1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-20 11:11:24 +00:00
freebsd/sbin/mdconfig/mdconfig.8
Bruce Evans 9e5ed8593f Use VOP_ADVISE() with POSIX_FADV_DONTNEED instead of IO_DIRECT to
implement not double-caching for reads from vnode-backed md devices.
Use VOP_ADVISE() similarly instead of !IO_DIRECT unsimilarly for writes.
Add a "cache" option to mdconfig to allow changing the default of not
caching.

This depends on a recent commit to fix VOP_ADVISE().  A previous version
had optimizations for sequential i/o's (merge the i/o's and only uncache
for discontiguous i/o's and for full blocks), but optimizations and
knowledge of block boundaries belong in VOP_ADVISE().  Read-ahead should
also be handled better, by supporting it in md and discarding it in
VOP_ADVISE().

POSIX_FADV_DONTNEED is ignored by zfs, but so is IO_DIRECT.

POSIX_FADV_DONTNEED works better than IO_DIRECT if it is not ignored,
since it only discards from the buffer cache immediately, while
IO_DIRECT also discards from the page cache immediately.

IO_DIRECT was not used for writes since it was claimed to be too slow,
but most of the slowness for writes is from doing them synchronously by
default.  Non-synchronous writes still deadlock in many cases.

IO_DIRECT only has a special implementation for ffs reads with DIRECTIO
configured.  Otherwise, if it is not ignored than it uses the buffer and
page caches normally except for discarding everything after each i/o,
and then it has much the same overheads as POSIX_FADV_DONTNEED.  The
overheads for reading with ffs and DIRECTIO were similar in tests of md.

Reviewed by:	kib
2018-12-21 08:15:31 +00:00

363 lines
9.1 KiB
Groff

.\" Copyright (c) 1993 University of Utah.
.\" Copyright (c) 1980, 1989, 1991, 1993
.\" The Regents of the University of California. All rights reserved.
.\" Copyright (c) 2000
.\" Poul-Henning Kamp All rights reserved.
.\"
.\" This code is derived from software contributed to Berkeley by
.\" the Systems Programming Group of the University of Utah Computer
.\" Science Department.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\" 3. Neither the name of the University nor the names of its contributors
.\" may be used to endorse or promote products derived from this software
.\" without specific prior written permission.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.\" @(#)vnconfig.8 8.1 (Berkeley) 6/5/93
.\" from: src/usr.sbin/vnconfig/vnconfig.8,v 1.19 2000/12/27 15:30:29
.\"
.\" $FreeBSD$
.\"
.Dd December 21, 2018
.Dt MDCONFIG 8
.Os
.Sh NAME
.Nm mdconfig
.Nd create and control memory disks
.Sh SYNOPSIS
.Nm
.Fl a
.Fl t Ar type
.Op Fl n
.Oo Fl o Oo Cm no Oc Ns Ar option Oc ...
.Op Fl f Ar file
.Op Fl s Ar size
.Op Fl S Ar sectorsize
.Op Fl u Ar unit
.Op Fl x Ar sectors/track
.Op Fl y Ar heads/cylinder
.Op Fl L Ar label
.Nm
.Fl d
.Fl u Ar unit
.Op Fl o Oo Cm no Oc Ns Ar force
.Nm
.Fl r
.Fl u Ar unit
.Fl s Ar size
.Op Fl o Oo Cm no Oc Ns Ar force
.Nm
.Fl l
.Op Fl n
.Op Fl v
.Op Fl f Ar file
.Op Fl u Ar unit
.Nm
.Ar file
.Sh DESCRIPTION
The
.Nm
utility creates and controls
.Xr md 4
devices.
.Pp
Options indicate an action to be performed:
.Bl -tag -width indent
.It Fl a
Attach a memory disk.
This will configure and attach a memory disk with the
parameters specified and attach it to the system.
If the
.Fl u Ar unit
option is not provided, the newly created device name will be printed on stdout.
.It Fl d
Detach a memory disk from the system and release all resources.
.It Fl r
Resize a memory disk.
.It Fl t Ar type
Select the type of the memory disk.
.Bl -tag -width "malloc"
.It Cm malloc
Storage for this type of memory disk is allocated with
.Xr malloc 9 .
This limits the size to the malloc bucket limit in the kernel.
If the
.Fl o Cm reserve
option is not set, creating and filling a large
malloc-backed memory disk is a very easy way to
panic the system.
.It Cm vnode
A file specified with
.Fl f Ar file
becomes the backing store for this memory disk.
.It Cm swap
Storage for this type of memory disk is allocated from buffer
memory.
Pages get pushed out to swap when the system is under memory
pressure, otherwise they stay in the operating memory.
Using
.Cm swap
backing is generally preferred instead of using
.Cm malloc
backing.
.It Cm null
Bitsink; all writes do nothing, all reads return zeroes.
.El
.It Fl f Ar file
Filename to use for the vnode type memory disk.
The
.Fl a
and
.Fl t Ar vnode
options are implied if not specified.
.It Fl l
List configured devices.
If given with
.Fl u ,
display details about that particular device.
If given with
.Fl f Ar file ,
display
.Xr md 4
device names of which
.Ar file
is used as the backing store.
If both of
.Fl u
and
.Fl f
options are specified,
display devices which match the two conditions.
If the
.Fl v
option is specified, show all details.
.It Fl n
When printing
.Xr md 4
device names, print only the unit number without the
.Xr md 4
prefix.
.It Fl s Ar size
Size of the memory disk.
.Ar Size
is the number of 512 byte sectors unless suffixed with a
.Cm b , k , m , g , t ,
or
.Cm p
which
denotes byte, kilobyte, megabyte, gigabyte, terabyte and petabyte respectively.
When used without the
.Fl r
option, the
.Fl a
and
.Fl t Ar swap
options are implied if not specified.
.It Fl S Ar sectorsize
Sectorsize to use for the memory disk, in bytes.
.It Fl x Ar sectors/track
See the description of the
.Fl y
option below.
.It Fl y Ar heads/cylinder
For
.Cm malloc
or
.Cm vnode
backed devices, the
.Fl x
and
.Fl y
options can be used to specify a synthetic geometry.
This is useful for constructing bootable images for later download to
other devices.
.It Fl L Ar label
Associate a label (arbitrary string) with the new memory disk.
The label can then be inspected with
.Bd -literal -offset indent
.Nm Fl l v
.Ed
.It Fl o Oo Cm no Oc Ns Ar option
Set or reset options.
.Bl -tag -width indent
.It Oo Cm no Oc Ns Cm async
For
.Cm vnode
backed devices: avoid
.Dv IO_SYNC
for increased performance but
at the risk of deadlocking the entire kernel.
.It Oo Cm no Oc Ns Cm cache
For
.Cm vnode
backed devices: enable/disable caching of data in system caches.
The default is to not cache.
.Pp
Accesses via the device are converted to accesses via the vnode.
The caching policy for the vnode is used initially.
This is normally to cache.
This caching policy is retained if the
.Cm cache
option is used.
Otherwise, caching is limited
by releasing data from caches soon after each access.
The release has the same semantics as the
.Dv POSIX_FADV_DONTNEED
feature of
.Xr posix_fadvise 2 .
The result is that with normal (non-zfs) caching,
buffers are released from the buffer cache soon after they are constructed,
but their data is kept in the page cache at lower priority.
.Pp
The
.Cm cache
option tends to waste memory by giving unwanted double caching,
but it saves time if there is memory to spare.
.It Oo Cm no Oc Ns Cm reserve
Allocate and reserve all needed storage from the start, rather than as needed.
.It Oo Cm no Oc Ns Cm cluster
Enable clustering on this disk.
.It Oo Cm no Oc Ns Cm compress
Enable/disable compression features to reduce memory usage.
.It Oo Cm no Oc Ns Cm force
Disable/enable extra sanity checks to prevent the user from doing something
that might adversely affect the system.
This can be used with the
.Fl d
flag to forcibly destroy an
.Xr md 4
disk that is still in use.
.It Oo Cm no Oc Ns Cm readonly
Enable/disable readonly mode.
.It Oo Cm no Oc Ns Cm verify
For
.Cm vnode
backed devices: enable/disable requesting verification of the
file used for backing store.
The type of verification depends on which security features are available.
One example of verification is testing file integrity with
checksums or cryptographic signatures.
.El
.It Fl u Ar unit
Request a specific unit number or device name for the
.Xr md 4
device instead of automatic allocation.
If a device name is specified, it must be start with
.Dq md
followed by the unit number.
.El
.Pp
The last form,
.Nm
.Ar file ,
is provided for convenience as an abbreviation of
.Nm
.Fl a
.Fl t Ar vnode
.Fl f Ar file .
.Sh EXAMPLES
Create a disk with
.Pa /tmp/boot.flp
as backing storage.
The name of the allocated unit will be printed on stdout, such as
.Dq Li md0 :
.Bd -literal -offset indent
mdconfig /tmp/boot.flp
.Ed
.Pp
Create a 1 gigabyte swap backed memory disk named
.Dq Li md3 :
.Bd -literal -offset indent
mdconfig -s 1g -u md3
.Ed
.Pp
Detach and free all resources used by
.Pa /dev/md3 :
.Bd -literal -offset indent
mdconfig -du md3
.Ed
.Pp
Show detailed information on current memory disks:
.Bd -literal -offset indent
mdconfig -lv
.Ed
.Pp
Resize the
.Dq Li md3
memory disk to 2 gigabytes:
.Bd -literal -offset indent
mdconfig -rs 2g -u md3
.Ed
.Pp
Create a 1 gigabyte swap backed disk, initialize an
.Xr ffs 7
file system on it, and mount it on
.Pa /tmp :
.Bd -literal -offset indent
mdconfig -s 1g -u md10
newfs -U /dev/md10
mount /dev/md10 /tmp
chmod 1777 /tmp
.Ed
.Pp
Create a memory disk out of an ISO 9660 CD image file,
using the first available
.Xr md 4
device, and then mount it:
.Bd -literal -offset indent
mount -t cd9660 /dev/`mdconfig -f cdimage.iso` /mnt
.Ed
.Pp
Create a file-backed device from a hard disk image that begins
with 512K of raw header information.
.Xr gnop 8
is used to skip over the header information, positioning
.Pa md1.nop
to the start of the filesystem in the image.
.Bd -literal -offset indent
mdconfig -u md1 -f diskimage.img
gnop create -o 512K md1
mount /dev/md1.nop /mnt
.Ed
.Sh SEE ALSO
.Xr open 2 ,
.Xr md 4 ,
.Xr ffs 7 ,
.Xr gpart 8 ,
.Xr mdmfs 8 ,
.Xr malloc 9
.Sh HISTORY
The
.Nm
utility first appeared in
.Fx 5.0
as a cleaner replacement for the
.Xr vn 4
and
.Xr vnconfig 8
combo.
.Sh AUTHORS
The
.Nm
utility was written by
.An Poul-Henning Kamp Aq Mt phk@FreeBSD.org .