1.\" Copyright (c) 1991 The Regents of the University of California. 2.\" All rights reserved. 3.\" 4.\" Redistribution and use in source and binary forms, with or without 5.\" modification, are permitted provided that the following conditions 6.\" are met: 7.\" 1. Redistributions of source code must retain the above copyright 8.\" notice, this list of conditions and the following disclaimer. 9.\" 2. Redistributions in binary form must reproduce the above copyright 10.\" notice, this list of conditions and the following disclaimer in the 11.\" documentation and/or other materials provided with the distribution. 12.\" 3. Neither the name of the University nor the names of its contributors 13.\" may be used to endorse or promote products derived from this software 14.\" without specific prior written permission. 15.\" 16.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 17.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 18.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 19.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 20.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 21.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 22.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 23.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26.\" SUCH DAMAGE. 27.\" 28.\" @(#)mem.4 5.3 (Berkeley) 5/2/91 29.\" $FreeBSD$ 30.\" 31.Dd March 11, 2022 32.Dt MEM 4 33.Os 34.Sh NAME 35.Nm mem , 36.Nm kmem 37.Nd memory files 38.Sh SYNOPSIS 39.Cd "device mem" 40.Sh DESCRIPTION 41The special file 42.Pa /dev/mem 43is an interface to the physical memory of the computer. 44Byte offsets in this file are interpreted as physical memory addresses. 45Reading and writing this file is equivalent to reading and writing 46memory itself. 47Only offsets within the bounds of 48.Pa /dev/mem 49are allowed. 50.Pp 51Kernel virtual memory is accessed through the interface 52.Pa /dev/kmem 53in the same manner as 54.Pa /dev/mem . 55Only kernel virtual addresses that are currently mapped to memory are allowed. 56.Pp 57On ISA the I/O memory space begins at physical address 0x000a0000 58and runs to 0x00100000. 59The 60per-process data 61size 62for the current process 63is 64.Dv UPAGES 65long, and ends at virtual 66address 0xf0000000. 67.Sh IOCTL INTERFACE 68.Ss Address Properties 69The 70.Dv MEM_EXTRACT_PADDR 71ioctl can be used to look up the physical address and NUMA domain of a given 72virtual address in the calling process' address space. 73The request is described by 74.Bd -literal 75struct mem_extract { 76 uint64_t me_vaddr; /* input */ 77 uint64_t me_paddr; /* output */ 78 int me_domain; /* output */ 79 int me_state; /* output */ 80}; 81.Ed 82.Pp 83The ioctl returns an error if the address is not valid. 84The information returned by 85.Dv MEM_EXTRACT_PADDR 86may be out of date by the time that the ioctl call returns. 87Specifically, concurrent system calls, page faults, or system page reclamation 88activity may have unmapped the virtual page or replaced the backing physical 89page before the ioctl call returns. 90Wired pages, e.g., those locked by 91.Xr mlock 2 , 92will not be reclaimed by the system. 93.Pp 94The 95.Fa me_state 96field provides information about the state of the virtual page: 97.Bl -tag -width indent 98.It Dv ME_STATE_INVALID 99The virtual address is invalid. 100.It Dv ME_STATE_VALID 101The virtual address is valid but is not mapped at the time of the ioctl call. 102.It Dv ME_STATE_MAPPED 103The virtual address corresponds to a physical page mapping, and the 104.Fa me_paddr 105and 106.Fa me_domain 107fields are valid. 108.El 109.Ss Memory Ranges 110.Pp 111Several architectures allow attributes to be associated with ranges of physical 112memory. 113These attributes can be manipulated via 114.Fn ioctl 115calls performed on 116.Pa /dev/mem . 117Declarations and data types are to be found in 118.In sys/memrange.h . 119.Pp 120The specific attributes, and number of programmable ranges may vary between 121architectures. 122The full set of supported attributes is: 123.Bl -tag -width indent 124.It Dv MDF_UNCACHEABLE 125The region is not cached. 126.It Dv MDF_WRITECOMBINE 127Writes to the region may be combined or performed out of order. 128.It Dv MDF_WRITETHROUGH 129Writes to the region are committed synchronously. 130.It Dv MDF_WRITEBACK 131Writes to the region are committed asynchronously. 132.It Dv MDF_WRITEPROTECT 133The region cannot be written to. 134.El 135.Pp 136Memory ranges are described by 137.Bd -literal 138struct mem_range_desc { 139 uint64_t mr_base; /* physical base address */ 140 uint64_t mr_len; /* physical length of region */ 141 int mr_flags; /* attributes of region */ 142 char mr_owner[8]; 143}; 144.Ed 145.Pp 146In addition to the region attributes listed above, the following flags 147may also be set in the 148.Fa mr_flags 149field: 150.Bl -tag -width indent 151.It MDF_FIXBASE 152The region's base address cannot be changed. 153.It MDF_FIXLEN 154The region's length cannot be changed. 155.It MDF_FIRMWARE 156The region is believed to have been established by the system firmware. 157.It MDF_ACTIVE 158The region is currently active. 159.It MDF_BOGUS 160We believe the region to be invalid or otherwise erroneous. 161.It MDF_FIXACTIVE 162The region cannot be disabled. 163.It MDF_BUSY 164The region is currently owned by another process and may not be 165altered. 166.El 167.Pp 168Operations are performed using 169.Bd -literal 170struct mem_range_op { 171 struct mem_range_desc *mo_desc; 172 int mo_arg[2]; 173}; 174.Ed 175.Pp 176The 177.Dv MEMRANGE_GET 178ioctl is used to retrieve current memory range attributes. 179If 180.Va mo_arg[0] 181is set to 0, it will be updated with the total number of memory range 182descriptors. 183If greater than 0, the array at 184.Va mo_desc 185will be filled with a corresponding number of descriptor structures, 186or the maximum, whichever is less. 187.Pp 188The 189.Dv MEMRANGE_SET 190ioctl is used to add, alter and remove memory range attributes. 191A range 192with the 193.Dv MDF_FIXACTIVE 194flag may not be removed; a range with the 195.Dv MDF_BUSY 196flag may not be removed or updated. 197.Pp 198.Va mo_arg[0] 199should be set to 200.Dv MEMRANGE_SET_UPDATE 201to update an existing or establish a new range, or to 202.Dv MEMRANGE_SET_REMOVE 203to remove a range. 204.El 205.Ss Live Kernel Dumps 206.Pp 207The 208.Dv MEM_KERNELDUMP 209ioctl will initiate a kernel dump against the running system, the contents of 210which will be written to a process-owned file descriptor. 211The resulting dump output will be in minidump format. 212The request is described by 213.Bd -literal 214struct mem_livedump_arg { 215 int fd; /* input */ 216 int flags /* input */ 217 uint8_t compression /* input */ 218}; 219.Ed 220.Pp 221The 222.Va fd 223field is used to pass the file descriptor. 224.Pp 225The 226.Va flags 227field is currently unused and must be set to zero. 228.Pp 229The 230.Va compression 231field can be used to specify the desired compression to 232be applied to the dump output. 233The supported values are defined in 234.In sys/kerneldump.h ; 235that is, 236.Dv KERNELDUMP_COMP_NONE , 237.Dv KERNELDUMP_COMP_GZIP , 238or 239.Dv KERNELDUMP_COMP_ZSTD . 240.Pp 241Kernel dumps taken against the running system may have inconsistent kernel data 242structures due to allocation, deallocation, or modification of memory 243concurrent to the dump procedure. 244Thus, the resulting core dump is not guaranteed to be usable. 245A system under load is more likely to produce an inconsistent result. 246Despite this, live kernel dumps can be useful for offline debugging of certain 247types of kernel bugs, such as deadlocks, or in inspecting a particular part of 248the system's state. 249.Sh RETURN VALUES 250.Ss MEM_EXTRACT_PADDR 251The 252.Dv MEM_EXTRACT_PADDR 253ioctl always returns a value of zero. 254.Ss MEMRANGE_GET/MEMRANGE_SET 255.Bl -tag -width Er 256.It Bq Er EOPNOTSUPP 257Memory range operations are not supported on this architecture. 258.It Bq Er ENXIO 259No memory range descriptors are available (e.g., firmware has not enabled 260any). 261.It Bq Er EINVAL 262The memory range supplied as an argument is invalid or overlaps another 263range in a fashion not supported by this architecture. 264.It Bq Er EBUSY 265An attempt to remove or update a range failed because the range is busy. 266.It Bq Er ENOSPC 267An attempt to create a new range failed due to a shortage of hardware 268resources (e.g., descriptor slots). 269.It Bq Er ENOENT 270An attempt to remove a range failed because no range matches the descriptor 271base/length supplied. 272.It Bq Er EPERM 273An attempt to remove a range failed because the range is permanently 274enabled. 275.El 276.Ss MEM_KERNELDUMP 277.Bl -tag -width Er 278.It Bq Er EOPNOTSUPP 279Kernel minidumps are not supported on this architecture. 280.It Bq Er EPERM 281An attempt to begin the kernel dump failed because the calling thread lacks the 282.It Bq Er EBADF 283The supplied file descriptor was invalid, or does not have write permission. 284.It Bq Er EBUSY 285An attempt to begin the kernel dump failed because one is already in progress. 286.It Bq Er EINVAL 287An invalid or unsupported value was specified in 288.Va flags . 289.It Bq Er EINVAL 290An invalid or unsupported compression type was specified. 291.Dv PRIV_KMEM_READ 292privilege. 293.El 294.Sh FILES 295.Bl -tag -width /dev/kmem -compact 296.It Pa /dev/mem 297.It Pa /dev/kmem 298.El 299.Sh SEE ALSO 300.Xr kvm 3 , 301.Xr memcontrol 8 302.Sh HISTORY 303The 304.Nm mem 305and 306.Nm kmem 307files appeared in 308.At v6 . 309The ioctl interface for memory range attributes was added in 310.Fx 3.2 . 311.Sh BUGS 312Busy range attributes are not yet managed correctly. 313.Pp 314This device is required for all users of 315.Xr kvm 3 316to operate. 317