Re: [PATCH v2] f2fs: remain written times to update inode during fsync

[Jaegeuk Kim]

On 04/03, Chao Yu wrote:
> On 2018/4/3 13:23, Jaegeuk Kim wrote:
> > On 04/03, Chao Yu wrote:
> >> On 2018/3/31 0:30, Jaegeuk Kim wrote:
> >>> Change log from v1:
> >>>  - add more description
> >>>
> >>> This fixes xfstests/generic/392.
> >>>
> >>> The failure was caused by different times between 1) one marked in the last
> >>> fsync(2) call and 2) the other given by roll-forward recovery after power-cut.
> >>> The reason was that we skipped updating inode block at 1), since its i_size
> >>> was recoverable along with 4KB-aligned data writes, which was fixed by:
> >>>   "f2fs: fix a wrong condition in f2fs_skip_inode_update"
> >>>
> >>> Signed-off-by: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
> >>> ---
> >>>  fs/f2fs/f2fs.h  | 15 +++++++++++++++
> >>>  fs/f2fs/inode.c |  4 ++++
> >>>  2 files changed, 19 insertions(+)
> >>>
> >>> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> >>> index 000f93f6767e..675c39d85111 100644
> >>> --- a/fs/f2fs/f2fs.h
> >>> +++ b/fs/f2fs/f2fs.h
> >>> @@ -664,6 +664,7 @@ struct f2fs_inode_info {
> >>>  	kprojid_t i_projid;		/* id for project quota */
> >>>  	int i_inline_xattr_size;	/* inline xattr size */
> >>>  	struct timespec i_crtime;	/* inode creation time */
> >>> +	struct timespec i_disk_time[4];	/* inode disk times */
> >>>  };
> >>>  
> >>>  static inline void get_extent_info(struct extent_info *ext,
> >>> @@ -2457,6 +2458,11 @@ static inline void clear_file(struct inode *inode, int type)
> >>>  	f2fs_mark_inode_dirty_sync(inode, true);
> >>>  }
> >>>  
> >>> +static inline bool time_equal(struct timespec a, struct timespec b)
> >>> +{
> >>> +	return (a.tv_sec == b.tv_sec) && (a.tv_nsec == b.tv_nsec);
> >>> +}
> >>
> >> We can use existing timespec_equal in <linux/time.h> instead of defining a
> >> duplicated one?
> > 
> > It is defined with const parameters.
> 
> I didn't get it, const keyword can used to protect parameter not to be changed
> in that function, that will be more safe, so it will be better to use that one?

Oh, I just made a mistake when testing timespec_equal().
I really need more recess time. :P

Change log from v3:
 - use timespec_equal()

This fixes xfstests/generic/392.

The failure was caused by different times between 1) one marked in the last
fsync(2) call and 2) the other given by roll-forward recovery after power-cut.
The reason was that we skipped updating inode block at 1), since its i_size
was recoverable along with 4KB-aligned data writes, which was fixed by:
  "f2fs: fix a wrong condition in f2fs_skip_inode_update"

Signed-off-by: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
---
 fs/f2fs/f2fs.h  | 11 +++++++++++
 fs/f2fs/inode.c |  8 ++++++++
 2 files changed, 19 insertions(+)

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 7102d3965fea..1df7f10476d6 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -664,6 +664,7 @@ struct f2fs_inode_info {
 	kprojid_t i_projid;		/* id for project quota */
 	int i_inline_xattr_size;	/* inline xattr size */
 	struct timespec i_crtime;	/* inode creation time */
+	struct timespec i_disk_time[4];	/* inode disk times */
 };
 
 static inline void get_extent_info(struct extent_info *ext,
@@ -2474,6 +2475,16 @@ static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync)
 			i_size_read(inode) & ~PAGE_MASK)
 		return false;
 
+	if (!timespec_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime))
+		return false;
+	if (!timespec_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime))
+		return false;
+	if (!timespec_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime))
+		return false;
+	if (!timespec_equal(F2FS_I(inode)->i_disk_time + 3,
+						&F2FS_I(inode)->i_crtime))
+		return false;
+
 	down_read(&F2FS_I(inode)->i_sem);
 	ret = F2FS_I(inode)->last_disk_size == i_size_read(inode);
 	up_read(&F2FS_I(inode)->i_sem);
diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c
index 401f09ccce7e..e0d9e8f27ed2 100644
--- a/fs/f2fs/inode.c
+++ b/fs/f2fs/inode.c
@@ -284,6 +284,10 @@ static int do_read_inode(struct inode *inode)
 		fi->i_crtime.tv_nsec = le32_to_cpu(ri->i_crtime_nsec);
 	}
 
+	F2FS_I(inode)->i_disk_time[0] = inode->i_atime;
+	F2FS_I(inode)->i_disk_time[1] = inode->i_ctime;
+	F2FS_I(inode)->i_disk_time[2] = inode->i_mtime;
+	F2FS_I(inode)->i_disk_time[3] = F2FS_I(inode)->i_crtime;
 	f2fs_put_page(node_page, 1);
 
 	stat_inc_inline_xattr(inode);
@@ -444,6 +448,10 @@ void update_inode(struct inode *inode, struct page *node_page)
 	if (inode->i_nlink == 0)
 		clear_inline_node(node_page);
 
+	F2FS_I(inode)->i_disk_time[0] = inode->i_atime;
+	F2FS_I(inode)->i_disk_time[1] = inode->i_ctime;
+	F2FS_I(inode)->i_disk_time[2] = inode->i_mtime;
+	F2FS_I(inode)->i_disk_time[3] = F2FS_I(inode)->i_crtime;
 }
 
 void update_inode_page(struct inode *inode)
-- 
2.15.0.531.g2ccb3012c9-goog