[Linux-NTFS-Dev] ntfs3 sparse file support status?

[Massimiliano G. <mas...@gm...>]

Hello everybody,

I am fstransform author (https://github.com/cosmos72/fstransform),
and I am currently testing it on ntfs3-mounted filesystems,
in order to understand whether fstransform is able to use the recently 
merged
ntfs3 kernel module to convert ntfs filesystems to ext4/xfs etc.

Previous attempts using fuse ntfs-3g or the old, read-only ntfs kernel 
module
had disappointing (read: negative) results due to their limitations, so 
I am hoping
that in-kernel read-write NTFS support would fare better.

My setup is:
* Linux/amd64 + Debian 11 (bookworm)
* kernel 5.17.2, compiled from sources to enable ntfs3 support (Debian 
kernels do not yet enable it)
* a large loop-mounted file containing an NTFS filesystem:

# ls -al ntfs.img
-rw-r----- 1 max users 53603205120 Apr  9 12:55 ntfs.img

# file ntfs.img
ntfs.img: DOS/MBR boot sector, code offset 0x52+2, OEM-ID "NTFS ", 
sectors/cluster 8, Media descriptor 0xf8, sectors/track 63, heads 255, 
hidden sectors 204800, dos < 4.0 BootSector (0x80), FAT (1Y bit by 
descriptor); NTFS, sectors/track 63, sectors 104693759, $MFT start 
cluster 786432, $MFTMirror start cluster 2, bytes/RecordSegment 
2^(-1*246), clusters/index block 1, serial number 060c8eafbc8eace7a; 
contains bootstrap BOOTMGR

# losetup /dev/loop0 ntfs.img
# mount /dev/loop0 /mnt -t ntfs3
# df /mnt
Filesystem     1K-blocks     Used Available Use% Mounted on
/dev/loop0      52346876 20930272  31416604  40% /mnt

# ls -al /mnt
total 262592
[.. lots of files and directories...]

---------------------------------------------------------
My problem: fstransform needs to create a sparse file as large as the 
device inside the
filesystem, so I tried to do the same:

# truncate --size=$((52346876*1024)) /mnt/sparse-file
truncate: failed to truncate '/mnt/sparse-file' at 53603201024 bytes: 
File too large

# rm /mnt/sparse-file
# truncate --size=$((40000000*1024)) /mnt/sparse-file
truncate: failed to truncate '/mnt/sparse-file' at 40960000000 bytes: No 
space left on device

# rm /mnt/sparse-file
# truncate --size=$((30000000*1024)) /mnt/sparse-file
# du /mnt/sparse-file
30000000        /mnt/sparse-file

# df /mnt
Filesystem     1K-blocks     Used Available Use% Mounted on
/dev/loop0      52346876 50930272   1416604  98% /mnt

So it seems that "truncate" creates non-sparse files inside 
NTFS3-mounted filesystems,
although from what I understand NTFS _does_ support sparse files, see 
for example http://ntfs.com/ntfs-sparse.htm

-------------------------------------------------------------
Questions:
1. Is "truncate" behaviour described above intentional/expected?
2. Is there a way to create sparse files inside NTFS3-mounted 
filesystems using Linux tools?

Thanks,

Massimiliano Ghilardi