The special file /dev/mem is an interface
to the physical memory of the computer. Byte offsets in this file are
interpreted as physical memory addresses. Reading and writing this file is
equivalent to reading and writing memory itself. Only offsets within the
bounds of /dev/mem are allowed.
Kernel virtual memory is accessed through the interface
/dev/kmem in the same manner as
/dev/mem. Only kernel virtual addresses that are
currently mapped to memory are allowed.
On ISA the I/O memory space begins at physical address 0x000a0000
and runs to 0x00100000. The per-process data size for the current process is
UPAGES
long, and ends at virtual address
0xf0000000.
The MEM_EXTRACT_PADDR
ioctl can be used to
look up the physical address and NUMA domain of a given virtual address in
the calling process' address space. The request is described by
struct mem_extract {
uint64_t me_vaddr; /* input */
uint64_t me_paddr; /* output */
int me_domain; /* output */
int me_state; /* output */
};
The ioctl returns an error if the address is not valid. The
information returned by MEM_EXTRACT_PADDR
may be out
of date by the time that the ioctl call returns. Specifically, concurrent
system calls, page faults, or system page reclamation activity may have
unmapped the virtual page or replaced the backing physical page before the
ioctl call returns. Wired pages, e.g., those locked by
mlock(2),
will not be reclaimed by the system.
The me_state field provides information
about the state of the virtual page:
ME_STATE_INVALID
- The virtual address is invalid.
ME_STATE_VALID
- The virtual address is valid but is not mapped at the time of the ioctl
call.
ME_STATE_MAPPED
- The virtual address corresponds to a physical page mapping, and the
me_paddr and me_domain fields
are valid.
Several architectures allow attributes to be associated with
ranges of physical memory. These attributes can be manipulated via
ioctl
()
calls performed on /dev/mem. Declarations and data
types are to be found in
<sys/memrange.h>
.
The specific attributes, and number of programmable ranges may
vary between architectures. The full set of supported attributes is:
MDF_UNCACHEABLE
- The region is not cached.
MDF_WRITECOMBINE
- Writes to the region may be combined or performed out of order.
MDF_WRITETHROUGH
- Writes to the region are committed synchronously.
MDF_WRITEBACK
- Writes to the region are committed asynchronously.
MDF_WRITEPROTECT
- The region cannot be written to.
Memory ranges are described by
struct mem_range_desc {
uint64_t mr_base; /* physical base address */
uint64_t mr_len; /* physical length of region */
int mr_flags; /* attributes of region */
char mr_owner[8];
};
In addition to the region attributes listed above, the following
flags may also be set in the mr_flags field:
- MDF_FIXBASE
- The region's base address cannot be changed.
- MDF_FIXLEN
- The region's length cannot be changed.
- MDF_FIRMWARE
- The region is believed to have been established by the system
firmware.
- MDF_ACTIVE
- The region is currently active.
- MDF_BOGUS
- We believe the region to be invalid or otherwise erroneous.
- MDF_FIXACTIVE
- The region cannot be disabled.
- MDF_BUSY
- The region is currently owned by another process and may not be
altered.
Operations are performed using
struct mem_range_op {
struct mem_range_desc *mo_desc;
int mo_arg[2];
};
The MEMRANGE_GET
ioctl is used to retrieve
current memory range attributes. If mo_arg[0] is set
to 0, it will be updated with the total number of memory range descriptors.
If greater than 0, the array at mo_desc will be filled
with a corresponding number of descriptor structures, or the maximum,
whichever is less.
The MEMRANGE_SET
ioctl is used to add,
alter and remove memory range attributes. A range with the
MDF_FIXACTIVE
flag may not be removed; a range with
the MDF_BUSY
flag may not be removed or updated.
mo_arg[0] should be set to
MEMRANGE_SET_UPDATE
to update an existing or
establish a new range, or to MEMRANGE_SET_REMOVE
to
remove a range.
The MEM_KERNELDUMP
ioctl will initiate a
kernel dump against the running system, the contents of which will be
written to a process-owned file descriptor. The resulting dump output will
be in minidump format. The request is described by
struct mem_livedump_arg {
int fd; /* input */
int flags /* input */
uint8_t compression /* input */
};
The fd field is used to pass the file
descriptor.
The flags field is currently unused and must
be set to zero.
The compression field can be used to specify
the desired compression to be applied to the dump output. The supported
values are defined in
<sys/kerneldump.h>
; that is,
KERNELDUMP_COMP_NONE
,
KERNELDUMP_COMP_GZIP
, or
KERNELDUMP_COMP_ZSTD
.
Kernel dumps taken against the running system may have
inconsistent kernel data structures due to allocation, deallocation, or
modification of memory concurrent to the dump procedure. Thus, the resulting
core dump is not guaranteed to be usable. A system under load is more likely
to produce an inconsistent result. Despite this, live kernel dumps can be
useful for offline debugging of certain types of kernel bugs, such as
deadlocks, or in inspecting a particular part of the system's state.
The MEM_EXTRACT_PADDR
ioctl always returns
a value of zero.
- [
EOPNOTSUPP
]
- Memory range operations are not supported on this architecture.
- [
ENXIO
]
- No memory range descriptors are available (e.g., firmware has not enabled
any).
- [
EINVAL
]
- The memory range supplied as an argument is invalid or overlaps another
range in a fashion not supported by this architecture.
- [
EBUSY
]
- An attempt to remove or update a range failed because the range is
busy.
- [
ENOSPC
]
- An attempt to create a new range failed due to a shortage of hardware
resources (e.g., descriptor slots).
- [
ENOENT
]
- An attempt to remove a range failed because no range matches the
descriptor base/length supplied.
- [
EPERM
]
- An attempt to remove a range failed because the range is permanently
enabled.
- [
EOPNOTSUPP
]
- Kernel minidumps are not supported on this architecture.
- [
EPERM
]
- An attempt to begin the kernel dump failed because the calling thread
lacks the
- [
EBADF
]
- The supplied file descriptor was invalid, or does not have write
permission.
- [
EBUSY
]
- An attempt to begin the kernel dump failed because one is already in
progress.
- [
EINVAL
]
- An invalid or unsupported value was specified in
flags.
- [
EINVAL
]
- An invalid or unsupported compression type was specified.
PRIV_KMEM_READ
privilege.
The /dev/mem file appeared in
Version 1 AT&T UNIX and
/dev/kmem in Version 5
AT&T UNIX. The ioctl interface for memory range attributes was
added in FreeBSD 3.2.
Busy range attributes are not yet managed correctly.
This device is required for all users of
kvm(3) to
operate.