what happens when copy files across two different file systems

Question

There is one problem after learning the file system in Linux kernel, what happens when copy files across two different file systems.

Since we could use mount -t ntfs DEV_BLK MOUNT_POINT to mount NTFS on Linux, the kernel must know how the data organized in disk, so if we try to copy a file from NTFS to ext4, the kernel should process data, attribute, permission, etc.

But NTFS and ext4 use different permission control, is it default behavior for Linux kernel to give 777 permission to the file copied from NTFS when it creates dentry, inode, etc.?

Answer 1

Just an extract of the Linux Kernel source 4.19 :

/* We do not support NTFS ACLs yet. */

Then most of permissions are set accordingly to this quite commented code :

        /* Everyone gets all permissions. */
        vi->i_mode |= S_IRWXUGO;
        /* If read-only, no one gets write permissions. */
        if (IS_RDONLY(vi))
                vi->i_mode &= ~S_IWUGO;
        if (m->flags & MFT_RECORD_IS_DIRECTORY) {
                vi->i_mode |= S_IFDIR;
                /*
                 * Apply the directory permissions mask set in the mount
                 * options.
                 */
                vi->i_mode &= ~vol->dmask;
                /* Things break without this kludge! */
                if (vi->i_nlink > 1)
                        set_nlink(vi, 1);
        } else {
                vi->i_mode |= S_IFREG;
                /* Apply the file permissions mask set in the mount options. */
                vi->i_mode &= ~vol->fmask;
        }

In other words, mainly the permissions of the mount options, not inherited by the ACLs from the file system.

Answer 2

Since we could use mount -t ntfs DEV_BLK MOUNT_POINT to mount NTFS on Linux, the kernel must know how the data organized in disk, so if we try to copy a file from NTFS to ext4, the kernel should process data, attribute, permission, etc.

exactly. This is the case for any file system: a file system driver is needed so that the operating system knows how to read and store data.

But NTFS and ext4 use different permission control, is it default behavior for Linux kernel to give 777 permission to the file copied from NTFS when it creates dentry, inode, etc.?

It's a pretty common phenomenon that different file systems support different access control semantics; that's why most file system drivers (and mount helpers) allow you to set up the mappings at mount time!

So: You can configure that for the ntfs / ntfs3g drivers at mount time. man mount.ntfs-3g will tell you what options it knows, how users and permissions are mapped and so on.