There's a suggested patch to not check for block devices on FreeBSD in ntfsprogs/mkntfs.c

See

https://bugs.freebsd.org/bugzilla/show_ ... ?id=197301

for details. I tried to attach the patch to this topic, but the forum software does not like .c, .diff or no extension.

Hi,

I do not understand the rationale for this proposed patch.

Formatting an ntfs device depends on the sector size. If the device is a block device, its sector size can be determined automatically. Otherwise the sector size is either supplied in an option or forced to 512. The original code requires option -F to force a sector size.

IMHO the proposed patch skips defining the sector size for non block devices.

What is the use case leading to this proposal ?


This forum requires the attachments to be compressed (.gz .zip, .bz2, etc.)

Regards

Jean-Pierre

The patch just skips the check for a block device - FreeBSD doesn't have them (only raw devices).
Without it the only way to use it on FreeBSD is with the -F option which isn't very user friendly.

The patch should probably use #ifdef __FreeBSD__ instead of #if 0 though (I wrote the patch originally for adding to the FreeBSD port not upstreaming)

As for checking the sector size - I think on FreeBSD the only way to do that is an ioctl(.., DIOCGSECTORSIZE, ..). I don't have a 4k device I want to put NTFS on to check though.

Hi,


The the corresponding check should be for a character device (is this what you call a raw device ?).

This is how the sector size is retrieved (in libntfs-3g/device.c), called from function mkntfs_override_vol_params() in mkntfs.c, assuming DIOCGSECTORSIZE is defined.

Can you check the attached patch, in particular please check the sector size is retrieved correctly.

Regards

Jean-Pierre