---- 在 星期三, 2021-02-24 16:51:03 Chengguang Xu <cgxu519@xxxxxxxxxxxx> 撰写 ----
> ---- 在 星期三, 2021-02-24 15:52:17 Su Yue <l@xxxxxxxxxx> 撰写 ----
> >
> > Cc to the author and linux-xfs, since it's xfsprogs related.
> >
> > On Tue 23 Feb 2021 at 21:40, Chengguang Xu <cgxu519@xxxxxxxxxxxx>
> > wrote:
> >
> > > It seems the expected result of testcase of "Hole + Data"
> > > in generic/473 is not correct, so just fix it properly.
> > >
> >
> > But it's not proper...
> >
> > > Signed-off-by: Chengguang Xu <cgxu519@xxxxxxxxxxxx>
> > > ---
> > > tests/generic/473.out | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/tests/generic/473.out b/tests/generic/473.out
> > > index 75816388..f1ee5805 100644
> > > --- a/tests/generic/473.out
> > > +++ b/tests/generic/473.out
> > > @@ -6,7 +6,7 @@ Data + Hole
> > > 1: [256..287]: hole
> > > Hole + Data
> > > 0: [0..127]: hole
> > > -1: [128..255]: data
> > > +1: [128..135]: data
> > >
> > The line is produced by `$XFS_IO_PROG -c "fiemap -v 0 65k" $file |
> > _filter_fiemap`.
> > 0-64k is a hole and 64k-128k is a data extent.
> > fiemap ioctl always returns *complete* ranges of extents.
Finally, I found btrfs returns *complete* rangne of extents but xfs/ext4 does not. :-/
[root@VM-89-226-centos /test]# xfs_io -c "fiemap 0 65k" a
a:
0: [0..127]: hole
1: [128..255]: 24576..24703
[root@VM-89-226-centos /test]# xfs_io -c "fiemap 0 128k" a
a:
0: [0..127]: hole
1: [128..255]: 24576..24703
>
> Manual testing result in latest kernel like below.
>
> [root@centos test]# uname -a
> Linux centos 5.11.0+ #5 SMP Tue Feb 23 21:02:27 CST 2021 x86_64 x86_64 x86_64 GNU/Linux
>
> [root@centos test]# xfs_io -V
> xfs_io version 5.0.0
>
> [root@centos test]# stat a
> File: a
> Size: 4194304 Blocks: 0 IO Block: 4096 regular file
> Device: fc01h/64513d Inode: 140 Links: 1
> Access: (0644/-rw-r--r--) Uid: ( 0/ root) Gid: ( 0/ root)
> Access: 2021-02-24 16:33:20.235654140 +0800
> Modify: 2021-02-24 16:33:25.070641521 +0800
> Change: 2021-02-24 16:33:25.070641521 +0800
> Birth: -
>
> [root@centos test]# xfs_io -c "pwrite 64k 64k" a
> wrote 65536/65536 bytes at offset 65536
> 64 KiB, 16 ops; 0.0000 sec (992.063 MiB/sec and 253968.2540 ops/sec)
>
> [root@VM-8-4-centos test]# xfs_io -c "fiemap -v 0 65k" a
> a:
> EXT: FILE-OFFSET BLOCK-RANGE TOTAL FLAGS
> 0: [0..127]: hole 128
> 1: [128..135]: 360..367 8 0x1
>
> [root@centos test]# xfs_io -c "fiemap -v 0 128k" a
> a:
> EXT: FILE-OFFSET BLOCK-RANGE TOTAL FLAGS
> 0: [0..127]: hole 128
> 1: [128..255]: 360..487 128 0x1
>
>
> >
> > You may ask why the ending hole range is not aligned to 128 in
> > 473.out. Because
> > fiemap ioctl returns nothing of querying holes. xfs_io does the
> > extra
> > print work for holes.
> >
> > xfsprogs-dev/io/fiemap.c:
> > for holes:
> > 153 if (lstart > llast) {
> > 154 print_hole(0, 0, 0, cur_extent, lflag, true, llast,
> > lstart);
> > 155 cur_extent++;
> > 156 num_printed++;
> > 157 }
> >
> > for the ending hole:
> > 381 if (cur_extent && last_logical < range_end)
> > 382 print_hole(foff_w, boff_w, tot_w, cur_extent, lflag,
> > !vflag,
> > 383 BTOBBT(last_logical), BTOBBT(range_end));
> >
> > > Hole + Data + Hole
> > > 0: [0..127]: hole
> > > 1: [128..255]: data
> >
>
