NetBSD/share/man/man9/pmap.9

1057 lines
31 KiB
Groff

.\" $NetBSD: pmap.9,v 1.4 2001/04/25 12:38:20 kleink Exp $
.\"
.\" Copyright (c) 2000, 2001 The NetBSD Foundation, Inc.
.\" All rights reserved.
.\"
.\" This code is derived from software contributed to The NetBSD Foundation
.\" by Jason R. Thorpe.
.\"
.\" 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. All advertising materials mentioning features or use of this software
.\" must display the following acknowledgement:
.\" This product includes software developed by the NetBSD
.\" Foundation, Inc. and its contributors.
.\" 4. Neither the name of The NetBSD Foundation 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 NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
.\"
.Dd April 16, 2001
.Dt PMAP 9
.Os
.Sh NAME
.Nm pmap
.Nd machine-dependent portion of the virtual memory system
.Sh SYNOPSIS
.Fd #include <sys/param.h>
.Fd #include <uvm/uvm_extern.h>
.Ft void
.Fn "pmap_init" "void"
.Ft void
.Fn "pmap_virtual_space" "vaddr_t *vstartp" "vaddr_t *vendp"
.Ft void
.Fn "pmap_steal_memory" "vsize_t size" "vaddr_t *vstartp" "vaddr_t *vendp"
.Ft pmap_t
.Fn "pmap_kernel" "void"
.Ft pmap_t
.Fn "pmap_create" "void"
.Ft void
.Fn "pmap_destroy" "pmap_t pmap"
.Ft void
.Fn "pmap_reference" "pmap_t pmap"
.Ft void
.Fn "pmap_fork" "pmap_t src_map" "pmap_t dst_map"
.Ft long
.Fn "pmap_resident_count" "pmap_t pmap"
.Ft long
.Fn "pmap_wired_count" "pmap_t pmap"
.Ft vaddr_t
.Fn "pmap_growkernel" "vaddr_t maxkvaddr"
.Ft int
.Fn "pmap_enter" "pmap_t pmap" "vaddr_t va" "paddr_t pa" "vm_prot_t prot" \
"int flags"
.Ft void
.Fn "pmap_remove" "pmap_t pmap" "vaddr_t sva" "vaddr_t eva"
.Ft void
.Fn "pmap_protect" "pmap_t pmap" "vaddr_t sva" "vaddr_t eva" "vm_prot_t prot"
.Ft void
.Fn "pmap_unwire" "pmap_t pmap" "vaddr_t va"
.Ft boolean_t
.Fn "pmap_extract" "pmap_t pmap" "vaddr_t va" "paddr_t *pap"
.Ft void
.Fn "pmap_kenter_pa" "vaddr_t va" "paddr_t pa" "vm_prot_t prot"
.Ft void
.Fn "pmap_kremove" "vaddr_t va" "vsize_t size"
.Ft void
.Fn "pmap_copy" "pmap_t dst_map" "pmap_t src_map" "vaddr_t dst_addr" \
"vsize_t len" "vaddr_t src_addr"
.Ft void
.Fn "pmap_collect" "pmap_t pmap"
.Ft void
.Fn "pmap_update" "void"
.Ft void
.Fn "pmap_activate" "struct proc *p"
.Ft void
.Fn "pmap_deactivate" "struct proc *p"
.Ft void
.Fn "pmap_zero_page" "paddr_t pa"
.Ft void
.Fn "pmap_copy_page" "paddr_t src" "paddr_t dst"
.Ft void
.Fn "pmap_page_protect" "vm_page_t pg" "vm_prot_t prot"
.Ft boolean_t
.Fn "pmap_clear_modify" "vm_page_t pg"
.Ft boolean_t
.Fn "pmap_clear_reference" "vm_page_t pg"
.Ft boolean_t
.Fn "pmap_is_modified" "vm_page_t pg"
.Ft boolean_t
.Fn "pmap_is_referenced" "vm_page_t pg"
.Ft paddr_t
.Fn "pmap_phys_address" "int cookie"
.Ft vaddr_t
.Fn "PMAP_MAP_POOLPAGE" "paddr_t pa"
.Ft paddr_t
.Fn "PMAP_UNMAP_POOLPAGE" "vaddr_t va"
.Sh DESCRIPTION
The
.Nm
module is the machine-dependent portion of the
.Nx
virtual memory system
.Xr uvm 9 .
The purpose of the
.Nm
module is to manage physical address maps, to program the
memory management hardware on the system, and perform any
cache operations necessary to ensure correct operation of
the virtual memory system. The
.Nm
module is also responsible for maintaining certain information
required by
.Xr uvm 9 .
.Pp
In order to cope with hardware architectures that make the
invalidation of virtual address mappings expensive (e.g.
TLB invalidations, TLB shootdown operations for multiple
processors), the
.Nm
module is allowed to delay mapping invalidation or protection
operations until such time as they are actually necessary. The
functions that are allowed to delay such actions are
.Fn pmap_enter ,
.Fn pmap_remove ,
.Fn pmap_protect ,
.Fn pmap_kenter_pa ,
and
.Fn pmap_kremove .
Callers of these functions must use the
.Fn pmap_update
function to notify the
.Nm
module that the mappings need to be made correct. Since the
.Nm
module is provided with information as to which processors are
using a given physical map, the
.Nm
module may use whatever optimizations it has available to reduce
the expense of virtual-to-physical mapping synchronization.
.Ss HEADER FILES AND DATA STRUCTURES
Machine-dependent code must provide the header file
.Pa <machine/pmap.h> .
This file contains the definition of the
.Dv pmap
structure:
.Bd -literal -offset indent
struct pmap {
/* Contents defined by pmap implementation. */
};
typedef struct pmap *pmap_t;
.Ed
.Pp
This header file may also define other data structures that the
.Nm
implementation uses.
.Pp
Note that all prototypes for
.Nm
interface functions are provided by the header file
.Pa <uvm/uvm_pmap.h> .
It is possible to override this behavior by defining the
C pre-processor macro
.Dq PMAP_EXCLUDE_DECLS .
This may be used to add a layer of indirection to
.Nm
API calls, for handling different MMU types in a single
.Nm
module, for example. If the
.Dq PMAP_EXCLUDE_DECLS
macro is defined,
.Pa <machine/pmap.h>
.Em must
provide function prototypes in a block like so:
.Bd -literal -offset indent
#ifdef _KERNEL /* not exposed to user namespace */
__BEGIN_DECLS /* make safe for C++ */
/* Prototypes go here. */
__END_DECLS
#endif /* _KERNEL */
.Ed
.Pp
The header file
.Pa <uvm/uvm_pmap.h>
defines a structure for tracking
.Nm
statistics (see below). This structure is defined as:
.Bd -literal -offset indent
struct pmap_statistics {
long resident_count; /* number of mapped pages */
long wired_count; /* number of wired pages */
};
.Ed
.Ss WIRED MAPPINGS
The
.Nm
module is based on the premise that all information contained
in the physical maps it manages is redundant. That is, physical
map information may be
.Dq forgotten
by the
.Nm
module in the event that it is necessary to do so; it can be rebuilt
by
.Xr uvm 9
by taking a page fault. There is one exception to this rule: so-called
.Dq wired
mappings may not be forgotten. Wired mappings are those for which either
no high-level information exists with which to rebuild the mapping, or
mappings which are needed by critical sections of code where taking a
page fault is unacceptable. Information about which mappings are wired
is provided to the
.Nm
module when a mapping is established.
.Ss MODIFIED/REFERENCED INFORMATION
The
.Nm
module is required to keep track of whether or not a page managed
by the virtual memory system has been referenced or modified. This
information is used by
.Xr uvm 9
to determine what happens to the page when scanned by the
pagedaemon.
.Pp
Many CPUs provide hardware support for tracking
modified/referenced information. However, many CPUs, particularly
modern RISC CPUs, do not. On CPUs which lack hardware support
for modified/referenced tracking, the
.Nm
module must emulate it in software. There are several strategies
for doing this, and the best strategy depends on the CPU.
.Pp
The
.Dq referenced
attribute is used by the pagedaemon to determine if a page is
.Dq active .
Active pages are not candidates for re-use in the page replacement
algorithm. Accurate referenced information is not required for
correct operation; if supplying referenced information for a page
is not feasible, then the
.Nm
implementation should always consider the
.Dq referenced
attribute to be FALSE.
.Pp
The
.Dq modified
attribute is used by the pagedaemon to determine if a page needs
to be cleaned (written to backing store; swap space, a regular file, etc.).
Accurate modified information
.Em must
be provided by the
.Nm
module for correct operation of the virtual memory system.
.Pp
Note that modified/referenced information is only tracked for
pages managed by the virtual memory system (i.e. pages for
which a vm_page structure exists). In addition, only
.Dq managed
mappings of those pages have modified/referenced tracking. Mappings
entered with the
.Fn pmap_enter
function are
.Dq managed
mappings. It is possible for
.Dq unmanaged
mappings of a page to be created, using the
.Fn pmap_kenter_pa
function. The use of
.Dq unmanaged
mappings should be limited to code which may execute in interrupt context
(for example, the kernel memory allocator), or to enter mappings for
physical addresses which are not managed by the virtual memory system.
.Dq Unmanaged
mappings may only be entered into the kernel's virtual address space.
This constraint is placed on the callers of the
.Fn pmap_kenter_pa
and
.Fn pmap_kremove
functions so that the
.Nm
implementation need not block interrupts when manipulating data
structures or holding locks.
.Pp
Also note that the modified/referenced information must be tracked
on a per-page basis; they are not attributes of a mapping, but attributes
of a page. Therefore, even after all mappings for a given page have
been removed, the modified/referenced information for that page
.Em must
be preserved. The only time the modified/referenced attributes may
be cleared is when the virtual memory system explicitly calls the
.Fn pmap_clear_modify
and
.Fn pmap_clear_reference
functions.
.Ss STATISTICS
The
.Nm
is required to keep statistics as to the number of
.Dq resident
pages and the number of
.Dq wired
pages.
.Pp
A
.Dq resident
page is one for which a mapping exists. This statistic is used to
compute the resident size of a process and enforce resource limits.
Only pages (whether managed by the virtual memory system or not)
which are mapped into a physical map should be counted in the resident
count.
.Pp
A
.Dq wired
page is one for which a wired mapping exists. This statistic is used to
enforce resource limits.
.Pp
Note that it is recommended (though not required) that the
.Nm
implementation utilize the
.Dv pmap_statistics
structure in the tracking of
.Nm
statistics by placing it inside the
.Dv pmap
structure and adjusting the counts when mappings are established, changed,
or removed. This avoids potentially expensive data structure traversals
when the statistics are queried.
.Ss REQUIRED FUNCTIONS
This section describes functions that a
.Nm
module must provide to the virtual memory system.
.Bl -tag -width offset -indent
.It void Fn "pmap_init" "void"
This function initializes the
.Nm
module. It is called by
.Fn uvm_init
to initialize any data structures that the module needs to
manage physical maps.
.It pmap_t Fn "pmap_kernel" "void"
Return a pointer to the
.Dv pmap
structure that maps the kernel virtual address space.
.Pp
Note that this function may be provided as a C pre-processor macro.
.It void Fn "pmap_virtual_space" "vaddr_t *vstartp" "vaddr_t *vendp"
The
.Fn pmap_virtual_space
function is called to determine the initial kernel virtual address
space beginning and end. These values are used to create the kernel's
virtual memory map. The function must set
.Fa *vstartp
to the first kernel virtual address that will be managed by
.Xr uvm 9 ,
and must set
.Fa *vendp
to the last kernel virtual address that will be managed by
.Xr uvm 9 .
.Pp
If the
.Fn pmap_growkernel
feature is used by a
.Nm
implementation, then
.Fa *vendp
should be set to the maximum kernel virtual address allowed by the
implementation. If
.Fn pmap_growkernel
is not used, then
.Fa *vendp
.Em must
be set to the maximum kernel virtual address that can be mapped with
the resources currently allocated to map the kernel virtual address
space.
.It pmap_t Fn "pmap_create" "void"
Create a physical map and return it to the caller. The reference
count on the new map is 1.
.It void Fn "pmap_destroy" "pmap_t pmap"
Drop the reference count on the specified physical map. If the
reference count drops to 0, all resources associated with the
physical map are released and the physical map destroyed. In the
case of a drop-to-0, no mappings will exist in the map. The
.Nm
implementation may assert this.
.It void Fn "pmap_reference" "pmap_t pmap"
Increment the reference count on the specified physical map.
.It long Fn "pmap_resident_count" "pmap_t pmap"
Query the
.Dq resident pages
statistic for
.Fa pmap .
.Pp
Note that this function may be provided as a C pre-processor macro.
.It long Fn "pmap_wired_count" "pmap_t pmap"
Query the
.Dq wired pages"
statistic for
.Fa pmap .
.Pp
Note that this function may be provided as a C pre-processor macro.
.It int Fn "pmap_enter" "pmap_t pmap" "vaddr_t va" "paddr_t pa" \
"vm_prot_t prot" "int flags"
Create a mapping in physical map
.Fa pmap
for the physical address
.Fa pa
at the virtual address
.Fa va
with protection specified by bits in
.Fa prot :
.Bl -tag -width "VM_PROT_EXECUTE " -indent
.It VM_PROT_READ
The mapping must allow reading.
.It VM_PROT_WRITE
The mapping must allow writing.
.It VM_PROT_EXECUTE
The page mapped contains instructions that will be executed by the
processor.
.El
.Pp
The
.Fa flags
argument contains protection bits (the same bits as used in the
.Fa prot
argument) indicating the type of access that caused the mapping to
be created. This information may be used to seed modified/referenced
information for the page being mapped, possibly avoiding redundant faults
on platforms that track modified/referenced information in software.
Other information provided by
.Fa flags :
.Bl -tag -width "PMAP_CANFAIL " -indent
.It PMAP_WIRED
The mapping being created is a wired mapping.
.It PMAP_CANFAIL
The call to
.Fn pmap_enter
is allowed to fail. If this flag is
.Em not
set, and the
.Fn pmap_enter
call is unable to create the mapping, perhaps due to insufficient
resources, the
.Nm
module must panic.
.El
.Pp
The access type provided in the
.Fa flags
argument will never exceed the protection specified by
.Fa prot .
The
.Nm
implementation may assert this.
Note that on systems that do not provide hardware support for
tracking modified/referenced information, modified/referenced
information for the page
.Em must
be seeded with the access type provided in
.Fa flags
if the PMAP_WIRED flag is set. This is to prevent a fault
for the purpose of tracking modified/referenced information
from occurring while the system is in a critical section where
a fault would be unacceptable.
.Pp
Note that
.Fn pmap_enter
is sometimes called to enter a mapping at a virtual address
for which a mapping already exists. In this situation, the
implementation must take whatever action is necessary to
invalidate the previous mapping before entering the new one.
.Pp
Also note that
.Fn pmap_enter
is sometimes called to change the protection for a pre-existing
mapping, or to change the
.Dq wired
attribute for a pre-existing mapping.
.Pp
The
.Fn pmap_enter
function returns 0 on success or an error code indicating the mode
of failure.
.It void Fn "pmap_remove" "pmap_t pmap" "vaddr_t sva" "vaddr_t eva"
Remove mappings from the virtual address range
.Fa sva
to
.Fa eva
from the specified physical map.
.It void Fn "pmap_protect" "pmap_t pmap" "vaddr_t sva" "vaddr_t eva" \
"vm_prot_t prot"
Set the protection of the mappings in the virtual address range
.Fa sva
to
.Fa eva
in the specified physical map.
.It void Fn "pmap_unwire" "pmap_t pmap" "vaddr_t va"
Clear the
.Dq wired
attribute on the mapping for virtual address
.Fa va .
.It boolean_t Fn "pmap_extract" "pmap_t pmap" "vaddr_t va" "paddr_t *pap"
This function extracts a mapping from the specified physical map. It
serves two purposes: to determine if a mapping exists for the specified
virtual address, and to determine what physical address is mapped at the
specified virtual address.
.Pp
The
.Fn pmap_extract
function returns FALSE if a mapping for
.Fa va
does not exist. Otherwise, it returns TRUE and places the
physical address mapped at
.Fa va
into
.Fa *pap
if the
.Fa pap
argument is non-NULL.
.It void Fn "pmap_kenter_pa" "vaddr_t va" "paddr_t pa" "vm_prot_t prot"
Enter an
.Dq unmanaged
mapping for physical address
.Fa pa
at virtual address
.Fa va
with protection
.Fa prot
into the kernel physical map. Mappings of this type are always
.Dq wired ,
and are unaffected by routines that alter the protection of pages
(such as
.Fn pmap_page_protect ) Ns .
Such mappings are also not included in the gathering of modified/referenced
information about a page. Mappings created with
.Fn pmap_kenter_pa
may be removed only with a call to
.Fn pmap_kremove .
.Pp
Note that
.Fn pmap_kenter_pa
must be safe for use in interrupt context.
.Fn splvm
blocks interrupts that might cause
.Fn pmap_kenter_pa
to be called.
.It void Fn "pmap_kremove" "vaddr_t va" "vsize_t size"
Remove all mappings starting at virtual address
.Fa va
for
.Fa size
bytes from the kernel physical map. All mappings that are
removed must be the
.Dq unmanaged
type created with
.Fn pmap_kenter_pa .
The implementation may assert this.
.It void Fn "pmap_copy" "pmap_t dst_map" "pmap_t src_map" "vaddr_t dst_addr" \
"vsize_t len" "vaddr_t src_addr"
This function copies the mappings starting at
.Fa src_addr
in
.Fa src_map
for
.Fa len
bytes into
.Fa dst_map
starting at
.Fa dst_addr .
.Pp
Note that while this function is required to be provided by a
.Nm
implementation, it is not actually required to do anything.
.Fn pmap_copy
is merely advisory (it is used in the
.Xr fork 2
path to
.Dq pre-fault
the child's address space).
.It void Fn "pmap_collect" "pmap_t pmap"
This function is called just before a process is swapped out to
allow the
.Nm
module to release resources used to map the process's address space.
The implementation may choose to remove physical mappings in order
to free e.g. page tables back to the system. Note, however, that
wired mappings must
.Em not
be removed when
.Fn pmap_collect
is called.
.Pp
Note that while this function is required to be provided by a
.Nm
implementation, it is not actually required to do anything.
.Fn pmap_collect
is merely advisory. It is recommended, however, that
.Fn pmap_collect
be fully implemented by a
.Nm
implementation.
.It void Fn "pmap_update" "void"
This function is used to inform the
.Nm
module that all physical mappings must now be correct. That is, all
delayed virtual-to-physical mappings updates (such as TLB invalidation
or address space identifier updates) must be completed. This
routine must be used after calls to
.Fn pmap_enter ,
.Fn pmap_remove ,
.Fn pmap_protect ,
.Fn pmap_kenter_pa ,
and
.Fn pmap_kremove
in order to ensure correct operation of the virtual memory system.
.Pp
If a
.Nm
implementation does not delay virtual-to-physical mapping updates,
.Fn pmap_update
has no operation. In this case, the call may be deleted using
a C pre-processor macro in
.Pa <machine/pmap.h> .
.It void Fn "pmap_activate" "struct proc *p"
Activate the physical map used by process
.Fa p .
This is called by the virtual memory system when the the
virtual memory context for a process is changed, and is also
often used by machine-dependent context switch code to program
the memory management hardware with the process's page table
base, etc. Note that
.Fn pmap_activate
may not always be called when
.Fa p
is the current process.
.Fn pmap_activate
must be able to handle this scenario.
.It void Fn "pmap_deactivate" "struct proc *p"
Deactivate the physical map used by process
.Fa p .
It is generally used in conjunction with
.Fn pmap_activate .
Like
.Fn pmap_activate ,
.Fn pmap_deactivate
may not always be called when
.Fa p
is the current process.
.It void Fn "pmap_zero_page" "paddr_t pa"
Zero the PAGE_SIZE sized region starting at physical address
.Fa pa .
The
.Nm
implementation must take whatever steps are necessary to map the
page to a kernel-accessible address and zero the page. It is
suggested that implementations use an optimized zeroing algorithm,
as the performance of this function directly impacts page fault
performance. The implementation may assume that the region is
PAGE_SIZE aligned and exactly PAGE_SIZE bytes in length.
.Pp
Note that the cache configuration of the platform should also be
considered in the implementation of
.Fn pmap_zero_page .
For example, on systems with a physically-addressed cache, the cache
load caused by zeroing the page will not be wasted, as the zeroing is
usually done on-demand. However, on systems with a virtually-addressed
cached, the cache load caused by zeroing the page
.Em will
be wasted, as the page will be mapped at a virtual address which is
different from that used to zero the page. In the virtually-addressed
cache case, care should also be taken to avoid cache alias problems.
.It void Fn "pmap_copy_page" "paddr_t src" "paddr_t dst"
Copy the PAGE_SIZE sized region starting at physical address
.Fa src
to the same sized region starting at physical address
.Fa dst .
The
.Nm
implementation must take whatever steps are necessary to map the
source and destination pages to a kernel-accessible address and
perform the copy. It is suggested that implementations use an
optimized copy algorithm, as the performance of this function
directly impacts page fault performance. The implementation may
assume that both regions are PAGE_SIZE aligned and exactly PAGE_SIZE
bytes in length.
.Pp
The same cache considerations that apply to
.Fn pmap_zero_page
apply to
.Fn pmap_copy_page .
.It void Fn "pmap_page_protect" "vm_page_t pg" "vm_prot_t prot"
Lower the permissions for all mappings of the page
.Fa pg
to
.Fa prot .
This function is used by the virtual memory system to implement
copy-on-write (called with VM_PROT_READ set in
.Fa prot )
and to revoke all mappings when cleaning a page (called with
no bits set in
.Fa prot ) Ns .
Access permissions must never be added to a page as a result of
this call.
.It boolean_t Fn "pmap_clear_modify" "vm_page_t pg"
Clear the
.Dq modified
attribute on the page
.Fa pg .
.Pp
The
.Fn pmap_clear_modify
function returns TRUE or FALSE indicating whether or not the
.Dq modified
attribute was set on the page before it was cleared.
.Pp
Note that this function may be provided as a C pre-processor macro.
.It boolean_t Fn "pmap_clear_reference" "vm_page_t pg"
Clear the
.Dq referenced
attribute on the page
.Fa pg .
.Pp
The
.Fn pmap_clear_reference
function returns TRUE or FALSE indicating whether or not the
.Dq referenced
attribute was set on the page before it was cleared.
.Pp
Note that this function may be provided as a C pre-processor macro.
.It boolean_t Fn "pmap_is_modified" "vm_page_t pg"
Test whether or not the
.Dq modified
attribute is set on page
.Fa pg .
.Pp
Note that this function may be provided as a C pre-processor macro.
.It boolean_t Fn "pmap_is_referenced" "vm_page_t pg"
Test whether or not the
.Dq referenced
attribute is set on page
.Fa pg .
.Pp
Note that this function may be provided as a C pre-processor macro.
.It paddr_t Fn "pmap_phys_address" "int cookie"
Convert a cookie returned by a device
.Fn mmap
function into a physical address. This function is provided to
accommodate systems which have physical address spaces larger than
can be directly addressed by the platform's
.Fa paddr_t
type. The existence of this function is highly dubious, and it is
expected that this function will be removed from the
.Nm pmap
API in a future release of
.Nx .
.Pp
Note that this function may be provided as a C pre-processor macro.
.El
.Ss OPTIONAL FUNCTIONS
This section describes several optional functions in the
.Nm
API.
.Bl -tag -width offset -indent
.It vaddr_t Fn "pmap_steal_memory" "vsize_t size" "vaddr_t *vstartp" \
"vaddr_t *vendp"
This function is a bootstrap memory allocator, which may be provided
as an alternative to the bootstrap memory allocator used within
.Xr uvm 9
itself. It is particularly useful on systems which provide e.g.
a direct-mapped memory segment. This function works by stealing
pages from the (to be) managed memory pool, which has already been
provided to
.Xr uvm 9
in the vm_physmem[] array. The pages are then mapped, or otherwise
made accessible to the kernel, in a machine-dependent way. The
memory must be zeroed by
.Fn pmap_steal_memory .
Note that memory allocated with
.Fn pmap_steal_memory
will never be freed, and mappings made by
.Fn pmap_steal_memory
must never be
.Dq forgotten .
.Pp
Note that
.Fn pmap_steal_memory
should not be used as a general-purpose early-startup memory
allocation routine. It is intended to be used only by the
.Fn uvm_pageboot_alloc
routine and its supporting routines. If you need to allocate
memory before the virtual memory system is initialized, use
.Fn uvm_pageboot_alloc .
See
.Xr uvm 9
for more information.
.Pp
The
.Fn pmap_steal_memory
function returns the kernel-accessible address of the allocated memory.
If no memory can be allocated, or if allocated memory cannot be mapped,
the function must panic.
.Pp
If the
.Fn pmap_steal_memory
function uses address space from the range provided to
.Xr uvm 9
by the
.Fn pmap_virtual_space
call, then
.Fn pmap_steal_memory
must adjust
.Fa *vstartp
and
.Fa *vendp
upon return.
.Pp
The
.Fn pmap_steal_memory
function is enabled by defining the C pre-processor macro
.Dq PMAP_STEAL_MEMORY
in
.Pa <machine/pmap.h> .
.It vaddr_t Fn "pmap_growkernel" "vaddr_t maxkvaddr"
Management of the kernel virtual address space is complicated by the
fact that it is not always safe to wait for resources with which to
map a kernel virtual address. However, it is not always desirable to
pre-allocate all resources necessary to map the entire kernel virtual
address space.
.Pp
The
.Fn pmap_growkernel
interface is designed to help alleviate this problem. The virtual
memory startup code may choose to allocate an initial set of mapping
resources (e.g. page tables) and set an internal variable indicating
how much kernel virtual address space can be mapped using those initial
resources. Then, when the virtual memory system wishes to map something
at an address beyond that initial limit, it calls
.Fn pmap_growkernel
to pre-allocate more sources with which to create the mapping. Note that
once additional kernel virtual address space mapping resources have been
allocated, they should not be freed; it is likely they will be needed
again.
.Pp
The
.Fn pmap_growkernel
function returns the new maximum kernel virtual address that can be mapped
with the resources it has available. If new resources cannot be allocated,
.Fn pmap_growkernel
must panic.
.Pp
The
.Fn pmap_growkernel
function is enabled by defining the C pre-processor macro
.Dq PMAP_GROWKERNEL
in
.Pa <machine/pmap.h> .
.It void Fn "pmap_fork" "pmap_t src_map" "pmap_t dst_map"
Some
.Nm
implementations may need to keep track of other information not
directly related to the virtual address space. For example, on
the i386 port, the Local Descriptor Table state of a process is
associated with the pmap (this is due to the fact that applications
manipulate the Local Descriptor Table directly expect it to be
logically associated with the virtual memory state of the process).
.Pp
The
.Fn pmap_fork
function is provided as a way to associate information from
.Fa src_map
with
.Fa dst_map
when a
.Dv vmspace
is forked.
.Fn pmap_fork
is called from
.Fn uvmspace_fork .
.Pp
The
.Fn pmap_fork
function is enabled by defining the C pre-processor macro
.Dq PMAP_FORK
in
.Pa <machine/pmap.h> .
.It vaddr_t Fn "PMAP_MAP_POOLPAGE" "paddr_t pa"
This function is used by the
.Xr pool 9
memory pool manager. Pools allocate backing pages one at a time. This
is provided as a means to use hardware features such as a direct-mapped
memory segment to map the pages used by the
.Xr pool 9
allocator. This can lead to better performance by e.g. reducing
TLB contention.
.Pp
.Fn PMAP_MAP_POOLPAGE
returns the kernel-accessible address of the page being mapped. It must
always succeed.
.Pp
The use of
.Fn PMAP_MAP_POOLPAGE
is enabled by defining it as a C pre-processor macro in
.Fa <machine/pmap.h> .
If
.Fn PMAP_MAP_POOLPAGE
is defined,
.Fn PMAP_UNMAP_POOLPAGE
must also be defined.
.Pp
The following is an example of how to define
.Fn PMAP_MAP_POOLPAGE :
.Bd -literal -offset indent
#define PMAP_MAP_POOLPAGE(pa) MIPS_PHYS_TO_KSEG0((pa))
.Ed
.Pp
This takes the physical address of a page and returns the KSEG0
address of that page on a MIPS processor.
.It paddr_t Fn "PMAP_UNMAP_POOLPAGE" "vaddr_t va"
This function is the inverse of
.Fn PMAP_MAP_POOLPAGE .
.Pp
.Fn PMAP_UNMAP_POOLPAGE
returns the physical address of the page corresponding to the
provided kernel-accessible address.
.Pp
The use of
.Fn PMAP_UNMAP_POOLPAGE
is enabled by defining it as a C pre-processor macro in
.Fa <machine/pmap.h> .
If
.Fn PMAP_UNMAP_POOLPAGE
is defined,
.Fn PMAP_MAP_POOLPAGE
must also be defined.
.Pp
The following is an example of how to define
.Fn PMAP_UNMAP_POOLPAGE :
.Bd -literal -offset indent
#define PMAP_UNMAP_POOLPAGE(pa) MIPS_KSEG0_TO_PHYS((va))
.Ed
.Pp
This takes the KSEG0 address of a previously-mapped pool page
and returns the physical address of that page on a MIPS processor.
.El
.Sh SEE ALSO
.Xr uvm 9
.Sh HISTORY
The
.Nm
module was originally part of the design of the virtual memory system
in the Mach Operating System. The goal was to provide a clean separation
between the machine-independent and the machine-dependent portions of
the virtual memory system, in stark contrast to the original
.Bx 3
virtual memory system, which was specific to the VAX.
.Pp
Between
.Bx 4.3
and
.Bx 4.4 ,
the Mach virtual memory system, including the
.Nm
API, was ported to
.Bx
and included in the
.Bx 4.4
release.
.Pp
.Nx
inherited the
.Bx
version of the Mach virtual memory system.
.Nx 1.4
was the first
.Nx
release with the new
.Xr uvm 9
virtual memory system, which included several changes to the
.Nm
API. Since the introduction of
.Xr uvm 9 ,
the
.Nm
API has evolved further.
.Sh BUGS
The use and definition of
.Fn pmap_activate
and
.Fn pmap_deactivate
needs to be reexamined.
.Pp
The use of
.Fn pmap_copy
needs to be reexamined. Empirical evidence suggests that performance
of the system suffers when
.Fn pmap_copy
actually performs its defined function. This is largely due to the
fact that the copy of the virtual-to-physical mappings is wasted if
the process calls
.Xr execve 2
after
.Xr fork 2 .
For this reason, it is recommended that
.Nm
implementations leave the body of the
.Fn pmap_copy
function empty for now.
.Sh AUTHOR
The original Mach VAX
.Nm
module was written by
.An Avadis Tevanian, Jr.
and
.An Michael Wayne Young .
.Pp
.An Mike Hibler
did the integration of the Mach virtual memory system into
.Bx 4.4
and implemented a
.Nm
module for the Motorola 68020+68851/68030/68040.
.Pp
The
.Nm
API as it exists in
.Nx
is derived from
.Bx 4.4 ,
and has been modified by
.An Chuck Cranor ,
.An Charles M. Hannum ,
.An Chuck Silvers ,
.An Wolfgang Solfrank ,
.An Bill Sommerfeld ,
and
.An Jason R. Thorpe .
.Pp
The author of this document is
.An Jason R. Thorpe
.Aq thorpej@netbsd.org .