mount.captive - mount(8) interface for NTFS disk access


mount.captive-ntfs {image-file|device} mountdir [-n] [-v] [-o options]

mount -t captive-ntfs [-n] [-v] [-o options] {image-file|device} dir


mount.captive(8) provides mount(8) interface to captive(7) library You should never call this command directly - use mount(8) instead.

mount.captive(8) (captive filesystem type) is never used - this command is provided just as a base mount(8) interface to captive(7). You must always use captive-fstype filesystem type such as captive-ntfs. Other supported filesystem types are: ntfs, fastfat, cdfs, ext2fsd



Pathname such as /dev/hda1 or /tmp/ntfs.bin. You should refer to the partition name, not the whole device (/dev/hda is forbidden). /dev/hda1 may correspond to /dev/ide/host0/bus0/target0/lun0/part1 on your system.


Existing empty target directory where {image-file|device} will be mounted.


No effect. The standard functionality of 'do not modify /etc/mtab' is not supported - /etc/mtab is always modified.


No effect. The standard functionality of 'verbose mode' is not supported.

-o options

Custom options passed to captive(7) separated by comma (,).

Options without double-dashes (--) will be used for FUSE, please see their description in the section ``Mount options'' of your local /usr/share/doc/fuse/README or the online:

Options prefixed by double-dashes (--) are used for libcaptive, the options list:


Pathname to .sys or .so filesystem module file. You will use exactly once this option. Possible choices are /usr/local/var/lib/captive/ntfs.sys etc.


Pathname to any W32 module to load w/o initialization. Multiple modules can be loaded although in common case you will use just /usr/local/var/lib/captive/ntoskrnl.exe here.


Pathname to the .captivemodid.xml database of existing W32 module identifications. The default used one is: /usr/local/etc/w32-mod-id.captivemodid.xml You must have this database update for any W32 binary module you are using. If you miss such database you may also try to use --load-untested below.


Load tthe W32 modules despite they may not match the current --modid-path identifications database. If you use this option Captive may fail very easily as such module was never tested before the release and may need some compatibility updates. Still no data should be corrupted even if using this --load-untested option.


Read/write mode: Any write access will be forbidden. You should set this mode for cdfs.sys (CD-ROM filesystem). This option is mutually exclusive with --blind and --rw.


Read/write mode: All writes are just simulated in memory (default). Microsoft Windows filesystem driver will see no difference between --blind and --rw although the UNIX image file/device will be open read/only as for --ro. All the changes get 'written' as long as captive(7) program runs - all the changes will be lost afterwards. This mode is the most suitable for debugging. This option is mutually exclusive with --ro and --rw.


Read/write mode: Write directly to the image file/device. Standard read/write disk mode. You should use --sandbox-server option in this case to have the disk protected against Microsoft Windows filesystem driver crashes. Modified disk image blocks are in --sandbox-server --rw mode buffered in the memory and they get reflected to the disk only after successful completion of all filesystem operations including filesystem unmount. This option is mutually exclusive with --ro and --blind.


Media type: CD-ROM. You must set this media type for cdfs.sys. Virtual Microsoft Windows block device driver used by Captive maps to \Device\CdRom0. This option is mutually exclusive with --disk.


Media type: Disk (default). You must set this media type for all the Microsoft Windows filesystem drivers except cdfs.sys. Virtual Microsoft Windows block device driver used by Captive maps to \Device\CaptiveHarddisk0. This option is mutually exclusive with --cdrom.


Turn on debugging messages. Be prepared for substation debug output. Use of --syslog feature is not recommended in this case.


Pathname to /usr/local/sbin/captive-sandbox-server program, turns on sandboxing. You should always use this option in conjunction with --rw, see it for details. Although this program is setuid root and it drops it privileges to captive user. Your system gets protected by chroot(2), setuid(2), setgid(2) and setrlimit(2) UNIX security features against malicious Microsoft Windows drivers. You should never use this option during debugging.

This option is turned on automatically during the mount operation by mount.captive-ntfs(8). Option needs to be used by hand for the captive-cmdline(2) client.


Specify CORBA IOR of /usr/local/sbin/captive-sandbox-server program, turns on sandboxing. Specified CORBA IOR should be the string starting by ``IOR:'' text. This option is useful only for debugging. No sandbox restarting is possible in this case.


Turn off sandboxing feature (default). No /usr/local/sbin/captive-sandbox-server is run. Microsoft Windows filesystem driver is run in native UNIX environment without any CORBA separation. This option is recommended only for debugging. It is dangerous to use --rw together, see its description for the details.


Pathname to strftime(3) for .captivebug.xml.gz bugreports. Every crash of sandbox child gets bugreported to the specified file. You should attempt to minimize the number of operations from the mount operation till the expected crash to minimize the snapshot file size. --sandbox-server option is required for --bug-pathname.

!!! Be aware '.captivebug.xml.gz' will contain data from your disk drive !!!


Messages sent to syslog(3) instead of stderr. This option gets handy for mount(8) operation as the messages would be lost otherway. Watch our for possible ``Filesystem crash broke dirty object'' messages where some written filesystem data got lost in the case of Microsoft Windows filesystem driver crash.


openlog(3) facility for --syslog. See facility section of openlog(3) man page for details. Lowercased values such as daemon or user are supported.




Jan Kratochvil <>,