.\" $OpenBSD: pmap.9,v 1.10 2007/12/23 15:03:50 jsing Exp $ .\" .\" Copyright (c) 2001, 2002, 2003 CubeSoft Communications, Inc. .\" .\" .\" 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. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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 $Mdocdate: December 23 2007 $ .Dt PMAP 9 .Os .Sh NAME .Nm pmap .Nd machine dependent interface to the MMU .Sh SYNOPSIS .Fd #include .Sh DESCRIPTION The architecture-dependent .Nm module describes how the physical mapping is done between the user-processes and kernel virtual addresses and the physical addresses of the main memory, providing machine-dependent translation and access tables that are used directly or indirectly by the memory-management hardware. The .Nm layer can be viewed as a big array of mapping entries that are indexed by virtual address to produce a physical address and flags. These flags describe the page's protection, whether the page has been referenced or modified and other characteristics. .Pp The .Nm interface is consistent across all platforms and hides the way page mappings are stored. .Sh INITIALIZATION .nr nS 1 .Ft void .Fn pmap_init "void" .nr nS 0 .Pp The .Fn pmap_init function is called from the machine-independent .Xr uvm 9 initialization code, when the MMU is enabled. .Sh PAGE MANAGEMENT Modified/referenced information is only tracked for pages managed by .Xr uvm 9 (pages for which a vm_page structure exists). Only managed mappings of those pages have modified/referenced tracking. The use of unmanaged mappings should be limited to code which may execute in interrupt context (such as .Xr malloc 9 ) or to enter mappings for physical addresses which are not managed by .Xr uvm 9 . This allows .Nm modules to avoid blocking interrupts when manipulating data structures or holding locks. Unmanaged mappings may only be entered into the kernel's virtual address space. The modified/referenced bits must be tracked on a per-page basis, as 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 bits for that page must be preserved. The only time the modified/referenced bits may be cleared is when .Xr uvm 9 explicitly calls the .Fn pmap_clear_modify and .Fn pmap_clear_reference functions. These functions must also change any internal state necessary to detect the page being modified or referenced again after the modified/referenced state is cleared. .Pp Mappings entered by .Fn pmap_enter are managed, mappings entered by .Fn pmap_kenter_pa are not. .Sh MAPPING ALLOCATION .nr nS 1 .Ft int .Fn pmap_enter "pmap_t pmap" "vaddr_t va" "paddr_t pa" "vm_prot_t prot" \ "int flags" .Ft void .Fn pmap_kenter_pa "vaddr_t va" "paddr_t pa" "vm_prot_t prot" .Ft void .Fn pmap_remove "pmap_t pmap" "vaddr_t sva" "paddr_t eva" .Ft void .Fn pmap_kremove "vaddr_t va" "vsize_t size" .nr nS 0 .Pp The .Fn pmap_enter function creates a managed mapping for physical page .Fa pa at the specified virtual address .Fa va in the target physical map .Fa pmap with protection specified by .Fa prot : .Bl -tag -width "VM_PROT_EXECUTE" .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 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" .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 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 . .Pp The .Fn pmap_enter function is called by the fault routine to establish a mapping for the page being faulted in. If .Fn pmap_enter is called to enter a mapping at a virtual address for which a mapping already exists, the previous mapping must be invalidated. .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_kenter_pa function creates an unmanaged mapping of physical address .Fa pa at the specified virtual address .Fa va with the protection specified by .Fa prot . .Pp The .Fn pmap_remove function removes the range of virtual addresses .Fa sva to .Fa eva from .Fa pmap , assuming proper alignment. .Fn pmap_remove is called during an unmap operation to remove low-level machine dependent mappings. .Pp The .Fn pmap_kremove function removes an unmanaged mapping at virtual address .Fa va of size .Fa size . .Pp A call to .Fn pmap_update must be made after .Fn pmap_kenter_pa or .Fn pmap_kremove to notify the .Nm layer that the mappings need to be made correct. .Sh ACCESS PROTECTION .nr nS 1 .Ft void .Fn pmap_unwire "pmap_t pmap" "vaddr_t va" .Ft void .Fn pmap_protect "pmap_t pmap" "vaddr_t sva" "vaddr_t eva" "vm_prot_t prot" .Ft void .Fn pmap_page_protect "struct vm_page *pg, vm_prot_t prot" .nr nS 0 .Pp The .Fn pmap_unwire function clears the wired attribute for a map/virtual-address pair. The mapping must already exist in .Fa pmap . .Pp The .Fn pmap_protect function sets the physical protection on range .Fa sva to .Fa eva , in .Fa pmap . .Pp The .Fn pmap_protect function is called during a copy-on-write operation to write protect copy-on-write memory, and when paging out a page to remove all mappings of a page. The .Fn pmap_page_protect function sets the permission for all mapping to page .Fa pg . The .Fn pmap_page_protect function is called before a pageout operation to ensure that all pmap references to a page are removed. .Sh PHYSICAL PAGE-USAGE INFORMATION .nr nS 1 .Ft boolean_t .Fn pmap_is_modified "struct vm_page *pg" .Ft boolean_t .Fn pmap_clear_modify "struct vm_page *pg" .Ft boolean_t .Fn pmap_is_referenced "struct vm_page *pg" .Ft boolean_t .Fn pmap_clear_reference "struct vm_page *pg" .nr nS 0 .Pp The .Fn pmap_is_modified and .Fn pmap_clear_modify functions read/set the modify bits on the specified physical page .Fa pg . The .Fn pmap_is_referenced and .Fn pmap_clear_reference functions read/set the reference bits on the specified physical page .Fa pg . .Pp The .Fn pmap_is_referenced and .Fn pmap_is_modified functions are called by the pagedaemon when looking for pages to free. The .Fn pmap_clear_referenced and .Fn pmap_clear_modify functions are called by the pagedaemon to help identification of pages that are no longer in demand. .Sh PHYSICAL PAGE INITIALIZATION .nr nS 1 .Ft void .Fn pmap_copy_page "struct vm_page *src" "struct vm_page *dst" .Ft void .Fn pmap_zero_page "struct vm_page *page" .nr nS 0 .Pp The .Fn pmap_copy_page function copies the content of the physical page .Fa src to physical page .Fa dst . .Pp The .Fn pmap_zero_page function fills .Fa page with zeroes. .Sh INTERNAL DATA STRUCTURE MANAGEMENT .nr nS 1 .Ft pmap_t .Fn pmap_create "void" .Ft void .Fn pmap_reference "pmap_t pmap" .Ft void .Fn pmap_destroy "pmap_t pmap" .nr nS 0 .Pp The .Fn pmap_create function creates an instance of the .Em pmap structure. .Pp The .Fn pmap_reference function increments the reference count on .Fa pmap . .Pp The .Fn pmap_destroy function decrements the reference count on physical map .Fa pmap and retires it from service if the count drops to zero, assuming it contains no valid mappings. .Sh OPTIONAL FUNCTIONS .nr nS 1 .Ft vaddr_t .Fn pmap_steal_memory "vsize_t size" "vaddr_t *vstartp" "vaddr_t *vendp" .Ft vaddr_t .Fn pmap_growkernel "vaddr_t maxkvaddr" .Ft void .Fn pmap_update "pmap_t pmap" .Ft void .Fn pmap_collect "pmap_t pmap" .Ft void .Fn pmap_virtual_space "vaddr_t *vstartp" "vaddr_t *vendp" .Ft void .Fn pmap_copy "pmap_t dst_pmap" "pmap_t src_pmap" "vaddr_t dst_addr" \ "vsize_t len" "vaddr_t src_addr" .nr nS 0 .Pp Wired memory allocation before the virtual memory system is bootstrapped is accomplished by the .Fn pmap_steal_memory function. After that point, the kernel memory allocation routines should be used. .Pp The .Fn pmap_growkernel function can preallocate kernel page tables to a specified virtual address. .Pp The .Fn pmap_update function notifies the .Nm module to force processing of all delayed actions for all pmaps. .Pp The .Fn pmap_collect function informs the .Nm module that the given .Em pmap is not expected to be used for some time, giving the .Nm module a chance to prioritize. The initial bounds of the kernel virtual address space are returned by .Fn pmap_virtual_space . .Pp The .Fn pmap_copy function copies the range specified by .Fa src_addr and .Fa src_len from .Fa src_pmap to the range described by .Fa dst_addr and .Fa dst_len in .Fa dst_map . .Fn pmap_copy is called during a .Xr fork 2 operation to give the child process an initial set of low-level mappings. .Sh SEE ALSO .Xr fork 2 , .Xr uvm 9 .Sh HISTORY The .Bx 4.4 .Nm module is based on Mach 3.0. The introduction of .Xr uvm 9 left the .Nm interface unchanged for the most part. .Sh BUGS Ifdefs must be documented. .Pp .Fn pmap_update should be mandatory.