Re: [PATCH 05/18] xfs: don't use vfs writeback for pure metadata modifications

[Christoph Hellwig <hch@xxxxxxxxxxxxx>]

As mentioned before I think xfs_trans_ichgtime should go into
xfs_trans_inode.c/xfs_trans.h.  Instead of keeping xfs_ichgtime around
for one caller it might be easier to just opencode it similar to
what other filesystems do, e.g. just:

        inode->i_mtime = inode->i_ctime = current_fs_time(inode->i_sb);
        mark_inode_dirty_sync(inode);

there's really no point in doing the check for an actual time change
first as it's a relatively rare operation.