Prompted by an article about Paragon’s recent submission of its own NTFS driver to the Linux kernel (It’ll likely be rejected because it’s 27,000 lines of code without organization), the article touched on the current status of NTFS kernel support.
Does not support write properly, only “read only”
Elsewhere, I saw this described in more detail, write is supported only if…
you’re over-writing the existing file, the over-write is smaller than what already exists
In summary,
The Linux NTFS driver is only fully ReadOnly, and largely does not support writing files or directories.
Wow.
I’ve been writing to mounting NTFS volumes without a problem for years, but has practically all been done from virtual machines to physical volumes. I assume that’s why i haven’t run into problems writing files and directories on NTFS volumes.
Somewhere along the line I already knew there was an unclear difference, so developed a practice to always format the volume using an OS where the format is “native” which would mean that I always used Windows to format NTFS, never Linux even when offered (like when using GParted or GParted live). And I always observed that whenever a Linux tool was used to alter an NTFS volume, the next time Windows used it, the volume was always detected “damaged” and would be fixed running chkdsk.
I thought I had left NTFS-3G long ago, but now I’m reading that is actually the solution to full Linux RW NTFS support… Install NTFS-3G which is a FUSE implementation that existed before the Linux NTFS kernel driver, and specify that when mounting the volume.
Maybe NTFS-3G might run a bit slower than a native kernel driver, but I’ll almost always prefer a fully complete solution with fewer problems.
NTFS-3G existed long before the acceptance of the NTFS kernel module.
My assumption is that new technology should be superior to whatever it replaces,
But inability to support writes (more specifically create new files and directories) seems to me a particularly grievous drawback.
Anyway,
Thanks to other technology I seem to have been shielded from this problem (virtualization drivers commonly provide full support for all expected functions).
Side Observation:
There is some discussion about benefits of native kernel support for NTFS, for example this makes it possible to install Linux on NTFS, theoretically maybe in a folder on a MSWindows system. Maybe something that could be useful for instance installing Linux in a MSWindows directory, and running Linux on Windows without a specially mounted file system (I’d still recommend isolating the kernel processes from Windows, though).