Re: [PATCH 1/2] ntfs: implement fileattr_get and fileattr_set

[liubaolin]

在 2026/9/2 09:22, Namjae Jeon 写道:
> On Tue, Sep 1, 2026 at 6:15 PM Baolin Liu <liubaolin12138@xxxxxxx> wrote:
> > From: Baolin Liu <liubaolin@xxxxxxxxxx>
> >
> > Wire up the fileattr inode operations so that lsattr(1) and chattr(1)
> > work on ntfs.
> >
> > fileattr_get reports the compressed and encrypted state of the attribute
> > and the immutable and append-only flags, the same states ntfs_getattr()
> > already reports through STATX_ATTR_*.
> >
> > fileattr_set accepts only FS_IMMUTABLE_FL and FS_APPEND_FL. Compression
> > and encryption are rejected because changing them would require
> > rewriting the attribute data. Both accepted flags go to inode->i_flags,
> > so the VFS enforces them.
> >
> > Neither flag has an on-disk representation of its own, so a value set
> > through fileattr_set only lives for the lifetime of the mount. Note also
> > that S_IMMUTABLE is re-derived on every mount, from FILE_ATTR_SYSTEM when
> > mounted with sys_immutable and unconditionally for system files, so
> > clearing it on such a file does not survive a remount either.
> NTFS forces S_IMMUTABLE on system metadata files such as $MFT, but
> this patch allows userspace to clear that protection. The patch should
> reject clearing immutable for these files and for files made immutable
> by the sys_immutable mount option.

Hi Namjae,

  You are right, that is a hole. The driver forces S_IMMUTABLE on those
  files for a reason and userspace should not be able to clear it.

  I will add the check and send a v2.

  Thanks for the review.

Thanks,
Baolin