6 +1872,8 @@ long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg) return f2fs_ioc_gc(filp, arg); case F2FS_IOC_WRITE_CHECKPOINT: return f2fs_ioc_write_checkpoint(filp, arg); + case F2FS_IOC_DEFRAGMENT: + return f2fs_ioc_defragment(filp, arg); default: return -ENOTTY; } -- cgit v1.2.2 From 04ef4b626c324a7529c80ffc45787b274a6fa68a Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 10 Nov 2015 18:44:20 +0800 Subject: f2fs: fix to enable missing ioctl interfaces in ->compat_ioctl In 64-bit kernel f2fs can supports 32-bit ioctl system call by identifying encoded code which is converted from 32-bit one to 64-bit one in ->compat_ioctl. When we introduced new interfaces in ->ioctl, we forgot to enable them in ->compat_ioctl, so enable them for fixing. Signed-off-by: Chao Yu [Jaegeuk Kim: fix wrongly added spaces together] Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 2f392982c597..3f1026caf807 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1901,6 +1901,22 @@ long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg) case F2FS_IOC32_SETFLAGS: cmd = F2FS_IOC_SETFLAGS; break; + case F2FS_IOC32_GETVERSION: + cmd = F2FS_IOC_GETVERSION; + break; + case F2FS_IOC_START_ATOMIC_WRITE: + case F2FS_IOC_COMMIT_ATOMIC_WRITE: + case F2FS_IOC_START_VOLATILE_WRITE: + case F2FS_IOC_RELEASE_VOLATILE_WRITE: + case F2FS_IOC_ABORT_VOLATILE_WRITE: + case F2FS_IOC_SHUTDOWN: + case F2FS_IOC_SET_ENCRYPTION_POLICY: + case F2FS_IOC_GET_ENCRYPTION_PWSALT: + case F2FS_IOC_GET_ENCRYPTION_POLICY: + case F2FS_IOC_GARBAGE_COLLECT: + case F2FS_IOC_WRITE_CHECKPOINT: + case F2FS_IOC_DEFRAGMENT: + break; default: return -ENOIOCTLCMD; } -- cgit v1.2.2 From 3519e3f992995d46c200134cfbf84c61b7a01f4c Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 1 Dec 2015 11:56:52 +0800 Subject: f2fs: use sbi->blocks_per_seg to avoid unnecessary calculation Use sbi->blocks_per_seg directly to avoid unnecessary calculation when using 1 << sbi->log_blocks_per_seg. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 3f1026caf807..96e4e048a8a1 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1654,10 +1654,9 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, struct f2fs_map_blocks map; struct extent_info ei; pgoff_t pg_start, pg_end; - unsigned int blk_per_seg = 1 << sbi->log_blocks_per_seg; + unsigned int blk_per_seg = sbi->blocks_per_seg; unsigned int total = 0, sec_num; - unsigned int pages_per_sec = sbi->segs_per_sec * - (1 << sbi->log_blocks_per_seg); + unsigned int pages_per_sec = sbi->segs_per_sec * blk_per_seg; block_t blk_end = 0; bool fragmented = false; int err; -- cgit v1.2.2 From 0cab80ee0c9e17337468c4c0f96786ccbca693d9 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 1 Dec 2015 11:36:16 +0800 Subject: f2fs: fix to convert inline inode in ->setattr In commit 3c4541452748 ("f2fs: do not trim preallocated blocks when truncating after i_size"), in order to follow the regulation: "truncate(x) where x > i_size will not trim all blocks past i_size." like other file systems, in ->setattr we invoked truncate_setsize instead of f2fs_truncate to avoid unneeded block trimming in such case, but forgot to call f2fs_convert_inline_inode keep consistency of inline data conversion rule. This patch fixes to convert inline data if necessary. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 96e4e048a8a1..a018ed327713 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -686,6 +686,14 @@ int f2fs_setattr(struct dentry *dentry, struct iattr *attr) * larger than i_size. */ truncate_setsize(inode, attr->ia_size); + + /* should convert inline inode here */ + if (f2fs_has_inline_data(inode) && + !f2fs_may_inline_data(inode)) { + err = f2fs_convert_inline_inode(inode); + if (err) + return err; + } inode->i_mtime = inode->i_ctime = CURRENT_TIME; } } -- cgit v1.2.2 From ea212a4a7a432e0ecd0f0f53971b70172b3e7f96 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Mon, 7 Dec 2015 10:18:54 -0800 Subject: f2fs: use lock_buffer when changing superblock When modifying sb contents, we need to use lock its buffer. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index a018ed327713..294e71576cec 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1591,14 +1591,18 @@ static int f2fs_ioc_get_encryption_pwsalt(struct file *filp, unsigned long arg) return err; /* update superblock with uuid */ + lock_buffer(sbi->raw_super_buf); generate_random_uuid(sbi->raw_super->encrypt_pw_salt); + unlock_buffer(sbi->raw_super_buf); err = f2fs_commit_super(sbi, false); mnt_drop_write_file(filp); if (err) { /* undo new data */ + lock_buffer(sbi->raw_super_buf); memset(sbi->raw_super->encrypt_pw_salt, 0, 16); + unlock_buffer(sbi->raw_super_buf); return err; } got_it: -- cgit v1.2.2 From d8fe4f0e74cb27e79b2b500ca6eae9f9b02748b4 Mon Sep 17 00:00:00 2001 From: Fan Li Date: Mon, 14 Dec 2015 13:34:00 +0800 Subject: f2fs: write only the pages in range during defragment @lend of filemap_write_and_wait_range is supposed to be a "offset in bytes where the range ends (inclusive)". Subtract 1 to avoid writing an extra page. Signed-off-by: Fan li Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 294e71576cec..5fac4f259965 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1686,7 +1686,7 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, /* writeback all dirty pages in the range */ err = filemap_write_and_wait_range(inode->i_mapping, range->start, - range->start + range->len); + range->start + range->len - 1); if (err) goto out; -- cgit v1.2.2 From e8931582bdaf4d4a6ee7e520ef0ccd481713a42a Mon Sep 17 00:00:00 2001 From: Fan Li Date: Mon, 14 Dec 2015 15:26:04 +0800 Subject: f2fs: fix to update variable correctly when skip a unmapped block map.m_len should be reduced after skip a block Signed-off-by: Fan li Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 5fac4f259965..9949d0f332c2 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1759,6 +1759,7 @@ do_map: if (!(map.m_flags & F2FS_MAP_FLAGS)) { map.m_lblk++; + map.m_len--; continue; } -- cgit v1.2.2 From a49324f127dec918f5a3b3f145d0bf2fb81f4588 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 15 Dec 2015 13:29:47 +0800 Subject: f2fs: rename {add,remove,release}_dirty_inode to {add,remove,release}_ino_entry remove_dirty_dir_inode will be renamed to remove_dirty_inode as a generic function in following patch for removing directory/regular/symlink inode in global dirty list. Here rename ino management related functions for readability, also in order to avoid name conflict. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 9949d0f332c2..c6d909e9661e 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -275,10 +275,10 @@ sync_nodes: goto out; /* once recovery info is written, don't need to tack this */ - remove_dirty_inode(sbi, ino, APPEND_INO); + remove_ino_entry(sbi, ino, APPEND_INO); clear_inode_flag(fi, FI_APPEND_WRITE); flush_out: - remove_dirty_inode(sbi, ino, UPDATE_INO); + remove_ino_entry(sbi, ino, UPDATE_INO); clear_inode_flag(fi, FI_UPDATE_WRITE); ret = f2fs_issue_flush(sbi); out: -- cgit v1.2.2 From a1c1e9b74ff380176bf3862c764061e0a7efd9bb Mon Sep 17 00:00:00 2001 From: Fan Li Date: Tue, 15 Dec 2015 17:02:41 +0800 Subject: f2fs: fix to reset variable correctlly f2fs_map_blocks will set m_flags and m_len to 0, so we don't need to reset m_flags ourselves, but have to reset m_len to correct value before use it again. Signed-off-by: Fan li Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index c6d909e9661e..1f5892f38018 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1700,7 +1700,6 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, } map.m_lblk = pg_start; - map.m_len = pg_end - pg_start; /* * lookup mapping info in dnode page cache, skip defragmenting if all @@ -1708,14 +1707,13 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, * in logical blocks. */ while (map.m_lblk < pg_end) { - map.m_flags = 0; + map.m_len = pg_end - map.m_lblk; err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_READ); if (err) goto out; if (!(map.m_flags & F2FS_MAP_FLAGS)) { map.m_lblk++; - map.m_len--; continue; } @@ -1726,7 +1724,6 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, blk_end = map.m_pblk + map.m_len; map.m_lblk += map.m_len; - map.m_len = pg_end - map.m_lblk; } if (!fragmented) @@ -1752,14 +1749,13 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, int cnt = 0; do_map: - map.m_flags = 0; + map.m_len = pg_end - map.m_lblk; err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_READ); if (err) goto clear_out; if (!(map.m_flags & F2FS_MAP_FLAGS)) { map.m_lblk++; - map.m_len--; continue; } @@ -1784,7 +1780,6 @@ do_map: } map.m_lblk = idx; - map.m_len = pg_end - idx; if (idx < pg_end && cnt < blk_per_seg) goto do_map; -- cgit v1.2.2 From e8240f656d4d5d718ce8cf6b4ea266d6719ef547 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Tue, 15 Dec 2015 17:19:26 +0800 Subject: f2fs: don't grab super block buffer header all the time We have already got one copy of valid super block in memory, do not grab buffer header of super block all the time. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 1f5892f38018..7f8ca47be0af 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1591,20 +1591,16 @@ static int f2fs_ioc_get_encryption_pwsalt(struct file *filp, unsigned long arg) return err; /* update superblock with uuid */ - lock_buffer(sbi->raw_super_buf); generate_random_uuid(sbi->raw_super->encrypt_pw_salt); - unlock_buffer(sbi->raw_super_buf); err = f2fs_commit_super(sbi, false); - - mnt_drop_write_file(filp); if (err) { /* undo new data */ - lock_buffer(sbi->raw_super_buf); memset(sbi->raw_super->encrypt_pw_salt, 0, 16); - unlock_buffer(sbi->raw_super_buf); + mnt_drop_write_file(filp); return err; } + mnt_drop_write_file(filp); got_it: if (copy_to_user((__u8 __user *)arg, sbi->raw_super->encrypt_pw_salt, 16)) -- cgit v1.2.2 From b9d777b85ff1ff79a1173190317b25bebc404ab4 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Tue, 22 Dec 2015 11:09:35 -0800 Subject: f2fs: check inline_data flag at converting time We can check inode's inline_data flag when calling to convert it. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 58 +++++++++++++++++++++++----------------------------------- 1 file changed, 23 insertions(+), 35 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 7f8ca47be0af..f2effe18d3c5 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -418,19 +418,18 @@ static loff_t f2fs_llseek(struct file *file, loff_t offset, int whence) static int f2fs_file_mmap(struct file *file, struct vm_area_struct *vma) { struct inode *inode = file_inode(file); + int err; if (f2fs_encrypted_inode(inode)) { - int err = f2fs_get_encryption_info(inode); + err = f2fs_get_encryption_info(inode); if (err) return 0; } /* we don't need to use inline_data strictly */ - if (f2fs_has_inline_data(inode)) { - int err = f2fs_convert_inline_inode(inode); - if (err) - return err; - } + err = f2fs_convert_inline_inode(inode); + if (err) + return err; file_accessed(file); vma->vm_ops = &f2fs_file_vm_ops; @@ -604,7 +603,7 @@ int f2fs_truncate(struct inode *inode, bool lock) trace_f2fs_truncate(inode); /* we should check inline_data size */ - if (f2fs_has_inline_data(inode) && !f2fs_may_inline_data(inode)) { + if (!f2fs_may_inline_data(inode)) { err = f2fs_convert_inline_inode(inode); if (err) return err; @@ -688,8 +687,7 @@ int f2fs_setattr(struct dentry *dentry, struct iattr *attr) truncate_setsize(inode, attr->ia_size); /* should convert inline inode here */ - if (f2fs_has_inline_data(inode) && - !f2fs_may_inline_data(inode)) { + if (!f2fs_may_inline_data(inode)) { err = f2fs_convert_inline_inode(inode); if (err) return err; @@ -786,13 +784,11 @@ static int punch_hole(struct inode *inode, loff_t offset, loff_t len) { pgoff_t pg_start, pg_end; loff_t off_start, off_end; - int ret = 0; + int ret; - if (f2fs_has_inline_data(inode)) { - ret = f2fs_convert_inline_inode(inode); - if (ret) - return ret; - } + ret = f2fs_convert_inline_inode(inode); + if (ret) + return ret; pg_start = ((unsigned long long) offset) >> PAGE_CACHE_SHIFT; pg_end = ((unsigned long long) offset + len) >> PAGE_CACHE_SHIFT; @@ -951,11 +947,9 @@ static int f2fs_collapse_range(struct inode *inode, loff_t offset, loff_t len) f2fs_balance_fs(F2FS_I_SB(inode)); - if (f2fs_has_inline_data(inode)) { - ret = f2fs_convert_inline_inode(inode); - if (ret) - return ret; - } + ret = f2fs_convert_inline_inode(inode); + if (ret) + return ret; pg_start = offset >> PAGE_CACHE_SHIFT; pg_end = (offset + len) >> PAGE_CACHE_SHIFT; @@ -1001,11 +995,9 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len, f2fs_balance_fs(sbi); - if (f2fs_has_inline_data(inode)) { - ret = f2fs_convert_inline_inode(inode); - if (ret) - return ret; - } + ret = f2fs_convert_inline_inode(inode); + if (ret) + return ret; ret = filemap_write_and_wait_range(mapping, offset, offset + len - 1); if (ret) @@ -1114,11 +1106,9 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len) f2fs_balance_fs(sbi); - if (f2fs_has_inline_data(inode)) { - ret = f2fs_convert_inline_inode(inode); - if (ret) - return ret; - } + ret = f2fs_convert_inline_inode(inode); + if (ret) + return ret; ret = truncate_blocks(inode, i_size_read(inode), true); if (ret) @@ -1168,11 +1158,9 @@ static int expand_inode_data(struct inode *inode, loff_t offset, if (ret) return ret; - if (f2fs_has_inline_data(inode)) { - ret = f2fs_convert_inline_inode(inode); - if (ret) - return ret; - } + ret = f2fs_convert_inline_inode(inode); + if (ret) + return ret; pg_start = ((unsigned long long) offset) >> PAGE_CACHE_SHIFT; pg_end = ((unsigned long long) offset + len) >> PAGE_CACHE_SHIFT; -- cgit v1.2.2 From 2a3407607028f7c780f1c20faa4e922bf631d340 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Tue, 22 Dec 2015 13:23:35 -0800 Subject: f2fs: call f2fs_balance_fs only when node was changed If user tries to update or read data, we don't need to call f2fs_balance_fs which triggers f2fs_gc, which increases unnecessary long latency. Reviewed-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 26 +++++++++----------------- 1 file changed, 9 insertions(+), 17 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index f2effe18d3c5..888ce4765779 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -40,8 +40,6 @@ static int f2fs_vm_page_mkwrite(struct vm_area_struct *vma, struct dnode_of_data dn; int err; - f2fs_balance_fs(sbi); - sb_start_pagefault(inode->i_sb); f2fs_bug_on(sbi, f2fs_has_inline_data(inode)); @@ -57,6 +55,9 @@ static int f2fs_vm_page_mkwrite(struct vm_area_struct *vma, f2fs_put_dnode(&dn); f2fs_unlock_op(sbi); + if (dn.node_changed) + f2fs_balance_fs(sbi); + file_update_time(vma->vm_file); lock_page(page); if (unlikely(page->mapping != inode->i_mapping || @@ -233,9 +234,6 @@ int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) goto out; } go_write: - /* guarantee free sections for fsync */ - f2fs_balance_fs(sbi); - /* * Both of fdatasync() and fsync() are able to be recovered from * sudden-power-off. @@ -267,6 +265,8 @@ sync_nodes: if (need_inode_block_update(sbi, ino)) { mark_inode_dirty_sync(inode); f2fs_write_inode(inode, NULL); + + f2fs_balance_fs(sbi); goto sync_nodes; } @@ -945,8 +945,6 @@ static int f2fs_collapse_range(struct inode *inode, loff_t offset, loff_t len) if (offset & (F2FS_BLKSIZE - 1) || len & (F2FS_BLKSIZE - 1)) return -EINVAL; - f2fs_balance_fs(F2FS_I_SB(inode)); - ret = f2fs_convert_inline_inode(inode); if (ret) return ret; @@ -993,8 +991,6 @@ static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len, if (ret) return ret; - f2fs_balance_fs(sbi); - ret = f2fs_convert_inline_inode(inode); if (ret) return ret; @@ -1104,12 +1100,12 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len) if (offset & (F2FS_BLKSIZE - 1) || len & (F2FS_BLKSIZE - 1)) return -EINVAL; - f2fs_balance_fs(sbi); - ret = f2fs_convert_inline_inode(inode); if (ret) return ret; + f2fs_balance_fs(sbi); + ret = truncate_blocks(inode, i_size_read(inode), true); if (ret) return ret; @@ -1152,8 +1148,6 @@ static int expand_inode_data(struct inode *inode, loff_t offset, loff_t off_start, off_end; int ret = 0; - f2fs_balance_fs(sbi); - ret = inode_newsize_ok(inode, (len + offset)); if (ret) return ret; @@ -1162,6 +1156,8 @@ static int expand_inode_data(struct inode *inode, loff_t offset, if (ret) return ret; + f2fs_balance_fs(sbi); + pg_start = ((unsigned long long) offset) >> PAGE_CACHE_SHIFT; pg_end = ((unsigned long long) offset + len) >> PAGE_CACHE_SHIFT; @@ -1349,8 +1345,6 @@ static int f2fs_ioc_start_atomic_write(struct file *filp) if (!inode_owner_or_capable(inode)) return -EACCES; - f2fs_balance_fs(F2FS_I_SB(inode)); - if (f2fs_is_atomic_file(inode)) return 0; @@ -1437,8 +1431,6 @@ static int f2fs_ioc_abort_volatile_write(struct file *filp) if (ret) return ret; - f2fs_balance_fs(F2FS_I_SB(inode)); - clear_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE); clear_inode_flag(F2FS_I(inode), FI_VOLATILE_FILE); commit_inmem_pages(inode, true); -- cgit v1.2.2 From c34f42e2cb2d27650549306de5ff36839e9177d6 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Wed, 23 Dec 2015 17:50:30 +0800 Subject: f2fs: report error of do_checkpoint do_checkpoint and write_checkpoint can fail due to reasons like triggering in a readonly fs or encountering IO error of storage device. So it's better to report such error info to user, let user be aware of failure of doing checkpoint. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 888ce4765779..780db8bd2451 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1618,6 +1618,7 @@ static int f2fs_ioc_write_checkpoint(struct file *filp, unsigned long arg) struct inode *inode = file_inode(filp); struct f2fs_sb_info *sbi = F2FS_I_SB(inode); struct cp_control cpc; + int err; if (!capable(CAP_SYS_ADMIN)) return -EPERM; @@ -1628,10 +1629,10 @@ static int f2fs_ioc_write_checkpoint(struct file *filp, unsigned long arg) cpc.reason = __get_cp_reason(sbi); mutex_lock(&sbi->gc_mutex); - write_checkpoint(sbi, &cpc); + err = write_checkpoint(sbi, &cpc); mutex_unlock(&sbi->gc_mutex); - return 0; + return err; } static int f2fs_defragment_range(struct f2fs_sb_info *sbi, -- cgit v1.2.2 From 6d5a1495eebd441216dc96913a4270100b26e104 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Thu, 24 Dec 2015 18:04:56 +0800 Subject: f2fs: let user being aware of IO error Sometimes we keep dumb when IO error occur in lower layer device, so user will not receive any error return value for some operation, but actually, the operation did not succeed. This sould be avoided, so this patch reports such kind of error to user. Signed-off-by: Chao Yu Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 780db8bd2451..2d87a3cf6768 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -259,8 +259,10 @@ sync_nodes: sync_node_pages(sbi, ino, &wbc); /* if cp_error was enabled, we should avoid infinite loop */ - if (unlikely(f2fs_cp_error(sbi))) + if (unlikely(f2fs_cp_error(sbi))) { + ret = -EIO; goto out; + } if (need_inode_block_update(sbi, ino)) { mark_inode_dirty_sync(inode); -- cgit v1.2.2 From e96248bb45d42375b23e1c083ec5a55151503e82 Mon Sep 17 00:00:00 2001 From: Chao Yu Date: Thu, 24 Dec 2015 18:11:32 +0800 Subject: f2fs: clean up f2fs_ioc_write_checkpoint Use f2fs_sync_fs to clean up codes in f2fs_ioc_write_checkpoint. Signed-off-by: Chao Yu [Jaegeuk Kim: remove unused err variable] Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 2d87a3cf6768..91f576a7903c 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1619,8 +1619,6 @@ static int f2fs_ioc_write_checkpoint(struct file *filp, unsigned long arg) { struct inode *inode = file_inode(filp); struct f2fs_sb_info *sbi = F2FS_I_SB(inode); - struct cp_control cpc; - int err; if (!capable(CAP_SYS_ADMIN)) return -EPERM; @@ -1628,13 +1626,7 @@ static int f2fs_ioc_write_checkpoint(struct file *filp, unsigned long arg) if (f2fs_readonly(sbi->sb)) return -EROFS; - cpc.reason = __get_cp_reason(sbi); - - mutex_lock(&sbi->gc_mutex); - err = write_checkpoint(sbi, &cpc); - mutex_unlock(&sbi->gc_mutex); - - return err; + return f2fs_sync_fs(sbi->sb, 1); } static int f2fs_defragment_range(struct f2fs_sb_info *sbi, -- cgit v1.2.2 From 732d56489f21c04f7bf60c675f7d152c9239a09c Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Tue, 29 Dec 2015 15:46:33 -0800 Subject: f2fs: fix f2fs_ioc_abort_volatile_write There are two rules to handle aborting volatile or atomic writes. 1. drop atomic writes - we don't need to keep any stale db data. 2. write journal data - we should keep the journal data with fsync for db recovery. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 91f576a7903c..b04ab40ddc73 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1433,9 +1433,14 @@ static int f2fs_ioc_abort_volatile_write(struct file *filp) if (ret) return ret; - clear_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE); - clear_inode_flag(F2FS_I(inode), FI_VOLATILE_FILE); - commit_inmem_pages(inode, true); + if (f2fs_is_atomic_file(inode)) { + clear_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE); + commit_inmem_pages(inode, true); + } + if (f2fs_is_volatile_file(inode)) { + clear_inode_flag(F2FS_I(inode), FI_VOLATILE_FILE); + ret = f2fs_sync_file(filp, 0, LLONG_MAX, 0); + } mnt_drop_write_file(filp); return ret; -- cgit v1.2.2 From c46a155bdf3c8877719aa63d1bf1d6e79e2a9764 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 31 Dec 2015 13:49:17 -0800 Subject: f2fs: use IPU for fdatasync This patch fixes missing IPU condition when fdatasync is called. With this patch, fdatasync is able to avoid additional node writes for recovery. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index b04ab40ddc73..e3d32f6b4b4f 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -202,7 +202,7 @@ int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) trace_f2fs_sync_file_enter(inode); /* if fdatasync is triggered, let's do in-place-update */ - if (get_dirty_pages(inode) <= SM_I(sbi)->min_fsync_blocks) + if (datasync || get_dirty_pages(inode) <= SM_I(sbi)->min_fsync_blocks) set_inode_flag(fi, FI_NEED_IPU); ret = filemap_write_and_wait_range(inode->i_mapping, start, end); clear_inode_flag(fi, FI_NEED_IPU); -- cgit v1.2.2 From 2a4b8e9fab9cea45d90179d9ee8e718c5ed26457 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 7 Jan 2016 13:52:34 -0800 Subject: f2fs: remove redundant calls This patch removes redundant calls. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index e3d32f6b4b4f..69bc65fd862c 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -267,8 +267,6 @@ sync_nodes: if (need_inode_block_update(sbi, ino)) { mark_inode_dirty_sync(inode); f2fs_write_inode(inode, NULL); - - f2fs_balance_fs(sbi); goto sync_nodes; } @@ -484,7 +482,6 @@ int truncate_data_blocks_range(struct dnode_of_data *dn, int count) F2FS_I(dn->inode)) + ofs; f2fs_update_extent_cache_range(dn, fofs, 0, len); dec_valid_block_count(sbi, dn->inode, nr_free); - set_page_dirty(dn->node_page); sync_inode_page(dn); } dn->ofs_in_node = ofs; -- cgit v1.2.2 From 2c4db1a6f6b42e2a9fb611cbbeb71a3a9a358ee0 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Thu, 7 Jan 2016 14:15:04 -0800 Subject: f2fs: clean up f2fs_balance_fs This patch adds one parameter to clean up all the callers of f2fs_balance_fs. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 69bc65fd862c..ff06827aa369 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -55,8 +55,7 @@ static int f2fs_vm_page_mkwrite(struct vm_area_struct *vma, f2fs_put_dnode(&dn); f2fs_unlock_op(sbi); - if (dn.node_changed) - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, dn.node_changed); file_update_time(vma->vm_file); lock_page(page); @@ -677,7 +676,7 @@ int f2fs_setattr(struct dentry *dentry, struct iattr *attr) err = f2fs_truncate(inode, true); if (err) return err; - f2fs_balance_fs(F2FS_I_SB(inode)); + f2fs_balance_fs(F2FS_I_SB(inode), true); } else { /* * do not trim all blocks after i_size if target size is @@ -732,7 +731,7 @@ static int fill_zero(struct inode *inode, pgoff_t index, if (!len) return 0; - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, true); f2fs_lock_op(sbi); page = get_new_data_page(inode, NULL, index, false); @@ -818,7 +817,7 @@ static int punch_hole(struct inode *inode, loff_t offset, loff_t len) loff_t blk_start, blk_end; struct f2fs_sb_info *sbi = F2FS_I_SB(inode); - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, true); blk_start = (loff_t)pg_start << PAGE_CACHE_SHIFT; blk_end = (loff_t)pg_end << PAGE_CACHE_SHIFT; @@ -921,7 +920,7 @@ static int f2fs_do_collapse(struct inode *inode, pgoff_t start, pgoff_t end) int ret = 0; for (; end < nrpages; start++, end++) { - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, true); f2fs_lock_op(sbi); ret = __exchange_data_block(inode, end, start, true); f2fs_unlock_op(sbi); @@ -1103,7 +1102,7 @@ static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len) if (ret) return ret; - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, true); ret = truncate_blocks(inode, i_size_read(inode), true); if (ret) @@ -1155,7 +1154,7 @@ static int expand_inode_data(struct inode *inode, loff_t offset, if (ret) return ret; - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, true); pg_start = ((unsigned long long) offset) >> PAGE_CACHE_SHIFT; pg_end = ((unsigned long long) offset + len) >> PAGE_CACHE_SHIFT; @@ -1653,7 +1652,7 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, pg_start = range->start >> PAGE_CACHE_SHIFT; pg_end = (range->start + range->len) >> PAGE_CACHE_SHIFT; - f2fs_balance_fs(sbi); + f2fs_balance_fs(sbi, true); mutex_lock(&inode->i_mutex); -- cgit v1.2.2 From d0239e1bf5204d602281f93c01d46bcf3531098d Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Fri, 8 Jan 2016 16:57:48 -0800 Subject: f2fs: detect idle time depending on user behavior This patch adds last time that user requested filesystem operations. This information is used to detect whether system is idle or not later. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index ff06827aa369..3d43857e9892 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -96,6 +96,7 @@ mapped: clear_cold_data(page); out: sb_end_pagefault(inode->i_sb); + f2fs_update_time(sbi, REQ_TIME); return block_page_mkwrite_return(err); } @@ -280,6 +281,7 @@ flush_out: remove_ino_entry(sbi, ino, UPDATE_INO); clear_inode_flag(fi, FI_UPDATE_WRITE); ret = f2fs_issue_flush(sbi); + f2fs_update_time(sbi, REQ_TIME); out: trace_f2fs_sync_file_exit(inode, need_cp, datasync, ret); f2fs_trace_ios(NULL, 1); @@ -485,6 +487,7 @@ int truncate_data_blocks_range(struct dnode_of_data *dn, int count) } dn->ofs_in_node = ofs; + f2fs_update_time(sbi, REQ_TIME); trace_f2fs_truncate_data_blocks_range(dn->inode, dn->nid, dn->ofs_in_node, nr_free); return nr_free; @@ -1236,6 +1239,7 @@ static long f2fs_fallocate(struct file *file, int mode, if (!ret) { inode->i_mtime = inode->i_ctime = CURRENT_TIME; mark_inode_dirty(inode); + f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); } out: @@ -1351,6 +1355,8 @@ static int f2fs_ioc_start_atomic_write(struct file *filp) return ret; set_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE); + f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); + return 0; } @@ -1398,6 +1404,7 @@ static int f2fs_ioc_start_volatile_write(struct file *filp) return ret; set_inode_flag(F2FS_I(inode), FI_VOLATILE_FILE); + f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); return 0; } @@ -1439,6 +1446,7 @@ static int f2fs_ioc_abort_volatile_write(struct file *filp) } mnt_drop_write_file(filp); + f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); return ret; } @@ -1478,6 +1486,7 @@ static int f2fs_ioc_shutdown(struct file *filp, unsigned long arg) default: return -EINVAL; } + f2fs_update_time(sbi, REQ_TIME); return 0; } @@ -1508,6 +1517,7 @@ static int f2fs_ioc_fitrim(struct file *filp, unsigned long arg) if (copy_to_user((struct fstrim_range __user *)arg, &range, sizeof(range))) return -EFAULT; + f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); return 0; } @@ -1531,6 +1541,7 @@ static int f2fs_ioc_set_encryption_policy(struct file *filp, unsigned long arg) sizeof(policy))) return -EFAULT; + f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); return f2fs_process_policy(&policy, inode); #else return -EOPNOTSUPP; @@ -1807,6 +1818,7 @@ static int f2fs_ioc_defragment(struct file *filp, unsigned long arg) } err = f2fs_defragment_range(sbi, filp, &range); + f2fs_update_time(sbi, REQ_TIME); if (err < 0) goto out; -- cgit v1.2.2 From 447135a86659c646017b8e707c1243c186bf2dff Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim Date: Sat, 9 Jan 2016 17:08:38 -0800 Subject: f2fs: should unset atomic flag after successful commit If there is an error during commit, we should keep the flag in order to abort it. Signed-off-by: Jaegeuk Kim --- fs/f2fs/file.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/f2fs/file.c') diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 3d43857e9892..18ddb1e5182a 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1378,8 +1378,10 @@ static int f2fs_ioc_commit_atomic_write(struct file *filp) if (f2fs_is_atomic_file(inode)) { clear_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE); ret = commit_inmem_pages(inode, false); - if (ret) + if (ret) { + set_inode_flag(F2FS_I(inode), FI_ATOMIC_FILE); goto err_out; + } } ret = f2fs_sync_file(filp, 0, LLONG_MAX, 0); -- cgit v1.2.2