Updated: April 19, 2023

Format a Power-Safe filesystem (QNX Neutrino)

Note: You must be logged in as root to run this utility.


mkqnx6fs [-Bqx] [-b blocksize] [-g groups]
         [-i inodes] [-I checksum] [-n blocks] [-O options] [-o options]
         [-r percent] [-T type] [-u uuid] [-v vol_name] host

Runs on:

QNX Neutrino


Write the fs-qnx6 filesystem boot loader only (ipl-diskpc2-fsq6). When you use this option, the utility does not reformat the filesystem or modify it in any other way. The boot loader that the utility writes is valid only for x86 64-bit targets.
-b blocksize
Set the logical blocksize of the filesystem. You can specify the size in bytes or in kilobytes, as follows:
Bytes Kilobytes
1024 1k
2048 2k
4096 4k
8192 8k
16384 16k
32768 32k

The default is 4096 bytes; the case of the “k” is ignored. Varying the blocksize can control various types of fragmentation as well as determine the maximum file size supported.

-g groups
Set the number of allocation groups to subdivide the filesystem. Valid values are in the range 1–64. If you do not specify this option, a value in the range 4–16 is selected automatically based on the filesystem size. An allocation group is a logical concept, not a physical segregation.
-i inodes
Set the maximum number of inodes in the filesystem. Each unique file or directory requires an inode.
-I checksum
The checksum used to protect inodes. Supported checksum types:
  • crc32 — POSIX CRC32 cksum in inodes.
  • f32 — Fletcher-32 cksum in inodes.

By default, no inode checksums are used.

-n blocks
Set the number of logical blocks in the filesystem. This is the total size of the filesystem, from which space is first allocated to the bitmap file and inodes file (so the number of user-accessible blocks will be slightly less than the specified value). For information about these files, see the Power-Safe filesystem chapter in the QNX Neutrino System Architecture Guide.

By default, mkqnx6fs makes the filesystem fully occupy the specified host if it's a physical device; in this case, mkqnx6fs looks at the device, determines its size, and uses all available space. The only difference when -n is used is that the bitmap file and inodes file will be sized to hold metadata for blocks number of blocks, instead of sized to hold metadata only for the current filesystem (meaning potentially less user-accessible space is available when -n is used). But in either case, the overall filesystem size will be exactly the same.

When host is a device and you set this option along with the -x option, mkqnx6fs creates an expandable filesystem, which is a filesystem with a maximum storage capacity that can exceed the size of the physical device (e.g., disk partition) on which it is currently stored.

Note: It's possible to boot from an expandable filesystem that doesn't entirely fill the device partition that hosts it. The image will expand to fill the partition upon first boot.
If host is a regular file, this file will be resized to what is appropriate for the specified number of blocks. If you want to fill an existing partition with this file, use the following formula to calculate the value of blocks:
  • If blocksize is less than or equals 4096:
    ((partition size - 16384) rounded down to a multiple of 4096) / blocksize
  • If blocksize is greater than 4096:
    ((partition size - (12288 + abs(blocksize - 12288) + blocksize) rounded down to a multiple of 4096) / blocksize
-O options
(“Oh”) Set(+) or unset(-) boot options:
  • quiet — stop the boot loader from doing any output, disable the boot image selection menu, and silently boot the default image.
  • cls — clear the screen first (in case the BIOS didn't do it earlier, and there isn't enough room for the menu).

The default is -O-quiet,-cls.

You can use this option with -B to just update the loader and options.

-o options
Set(+) or unset(-) filesystem options:
  • icheck — perform integrity checks on the inodes.
  • lfncksum — enable a cksum algorithm on long filenames (longer than 27 characters), which greatly improves their lookup performance.

The default is -o+lfncksum.

This default is incompatible with the 6.4.0 version of the Power-Safe filesystem. If you wish to format a filesystem that can be mounted read-write by 6.4.0, you must specify -o-lfncksum; otherwise it will allow only read-only mounting.
Operate quietly; don't prompt for confirmation and don't display the resulting configuration of the new filesystem. Without this option, mkqnx6fs will confirm that you meant to format if the host is a block-special device or is currently mounted.
-r percent
Set the percentage of the filesystem to reserve to prevent it from becoming completely full. In general, filesystem performance degrades when the disk is nearly full; this option just makes ENOSPC happen prematurely to stop this. The default is 3%.
-T type
Set the expected usage type of the filesystem; valid values are desktop, runtime, and media. This type is used to pick the appropriate blocksize, number of allocation groups, and number of inodes. It's a hint that's intended to replace explicit -b, -g, -i, and -r values.
-u uuid
Specify a 128-bit UUID for the filesystem, in the UUID “8-4-4-4-12” format. If you don't specify a UUID, mkqnx6fs generates a random, time-based (version 4 UUID) value.
-v vol_name
Specify a volume name of up to 16 characters.
Note: You can't specify both the -u and -v options.
Create a filesystem that can expand to fit a partition. This option is effective only when the host is a physical device and the -n option is used. In this case, -n blocks determines the maximum partition size.

You can then copy the partition to a file (for example, using dd) and transfer the file to the destination partition (a filesystem image). For information on expanding a filesystem to fit the partition it is installed on, see the chkqnx6fs -x option. You can expand the filesystem in other ways also; for example, by copying the partition onto a larger device or appending additional physical devices to the logical partition that stores the filesystem.

The host of the new filesystem. You can specify this as a block-special device or partition (e.g., /dev/hd0t76), as a regular file, or as the root directory of a mounted fs-qnx6 filesystem (which will be resolved to the real host device). If you specify the host using a regular file, that regular file must exist before you run the mkqnx6fs command. For example, you can run the touch utility to create the regular file before you run the command.


The mkqnx6fs utility creates a fresh Power-Safe (fs-qnx6) filesystem on the specified host (typically a hard disk partition, although you can create an image inside a regular file).

The integer fields of the filesystem are maintained as either all little-endian or all big-endian, as dictated by the -e option. Thus no CPU architecture pays a byte-swapping penalty for local disks. The filesystem detects the endianness and swaps if necessary, so you can move a disk across platforms (with a slight penalty in performance).

Summary of filesystem commands

The following table shows the shared objects and related commands for the filesystems:

Partition type Filesystem Shared object Initialize with: Check with:
1, 4, or 6 DOS mkdosfs chkdosfs
7 Windows NTa N/A N/A
11, 12, or 14 FAT32 mkdosfs chkdosfs
77, 78, or 79 QNX 4 N/A N/A N/A
131 Linux (Ext2) N/A N/A
175 Apple Macintosh HFS or HFS Plusa N/A N/A
177, 178, or 179 Power-Safe mkqnx6fs chkqnx6fsb
185 QNX Trusted Disk mkqfs N/A

a Read-only.

b Not usually necessary.

For more information, see the Filesystems chapter of the System Architecture guide.


# mkqnx6fs /dev/hd0t177
All files on /dev/hd0t177 will be lost!
Confirm filesystem re-format (y) or (n): y
Format fs-qnx6: 8040524 blocks, 62816 inodes, 8 groups

Exit status:

The filesystem was formatted successfully.
An error occurred (a descriptive message is written to stderr).