Messages in this thread Patch in this message | | | From | Chao Yu <> | Subject | [PATCH 3/3] f2fs: keep isize once block is reserved cross EOF | Date | Sun, 5 Nov 2017 21:53:30 +0800 |
| |
From: Chao Yu <yuchao0@huawei.com>
Without FADVISE_KEEP_SIZE_BIT, we will try to recover file size according to last non-hole block, so in fallocate(), we must set FADVISE_KEEP_SIZE_BIT flag once we have preallocated block cross EOF, instead of when all preallocation is success. Otherwise, file size will be incorrect due to lack of this flag.
Simple testcase to reproduce this:
1. echo 2 > /sys/fs/f2fs/<device>/inject_type 2. echo 10 > /sys/fs/f2fs/<device>/inject_rate 3. run tests/generic/392 4. disable fault injection 5. do remount
Signed-off-by: Chao Yu <yuchao0@huawei.com> --- fs/f2fs/file.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index d9d8b1c372f8..e67f03546391 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -1472,8 +1472,12 @@ static int expand_inode_data(struct inode *inode, loff_t offset, new_size = ((loff_t)pg_end << PAGE_SHIFT) + off_end; } - if (!(mode & FALLOC_FL_KEEP_SIZE) && i_size_read(inode) < new_size) - f2fs_i_size_write(inode, new_size); + if (new_size > i_size_read(inode)) { + if (mode & FALLOC_FL_KEEP_SIZE) + file_set_keep_isize(inode); + else + f2fs_i_size_write(inode, new_size); + } return err; } @@ -1520,8 +1524,6 @@ static long f2fs_fallocate(struct file *file, int mode, if (!ret) { inode->i_mtime = inode->i_ctime = current_time(inode); f2fs_mark_inode_dirty_sync(inode, false); - if (mode & FALLOC_FL_KEEP_SIZE) - file_set_keep_isize(inode); f2fs_update_time(F2FS_I_SB(inode), REQ_TIME); } -- 2.14.1.145.gb3622a4ee
| |