MDMFS(8)

💅 MDMFS(8) — the mdmfs / mount_mfs glow-up guide 🎀

okay so bestie, you're reading the MDMFS(8) page from the System Manager's Manual, section 8 (that's the admin-commands girlies). grab a strawberry drink, we're configuring an in-memory file system together and it's gonna be iconic. ✨

---

🌸 NAME 🌸

mdmfs, mount_mfs — configure and mount an in-memory file system, either using the md(4) driver or using the tmpfs(4) filesystem. two paths, one destiny, no cap. 💫

---

🦋 SYNOPSIS 🦋

mdmfs [-DLlMNnPStTUX] [-a maxcontig] [-b block-size]
      [-c blocks-per-cylinder-group] [-d max-extent-size]
      [-E path-mdconfig] [-e maxbpg] [-F file] [-f frag-size] [-i bytes]
      [-k skel] [-m percent-free] [-O optimization] [-o mount-options]
      [-p permissions] [-s size] [-T fstype] [-v version]
      [-w user:group] md-device mount-point

like, yeah, that's a lot of letters. it's giving alphabet soup but make it powerful. 🫶

---

💜 DESCRIPTION 💜

so the mdmfs utility is literally designed to be a work-alike and a look-alike of the deprecated mount_mfs. the end result? essentially the same. the way it gets there? completely different, main character energy only. 🔥

working off md-device, mdmfs either:

• creates a tmpfs(4) filesystem, or
• configures an md(4) disk using mdconfig(8), slaps a UFS file system on it (unless -P was specified) using newfs(8), and mounts it using mount(8).

it can also handle geom_uzip(4) compressed disk images — as long as your kernel supports that GEOM class. all the command line options get passed to the appropriate program at the appropriate stage so you get exactly the effect you asked for. it's a whole relay race and everyone ate. 🏃‍♀️💨

when md-device is `auto', mdmfs uses tmpfs(4) if it's present in the kernel or can be loaded as a module. otherwise it falls back to using the md(4) auto-unit, exactly as if you'd specified `md'.

when md-device is `tmpfs', mdmfs mounts a tmpfs(4) filesystem, translating the -s size option (if present) into a `-o size=' mount option. any -o options on the command line get passed straight through to the tmpfs(4) mount. options that are specific to mdconfig(8) or newfs(8)? ignored. bye. 💀

when md-device does not result in tmpfs(4) being used, then an md(4) device gets configured instead. by default, mdmfs creates a swap-based (MD_SWAP) disk with soft-updates enabled and mounts it on mount-point. it uses the md(4) device specified by md-device. if md-device is ‘md’ (no unit number), it'll use md(4)'s auto-unit feature to automatically pick an unused device. unless you say otherwise with one of the options below, it uses the default arguments for all the helper programs. defaults are cute like that. 🎀

the following options are available. where possible, the option letter matches the one mount_mfs used for the same thing — consistency queen. 👑

---

✨ OPTIONS ✨

• -a maxcontig — specify the maximum number of contiguous blocks that get laid out before forcing a rotational delay (check the -d option for the related tea ☕).
• -b block-size — the block size of the file system, in bytes.
• -c blocks-per-cylinder-group — the number of blocks per cylinder group in the file system.
• -D — if you're not using auto-unit, don't run mdconfig(8) to try to detach the unit before attaching it. skip the breakup, go straight to the situationship. 🙈
• -d max-extent-size — the file system may choose to store large files using extents. this parameter specifies the largest extent size that may be used. it is presently limited to its default value, which is 16 times the file system blocksize. (so like, don't get creative with it, bestie. it's locked in. 🔒)
• -E path-mdconfig — use path-mdconfig as the location of the mdconfig(8) utility.
• -e maxbpg — indicate the maximum number of blocks any single file can allocate out of a cylinder group before it's forced to start allocating blocks from another cylinder group. boundaries, we love to see them. 💅
• -F file — create a vnode-backed (MD_VNODE) memory disk backed by file.
• -f frag-size — the fragment size of the file system in bytes.
• -i bytes — number of bytes per inode.
• -k skel — copy the content of directory skel into mount-point.
• -l — enable multilabel MAC on the new file system.
• -L — show the output of the helper programs. by default it's sent to /dev/null (silenced, ghosted, gone 😶‍🌫️).
• -M — create a malloc(9)-backed disk (MD_MALLOC) instead of a swap-backed disk.
• -m percent-free — the percentage of space reserved for the superuser.
• -N — do not actually run the helper programs. this is most useful in conjunction with -X.
• -n — do not create a .snap directory on the new file system.
• -O optimization — select the optimization preference; valid choices are space and time, which optimize for minimum space fragmentation and minimum time spent allocating blocks, respectively. pick your fighter. ⚔️
• -o mount-options — specify the mount options with which to mount the file system. see mount(8) for more info.
• -P — preserve the existing file system; do not run newfs(8). this only makes sense if -F is specified to create a vnode-backed disk.
• -p permissions — set the file (directory) permissions of the mount point mount-point to permissions. the permissions argument can be in any of the mode formats recognized by chmod(1). if symbolic permissions are specified, the operation characters “+” and “-” are interpreted relative to the initial permissions of “a=rwx”.
• -S — do not enable soft-updates on the file system.
• -s size — specify the size of the disk to create. this only makes sense if -F is not specified. that is, it'll work when the backing storage is some form of memory, as opposed to a fixed-size file. the size may include the usual SI suffixes (k, m, g, t, p). a number without a suffix is interpreted as a count of 512-byte sectors. (so 32m slaps, and a bare number means sectors, iykyk 🧚)
• -t — turn on the TRIM enable flag for newfs(8). when used with a file system that issues BIO_DELETE bio requests, md(4) returns deleted blocks to the system memory pool. recycling era ♻️✨
• -T fstype — specify a file system type for a vnode-backed memory disk. any file system supported by the mount(8) command can be specified. this option only makes sense when -F and -P are used.
• -U — enable soft-updates on the file system. this is the default, and is accepted only for compatibility. it's only really useful to negate the -S flag, should such a need occur. (yes she's redundant, she's still invited to the party 🎉)
• -v version — specify the UFS version number for use on the file system; it may be either 1 or 2. the default is derived from the default of the newfs(8) command.
• -w user:group — set the owner and group to user and group, respectively. the arguments have the same semantics as with chown(8), but specifying just a user or just a group is not supported. (both or nothing, babe. 💔)
• -X — print what command will be run before running it, plus other assorted debugging information. narrating your own life, honestly iconic. 🎤

so here's the passing-the-notes situation, bestie 📝:

• the -F and -s options are passed to mdconfig(8) as -f and -s, respectively.
• the -a, -b, -c, -d, -e, -f, -i, -m and -n options are passed to newfs(8) with the same letter.
• the -O option is passed to newfs(8) as -o.
• the -o option is passed to mount(8) with the same letter.
• the -T option is passed to mount(8) as -t.

for info on semantics, refer to the documentation of the programs the options are passed to. they said what they said. 💬

---

🌈 EXAMPLES 🌈

create and mount a 32 megabyte swap-backed file system on /tmp:

mdmfs -s 32m md /tmp

the same file system, created as an entry in /etc/fstab:

md /tmp mfs rw,-s32m 2 0

create and mount a 16 megabyte malloc-backed file system on /tmp using the /dev/md1 device; furthermore, do not use soft-updates on it and mount it async:

mdmfs -M -S -o async -s 16m md1 /tmp

create and mount a geom_uzip(4)-based compressed disk image:

mdmfs -P -F foo.uzip -oro md.uzip /tmp/

mount the same image, specifying the /dev/md1 device:

mdmfs -P -F foo.uzip -oro md1.uzip /tmp/

configure a vnode-backed file system and mount its first partition, using automatic device numbering:

mdmfs -P -F foo.img mds1a /tmp/

mount a vnode-backed cd9660 file system using automatic device numbering:

mdmfs -T cd9660 -P -F foo.iso md /tmp

---

🎀 COMPATIBILITY 🎀

the mdmfs utility, while designed to be compatible with mount_mfs, can absolutely be useful all by itself. but since mount_mfs had some silly defaults, a “compatibility” mode is provided for when bug-to-bug compatibility is what you actually want. (yes, bug-to-bug. we're honoring the mess. 😭)

compatibility is enabled by starting mdmfs with the name mount_mfs or mfs (as returned by getprogname(3)). in this mode, the following behavior, as done by mount_mfs, is duplicated:

• the file mode of mount-point is set by default to 01777 as if -p 1777 was given on the command line.

---

🦋 SEE ALSO 🦋

md(4), tmpfs(4), fstab(5), mdconfig(8), mount(8), newfs(8) — go read the whole friend group, they're all linked. 🔗

---

⭐ HISTORY ⭐

the mdmfs utility appeared in FreeBSD 5.0. a legend was born. 👑

---

💖 AUTHORS 💖

Dima Dorfman — thank you for your service, king. 🫶

---

and that's the whole page, bestie! every flag, every caveat, every default — nothing dropped, nothing skipped, no cap. 🧚‍♀️✨

---

this lil rewrite is serving the Debian flavor of the page, dated October 31, 2019, and it's called MDMFS(8) — mwah, go mount something cute. 💅🎀