pte.h revision 1.22
1/* $NetBSD: pte.h,v 1.22 2010/04/06 20:43:57 jld Exp $ */ 2 3/* 4 * Copyright (c) 2001 Wasabi Systems, Inc. 5 * All rights reserved. 6 * 7 * Written by Frank van der Linden for Wasabi Systems, Inc. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 3. All advertising materials mentioning features or use of this software 18 * must display the following acknowledgement: 19 * This product includes software developed for the NetBSD Project by 20 * Wasabi Systems, Inc. 21 * 4. The name of Wasabi Systems, Inc. may not be used to endorse 22 * or promote products derived from this software without specific prior 23 * written permission. 24 * 25 * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND 26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 27 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 28 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC 29 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 35 * POSSIBILITY OF SUCH DAMAGE. 36 */ 37 38/* 39 * 40 * Copyright (c) 1997 Charles D. Cranor and Washington University. 41 * All rights reserved. 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 3. All advertising materials mentioning features or use of this software 52 * must display the following acknowledgment: 53 * This product includes software developed by Charles D. Cranor and 54 * Washington University. 55 * 4. The name of the author may not be used to endorse or promote products 56 * derived from this software without specific prior written permission. 57 * 58 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 59 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 60 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 61 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 62 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 63 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 64 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 65 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 66 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 67 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 68 */ 69 70/* 71 * pte.h rewritten by chuck based on the jolitz version, plus random 72 * info on the pentium and other processors found on the net. the 73 * goal of this rewrite is to provide enough documentation on the MMU 74 * hardware that the reader will be able to understand it without having 75 * to refer to a hardware manual. 76 */ 77 78#ifndef _I386_PTE_H_ 79#define _I386_PTE_H_ 80#ifdef _KERNEL_OPT 81#include "opt_xen.h" 82#endif 83 84/* 85 * i386 MMU hardware structure (without PAE extension): 86 * 87 * the i386 MMU is a two-level MMU which maps 4GB of virtual memory. 88 * the pagesize is 4K (4096 [0x1000] bytes), although newer pentium 89 * processors can support a 4MB pagesize as well. 90 * 91 * the first level table (segment table?) is called a "page directory" 92 * and it contains 1024 page directory entries (PDEs). each PDE is 93 * 4 bytes (an int), so a PD fits in a single 4K page. this page is 94 * the page directory page (PDP). each PDE in a PDP maps 4MB of space 95 * (1024 * 4MB = 4GB). a PDE contains the physical address of the 96 * second level table: the page table. or, if 4MB pages are being used, 97 * then the PDE contains the PA of the 4MB page being mapped. 98 * 99 * a page table consists of 1024 page table entries (PTEs). each PTE is 100 * 4 bytes (an int), so a page table also fits in a single 4K page. a 101 * 4K page being used as a page table is called a page table page (PTP). 102 * each PTE in a PTP maps one 4K page (1024 * 4K = 4MB). a PTE contains 103 * the physical address of the page it maps and some flag bits (described 104 * below). 105 * 106 * the processor has a special register, "cr3", which points to the 107 * the PDP which is currently controlling the mappings of the virtual 108 * address space. 109 * 110 * the following picture shows the translation process for a 4K page: 111 * 112 * %cr3 register [PA of PDP] 113 * | 114 * | 115 * | bits <31-22> of VA bits <21-12> of VA bits <11-0> 116 * | index the PDP (0 - 1023) index the PTP are the page offset 117 * | | | | 118 * | v | | 119 * +--->+----------+ | | 120 * | PD Page | PA of v | 121 * | |---PTP-------->+------------+ | 122 * | 1024 PDE | | page table |--PTE--+ | 123 * | entries | | (aka PTP) | | | 124 * +----------+ | 1024 PTE | | | 125 * | entries | | | 126 * +------------+ | | 127 * | | 128 * bits <31-12> bits <11-0> 129 * p h y s i c a l a d d r 130 * 131 * the i386 caches PTEs in a TLB. it is important to flush out old 132 * TLB mappings when making a change to a mappings. writing to the 133 * %cr3 will flush the entire TLB. newer processors also have an 134 * instruction that will invalidate the mapping of a single page (which 135 * is useful if you are changing a single mappings because it preserves 136 * all the cached TLB entries). 137 * 138 * as shows, bits 31-12 of the PTE contain PA of the page being mapped. 139 * the rest of the PTE is defined as follows: 140 * bit# name use 141 * 11 n/a available for OS use, hardware ignores it 142 * 10 n/a available for OS use, hardware ignores it 143 * 9 n/a available for OS use, hardware ignores it 144 * 8 G global bit (see discussion below) 145 * 7 PS page size [for PDEs] (0=4k, 1=4M <if supported>) 146 * 6 D dirty (modified) page 147 * 5 A accessed (referenced) page 148 * 4 PCD cache disable 149 * 3 PWT prevent write through (cache) 150 * 2 U/S user/supervisor bit (0=supervisor only, 1=both u&s) 151 * 1 R/W read/write bit (0=read only, 1=read-write) 152 * 0 P present (valid) 153 * 154 * notes: 155 * - PS is only supported on newer processors 156 * - PTEs with the G bit are global in the sense that they are not 157 * flushed from the TLB when %cr3 is written (to flush, use the 158 * "flush single page" instruction). this is only supported on 159 * newer processors. this bit can be used to keep the kernel's 160 * TLB entries around while context switching. since the kernel 161 * is mapped into all processes at the same place it does not make 162 * sense to flush these entries when switching from one process' 163 * pmap to another. 164 * 165 * The PAE extension extends the size of the PTE to 64 bits (52bits physical 166 * address) and is compatible with the amd64 PTE format. The first level 167 * maps 2M, the second 1G, so a third level page table is introduced to 168 * map the 4GB virtual address space. This PD has only 4 entries. 169 * We can't use recursive mapping at level 3 to map the PD pages, as this 170 * would eat one GB of address space. In addition, Xen imposes restrictions 171 * on the entries we put in the L3 page (for example, the page pointed to by 172 * the last slot can't be shared among different L3 pages), which makes 173 * handling this L3 page in the same way we do for L2 on i386 (or L4 on amd64) 174 * difficult. For most things we'll just pretend to have only 2 levels, 175 * with the 2 high bits of the L2 index being in fact the index in the 176 * L3. 177 */ 178 179#if !defined(_LOCORE) 180 181/* 182 * here we define the data types for PDEs and PTEs 183 */ 184#ifdef PAE 185typedef uint64_t pd_entry_t; /* PDE */ 186typedef uint64_t pt_entry_t; /* PTE */ 187#else 188typedef uint32_t pd_entry_t; /* PDE */ 189typedef uint32_t pt_entry_t; /* PTE */ 190#endif 191 192#endif 193 194/* 195 * now we define various for playing with virtual addresses 196 */ 197 198#ifdef PAE 199#define L1_SHIFT 12 200#define L2_SHIFT 21 201#define L3_SHIFT 30 202#define NBPD_L1 (1ULL << L1_SHIFT) /* # bytes mapped by L1 ent (4K) */ 203#define NBPD_L2 (1ULL << L2_SHIFT) /* # bytes mapped by L2 ent (2MB) */ 204#define NBPD_L3 (1ULL << L3_SHIFT) /* # bytes mapped by L3 ent (1GB) */ 205 206#define L3_MASK 0xc0000000 207#define L2_REALMASK 0x3fe00000 208#define L2_MASK (L2_REALMASK | L3_MASK) 209#define L1_MASK 0x001ff000 210 211#define L3_FRAME (L3_MASK) 212#define L2_FRAME (L3_FRAME | L2_MASK) 213#define L1_FRAME (L2_FRAME|L1_MASK) 214 215#define PG_FRAME 0x000ffffffffff000ULL /* page frame mask */ 216#define PG_LGFRAME 0x000fffffffe00000ULL /* large (2MB) page frame mask */ 217 218/* macros to get real L2 and L3 index, from our "extended" L2 index */ 219#define l2tol3(idx) ((idx) >> (L3_SHIFT - L2_SHIFT)) 220#define l2tol2(idx) ((idx) & (L2_REALMASK >> L2_SHIFT)) 221#else /* PAE */ 222#define L1_SHIFT 12 223#define L2_SHIFT 22 224#define NBPD_L1 (1UL << L1_SHIFT) /* # bytes mapped by L1 ent (4K) */ 225#define NBPD_L2 (1UL << L2_SHIFT) /* # bytes mapped by L2 ent (4MB) */ 226 227#define L2_MASK 0xffc00000 228#define L1_MASK 0x003ff000 229 230#define L2_FRAME (L2_MASK) 231#define L1_FRAME (L2_FRAME|L1_MASK) 232 233#define PG_FRAME 0xfffff000 /* page frame mask */ 234#define PG_LGFRAME 0xffc00000 /* large (4MB) page frame mask */ 235 236#endif /* PAE */ 237/* 238 * here we define the bits of the PDE/PTE, as described above: 239 * 240 * XXXCDC: need to rename these (PG_u == ugly). 241 */ 242 243#define PG_V 0x00000001 /* valid entry */ 244#define PG_RO 0x00000000 /* read-only page */ 245#define PG_RW 0x00000002 /* read-write page */ 246#define PG_u 0x00000004 /* user accessible page */ 247#define PG_PROT 0x00000806 /* all protection bits */ 248#define PG_N 0x00000018 /* non-cacheable */ 249#define PG_U 0x00000020 /* has been used */ 250#define PG_M 0x00000040 /* has been modified */ 251#define PG_PS 0x00000080 /* 4MB page size */ 252#define PG_G 0x00000100 /* global, don't TLB flush */ 253#define PG_AVAIL1 0x00000200 /* ignored by hardware */ 254#define PG_AVAIL2 0x00000400 /* ignored by hardware */ 255#define PG_AVAIL3 0x00000800 /* ignored by hardware */ 256 257/* 258 * various short-hand protection codes 259 */ 260 261#define PG_KR 0x00000000 /* kernel read-only */ 262#define PG_KW 0x00000002 /* kernel read-write */ 263 264#ifdef PAE 265#define PG_NX 0x8000000000000000 /* No-execute */ 266#else 267#define PG_NX 0 /* dummy */ 268#endif 269 270/* 271 * page protection exception bits 272 */ 273 274#define PGEX_P 0x01 /* protection violation (vs. no mapping) */ 275#define PGEX_W 0x02 /* exception during a write cycle */ 276#define PGEX_U 0x04 /* exception while in user mode (upl) */ 277 278#endif /* _I386_PTE_H_ */ 279