pmap.c revision 1.23 1 1.23 jeremy /* $NetBSD: pmap.c,v 1.23 1997/05/28 04:28:52 jeremy Exp $ */
2 1.1 gwr
3 1.1 gwr /*-
4 1.10 jeremy * Copyright (c) 1996, 1997 The NetBSD Foundation, Inc.
5 1.1 gwr * All rights reserved.
6 1.1 gwr *
7 1.1 gwr * This code is derived from software contributed to The NetBSD Foundation
8 1.1 gwr * by Jeremy Cooper.
9 1.1 gwr *
10 1.1 gwr * Redistribution and use in source and binary forms, with or without
11 1.1 gwr * modification, are permitted provided that the following conditions
12 1.1 gwr * are met:
13 1.1 gwr * 1. Redistributions of source code must retain the above copyright
14 1.1 gwr * notice, this list of conditions and the following disclaimer.
15 1.1 gwr * 2. Redistributions in binary form must reproduce the above copyright
16 1.1 gwr * notice, this list of conditions and the following disclaimer in the
17 1.1 gwr * documentation and/or other materials provided with the distribution.
18 1.1 gwr * 3. All advertising materials mentioning features or use of this software
19 1.1 gwr * must display the following acknowledgement:
20 1.1 gwr * This product includes software developed by the NetBSD
21 1.1 gwr * Foundation, Inc. and its contributors.
22 1.1 gwr * 4. Neither the name of The NetBSD Foundation nor the names of its
23 1.1 gwr * contributors may be used to endorse or promote products derived
24 1.1 gwr * from this software without specific prior written permission.
25 1.1 gwr *
26 1.1 gwr * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27 1.1 gwr * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28 1.1 gwr * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29 1.1 gwr * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30 1.1 gwr * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31 1.1 gwr * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32 1.1 gwr * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33 1.1 gwr * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34 1.1 gwr * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35 1.1 gwr * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36 1.1 gwr * POSSIBILITY OF SUCH DAMAGE.
37 1.1 gwr */
38 1.1 gwr
39 1.1 gwr /*
40 1.1 gwr * XXX These comments aren't quite accurate. Need to change.
41 1.1 gwr * The sun3x uses the MC68851 Memory Management Unit, which is built
42 1.1 gwr * into the CPU. The 68851 maps virtual to physical addresses using
43 1.1 gwr * a multi-level table lookup, which is stored in the very memory that
44 1.1 gwr * it maps. The number of levels of lookup is configurable from one
45 1.1 gwr * to four. In this implementation, we use three, named 'A' through 'C'.
46 1.1 gwr *
47 1.1 gwr * The MMU translates virtual addresses into physical addresses by
48 1.1 gwr * traversing these tables in a proccess called a 'table walk'. The most
49 1.1 gwr * significant 7 bits of the Virtual Address ('VA') being translated are
50 1.1 gwr * used as an index into the level A table, whose base in physical memory
51 1.1 gwr * is stored in a special MMU register, the 'CPU Root Pointer' or CRP. The
52 1.1 gwr * address found at that index in the A table is used as the base
53 1.1 gwr * address for the next table, the B table. The next six bits of the VA are
54 1.1 gwr * used as an index into the B table, which in turn gives the base address
55 1.1 gwr * of the third and final C table.
56 1.1 gwr *
57 1.1 gwr * The next six bits of the VA are used as an index into the C table to
58 1.1 gwr * locate a Page Table Entry (PTE). The PTE is a physical address in memory
59 1.1 gwr * to which the remaining 13 bits of the VA are added, producing the
60 1.1 gwr * mapped physical address.
61 1.1 gwr *
62 1.1 gwr * To map the entire memory space in this manner would require 2114296 bytes
63 1.1 gwr * of page tables per process - quite expensive. Instead we will
64 1.1 gwr * allocate a fixed but considerably smaller space for the page tables at
65 1.1 gwr * the time the VM system is initialized. When the pmap code is asked by
66 1.1 gwr * the kernel to map a VA to a PA, it allocates tables as needed from this
67 1.1 gwr * pool. When there are no more tables in the pool, tables are stolen
68 1.1 gwr * from the oldest mapped entries in the tree. This is only possible
69 1.1 gwr * because all memory mappings are stored in the kernel memory map
70 1.1 gwr * structures, independent of the pmap structures. A VA which references
71 1.1 gwr * one of these invalidated maps will cause a page fault. The kernel
72 1.1 gwr * will determine that the page fault was caused by a task using a valid
73 1.1 gwr * VA, but for some reason (which does not concern it), that address was
74 1.1 gwr * not mapped. It will ask the pmap code to re-map the entry and then
75 1.1 gwr * it will resume executing the faulting task.
76 1.1 gwr *
77 1.1 gwr * In this manner the most efficient use of the page table space is
78 1.1 gwr * achieved. Tasks which do not execute often will have their tables
79 1.1 gwr * stolen and reused by tasks which execute more frequently. The best
80 1.1 gwr * size for the page table pool will probably be determined by
81 1.1 gwr * experimentation.
82 1.1 gwr *
83 1.1 gwr * You read all of the comments so far. Good for you.
84 1.1 gwr * Now go play!
85 1.1 gwr */
86 1.1 gwr
87 1.1 gwr /*** A Note About the 68851 Address Translation Cache
88 1.1 gwr * The MC68851 has a 64 entry cache, called the Address Translation Cache
89 1.1 gwr * or 'ATC'. This cache stores the most recently used page descriptors
90 1.1 gwr * accessed by the MMU when it does translations. Using a marker called a
91 1.1 gwr * 'task alias' the MMU can store the descriptors from 8 different table
92 1.1 gwr * spaces concurrently. The task alias is associated with the base
93 1.1 gwr * address of the level A table of that address space. When an address
94 1.1 gwr * space is currently active (the CRP currently points to its A table)
95 1.1 gwr * the only cached descriptors that will be obeyed are ones which have a
96 1.1 gwr * matching task alias of the current space associated with them.
97 1.1 gwr *
98 1.1 gwr * Since the cache is always consulted before any table lookups are done,
99 1.1 gwr * it is important that it accurately reflect the state of the MMU tables.
100 1.1 gwr * Whenever a change has been made to a table that has been loaded into
101 1.1 gwr * the MMU, the code must be sure to flush any cached entries that are
102 1.1 gwr * affected by the change. These instances are documented in the code at
103 1.1 gwr * various points.
104 1.1 gwr */
105 1.1 gwr /*** A Note About the Note About the 68851 Address Translation Cache
106 1.1 gwr * 4 months into this code I discovered that the sun3x does not have
107 1.1 gwr * a MC68851 chip. Instead, it has a version of this MMU that is part of the
108 1.1 gwr * the 68030 CPU.
109 1.1 gwr * All though it behaves very similarly to the 68851, it only has 1 task
110 1.8 gwr * alias and a 22 entry cache. So sadly (or happily), the first paragraph
111 1.8 gwr * of the previous note does not apply to the sun3x pmap.
112 1.1 gwr */
113 1.1 gwr
114 1.1 gwr #include <sys/param.h>
115 1.1 gwr #include <sys/systm.h>
116 1.1 gwr #include <sys/proc.h>
117 1.1 gwr #include <sys/malloc.h>
118 1.1 gwr #include <sys/user.h>
119 1.1 gwr #include <sys/queue.h>
120 1.20 thorpej #include <sys/kcore.h>
121 1.1 gwr
122 1.1 gwr #include <vm/vm.h>
123 1.1 gwr #include <vm/vm_kern.h>
124 1.1 gwr #include <vm/vm_page.h>
125 1.1 gwr
126 1.1 gwr #include <machine/cpu.h>
127 1.17 gwr #include <machine/kcore.h>
128 1.1 gwr #include <machine/pmap.h>
129 1.1 gwr #include <machine/pte.h>
130 1.5 gwr #include <machine/machdep.h>
131 1.1 gwr #include <machine/mon.h>
132 1.1 gwr
133 1.1 gwr #include "pmap_pvt.h"
134 1.1 gwr
135 1.1 gwr /* XXX - What headers declare these? */
136 1.1 gwr extern struct pcb *curpcb;
137 1.1 gwr extern int physmem;
138 1.1 gwr
139 1.7 gwr extern void copypage __P((const void*, void*));
140 1.7 gwr extern void zeropage __P((void*));
141 1.7 gwr
142 1.1 gwr /* Defined in locore.s */
143 1.1 gwr extern char kernel_text[];
144 1.1 gwr
145 1.1 gwr /* Defined by the linker */
146 1.1 gwr extern char etext[], edata[], end[];
147 1.1 gwr extern char *esym; /* DDB */
148 1.1 gwr
149 1.7 gwr /*************************** DEBUGGING DEFINITIONS ***********************
150 1.7 gwr * Macros, preprocessor defines and variables used in debugging can make *
151 1.7 gwr * code hard to read. Anything used exclusively for debugging purposes *
152 1.7 gwr * is defined here to avoid having such mess scattered around the file. *
153 1.7 gwr *************************************************************************/
154 1.8 gwr #ifdef PMAP_DEBUG
155 1.7 gwr /*
156 1.7 gwr * To aid the debugging process, macros should be expanded into smaller steps
157 1.7 gwr * that accomplish the same goal, yet provide convenient places for placing
158 1.8 gwr * breakpoints. When this code is compiled with PMAP_DEBUG mode defined, the
159 1.7 gwr * 'INLINE' keyword is defined to an empty string. This way, any function
160 1.7 gwr * defined to be a 'static INLINE' will become 'outlined' and compiled as
161 1.7 gwr * a separate function, which is much easier to debug.
162 1.7 gwr */
163 1.7 gwr #define INLINE /* nothing */
164 1.7 gwr
165 1.1 gwr /*
166 1.7 gwr * It is sometimes convenient to watch the activity of a particular table
167 1.7 gwr * in the system. The following variables are used for that purpose.
168 1.1 gwr */
169 1.7 gwr a_tmgr_t *pmap_watch_atbl = 0;
170 1.7 gwr b_tmgr_t *pmap_watch_btbl = 0;
171 1.7 gwr c_tmgr_t *pmap_watch_ctbl = 0;
172 1.1 gwr
173 1.7 gwr int pmap_debug = 0;
174 1.7 gwr #define DPRINT(args) if (pmap_debug) printf args
175 1.7 gwr
176 1.7 gwr #else /********** Stuff below is defined if NOT debugging **************/
177 1.7 gwr
178 1.7 gwr #define INLINE inline
179 1.10 jeremy #define DPRINT(args) /* nada */
180 1.7 gwr
181 1.10 jeremy #endif /* PMAP_DEBUG */
182 1.7 gwr /*********************** END OF DEBUGGING DEFINITIONS ********************/
183 1.1 gwr
184 1.1 gwr /*** Management Structure - Memory Layout
185 1.1 gwr * For every MMU table in the sun3x pmap system there must be a way to
186 1.1 gwr * manage it; we must know which process is using it, what other tables
187 1.1 gwr * depend on it, and whether or not it contains any locked pages. This
188 1.1 gwr * is solved by the creation of 'table management' or 'tmgr'
189 1.1 gwr * structures. One for each MMU table in the system.
190 1.1 gwr *
191 1.1 gwr * MAP OF MEMORY USED BY THE PMAP SYSTEM
192 1.1 gwr *
193 1.1 gwr * towards lower memory
194 1.1 gwr * kernAbase -> +-------------------------------------------------------+
195 1.1 gwr * | Kernel MMU A level table |
196 1.1 gwr * kernBbase -> +-------------------------------------------------------+
197 1.1 gwr * | Kernel MMU B level tables |
198 1.1 gwr * kernCbase -> +-------------------------------------------------------+
199 1.1 gwr * | |
200 1.1 gwr * | Kernel MMU C level tables |
201 1.1 gwr * | |
202 1.7 gwr * mmuCbase -> +-------------------------------------------------------+
203 1.7 gwr * | User MMU C level tables |
204 1.1 gwr * mmuAbase -> +-------------------------------------------------------+
205 1.1 gwr * | |
206 1.1 gwr * | User MMU A level tables |
207 1.1 gwr * | |
208 1.1 gwr * mmuBbase -> +-------------------------------------------------------+
209 1.1 gwr * | User MMU B level tables |
210 1.1 gwr * tmgrAbase -> +-------------------------------------------------------+
211 1.1 gwr * | TMGR A level table structures |
212 1.1 gwr * tmgrBbase -> +-------------------------------------------------------+
213 1.1 gwr * | TMGR B level table structures |
214 1.1 gwr * tmgrCbase -> +-------------------------------------------------------+
215 1.1 gwr * | TMGR C level table structures |
216 1.1 gwr * pvbase -> +-------------------------------------------------------+
217 1.1 gwr * | Physical to Virtual mapping table (list heads) |
218 1.1 gwr * pvebase -> +-------------------------------------------------------+
219 1.1 gwr * | Physical to Virtual mapping table (list elements) |
220 1.1 gwr * | |
221 1.1 gwr * +-------------------------------------------------------+
222 1.1 gwr * towards higher memory
223 1.1 gwr *
224 1.1 gwr * For every A table in the MMU A area, there will be a corresponding
225 1.1 gwr * a_tmgr structure in the TMGR A area. The same will be true for
226 1.1 gwr * the B and C tables. This arrangement will make it easy to find the
227 1.1 gwr * controling tmgr structure for any table in the system by use of
228 1.1 gwr * (relatively) simple macros.
229 1.1 gwr */
230 1.7 gwr
231 1.7 gwr /*
232 1.8 gwr * Global variables for storing the base addresses for the areas
233 1.1 gwr * labeled above.
234 1.1 gwr */
235 1.7 gwr static vm_offset_t kernAphys;
236 1.1 gwr static mmu_long_dte_t *kernAbase;
237 1.1 gwr static mmu_short_dte_t *kernBbase;
238 1.1 gwr static mmu_short_pte_t *kernCbase;
239 1.15 gwr static mmu_short_pte_t *mmuCbase;
240 1.15 gwr static mmu_short_dte_t *mmuBbase;
241 1.1 gwr static mmu_long_dte_t *mmuAbase;
242 1.1 gwr static a_tmgr_t *Atmgrbase;
243 1.1 gwr static b_tmgr_t *Btmgrbase;
244 1.1 gwr static c_tmgr_t *Ctmgrbase;
245 1.15 gwr static pv_t *pvbase;
246 1.1 gwr static pv_elem_t *pvebase;
247 1.15 gwr struct pmap kernel_pmap;
248 1.1 gwr
249 1.8 gwr /*
250 1.8 gwr * This holds the CRP currently loaded into the MMU.
251 1.8 gwr */
252 1.8 gwr struct mmu_rootptr kernel_crp;
253 1.8 gwr
254 1.8 gwr /*
255 1.8 gwr * Just all around global variables.
256 1.1 gwr */
257 1.1 gwr static TAILQ_HEAD(a_pool_head_struct, a_tmgr_struct) a_pool;
258 1.1 gwr static TAILQ_HEAD(b_pool_head_struct, b_tmgr_struct) b_pool;
259 1.1 gwr static TAILQ_HEAD(c_pool_head_struct, c_tmgr_struct) c_pool;
260 1.7 gwr
261 1.7 gwr
262 1.7 gwr /*
263 1.7 gwr * Flags used to mark the safety/availability of certain operations or
264 1.7 gwr * resources.
265 1.7 gwr */
266 1.23 jeremy static boolean_t pv_initialized = FALSE, /* PV system has been initialized. */
267 1.23 jeremy bootstrap_alloc_enabled = FALSE; /*Safe to use pmap_bootstrap_alloc().*/
268 1.23 jeremy u_char tmp_vpage0_inuse, /* Temporary virtual page 0 is in use */
269 1.23 jeremy tmp_vpage1_inuse; /* Temporary virtual page 1 is in use */
270 1.1 gwr
271 1.1 gwr /*
272 1.1 gwr * XXX: For now, retain the traditional variables that were
273 1.1 gwr * used in the old pmap/vm interface (without NONCONTIG).
274 1.1 gwr */
275 1.1 gwr /* Kernel virtual address space available: */
276 1.1 gwr vm_offset_t virtual_avail, virtual_end;
277 1.1 gwr /* Physical address space available: */
278 1.1 gwr vm_offset_t avail_start, avail_end;
279 1.1 gwr
280 1.7 gwr /* This keep track of the end of the contiguously mapped range. */
281 1.7 gwr vm_offset_t virtual_contig_end;
282 1.7 gwr
283 1.7 gwr /* Physical address used by pmap_next_page() */
284 1.7 gwr vm_offset_t avail_next;
285 1.7 gwr
286 1.7 gwr /* These are used by pmap_copy_page(), etc. */
287 1.23 jeremy vm_offset_t tmp_vpages[2]; /* Note: tmp_vpage[0] MUST be mapped R/O */
288 1.23 jeremy /* tmp_vpage[1] MUST be mapped R/W */
289 1.1 gwr
290 1.7 gwr /*
291 1.7 gwr * The 3/80 is the only member of the sun3x family that has non-contiguous
292 1.1 gwr * physical memory. Memory is divided into 4 banks which are physically
293 1.1 gwr * locatable on the system board. Although the size of these banks varies
294 1.1 gwr * with the size of memory they contain, their base addresses are
295 1.1 gwr * permenently fixed. The following structure, which describes these
296 1.1 gwr * banks, is initialized by pmap_bootstrap() after it reads from a similar
297 1.1 gwr * structure provided by the ROM Monitor.
298 1.1 gwr *
299 1.1 gwr * For the other machines in the sun3x architecture which do have contiguous
300 1.1 gwr * RAM, this list will have only one entry, which will describe the entire
301 1.1 gwr * range of available memory.
302 1.1 gwr */
303 1.20 thorpej struct pmap_physmem_struct avail_mem[SUN3X_NPHYS_RAM_SEGS];
304 1.1 gwr u_int total_phys_mem;
305 1.1 gwr
306 1.7 gwr /*************************************************************************/
307 1.7 gwr
308 1.7 gwr /*
309 1.7 gwr * XXX - Should "tune" these based on statistics.
310 1.7 gwr *
311 1.7 gwr * My first guess about the relative numbers of these needed is
312 1.7 gwr * based on the fact that a "typical" process will have several
313 1.7 gwr * pages mapped at low virtual addresses (text, data, bss), then
314 1.7 gwr * some mapped shared libraries, and then some stack pages mapped
315 1.7 gwr * near the high end of the VA space. Each process can use only
316 1.7 gwr * one A table, and most will use only two B tables (maybe three)
317 1.7 gwr * and probably about four C tables. Therefore, the first guess
318 1.7 gwr * at the relative numbers of these needed is 1:2:4 -gwr
319 1.7 gwr *
320 1.7 gwr * The number of C tables needed is closely related to the amount
321 1.7 gwr * of physical memory available plus a certain amount attributable
322 1.7 gwr * to the use of double mappings. With a few simulation statistics
323 1.7 gwr * we can find a reasonably good estimation of this unknown value.
324 1.7 gwr * Armed with that and the above ratios, we have a good idea of what
325 1.7 gwr * is needed at each level. -j
326 1.7 gwr *
327 1.7 gwr * Note: It is not physical memory memory size, but the total mapped
328 1.7 gwr * virtual space required by the combined working sets of all the
329 1.7 gwr * currently _runnable_ processes. (Sleeping ones don't count.)
330 1.7 gwr * The amount of physical memory should be irrelevant. -gwr
331 1.7 gwr */
332 1.22 jeremy #ifdef FIXED_NTABLES
333 1.7 gwr #define NUM_A_TABLES 16
334 1.7 gwr #define NUM_B_TABLES 32
335 1.7 gwr #define NUM_C_TABLES 64
336 1.22 jeremy #else
337 1.22 jeremy unsigned int NUM_A_TABLES, NUM_B_TABLES, NUM_C_TABLES;
338 1.22 jeremy #endif /* FIXED_NTABLES */
339 1.7 gwr
340 1.7 gwr /*
341 1.7 gwr * This determines our total virtual mapping capacity.
342 1.7 gwr * Yes, it is a FIXED value so we can pre-allocate.
343 1.7 gwr */
344 1.7 gwr #define NUM_USER_PTES (NUM_C_TABLES * MMU_C_TBL_SIZE)
345 1.15 gwr
346 1.15 gwr /*
347 1.15 gwr * The size of the Kernel Virtual Address Space (KVAS)
348 1.15 gwr * for purposes of MMU table allocation is -KERNBASE
349 1.15 gwr * (length from KERNBASE to 0xFFFFffff)
350 1.15 gwr */
351 1.15 gwr #define KVAS_SIZE (-KERNBASE)
352 1.15 gwr
353 1.15 gwr /* Numbers of kernel MMU tables to support KVAS_SIZE. */
354 1.15 gwr #define KERN_B_TABLES (KVAS_SIZE >> MMU_TIA_SHIFT)
355 1.15 gwr #define KERN_C_TABLES (KVAS_SIZE >> MMU_TIB_SHIFT)
356 1.15 gwr #define NUM_KERN_PTES (KVAS_SIZE >> MMU_TIC_SHIFT)
357 1.7 gwr
358 1.7 gwr /*************************** MISCELANEOUS MACROS *************************/
359 1.7 gwr #define PMAP_LOCK() ; /* Nothing, for now */
360 1.7 gwr #define PMAP_UNLOCK() ; /* same. */
361 1.7 gwr #define NULL 0
362 1.7 gwr
363 1.7 gwr static INLINE void * mmu_ptov __P((vm_offset_t pa));
364 1.7 gwr static INLINE vm_offset_t mmu_vtop __P((void * va));
365 1.7 gwr
366 1.7 gwr #if 0
367 1.7 gwr static INLINE a_tmgr_t * mmuA2tmgr __P((mmu_long_dte_t *));
368 1.7 gwr #endif
369 1.7 gwr static INLINE b_tmgr_t * mmuB2tmgr __P((mmu_short_dte_t *));
370 1.7 gwr static INLINE c_tmgr_t * mmuC2tmgr __P((mmu_short_pte_t *));
371 1.7 gwr
372 1.7 gwr static INLINE pv_t *pa2pv __P((vm_offset_t pa));
373 1.7 gwr static INLINE int pteidx __P((mmu_short_pte_t *));
374 1.7 gwr static INLINE pmap_t current_pmap __P((void));
375 1.7 gwr
376 1.7 gwr /*
377 1.7 gwr * We can always convert between virtual and physical addresses
378 1.7 gwr * for anything in the range [KERNBASE ... avail_start] because
379 1.7 gwr * that range is GUARANTEED to be mapped linearly.
380 1.7 gwr * We rely heavily upon this feature!
381 1.7 gwr */
382 1.7 gwr static INLINE void *
383 1.7 gwr mmu_ptov(pa)
384 1.7 gwr vm_offset_t pa;
385 1.7 gwr {
386 1.7 gwr register vm_offset_t va;
387 1.7 gwr
388 1.7 gwr va = (pa + KERNBASE);
389 1.8 gwr #ifdef PMAP_DEBUG
390 1.7 gwr if ((va < KERNBASE) || (va >= virtual_contig_end))
391 1.7 gwr panic("mmu_ptov");
392 1.7 gwr #endif
393 1.7 gwr return ((void*)va);
394 1.7 gwr }
395 1.7 gwr static INLINE vm_offset_t
396 1.7 gwr mmu_vtop(vva)
397 1.7 gwr void *vva;
398 1.7 gwr {
399 1.7 gwr register vm_offset_t va;
400 1.7 gwr
401 1.7 gwr va = (vm_offset_t)vva;
402 1.8 gwr #ifdef PMAP_DEBUG
403 1.7 gwr if ((va < KERNBASE) || (va >= virtual_contig_end))
404 1.7 gwr panic("mmu_ptov");
405 1.7 gwr #endif
406 1.7 gwr return (va - KERNBASE);
407 1.7 gwr }
408 1.7 gwr
409 1.7 gwr /*
410 1.7 gwr * These macros map MMU tables to their corresponding manager structures.
411 1.1 gwr * They are needed quite often because many of the pointers in the pmap
412 1.1 gwr * system reference MMU tables and not the structures that control them.
413 1.1 gwr * There needs to be a way to find one when given the other and these
414 1.1 gwr * macros do so by taking advantage of the memory layout described above.
415 1.1 gwr * Here's a quick step through the first macro, mmuA2tmgr():
416 1.1 gwr *
417 1.1 gwr * 1) find the offset of the given MMU A table from the base of its table
418 1.1 gwr * pool (table - mmuAbase).
419 1.1 gwr * 2) convert this offset into a table index by dividing it by the
420 1.1 gwr * size of one MMU 'A' table. (sizeof(mmu_long_dte_t) * MMU_A_TBL_SIZE)
421 1.1 gwr * 3) use this index to select the corresponding 'A' table manager
422 1.1 gwr * structure from the 'A' table manager pool (Atmgrbase[index]).
423 1.1 gwr */
424 1.7 gwr /* This function is not currently used. */
425 1.7 gwr #if 0
426 1.7 gwr static INLINE a_tmgr_t *
427 1.7 gwr mmuA2tmgr(mmuAtbl)
428 1.7 gwr mmu_long_dte_t *mmuAtbl;
429 1.7 gwr {
430 1.7 gwr register int idx;
431 1.7 gwr
432 1.7 gwr /* Which table is this in? */
433 1.7 gwr idx = (mmuAtbl - mmuAbase) / MMU_A_TBL_SIZE;
434 1.8 gwr #ifdef PMAP_DEBUG
435 1.7 gwr if ((idx < 0) || (idx >= NUM_A_TABLES))
436 1.7 gwr panic("mmuA2tmgr");
437 1.7 gwr #endif
438 1.7 gwr return (&Atmgrbase[idx]);
439 1.7 gwr }
440 1.7 gwr #endif /* 0 */
441 1.7 gwr
442 1.7 gwr static INLINE b_tmgr_t *
443 1.7 gwr mmuB2tmgr(mmuBtbl)
444 1.7 gwr mmu_short_dte_t *mmuBtbl;
445 1.7 gwr {
446 1.7 gwr register int idx;
447 1.7 gwr
448 1.7 gwr /* Which table is this in? */
449 1.7 gwr idx = (mmuBtbl - mmuBbase) / MMU_B_TBL_SIZE;
450 1.8 gwr #ifdef PMAP_DEBUG
451 1.7 gwr if ((idx < 0) || (idx >= NUM_B_TABLES))
452 1.7 gwr panic("mmuB2tmgr");
453 1.7 gwr #endif
454 1.7 gwr return (&Btmgrbase[idx]);
455 1.7 gwr }
456 1.7 gwr
457 1.7 gwr /* mmuC2tmgr INTERNAL
458 1.7 gwr **
459 1.7 gwr * Given a pte known to belong to a C table, return the address of
460 1.7 gwr * that table's management structure.
461 1.7 gwr */
462 1.7 gwr static INLINE c_tmgr_t *
463 1.7 gwr mmuC2tmgr(mmuCtbl)
464 1.7 gwr mmu_short_pte_t *mmuCtbl;
465 1.7 gwr {
466 1.7 gwr register int idx;
467 1.7 gwr
468 1.7 gwr /* Which table is this in? */
469 1.7 gwr idx = (mmuCtbl - mmuCbase) / MMU_C_TBL_SIZE;
470 1.8 gwr #ifdef PMAP_DEBUG
471 1.7 gwr if ((idx < 0) || (idx >= NUM_C_TABLES))
472 1.7 gwr panic("mmuC2tmgr");
473 1.7 gwr #endif
474 1.7 gwr return (&Ctmgrbase[idx]);
475 1.7 gwr }
476 1.7 gwr
477 1.8 gwr /* This is now a function call below.
478 1.1 gwr * #define pa2pv(pa) \
479 1.1 gwr * (&pvbase[(unsigned long)\
480 1.16 gwr * _btop(pa)\
481 1.1 gwr * ])
482 1.1 gwr */
483 1.1 gwr
484 1.7 gwr /* pa2pv INTERNAL
485 1.7 gwr **
486 1.7 gwr * Return the pv_list_head element which manages the given physical
487 1.7 gwr * address.
488 1.7 gwr */
489 1.7 gwr static INLINE pv_t *
490 1.7 gwr pa2pv(pa)
491 1.7 gwr vm_offset_t pa;
492 1.7 gwr {
493 1.7 gwr register struct pmap_physmem_struct *bank;
494 1.7 gwr register int idx;
495 1.7 gwr
496 1.7 gwr bank = &avail_mem[0];
497 1.7 gwr while (pa >= bank->pmem_end)
498 1.7 gwr bank = bank->pmem_next;
499 1.7 gwr
500 1.7 gwr pa -= bank->pmem_start;
501 1.16 gwr idx = bank->pmem_pvbase + _btop(pa);
502 1.8 gwr #ifdef PMAP_DEBUG
503 1.7 gwr if ((idx < 0) || (idx >= physmem))
504 1.7 gwr panic("pa2pv");
505 1.7 gwr #endif
506 1.7 gwr return &pvbase[idx];
507 1.7 gwr }
508 1.7 gwr
509 1.7 gwr /* pteidx INTERNAL
510 1.7 gwr **
511 1.7 gwr * Return the index of the given PTE within the entire fixed table of
512 1.7 gwr * PTEs.
513 1.7 gwr */
514 1.7 gwr static INLINE int
515 1.7 gwr pteidx(pte)
516 1.7 gwr mmu_short_pte_t *pte;
517 1.7 gwr {
518 1.7 gwr return (pte - kernCbase);
519 1.7 gwr }
520 1.7 gwr
521 1.7 gwr /*
522 1.8 gwr * This just offers a place to put some debugging checks,
523 1.8 gwr * and reduces the number of places "curproc" appears...
524 1.7 gwr */
525 1.7 gwr static INLINE pmap_t
526 1.7 gwr current_pmap()
527 1.7 gwr {
528 1.7 gwr struct proc *p;
529 1.7 gwr struct vmspace *vm;
530 1.7 gwr vm_map_t map;
531 1.7 gwr pmap_t pmap;
532 1.7 gwr
533 1.7 gwr p = curproc; /* XXX */
534 1.9 gwr if (p == NULL)
535 1.9 gwr pmap = &kernel_pmap;
536 1.9 gwr else {
537 1.9 gwr vm = p->p_vmspace;
538 1.9 gwr map = &vm->vm_map;
539 1.9 gwr pmap = vm_map_pmap(map);
540 1.9 gwr }
541 1.7 gwr
542 1.7 gwr return (pmap);
543 1.7 gwr }
544 1.7 gwr
545 1.7 gwr
546 1.1 gwr /*************************** FUNCTION DEFINITIONS ************************
547 1.1 gwr * These appear here merely for the compiler to enforce type checking on *
548 1.1 gwr * all function calls. *
549 1.7 gwr *************************************************************************/
550 1.1 gwr
551 1.1 gwr /** External functions
552 1.1 gwr ** - functions used within this module but written elsewhere.
553 1.1 gwr ** both of these functions are in locore.s
554 1.7 gwr ** XXX - These functions were later replaced with their more cryptic
555 1.7 gwr ** hp300 counterparts. They may be removed now.
556 1.7 gwr **/
557 1.7 gwr #if 0 /* deprecated mmu */
558 1.1 gwr void mmu_seturp __P((vm_offset_t));
559 1.1 gwr void mmu_flush __P((int, vm_offset_t));
560 1.1 gwr void mmu_flusha __P((void));
561 1.7 gwr #endif /* 0 */
562 1.1 gwr
563 1.1 gwr /** Internal functions
564 1.1 gwr ** - all functions used only within this module are defined in
565 1.1 gwr ** pmap_pvt.h
566 1.1 gwr **/
567 1.1 gwr
568 1.1 gwr /** Interface functions
569 1.1 gwr ** - functions required by the Mach VM Pmap interface, with MACHINE_CONTIG
570 1.1 gwr ** defined.
571 1.1 gwr **/
572 1.1 gwr #ifdef INCLUDED_IN_PMAP_H
573 1.1 gwr void pmap_bootstrap __P((void));
574 1.1 gwr void *pmap_bootstrap_alloc __P((int));
575 1.1 gwr void pmap_enter __P((pmap_t, vm_offset_t, vm_offset_t, vm_prot_t, boolean_t));
576 1.1 gwr pmap_t pmap_create __P((vm_size_t));
577 1.1 gwr void pmap_destroy __P((pmap_t));
578 1.1 gwr void pmap_reference __P((pmap_t));
579 1.1 gwr boolean_t pmap_is_referenced __P((vm_offset_t));
580 1.1 gwr boolean_t pmap_is_modified __P((vm_offset_t));
581 1.1 gwr void pmap_clear_modify __P((vm_offset_t));
582 1.1 gwr vm_offset_t pmap_extract __P((pmap_t, vm_offset_t));
583 1.7 gwr void pmap_activate __P((pmap_t));
584 1.1 gwr int pmap_page_index __P((vm_offset_t));
585 1.1 gwr u_int pmap_free_pages __P((void));
586 1.1 gwr #endif /* INCLUDED_IN_PMAP_H */
587 1.1 gwr
588 1.1 gwr /********************************** CODE ********************************
589 1.1 gwr * Functions that are called from other parts of the kernel are labeled *
590 1.1 gwr * as 'INTERFACE' functions. Functions that are only called from *
591 1.1 gwr * within the pmap module are labeled as 'INTERNAL' functions. *
592 1.1 gwr * Functions that are internal, but are not (currently) used at all are *
593 1.1 gwr * labeled 'INTERNAL_X'. *
594 1.1 gwr ************************************************************************/
595 1.1 gwr
596 1.1 gwr /* pmap_bootstrap INTERNAL
597 1.1 gwr **
598 1.16 gwr * Initializes the pmap system. Called at boot time from _vm_init()
599 1.1 gwr * in _startup.c.
600 1.1 gwr *
601 1.1 gwr * Reminder: having a pmap_bootstrap_alloc() and also having the VM
602 1.1 gwr * system implement pmap_steal_memory() is redundant.
603 1.1 gwr * Don't release this code without removing one or the other!
604 1.1 gwr */
605 1.1 gwr void
606 1.1 gwr pmap_bootstrap(nextva)
607 1.1 gwr vm_offset_t nextva;
608 1.1 gwr {
609 1.1 gwr struct physmemory *membank;
610 1.1 gwr struct pmap_physmem_struct *pmap_membank;
611 1.1 gwr vm_offset_t va, pa, eva;
612 1.1 gwr int b, c, i, j; /* running table counts */
613 1.1 gwr int size;
614 1.1 gwr
615 1.1 gwr /*
616 1.1 gwr * This function is called by __bootstrap after it has
617 1.1 gwr * determined the type of machine and made the appropriate
618 1.1 gwr * patches to the ROM vectors (XXX- I don't quite know what I meant
619 1.1 gwr * by that.) It allocates and sets up enough of the pmap system
620 1.1 gwr * to manage the kernel's address space.
621 1.1 gwr */
622 1.1 gwr
623 1.1 gwr /*
624 1.7 gwr * Determine the range of kernel virtual and physical
625 1.7 gwr * space available. Note that we ABSOLUTELY DEPEND on
626 1.7 gwr * the fact that the first bank of memory (4MB) is
627 1.7 gwr * mapped linearly to KERNBASE (which we guaranteed in
628 1.7 gwr * the first instructions of locore.s).
629 1.7 gwr * That is plenty for our bootstrap work.
630 1.1 gwr */
631 1.16 gwr virtual_avail = _round_page(nextva);
632 1.7 gwr virtual_contig_end = KERNBASE + 0x400000; /* +4MB */
633 1.1 gwr virtual_end = VM_MAX_KERNEL_ADDRESS;
634 1.7 gwr /* Don't need avail_start til later. */
635 1.1 gwr
636 1.7 gwr /* We may now call pmap_bootstrap_alloc(). */
637 1.7 gwr bootstrap_alloc_enabled = TRUE;
638 1.1 gwr
639 1.1 gwr /*
640 1.1 gwr * This is a somewhat unwrapped loop to deal with
641 1.1 gwr * copying the PROM's 'phsymem' banks into the pmap's
642 1.1 gwr * banks. The following is always assumed:
643 1.1 gwr * 1. There is always at least one bank of memory.
644 1.1 gwr * 2. There is always a last bank of memory, and its
645 1.1 gwr * pmem_next member must be set to NULL.
646 1.1 gwr * XXX - Use: do { ... } while (membank->next) instead?
647 1.1 gwr * XXX - Why copy this stuff at all? -gwr
648 1.8 gwr * - It is needed in pa2pv().
649 1.1 gwr */
650 1.1 gwr membank = romVectorPtr->v_physmemory;
651 1.1 gwr pmap_membank = avail_mem;
652 1.1 gwr total_phys_mem = 0;
653 1.1 gwr
654 1.1 gwr while (membank->next) {
655 1.1 gwr pmap_membank->pmem_start = membank->address;
656 1.1 gwr pmap_membank->pmem_end = membank->address + membank->size;
657 1.1 gwr total_phys_mem += membank->size;
658 1.1 gwr /* This silly syntax arises because pmap_membank
659 1.1 gwr * is really a pre-allocated array, but it is put into
660 1.1 gwr * use as a linked list.
661 1.1 gwr */
662 1.1 gwr pmap_membank->pmem_next = pmap_membank + 1;
663 1.1 gwr pmap_membank = pmap_membank->pmem_next;
664 1.1 gwr membank = membank->next;
665 1.1 gwr }
666 1.1 gwr
667 1.1 gwr /*
668 1.1 gwr * XXX The last bank of memory should be reduced to exclude the
669 1.1 gwr * physical pages needed by the PROM monitor from being used
670 1.1 gwr * in the VM system. XXX - See below - Fix!
671 1.1 gwr */
672 1.1 gwr pmap_membank->pmem_start = membank->address;
673 1.1 gwr pmap_membank->pmem_end = membank->address + membank->size;
674 1.1 gwr pmap_membank->pmem_next = NULL;
675 1.1 gwr
676 1.1 gwr #if 0 /* XXX - Need to integrate this! */
677 1.1 gwr /*
678 1.1 gwr * The last few pages of physical memory are "owned" by
679 1.1 gwr * the PROM. The total amount of memory we are allowed
680 1.1 gwr * to use is given by the romvec pointer. -gwr
681 1.1 gwr *
682 1.1 gwr * We should dedicate different variables for 'useable'
683 1.1 gwr * and 'physically available'. Most users are used to the
684 1.1 gwr * kernel reporting the amount of memory 'physically available'
685 1.1 gwr * as opposed to 'useable by the kernel' at boot time. -j
686 1.1 gwr */
687 1.1 gwr total_phys_mem = *romVectorPtr->memoryAvail;
688 1.1 gwr #endif /* XXX */
689 1.1 gwr
690 1.1 gwr total_phys_mem += membank->size; /* XXX see above */
691 1.1 gwr physmem = btoc(total_phys_mem);
692 1.7 gwr
693 1.7 gwr /*
694 1.7 gwr * Avail_end is set to the first byte of physical memory
695 1.7 gwr * after the end of the last bank. We use this only to
696 1.7 gwr * determine if a physical address is "managed" memory.
697 1.8 gwr *
698 1.8 gwr * XXX - The setting of avail_end is a temporary ROM saving hack.
699 1.7 gwr */
700 1.8 gwr avail_end = pmap_membank->pmem_end -
701 1.8 gwr (total_phys_mem - *romVectorPtr->memoryAvail);
702 1.16 gwr avail_end = _trunc_page(avail_end);
703 1.1 gwr
704 1.1 gwr /*
705 1.15 gwr * First allocate enough kernel MMU tables to map all
706 1.15 gwr * of kernel virtual space from KERNBASE to 0xFFFFFFFF.
707 1.1 gwr * Note: All must be aligned on 256 byte boundaries.
708 1.15 gwr * Start with the level-A table (one of those).
709 1.1 gwr */
710 1.15 gwr size = sizeof(mmu_long_dte_t) * MMU_A_TBL_SIZE;
711 1.7 gwr kernAbase = pmap_bootstrap_alloc(size);
712 1.1 gwr bzero(kernAbase, size);
713 1.1 gwr
714 1.15 gwr /* Now the level-B kernel tables... */
715 1.15 gwr size = sizeof(mmu_short_dte_t) * MMU_B_TBL_SIZE * KERN_B_TABLES;
716 1.7 gwr kernBbase = pmap_bootstrap_alloc(size);
717 1.1 gwr bzero(kernBbase, size);
718 1.1 gwr
719 1.15 gwr /* Now the level-C kernel tables... */
720 1.15 gwr size = sizeof(mmu_short_pte_t) * MMU_C_TBL_SIZE * KERN_C_TABLES;
721 1.15 gwr kernCbase = pmap_bootstrap_alloc(size);
722 1.15 gwr bzero(kernCbase, size);
723 1.7 gwr /*
724 1.7 gwr * Note: In order for the PV system to work correctly, the kernel
725 1.7 gwr * and user-level C tables must be allocated contiguously.
726 1.7 gwr * Nothing should be allocated between here and the allocation of
727 1.7 gwr * mmuCbase below. XXX: Should do this as one allocation, and
728 1.7 gwr * then compute a pointer for mmuCbase instead of this...
729 1.15 gwr *
730 1.15 gwr * Allocate user MMU tables.
731 1.15 gwr * These must be contiguous with the preceeding.
732 1.7 gwr */
733 1.22 jeremy
734 1.22 jeremy #ifndef FIXED_NTABLES
735 1.22 jeremy /*
736 1.22 jeremy * The number of user-level C tables that should be allocated is
737 1.22 jeremy * related to the size of physical memory. In general, there should
738 1.22 jeremy * be enough tables to map four times the amount of available RAM.
739 1.22 jeremy * The extra amount is needed because some table space is wasted by
740 1.22 jeremy * fragmentation.
741 1.22 jeremy */
742 1.22 jeremy NUM_C_TABLES = (total_phys_mem * 4) / (MMU_C_TBL_SIZE * MMU_PAGE_SIZE);
743 1.22 jeremy NUM_B_TABLES = NUM_C_TABLES / 2;
744 1.22 jeremy NUM_A_TABLES = NUM_B_TABLES / 2;
745 1.22 jeremy #endif /* !FIXED_NTABLES */
746 1.22 jeremy
747 1.15 gwr size = sizeof(mmu_short_pte_t) * MMU_C_TBL_SIZE * NUM_C_TABLES;
748 1.15 gwr mmuCbase = pmap_bootstrap_alloc(size);
749 1.15 gwr
750 1.15 gwr size = sizeof(mmu_short_dte_t) * MMU_B_TBL_SIZE * NUM_B_TABLES;
751 1.15 gwr mmuBbase = pmap_bootstrap_alloc(size);
752 1.1 gwr
753 1.15 gwr size = sizeof(mmu_long_dte_t) * MMU_A_TBL_SIZE * NUM_A_TABLES;
754 1.15 gwr mmuAbase = pmap_bootstrap_alloc(size);
755 1.7 gwr
756 1.7 gwr /*
757 1.7 gwr * Fill in the never-changing part of the kernel tables.
758 1.7 gwr * For simplicity, the kernel's mappings will be editable as a
759 1.1 gwr * flat array of page table entries at kernCbase. The
760 1.1 gwr * higher level 'A' and 'B' tables must be initialized to point
761 1.1 gwr * to this lower one.
762 1.1 gwr */
763 1.1 gwr b = c = 0;
764 1.1 gwr
765 1.7 gwr /*
766 1.7 gwr * Invalidate all mappings below KERNBASE in the A table.
767 1.1 gwr * This area has already been zeroed out, but it is good
768 1.1 gwr * practice to explicitly show that we are interpreting
769 1.1 gwr * it as a list of A table descriptors.
770 1.1 gwr */
771 1.1 gwr for (i = 0; i < MMU_TIA(KERNBASE); i++) {
772 1.1 gwr kernAbase[i].addr.raw = 0;
773 1.1 gwr }
774 1.1 gwr
775 1.7 gwr /*
776 1.7 gwr * Set up the kernel A and B tables so that they will reference the
777 1.1 gwr * correct spots in the contiguous table of PTEs allocated for the
778 1.1 gwr * kernel's virtual memory space.
779 1.1 gwr */
780 1.1 gwr for (i = MMU_TIA(KERNBASE); i < MMU_A_TBL_SIZE; i++) {
781 1.1 gwr kernAbase[i].attr.raw =
782 1.1 gwr MMU_LONG_DTE_LU | MMU_LONG_DTE_SUPV | MMU_DT_SHORT;
783 1.7 gwr kernAbase[i].addr.raw = mmu_vtop(&kernBbase[b]);
784 1.1 gwr
785 1.1 gwr for (j=0; j < MMU_B_TBL_SIZE; j++) {
786 1.7 gwr kernBbase[b + j].attr.raw = mmu_vtop(&kernCbase[c])
787 1.1 gwr | MMU_DT_SHORT;
788 1.1 gwr c += MMU_C_TBL_SIZE;
789 1.1 gwr }
790 1.1 gwr b += MMU_B_TBL_SIZE;
791 1.1 gwr }
792 1.1 gwr
793 1.8 gwr /* XXX - Doing kernel_pmap a little further down. */
794 1.8 gwr
795 1.7 gwr pmap_alloc_usermmu(); /* Allocate user MMU tables. */
796 1.7 gwr pmap_alloc_usertmgr(); /* Allocate user MMU table managers.*/
797 1.7 gwr pmap_alloc_pv(); /* Allocate physical->virtual map. */
798 1.7 gwr
799 1.7 gwr /*
800 1.7 gwr * We are now done with pmap_bootstrap_alloc(). Round up
801 1.7 gwr * `virtual_avail' to the nearest page, and set the flag
802 1.7 gwr * to prevent use of pmap_bootstrap_alloc() hereafter.
803 1.7 gwr */
804 1.7 gwr pmap_bootstrap_aalign(NBPG);
805 1.7 gwr bootstrap_alloc_enabled = FALSE;
806 1.7 gwr
807 1.7 gwr /*
808 1.7 gwr * Now that we are done with pmap_bootstrap_alloc(), we
809 1.7 gwr * must save the virtual and physical addresses of the
810 1.7 gwr * end of the linearly mapped range, which are stored in
811 1.7 gwr * virtual_contig_end and avail_start, respectively.
812 1.7 gwr * These variables will never change after this point.
813 1.7 gwr */
814 1.7 gwr virtual_contig_end = virtual_avail;
815 1.7 gwr avail_start = virtual_avail - KERNBASE;
816 1.7 gwr
817 1.7 gwr /*
818 1.7 gwr * `avail_next' is a running pointer used by pmap_next_page() to
819 1.7 gwr * keep track of the next available physical page to be handed
820 1.7 gwr * to the VM system during its initialization, in which it
821 1.7 gwr * asks for physical pages, one at a time.
822 1.7 gwr */
823 1.7 gwr avail_next = avail_start;
824 1.7 gwr
825 1.7 gwr /*
826 1.7 gwr * Now allocate some virtual addresses, but not the physical pages
827 1.7 gwr * behind them. Note that virtual_avail is already page-aligned.
828 1.7 gwr *
829 1.7 gwr * tmp_vpages[] is an array of two virtual pages used for temporary
830 1.7 gwr * kernel mappings in the pmap module to facilitate various physical
831 1.7 gwr * address-oritented operations.
832 1.7 gwr */
833 1.7 gwr tmp_vpages[0] = virtual_avail;
834 1.7 gwr virtual_avail += NBPG;
835 1.7 gwr tmp_vpages[1] = virtual_avail;
836 1.7 gwr virtual_avail += NBPG;
837 1.23 jeremy tmp_vpage0_inuse = tmp_vpage1_inuse = 0;
838 1.7 gwr
839 1.7 gwr /** Initialize the PV system **/
840 1.7 gwr pmap_init_pv();
841 1.7 gwr
842 1.7 gwr /*
843 1.7 gwr * Fill in the kernel_pmap structure and kernel_crp.
844 1.7 gwr */
845 1.7 gwr kernAphys = mmu_vtop(kernAbase);
846 1.7 gwr kernel_pmap.pm_a_tmgr = NULL;
847 1.7 gwr kernel_pmap.pm_a_phys = kernAphys;
848 1.7 gwr kernel_pmap.pm_refcount = 1; /* always in use */
849 1.7 gwr
850 1.7 gwr kernel_crp.rp_attr = MMU_LONG_DTE_LU | MMU_DT_LONG;
851 1.7 gwr kernel_crp.rp_addr = kernAphys;
852 1.7 gwr
853 1.1 gwr /*
854 1.1 gwr * Now pmap_enter_kernel() may be used safely and will be
855 1.7 gwr * the main interface used hereafter to modify the kernel's
856 1.7 gwr * virtual address space. Note that since we are still running
857 1.7 gwr * under the PROM's address table, none of these table modifications
858 1.7 gwr * actually take effect until pmap_takeover_mmu() is called.
859 1.1 gwr *
860 1.7 gwr * Note: Our tables do NOT have the PROM linear mappings!
861 1.7 gwr * Only the mappings created here exist in our tables, so
862 1.7 gwr * remember to map anything we expect to use.
863 1.1 gwr */
864 1.1 gwr va = (vm_offset_t) KERNBASE;
865 1.7 gwr pa = 0;
866 1.1 gwr
867 1.1 gwr /*
868 1.7 gwr * The first page of the kernel virtual address space is the msgbuf
869 1.7 gwr * page. The page attributes (data, non-cached) are set here, while
870 1.7 gwr * the address is assigned to this global pointer in cpu_startup().
871 1.1 gwr * XXX - Make it non-cached?
872 1.1 gwr */
873 1.1 gwr pmap_enter_kernel(va, pa|PMAP_NC, VM_PROT_ALL);
874 1.1 gwr va += NBPG; pa += NBPG;
875 1.1 gwr
876 1.7 gwr /* Next page is used as the temporary stack. */
877 1.1 gwr pmap_enter_kernel(va, pa, VM_PROT_ALL);
878 1.1 gwr va += NBPG; pa += NBPG;
879 1.1 gwr
880 1.1 gwr /*
881 1.1 gwr * Map all of the kernel's text segment as read-only and cacheable.
882 1.1 gwr * (Cacheable is implied by default). Unfortunately, the last bytes
883 1.1 gwr * of kernel text and the first bytes of kernel data will often be
884 1.1 gwr * sharing the same page. Therefore, the last page of kernel text
885 1.1 gwr * has to be mapped as read/write, to accomodate the data.
886 1.1 gwr */
887 1.16 gwr eva = _trunc_page((vm_offset_t)etext);
888 1.7 gwr for (; va < eva; va += NBPG, pa += NBPG)
889 1.1 gwr pmap_enter_kernel(va, pa, VM_PROT_READ|VM_PROT_EXECUTE);
890 1.1 gwr
891 1.7 gwr /*
892 1.7 gwr * Map all of the kernel's data as read/write and cacheable.
893 1.7 gwr * This includes: data, BSS, symbols, and everything in the
894 1.7 gwr * contiguous memory used by pmap_bootstrap_alloc()
895 1.1 gwr */
896 1.7 gwr for (; pa < avail_start; va += NBPG, pa += NBPG)
897 1.1 gwr pmap_enter_kernel(va, pa, VM_PROT_READ|VM_PROT_WRITE);
898 1.1 gwr
899 1.7 gwr /*
900 1.7 gwr * At this point we are almost ready to take over the MMU. But first
901 1.7 gwr * we must save the PROM's address space in our map, as we call its
902 1.7 gwr * routines and make references to its data later in the kernel.
903 1.1 gwr */
904 1.7 gwr pmap_bootstrap_copyprom();
905 1.7 gwr pmap_takeover_mmu();
906 1.13 gwr pmap_bootstrap_setprom();
907 1.1 gwr
908 1.1 gwr /* Notify the VM system of our page size. */
909 1.1 gwr PAGE_SIZE = NBPG;
910 1.1 gwr vm_set_page_size();
911 1.1 gwr }
912 1.1 gwr
913 1.1 gwr
914 1.1 gwr /* pmap_alloc_usermmu INTERNAL
915 1.1 gwr **
916 1.1 gwr * Called from pmap_bootstrap() to allocate MMU tables that will
917 1.1 gwr * eventually be used for user mappings.
918 1.1 gwr */
919 1.1 gwr void
920 1.1 gwr pmap_alloc_usermmu()
921 1.1 gwr {
922 1.7 gwr /* XXX: Moved into caller. */
923 1.1 gwr }
924 1.1 gwr
925 1.1 gwr /* pmap_alloc_pv INTERNAL
926 1.1 gwr **
927 1.1 gwr * Called from pmap_bootstrap() to allocate the physical
928 1.1 gwr * to virtual mapping list. Each physical page of memory
929 1.1 gwr * in the system has a corresponding element in this list.
930 1.1 gwr */
931 1.1 gwr void
932 1.1 gwr pmap_alloc_pv()
933 1.1 gwr {
934 1.1 gwr int i;
935 1.1 gwr unsigned int total_mem;
936 1.1 gwr
937 1.7 gwr /*
938 1.7 gwr * Allocate a pv_head structure for every page of physical
939 1.1 gwr * memory that will be managed by the system. Since memory on
940 1.1 gwr * the 3/80 is non-contiguous, we cannot arrive at a total page
941 1.1 gwr * count by subtraction of the lowest available address from the
942 1.1 gwr * highest, but rather we have to step through each memory
943 1.1 gwr * bank and add the number of pages in each to the total.
944 1.1 gwr *
945 1.1 gwr * At this time we also initialize the offset of each bank's
946 1.1 gwr * starting pv_head within the pv_head list so that the physical
947 1.1 gwr * memory state routines (pmap_is_referenced(),
948 1.1 gwr * pmap_is_modified(), et al.) can quickly find coresponding
949 1.1 gwr * pv_heads in spite of the non-contiguity.
950 1.1 gwr */
951 1.1 gwr total_mem = 0;
952 1.20 thorpej for (i = 0; i < SUN3X_NPHYS_RAM_SEGS; i++) {
953 1.16 gwr avail_mem[i].pmem_pvbase = _btop(total_mem);
954 1.1 gwr total_mem += avail_mem[i].pmem_end -
955 1.1 gwr avail_mem[i].pmem_start;
956 1.1 gwr if (avail_mem[i].pmem_next == NULL)
957 1.1 gwr break;
958 1.1 gwr }
959 1.1 gwr #ifdef PMAP_DEBUG
960 1.1 gwr if (total_mem != total_phys_mem)
961 1.1 gwr panic("pmap_alloc_pv did not arrive at correct page count");
962 1.1 gwr #endif
963 1.1 gwr
964 1.1 gwr pvbase = (pv_t *) pmap_bootstrap_alloc(sizeof(pv_t) *
965 1.16 gwr _btop(total_phys_mem));
966 1.1 gwr }
967 1.1 gwr
968 1.1 gwr /* pmap_alloc_usertmgr INTERNAL
969 1.1 gwr **
970 1.1 gwr * Called from pmap_bootstrap() to allocate the structures which
971 1.1 gwr * facilitate management of user MMU tables. Each user MMU table
972 1.1 gwr * in the system has one such structure associated with it.
973 1.1 gwr */
974 1.1 gwr void
975 1.1 gwr pmap_alloc_usertmgr()
976 1.1 gwr {
977 1.1 gwr /* Allocate user MMU table managers */
978 1.7 gwr /* It would be a lot simpler to just make these BSS, but */
979 1.7 gwr /* we may want to change their size at boot time... -j */
980 1.1 gwr Atmgrbase = (a_tmgr_t *) pmap_bootstrap_alloc(sizeof(a_tmgr_t)
981 1.1 gwr * NUM_A_TABLES);
982 1.1 gwr Btmgrbase = (b_tmgr_t *) pmap_bootstrap_alloc(sizeof(b_tmgr_t)
983 1.1 gwr * NUM_B_TABLES);
984 1.1 gwr Ctmgrbase = (c_tmgr_t *) pmap_bootstrap_alloc(sizeof(c_tmgr_t)
985 1.1 gwr * NUM_C_TABLES);
986 1.1 gwr
987 1.7 gwr /*
988 1.7 gwr * Allocate PV list elements for the physical to virtual
989 1.1 gwr * mapping system.
990 1.1 gwr */
991 1.1 gwr pvebase = (pv_elem_t *) pmap_bootstrap_alloc(
992 1.7 gwr sizeof(pv_elem_t) * (NUM_USER_PTES + NUM_KERN_PTES));
993 1.1 gwr }
994 1.1 gwr
995 1.1 gwr /* pmap_bootstrap_copyprom() INTERNAL
996 1.1 gwr **
997 1.1 gwr * Copy the PROM mappings into our own tables. Note, we
998 1.1 gwr * can use physical addresses until __bootstrap returns.
999 1.1 gwr */
1000 1.1 gwr void
1001 1.1 gwr pmap_bootstrap_copyprom()
1002 1.1 gwr {
1003 1.1 gwr MachMonRomVector *romp;
1004 1.1 gwr int *mon_ctbl;
1005 1.1 gwr mmu_short_pte_t *kpte;
1006 1.1 gwr int i, len;
1007 1.1 gwr
1008 1.1 gwr romp = romVectorPtr;
1009 1.1 gwr
1010 1.1 gwr /*
1011 1.1 gwr * Copy the mappings in MON_KDB_START...MONEND
1012 1.1 gwr * Note: mon_ctbl[0] maps MON_KDB_START
1013 1.1 gwr */
1014 1.1 gwr mon_ctbl = *romp->monptaddr;
1015 1.16 gwr i = _btop(MON_KDB_START - KERNBASE);
1016 1.1 gwr kpte = &kernCbase[i];
1017 1.16 gwr len = _btop(MONEND - MON_KDB_START);
1018 1.1 gwr
1019 1.1 gwr for (i = 0; i < len; i++) {
1020 1.1 gwr kpte[i].attr.raw = mon_ctbl[i];
1021 1.1 gwr }
1022 1.1 gwr
1023 1.1 gwr /*
1024 1.1 gwr * Copy the mappings at MON_DVMA_BASE (to the end).
1025 1.1 gwr * Note, in here, mon_ctbl[0] maps MON_DVMA_BASE.
1026 1.1 gwr * XXX - This does not appear to be necessary, but
1027 1.1 gwr * I'm not sure yet if it is or not. -gwr
1028 1.1 gwr */
1029 1.1 gwr mon_ctbl = *romp->shadowpteaddr;
1030 1.16 gwr i = _btop(MON_DVMA_BASE - KERNBASE);
1031 1.1 gwr kpte = &kernCbase[i];
1032 1.16 gwr len = _btop(MON_DVMA_SIZE);
1033 1.1 gwr
1034 1.1 gwr for (i = 0; i < len; i++) {
1035 1.1 gwr kpte[i].attr.raw = mon_ctbl[i];
1036 1.1 gwr }
1037 1.1 gwr }
1038 1.1 gwr
1039 1.1 gwr /* pmap_takeover_mmu INTERNAL
1040 1.1 gwr **
1041 1.1 gwr * Called from pmap_bootstrap() after it has copied enough of the
1042 1.1 gwr * PROM mappings into the kernel map so that we can use our own
1043 1.1 gwr * MMU table.
1044 1.1 gwr */
1045 1.1 gwr void
1046 1.1 gwr pmap_takeover_mmu()
1047 1.1 gwr {
1048 1.1 gwr
1049 1.13 gwr loadcrp(&kernel_crp);
1050 1.1 gwr }
1051 1.1 gwr
1052 1.13 gwr /* pmap_bootstrap_setprom() INTERNAL
1053 1.13 gwr **
1054 1.13 gwr * Set the PROM mappings so it can see kernel space.
1055 1.13 gwr * Note that physical addresses are used here, which
1056 1.13 gwr * we can get away with because this runs with the
1057 1.13 gwr * low 1GB set for transparent translation.
1058 1.13 gwr */
1059 1.13 gwr void
1060 1.13 gwr pmap_bootstrap_setprom()
1061 1.13 gwr {
1062 1.13 gwr mmu_long_dte_t *mon_dte;
1063 1.13 gwr extern struct mmu_rootptr mon_crp;
1064 1.13 gwr int i;
1065 1.13 gwr
1066 1.13 gwr mon_dte = (mmu_long_dte_t *) mon_crp.rp_addr;
1067 1.13 gwr for (i = MMU_TIA(KERNBASE); i < MMU_TIA(KERN_END); i++) {
1068 1.13 gwr mon_dte[i].attr.raw = kernAbase[i].attr.raw;
1069 1.13 gwr mon_dte[i].addr.raw = kernAbase[i].addr.raw;
1070 1.13 gwr }
1071 1.13 gwr }
1072 1.13 gwr
1073 1.13 gwr
1074 1.1 gwr /* pmap_init INTERFACE
1075 1.1 gwr **
1076 1.1 gwr * Called at the end of vm_init() to set up the pmap system to go
1077 1.7 gwr * into full time operation. All initialization of kernel_pmap
1078 1.7 gwr * should be already done by now, so this should just do things
1079 1.7 gwr * needed for user-level pmaps to work.
1080 1.1 gwr */
1081 1.1 gwr void
1082 1.1 gwr pmap_init()
1083 1.1 gwr {
1084 1.1 gwr /** Initialize the manager pools **/
1085 1.1 gwr TAILQ_INIT(&a_pool);
1086 1.1 gwr TAILQ_INIT(&b_pool);
1087 1.1 gwr TAILQ_INIT(&c_pool);
1088 1.1 gwr
1089 1.1 gwr /**************************************************************
1090 1.1 gwr * Initialize all tmgr structures and MMU tables they manage. *
1091 1.1 gwr **************************************************************/
1092 1.1 gwr /** Initialize A tables **/
1093 1.1 gwr pmap_init_a_tables();
1094 1.1 gwr /** Initialize B tables **/
1095 1.1 gwr pmap_init_b_tables();
1096 1.1 gwr /** Initialize C tables **/
1097 1.1 gwr pmap_init_c_tables();
1098 1.1 gwr }
1099 1.1 gwr
1100 1.1 gwr /* pmap_init_a_tables() INTERNAL
1101 1.1 gwr **
1102 1.1 gwr * Initializes all A managers, their MMU A tables, and inserts
1103 1.1 gwr * them into the A manager pool for use by the system.
1104 1.1 gwr */
1105 1.1 gwr void
1106 1.1 gwr pmap_init_a_tables()
1107 1.1 gwr {
1108 1.1 gwr int i;
1109 1.1 gwr a_tmgr_t *a_tbl;
1110 1.1 gwr
1111 1.1 gwr for (i=0; i < NUM_A_TABLES; i++) {
1112 1.1 gwr /* Select the next available A manager from the pool */
1113 1.1 gwr a_tbl = &Atmgrbase[i];
1114 1.1 gwr
1115 1.7 gwr /*
1116 1.7 gwr * Clear its parent entry. Set its wired and valid
1117 1.1 gwr * entry count to zero.
1118 1.1 gwr */
1119 1.1 gwr a_tbl->at_parent = NULL;
1120 1.1 gwr a_tbl->at_wcnt = a_tbl->at_ecnt = 0;
1121 1.1 gwr
1122 1.1 gwr /* Assign it the next available MMU A table from the pool */
1123 1.1 gwr a_tbl->at_dtbl = &mmuAbase[i * MMU_A_TBL_SIZE];
1124 1.1 gwr
1125 1.7 gwr /*
1126 1.7 gwr * Initialize the MMU A table with the table in the `proc0',
1127 1.1 gwr * or kernel, mapping. This ensures that every process has
1128 1.1 gwr * the kernel mapped in the top part of its address space.
1129 1.1 gwr */
1130 1.1 gwr bcopy(kernAbase, a_tbl->at_dtbl, MMU_A_TBL_SIZE *
1131 1.1 gwr sizeof(mmu_long_dte_t));
1132 1.1 gwr
1133 1.7 gwr /*
1134 1.7 gwr * Finally, insert the manager into the A pool,
1135 1.1 gwr * making it ready to be used by the system.
1136 1.1 gwr */
1137 1.1 gwr TAILQ_INSERT_TAIL(&a_pool, a_tbl, at_link);
1138 1.1 gwr }
1139 1.1 gwr }
1140 1.1 gwr
1141 1.1 gwr /* pmap_init_b_tables() INTERNAL
1142 1.1 gwr **
1143 1.1 gwr * Initializes all B table managers, their MMU B tables, and
1144 1.1 gwr * inserts them into the B manager pool for use by the system.
1145 1.1 gwr */
1146 1.1 gwr void
1147 1.1 gwr pmap_init_b_tables()
1148 1.1 gwr {
1149 1.1 gwr int i,j;
1150 1.1 gwr b_tmgr_t *b_tbl;
1151 1.1 gwr
1152 1.1 gwr for (i=0; i < NUM_B_TABLES; i++) {
1153 1.1 gwr /* Select the next available B manager from the pool */
1154 1.1 gwr b_tbl = &Btmgrbase[i];
1155 1.1 gwr
1156 1.1 gwr b_tbl->bt_parent = NULL; /* clear its parent, */
1157 1.1 gwr b_tbl->bt_pidx = 0; /* parent index, */
1158 1.1 gwr b_tbl->bt_wcnt = 0; /* wired entry count, */
1159 1.1 gwr b_tbl->bt_ecnt = 0; /* valid entry count. */
1160 1.1 gwr
1161 1.1 gwr /* Assign it the next available MMU B table from the pool */
1162 1.1 gwr b_tbl->bt_dtbl = &mmuBbase[i * MMU_B_TBL_SIZE];
1163 1.1 gwr
1164 1.1 gwr /* Invalidate every descriptor in the table */
1165 1.1 gwr for (j=0; j < MMU_B_TBL_SIZE; j++)
1166 1.1 gwr b_tbl->bt_dtbl[j].attr.raw = MMU_DT_INVALID;
1167 1.1 gwr
1168 1.1 gwr /* Insert the manager into the B pool */
1169 1.1 gwr TAILQ_INSERT_TAIL(&b_pool, b_tbl, bt_link);
1170 1.1 gwr }
1171 1.1 gwr }
1172 1.1 gwr
1173 1.1 gwr /* pmap_init_c_tables() INTERNAL
1174 1.1 gwr **
1175 1.1 gwr * Initializes all C table managers, their MMU C tables, and
1176 1.1 gwr * inserts them into the C manager pool for use by the system.
1177 1.1 gwr */
1178 1.1 gwr void
1179 1.1 gwr pmap_init_c_tables()
1180 1.1 gwr {
1181 1.1 gwr int i,j;
1182 1.1 gwr c_tmgr_t *c_tbl;
1183 1.1 gwr
1184 1.1 gwr for (i=0; i < NUM_C_TABLES; i++) {
1185 1.1 gwr /* Select the next available C manager from the pool */
1186 1.1 gwr c_tbl = &Ctmgrbase[i];
1187 1.1 gwr
1188 1.1 gwr c_tbl->ct_parent = NULL; /* clear its parent, */
1189 1.1 gwr c_tbl->ct_pidx = 0; /* parent index, */
1190 1.1 gwr c_tbl->ct_wcnt = 0; /* wired entry count, */
1191 1.1 gwr c_tbl->ct_ecnt = 0; /* valid entry count. */
1192 1.1 gwr
1193 1.1 gwr /* Assign it the next available MMU C table from the pool */
1194 1.1 gwr c_tbl->ct_dtbl = &mmuCbase[i * MMU_C_TBL_SIZE];
1195 1.1 gwr
1196 1.1 gwr for (j=0; j < MMU_C_TBL_SIZE; j++)
1197 1.1 gwr c_tbl->ct_dtbl[j].attr.raw = MMU_DT_INVALID;
1198 1.1 gwr
1199 1.1 gwr TAILQ_INSERT_TAIL(&c_pool, c_tbl, ct_link);
1200 1.1 gwr }
1201 1.1 gwr }
1202 1.1 gwr
1203 1.1 gwr /* pmap_init_pv() INTERNAL
1204 1.1 gwr **
1205 1.1 gwr * Initializes the Physical to Virtual mapping system.
1206 1.1 gwr */
1207 1.1 gwr void
1208 1.1 gwr pmap_init_pv()
1209 1.1 gwr {
1210 1.7 gwr int i;
1211 1.7 gwr
1212 1.7 gwr /* Initialize every PV head. */
1213 1.16 gwr for (i = 0; i < _btop(total_phys_mem); i++) {
1214 1.7 gwr pvbase[i].pv_idx = PVE_EOL; /* Indicate no mappings */
1215 1.7 gwr pvbase[i].pv_flags = 0; /* Zero out page flags */
1216 1.7 gwr }
1217 1.7 gwr
1218 1.1 gwr pv_initialized = TRUE;
1219 1.1 gwr }
1220 1.1 gwr
1221 1.1 gwr /* get_a_table INTERNAL
1222 1.1 gwr **
1223 1.1 gwr * Retrieve and return a level A table for use in a user map.
1224 1.1 gwr */
1225 1.1 gwr a_tmgr_t *
1226 1.1 gwr get_a_table()
1227 1.1 gwr {
1228 1.1 gwr a_tmgr_t *tbl;
1229 1.7 gwr pmap_t pmap;
1230 1.1 gwr
1231 1.1 gwr /* Get the top A table in the pool */
1232 1.1 gwr tbl = a_pool.tqh_first;
1233 1.7 gwr if (tbl == NULL) {
1234 1.7 gwr /*
1235 1.7 gwr * XXX - Instead of panicing here and in other get_x_table
1236 1.7 gwr * functions, we do have the option of sleeping on the head of
1237 1.7 gwr * the table pool. Any function which updates the table pool
1238 1.7 gwr * would then issue a wakeup() on the head, thus waking up any
1239 1.7 gwr * processes waiting for a table.
1240 1.7 gwr *
1241 1.7 gwr * Actually, the place to sleep would be when some process
1242 1.7 gwr * asks for a "wired" mapping that would run us short of
1243 1.7 gwr * mapping resources. This design DEPENDS on always having
1244 1.7 gwr * some mapping resources in the pool for stealing, so we
1245 1.7 gwr * must make sure we NEVER let the pool become empty. -gwr
1246 1.7 gwr */
1247 1.1 gwr panic("get_a_table: out of A tables.");
1248 1.7 gwr }
1249 1.7 gwr
1250 1.1 gwr TAILQ_REMOVE(&a_pool, tbl, at_link);
1251 1.7 gwr /*
1252 1.7 gwr * If the table has a non-null parent pointer then it is in use.
1253 1.1 gwr * Forcibly abduct it from its parent and clear its entries.
1254 1.1 gwr * No re-entrancy worries here. This table would not be in the
1255 1.1 gwr * table pool unless it was available for use.
1256 1.7 gwr *
1257 1.7 gwr * Note that the second argument to free_a_table() is FALSE. This
1258 1.7 gwr * indicates that the table should not be relinked into the A table
1259 1.7 gwr * pool. That is a job for the function that called us.
1260 1.1 gwr */
1261 1.1 gwr if (tbl->at_parent) {
1262 1.7 gwr pmap = tbl->at_parent;
1263 1.8 gwr free_a_table(tbl, FALSE);
1264 1.7 gwr pmap->pm_a_tmgr = NULL;
1265 1.7 gwr pmap->pm_a_phys = kernAphys;
1266 1.1 gwr }
1267 1.1 gwr #ifdef NON_REENTRANT
1268 1.7 gwr /*
1269 1.7 gwr * If the table isn't to be wired down, re-insert it at the
1270 1.1 gwr * end of the pool.
1271 1.1 gwr */
1272 1.1 gwr if (!wired)
1273 1.7 gwr /*
1274 1.7 gwr * Quandary - XXX
1275 1.1 gwr * Would it be better to let the calling function insert this
1276 1.1 gwr * table into the queue? By inserting it here, we are allowing
1277 1.1 gwr * it to be stolen immediately. The calling function is
1278 1.1 gwr * probably not expecting to use a table that it is not
1279 1.1 gwr * assured full control of.
1280 1.1 gwr * Answer - In the intrest of re-entrancy, it is best to let
1281 1.1 gwr * the calling function determine when a table is available
1282 1.1 gwr * for use. Therefore this code block is not used.
1283 1.1 gwr */
1284 1.1 gwr TAILQ_INSERT_TAIL(&a_pool, tbl, at_link);
1285 1.1 gwr #endif /* NON_REENTRANT */
1286 1.1 gwr return tbl;
1287 1.1 gwr }
1288 1.1 gwr
1289 1.1 gwr /* get_b_table INTERNAL
1290 1.1 gwr **
1291 1.1 gwr * Return a level B table for use.
1292 1.1 gwr */
1293 1.1 gwr b_tmgr_t *
1294 1.1 gwr get_b_table()
1295 1.1 gwr {
1296 1.1 gwr b_tmgr_t *tbl;
1297 1.1 gwr
1298 1.1 gwr /* See 'get_a_table' for comments. */
1299 1.1 gwr tbl = b_pool.tqh_first;
1300 1.1 gwr if (tbl == NULL)
1301 1.1 gwr panic("get_b_table: out of B tables.");
1302 1.1 gwr TAILQ_REMOVE(&b_pool, tbl, bt_link);
1303 1.1 gwr if (tbl->bt_parent) {
1304 1.1 gwr tbl->bt_parent->at_dtbl[tbl->bt_pidx].attr.raw = MMU_DT_INVALID;
1305 1.1 gwr tbl->bt_parent->at_ecnt--;
1306 1.8 gwr free_b_table(tbl, FALSE);
1307 1.1 gwr }
1308 1.1 gwr #ifdef NON_REENTRANT
1309 1.1 gwr if (!wired)
1310 1.1 gwr /* XXX see quandary in get_b_table */
1311 1.1 gwr /* XXX start lock */
1312 1.1 gwr TAILQ_INSERT_TAIL(&b_pool, tbl, bt_link);
1313 1.1 gwr /* XXX end lock */
1314 1.1 gwr #endif /* NON_REENTRANT */
1315 1.1 gwr return tbl;
1316 1.1 gwr }
1317 1.1 gwr
1318 1.1 gwr /* get_c_table INTERNAL
1319 1.1 gwr **
1320 1.1 gwr * Return a level C table for use.
1321 1.1 gwr */
1322 1.1 gwr c_tmgr_t *
1323 1.1 gwr get_c_table()
1324 1.1 gwr {
1325 1.1 gwr c_tmgr_t *tbl;
1326 1.1 gwr
1327 1.1 gwr /* See 'get_a_table' for comments */
1328 1.1 gwr tbl = c_pool.tqh_first;
1329 1.1 gwr if (tbl == NULL)
1330 1.1 gwr panic("get_c_table: out of C tables.");
1331 1.1 gwr TAILQ_REMOVE(&c_pool, tbl, ct_link);
1332 1.1 gwr if (tbl->ct_parent) {
1333 1.1 gwr tbl->ct_parent->bt_dtbl[tbl->ct_pidx].attr.raw = MMU_DT_INVALID;
1334 1.1 gwr tbl->ct_parent->bt_ecnt--;
1335 1.8 gwr free_c_table(tbl, FALSE);
1336 1.1 gwr }
1337 1.1 gwr #ifdef NON_REENTRANT
1338 1.1 gwr if (!wired)
1339 1.1 gwr /* XXX See quandary in get_a_table */
1340 1.1 gwr /* XXX start lock */
1341 1.1 gwr TAILQ_INSERT_TAIL(&c_pool, tbl, c_link);
1342 1.1 gwr /* XXX end lock */
1343 1.1 gwr #endif /* NON_REENTRANT */
1344 1.1 gwr
1345 1.1 gwr return tbl;
1346 1.1 gwr }
1347 1.1 gwr
1348 1.7 gwr /*
1349 1.7 gwr * The following 'free_table' and 'steal_table' functions are called to
1350 1.1 gwr * detach tables from their current obligations (parents and children) and
1351 1.1 gwr * prepare them for reuse in another mapping.
1352 1.1 gwr *
1353 1.1 gwr * Free_table is used when the calling function will handle the fate
1354 1.1 gwr * of the parent table, such as returning it to the free pool when it has
1355 1.1 gwr * no valid entries. Functions that do not want to handle this should
1356 1.1 gwr * call steal_table, in which the parent table's descriptors and entry
1357 1.1 gwr * count are automatically modified when this table is removed.
1358 1.1 gwr */
1359 1.1 gwr
1360 1.1 gwr /* free_a_table INTERNAL
1361 1.1 gwr **
1362 1.1 gwr * Unmaps the given A table and all child tables from their current
1363 1.1 gwr * mappings. Returns the number of pages that were invalidated.
1364 1.7 gwr * If 'relink' is true, the function will return the table to the head
1365 1.7 gwr * of the available table pool.
1366 1.1 gwr *
1367 1.1 gwr * Cache note: The MC68851 will automatically flush all
1368 1.1 gwr * descriptors derived from a given A table from its
1369 1.1 gwr * Automatic Translation Cache (ATC) if we issue a
1370 1.1 gwr * 'PFLUSHR' instruction with the base address of the
1371 1.1 gwr * table. This function should do, and does so.
1372 1.1 gwr * Note note: We are using an MC68030 - there is no
1373 1.1 gwr * PFLUSHR.
1374 1.1 gwr */
1375 1.1 gwr int
1376 1.7 gwr free_a_table(a_tbl, relink)
1377 1.1 gwr a_tmgr_t *a_tbl;
1378 1.7 gwr boolean_t relink;
1379 1.1 gwr {
1380 1.1 gwr int i, removed_cnt;
1381 1.1 gwr mmu_long_dte_t *dte;
1382 1.1 gwr mmu_short_dte_t *dtbl;
1383 1.1 gwr b_tmgr_t *tmgr;
1384 1.1 gwr
1385 1.7 gwr /*
1386 1.7 gwr * Flush the ATC cache of all cached descriptors derived
1387 1.1 gwr * from this table.
1388 1.22 jeremy * Sun3x does not use 68851's cached table feature
1389 1.1 gwr * flush_atc_crp(mmu_vtop(a_tbl->dte));
1390 1.1 gwr */
1391 1.1 gwr
1392 1.7 gwr /*
1393 1.7 gwr * Remove any pending cache flushes that were designated
1394 1.1 gwr * for the pmap this A table belongs to.
1395 1.1 gwr * a_tbl->parent->atc_flushq[0] = 0;
1396 1.22 jeremy * Not implemented in sun3x.
1397 1.1 gwr */
1398 1.1 gwr
1399 1.7 gwr /*
1400 1.7 gwr * All A tables in the system should retain a map for the
1401 1.1 gwr * kernel. If the table contains any valid descriptors
1402 1.1 gwr * (other than those for the kernel area), invalidate them all,
1403 1.1 gwr * stopping short of the kernel's entries.
1404 1.1 gwr */
1405 1.1 gwr removed_cnt = 0;
1406 1.1 gwr if (a_tbl->at_ecnt) {
1407 1.1 gwr dte = a_tbl->at_dtbl;
1408 1.8 gwr for (i=0; i < MMU_TIA(KERNBASE); i++) {
1409 1.7 gwr /*
1410 1.7 gwr * If a table entry points to a valid B table, free
1411 1.1 gwr * it and its children.
1412 1.1 gwr */
1413 1.1 gwr if (MMU_VALID_DT(dte[i])) {
1414 1.7 gwr /*
1415 1.7 gwr * The following block does several things,
1416 1.1 gwr * from innermost expression to the
1417 1.1 gwr * outermost:
1418 1.1 gwr * 1) It extracts the base (cc 1996)
1419 1.1 gwr * address of the B table pointed
1420 1.1 gwr * to in the A table entry dte[i].
1421 1.1 gwr * 2) It converts this base address into
1422 1.1 gwr * the virtual address it can be
1423 1.1 gwr * accessed with. (all MMU tables point
1424 1.1 gwr * to physical addresses.)
1425 1.1 gwr * 3) It finds the corresponding manager
1426 1.1 gwr * structure which manages this MMU table.
1427 1.1 gwr * 4) It frees the manager structure.
1428 1.1 gwr * (This frees the MMU table and all
1429 1.1 gwr * child tables. See 'free_b_table' for
1430 1.1 gwr * details.)
1431 1.1 gwr */
1432 1.7 gwr dtbl = mmu_ptov(dte[i].addr.raw);
1433 1.1 gwr tmgr = mmuB2tmgr(dtbl);
1434 1.7 gwr removed_cnt += free_b_table(tmgr, TRUE);
1435 1.8 gwr dte[i].attr.raw = MMU_DT_INVALID;
1436 1.1 gwr }
1437 1.8 gwr }
1438 1.8 gwr a_tbl->at_ecnt = 0;
1439 1.1 gwr }
1440 1.7 gwr if (relink) {
1441 1.7 gwr a_tbl->at_parent = NULL;
1442 1.7 gwr TAILQ_REMOVE(&a_pool, a_tbl, at_link);
1443 1.7 gwr TAILQ_INSERT_HEAD(&a_pool, a_tbl, at_link);
1444 1.7 gwr }
1445 1.1 gwr return removed_cnt;
1446 1.1 gwr }
1447 1.1 gwr
1448 1.1 gwr /* free_b_table INTERNAL
1449 1.1 gwr **
1450 1.1 gwr * Unmaps the given B table and all its children from their current
1451 1.1 gwr * mappings. Returns the number of pages that were invalidated.
1452 1.1 gwr * (For comments, see 'free_a_table()').
1453 1.1 gwr */
1454 1.1 gwr int
1455 1.7 gwr free_b_table(b_tbl, relink)
1456 1.1 gwr b_tmgr_t *b_tbl;
1457 1.7 gwr boolean_t relink;
1458 1.1 gwr {
1459 1.1 gwr int i, removed_cnt;
1460 1.1 gwr mmu_short_dte_t *dte;
1461 1.1 gwr mmu_short_pte_t *dtbl;
1462 1.1 gwr c_tmgr_t *tmgr;
1463 1.1 gwr
1464 1.1 gwr removed_cnt = 0;
1465 1.1 gwr if (b_tbl->bt_ecnt) {
1466 1.1 gwr dte = b_tbl->bt_dtbl;
1467 1.8 gwr for (i=0; i < MMU_B_TBL_SIZE; i++) {
1468 1.1 gwr if (MMU_VALID_DT(dte[i])) {
1469 1.7 gwr dtbl = mmu_ptov(MMU_DTE_PA(dte[i]));
1470 1.1 gwr tmgr = mmuC2tmgr(dtbl);
1471 1.7 gwr removed_cnt += free_c_table(tmgr, TRUE);
1472 1.8 gwr dte[i].attr.raw = MMU_DT_INVALID;
1473 1.1 gwr }
1474 1.8 gwr }
1475 1.8 gwr b_tbl->bt_ecnt = 0;
1476 1.1 gwr }
1477 1.1 gwr
1478 1.7 gwr if (relink) {
1479 1.7 gwr b_tbl->bt_parent = NULL;
1480 1.7 gwr TAILQ_REMOVE(&b_pool, b_tbl, bt_link);
1481 1.7 gwr TAILQ_INSERT_HEAD(&b_pool, b_tbl, bt_link);
1482 1.7 gwr }
1483 1.1 gwr return removed_cnt;
1484 1.1 gwr }
1485 1.1 gwr
1486 1.1 gwr /* free_c_table INTERNAL
1487 1.1 gwr **
1488 1.1 gwr * Unmaps the given C table from use and returns it to the pool for
1489 1.1 gwr * re-use. Returns the number of pages that were invalidated.
1490 1.1 gwr *
1491 1.1 gwr * This function preserves any physical page modification information
1492 1.1 gwr * contained in the page descriptors within the C table by calling
1493 1.1 gwr * 'pmap_remove_pte().'
1494 1.1 gwr */
1495 1.1 gwr int
1496 1.7 gwr free_c_table(c_tbl, relink)
1497 1.1 gwr c_tmgr_t *c_tbl;
1498 1.7 gwr boolean_t relink;
1499 1.1 gwr {
1500 1.1 gwr int i, removed_cnt;
1501 1.1 gwr
1502 1.1 gwr removed_cnt = 0;
1503 1.8 gwr if (c_tbl->ct_ecnt) {
1504 1.8 gwr for (i=0; i < MMU_C_TBL_SIZE; i++) {
1505 1.1 gwr if (MMU_VALID_DT(c_tbl->ct_dtbl[i])) {
1506 1.1 gwr pmap_remove_pte(&c_tbl->ct_dtbl[i]);
1507 1.1 gwr removed_cnt++;
1508 1.1 gwr }
1509 1.8 gwr }
1510 1.8 gwr c_tbl->ct_ecnt = 0;
1511 1.8 gwr }
1512 1.8 gwr
1513 1.7 gwr if (relink) {
1514 1.7 gwr c_tbl->ct_parent = NULL;
1515 1.7 gwr TAILQ_REMOVE(&c_pool, c_tbl, ct_link);
1516 1.7 gwr TAILQ_INSERT_HEAD(&c_pool, c_tbl, ct_link);
1517 1.7 gwr }
1518 1.1 gwr return removed_cnt;
1519 1.1 gwr }
1520 1.1 gwr
1521 1.8 gwr #if 0
1522 1.1 gwr /* free_c_table_novalid INTERNAL
1523 1.1 gwr **
1524 1.1 gwr * Frees the given C table manager without checking to see whether
1525 1.1 gwr * or not it contains any valid page descriptors as it is assumed
1526 1.1 gwr * that it does not.
1527 1.1 gwr */
1528 1.1 gwr void
1529 1.1 gwr free_c_table_novalid(c_tbl)
1530 1.1 gwr c_tmgr_t *c_tbl;
1531 1.1 gwr {
1532 1.1 gwr TAILQ_REMOVE(&c_pool, c_tbl, ct_link);
1533 1.1 gwr TAILQ_INSERT_HEAD(&c_pool, c_tbl, ct_link);
1534 1.1 gwr c_tbl->ct_parent->bt_dtbl[c_tbl->ct_pidx].attr.raw = MMU_DT_INVALID;
1535 1.7 gwr c_tbl->ct_parent->bt_ecnt--;
1536 1.7 gwr /*
1537 1.7 gwr * XXX - Should call equiv. of 'free_b_table_novalid' here if
1538 1.7 gwr * we just removed the last entry of the parent B table.
1539 1.7 gwr * But I want to insure that this will not endanger pmap_enter()
1540 1.7 gwr * with sudden removal of tables it is working with.
1541 1.7 gwr *
1542 1.7 gwr * We should probably add another field to each table, indicating
1543 1.7 gwr * whether or not it is 'locked', ie. in the process of being
1544 1.7 gwr * modified.
1545 1.7 gwr */
1546 1.7 gwr c_tbl->ct_parent = NULL;
1547 1.1 gwr }
1548 1.8 gwr #endif
1549 1.1 gwr
1550 1.1 gwr /* pmap_remove_pte INTERNAL
1551 1.1 gwr **
1552 1.1 gwr * Unmap the given pte and preserve any page modification
1553 1.1 gwr * information by transfering it to the pv head of the
1554 1.1 gwr * physical page it maps to. This function does not update
1555 1.1 gwr * any reference counts because it is assumed that the calling
1556 1.8 gwr * function will do so.
1557 1.1 gwr */
1558 1.1 gwr void
1559 1.1 gwr pmap_remove_pte(pte)
1560 1.1 gwr mmu_short_pte_t *pte;
1561 1.1 gwr {
1562 1.7 gwr u_short pv_idx, targ_idx;
1563 1.7 gwr int s;
1564 1.1 gwr vm_offset_t pa;
1565 1.1 gwr pv_t *pv;
1566 1.1 gwr
1567 1.1 gwr pa = MMU_PTE_PA(*pte);
1568 1.1 gwr if (is_managed(pa)) {
1569 1.1 gwr pv = pa2pv(pa);
1570 1.7 gwr targ_idx = pteidx(pte); /* Index of PTE being removed */
1571 1.7 gwr
1572 1.7 gwr /*
1573 1.7 gwr * If the PTE being removed is the first (or only) PTE in
1574 1.7 gwr * the list of PTEs currently mapped to this page, remove the
1575 1.7 gwr * PTE by changing the index found on the PV head. Otherwise
1576 1.7 gwr * a linear search through the list will have to be executed
1577 1.7 gwr * in order to find the PVE which points to the PTE being
1578 1.7 gwr * removed, so that it may be modified to point to its new
1579 1.7 gwr * neighbor.
1580 1.7 gwr */
1581 1.7 gwr s = splimp();
1582 1.7 gwr pv_idx = pv->pv_idx; /* Index of first PTE in PV list */
1583 1.7 gwr if (pv_idx == targ_idx) {
1584 1.7 gwr pv->pv_idx = pvebase[targ_idx].pve_next;
1585 1.7 gwr } else {
1586 1.7 gwr /*
1587 1.7 gwr * Find the PV element which points to the target
1588 1.7 gwr * element.
1589 1.7 gwr */
1590 1.7 gwr while (pvebase[pv_idx].pve_next != targ_idx) {
1591 1.7 gwr pv_idx = pvebase[pv_idx].pve_next;
1592 1.7 gwr #ifdef DIAGNOSTIC
1593 1.7 gwr if (pv_idx == PVE_EOL)
1594 1.7 gwr panic("pmap_remove_pte: pv list end!");
1595 1.7 gwr #endif
1596 1.7 gwr }
1597 1.7 gwr
1598 1.7 gwr /*
1599 1.7 gwr * At this point, pv_idx is the index of the PV
1600 1.7 gwr * element just before the target element in the list.
1601 1.7 gwr * Unlink the target.
1602 1.7 gwr */
1603 1.7 gwr pvebase[pv_idx].pve_next = pvebase[targ_idx].pve_next;
1604 1.7 gwr }
1605 1.7 gwr /*
1606 1.7 gwr * Save the mod/ref bits of the pte by simply
1607 1.1 gwr * ORing the entire pte onto the pv_flags member
1608 1.1 gwr * of the pv structure.
1609 1.1 gwr * There is no need to use a separate bit pattern
1610 1.1 gwr * for usage information on the pv head than that
1611 1.1 gwr * which is used on the MMU ptes.
1612 1.1 gwr */
1613 1.7 gwr pv->pv_flags |= (u_short) pte->attr.raw;
1614 1.7 gwr splx(s);
1615 1.1 gwr }
1616 1.1 gwr
1617 1.1 gwr pte->attr.raw = MMU_DT_INVALID;
1618 1.1 gwr }
1619 1.1 gwr
1620 1.1 gwr /* pmap_stroll INTERNAL
1621 1.1 gwr **
1622 1.1 gwr * Retrieve the addresses of all table managers involved in the mapping of
1623 1.1 gwr * the given virtual address. If the table walk completed sucessfully,
1624 1.7 gwr * return TRUE. If it was only partially sucessful, return FALSE.
1625 1.1 gwr * The table walk performed by this function is important to many other
1626 1.1 gwr * functions in this module.
1627 1.7 gwr *
1628 1.7 gwr * Note: This function ought to be easier to read.
1629 1.1 gwr */
1630 1.1 gwr boolean_t
1631 1.1 gwr pmap_stroll(pmap, va, a_tbl, b_tbl, c_tbl, pte, a_idx, b_idx, pte_idx)
1632 1.1 gwr pmap_t pmap;
1633 1.1 gwr vm_offset_t va;
1634 1.1 gwr a_tmgr_t **a_tbl;
1635 1.1 gwr b_tmgr_t **b_tbl;
1636 1.1 gwr c_tmgr_t **c_tbl;
1637 1.1 gwr mmu_short_pte_t **pte;
1638 1.1 gwr int *a_idx, *b_idx, *pte_idx;
1639 1.1 gwr {
1640 1.1 gwr mmu_long_dte_t *a_dte; /* A: long descriptor table */
1641 1.1 gwr mmu_short_dte_t *b_dte; /* B: short descriptor table */
1642 1.1 gwr
1643 1.1 gwr if (pmap == pmap_kernel())
1644 1.1 gwr return FALSE;
1645 1.1 gwr
1646 1.7 gwr /* Does the given pmap have its own A table? */
1647 1.7 gwr *a_tbl = pmap->pm_a_tmgr;
1648 1.1 gwr if (*a_tbl == NULL)
1649 1.1 gwr return FALSE; /* No. Return unknown. */
1650 1.1 gwr /* Does the A table have a valid B table
1651 1.1 gwr * under the corresponding table entry?
1652 1.1 gwr */
1653 1.1 gwr *a_idx = MMU_TIA(va);
1654 1.1 gwr a_dte = &((*a_tbl)->at_dtbl[*a_idx]);
1655 1.1 gwr if (!MMU_VALID_DT(*a_dte))
1656 1.1 gwr return FALSE; /* No. Return unknown. */
1657 1.1 gwr /* Yes. Extract B table from the A table. */
1658 1.7 gwr *b_tbl = mmuB2tmgr(mmu_ptov(a_dte->addr.raw));
1659 1.1 gwr /* Does the B table have a valid C table
1660 1.1 gwr * under the corresponding table entry?
1661 1.1 gwr */
1662 1.1 gwr *b_idx = MMU_TIB(va);
1663 1.1 gwr b_dte = &((*b_tbl)->bt_dtbl[*b_idx]);
1664 1.1 gwr if (!MMU_VALID_DT(*b_dte))
1665 1.1 gwr return FALSE; /* No. Return unknown. */
1666 1.1 gwr /* Yes. Extract C table from the B table. */
1667 1.7 gwr *c_tbl = mmuC2tmgr(mmu_ptov(MMU_DTE_PA(*b_dte)));
1668 1.1 gwr *pte_idx = MMU_TIC(va);
1669 1.1 gwr *pte = &((*c_tbl)->ct_dtbl[*pte_idx]);
1670 1.1 gwr
1671 1.1 gwr return TRUE;
1672 1.1 gwr }
1673 1.1 gwr
1674 1.1 gwr /* pmap_enter INTERFACE
1675 1.1 gwr **
1676 1.1 gwr * Called by the kernel to map a virtual address
1677 1.1 gwr * to a physical address in the given process map.
1678 1.1 gwr *
1679 1.1 gwr * Note: this function should apply an exclusive lock
1680 1.1 gwr * on the pmap system for its duration. (it certainly
1681 1.1 gwr * would save my hair!!)
1682 1.7 gwr * This function ought to be easier to read.
1683 1.1 gwr */
1684 1.1 gwr void
1685 1.1 gwr pmap_enter(pmap, va, pa, prot, wired)
1686 1.1 gwr pmap_t pmap;
1687 1.1 gwr vm_offset_t va;
1688 1.1 gwr vm_offset_t pa;
1689 1.1 gwr vm_prot_t prot;
1690 1.1 gwr boolean_t wired;
1691 1.1 gwr {
1692 1.7 gwr boolean_t insert, managed; /* Marks the need for PV insertion.*/
1693 1.7 gwr u_short nidx; /* PV list index */
1694 1.7 gwr int s; /* Used for splimp()/splx() */
1695 1.7 gwr int flags; /* Mapping flags. eg. Cache inhibit */
1696 1.8 gwr u_int a_idx, b_idx, pte_idx; /* table indices */
1697 1.1 gwr a_tmgr_t *a_tbl; /* A: long descriptor table manager */
1698 1.1 gwr b_tmgr_t *b_tbl; /* B: short descriptor table manager */
1699 1.1 gwr c_tmgr_t *c_tbl; /* C: short page table manager */
1700 1.1 gwr mmu_long_dte_t *a_dte; /* A: long descriptor table */
1701 1.1 gwr mmu_short_dte_t *b_dte; /* B: short descriptor table */
1702 1.1 gwr mmu_short_pte_t *c_pte; /* C: short page descriptor table */
1703 1.1 gwr pv_t *pv; /* pv list head */
1704 1.1 gwr enum {NONE, NEWA, NEWB, NEWC} llevel; /* used at end */
1705 1.1 gwr
1706 1.1 gwr if (pmap == NULL)
1707 1.1 gwr return;
1708 1.1 gwr if (pmap == pmap_kernel()) {
1709 1.1 gwr pmap_enter_kernel(va, pa, prot);
1710 1.1 gwr return;
1711 1.1 gwr }
1712 1.7 gwr
1713 1.7 gwr flags = (pa & ~MMU_PAGE_MASK);
1714 1.7 gwr pa &= MMU_PAGE_MASK;
1715 1.7 gwr
1716 1.7 gwr /*
1717 1.22 jeremy * Determine if the physical address being mapped is on-board RAM.
1718 1.22 jeremy * Any other area of the address space is likely to belong to a
1719 1.22 jeremy * device and hence it would be disasterous to cache its contents.
1720 1.7 gwr */
1721 1.7 gwr if ((managed = is_managed(pa)) == FALSE)
1722 1.7 gwr flags |= PMAP_NC;
1723 1.7 gwr
1724 1.7 gwr /*
1725 1.7 gwr * For user mappings we walk along the MMU tables of the given
1726 1.1 gwr * pmap, reaching a PTE which describes the virtual page being
1727 1.1 gwr * mapped or changed. If any level of the walk ends in an invalid
1728 1.1 gwr * entry, a table must be allocated and the entry must be updated
1729 1.1 gwr * to point to it.
1730 1.1 gwr * There is a bit of confusion as to whether this code must be
1731 1.1 gwr * re-entrant. For now we will assume it is. To support
1732 1.1 gwr * re-entrancy we must unlink tables from the table pool before
1733 1.1 gwr * we assume we may use them. Tables are re-linked into the pool
1734 1.1 gwr * when we are finished with them at the end of the function.
1735 1.1 gwr * But I don't feel like doing that until we have proof that this
1736 1.1 gwr * needs to be re-entrant.
1737 1.1 gwr * 'llevel' records which tables need to be relinked.
1738 1.1 gwr */
1739 1.1 gwr llevel = NONE;
1740 1.1 gwr
1741 1.7 gwr /*
1742 1.7 gwr * Step 1 - Retrieve the A table from the pmap. If it has no
1743 1.7 gwr * A table, allocate a new one from the available pool.
1744 1.1 gwr */
1745 1.1 gwr
1746 1.7 gwr a_tbl = pmap->pm_a_tmgr;
1747 1.7 gwr if (a_tbl == NULL) {
1748 1.7 gwr /*
1749 1.7 gwr * This pmap does not currently have an A table. Allocate
1750 1.7 gwr * a new one.
1751 1.7 gwr */
1752 1.7 gwr a_tbl = get_a_table();
1753 1.7 gwr a_tbl->at_parent = pmap;
1754 1.7 gwr
1755 1.7 gwr /*
1756 1.7 gwr * Assign this new A table to the pmap, and calculate its
1757 1.7 gwr * physical address so that loadcrp() can be used to make
1758 1.7 gwr * the table active.
1759 1.7 gwr */
1760 1.7 gwr pmap->pm_a_tmgr = a_tbl;
1761 1.7 gwr pmap->pm_a_phys = mmu_vtop(a_tbl->at_dtbl);
1762 1.7 gwr
1763 1.7 gwr /*
1764 1.7 gwr * If the process receiving a new A table is the current
1765 1.7 gwr * process, we are responsible for setting the MMU so that
1766 1.9 gwr * it becomes the current address space. This only adds
1767 1.9 gwr * new mappings, so no need to flush anything.
1768 1.7 gwr */
1769 1.9 gwr if (pmap == current_pmap()) {
1770 1.9 gwr kernel_crp.rp_addr = pmap->pm_a_phys;
1771 1.9 gwr loadcrp(&kernel_crp);
1772 1.9 gwr }
1773 1.7 gwr
1774 1.1 gwr if (!wired)
1775 1.1 gwr llevel = NEWA;
1776 1.1 gwr } else {
1777 1.7 gwr /*
1778 1.7 gwr * Use the A table already allocated for this pmap.
1779 1.1 gwr * Unlink it from the A table pool if necessary.
1780 1.1 gwr */
1781 1.1 gwr if (wired && !a_tbl->at_wcnt)
1782 1.1 gwr TAILQ_REMOVE(&a_pool, a_tbl, at_link);
1783 1.1 gwr }
1784 1.1 gwr
1785 1.7 gwr /*
1786 1.7 gwr * Step 2 - Walk into the B table. If there is no valid B table,
1787 1.1 gwr * allocate one.
1788 1.1 gwr */
1789 1.1 gwr
1790 1.1 gwr a_idx = MMU_TIA(va); /* Calculate the TIA of the VA. */
1791 1.1 gwr a_dte = &a_tbl->at_dtbl[a_idx]; /* Retrieve descriptor from table */
1792 1.1 gwr if (MMU_VALID_DT(*a_dte)) { /* Is the descriptor valid? */
1793 1.7 gwr /* The descriptor is valid. Use the B table it points to. */
1794 1.1 gwr /*************************************
1795 1.1 gwr * a_idx *
1796 1.1 gwr * v *
1797 1.1 gwr * a_tbl -> +-+-+-+-+-+-+-+-+-+-+-+- *
1798 1.1 gwr * | | | | | | | | | | | | *
1799 1.1 gwr * +-+-+-+-+-+-+-+-+-+-+-+- *
1800 1.1 gwr * | *
1801 1.1 gwr * \- b_tbl -> +-+- *
1802 1.1 gwr * | | *
1803 1.1 gwr * +-+- *
1804 1.1 gwr *************************************/
1805 1.7 gwr b_dte = mmu_ptov(a_dte->addr.raw);
1806 1.1 gwr b_tbl = mmuB2tmgr(b_dte);
1807 1.7 gwr
1808 1.7 gwr /*
1809 1.7 gwr * If the requested mapping must be wired, but this table
1810 1.7 gwr * being used to map it is not, the table must be removed
1811 1.7 gwr * from the available pool and its wired entry count
1812 1.7 gwr * incremented.
1813 1.7 gwr */
1814 1.1 gwr if (wired && !b_tbl->bt_wcnt) {
1815 1.1 gwr TAILQ_REMOVE(&b_pool, b_tbl, bt_link);
1816 1.7 gwr a_tbl->at_wcnt++;
1817 1.1 gwr }
1818 1.1 gwr } else {
1819 1.7 gwr /* The descriptor is invalid. Allocate a new B table. */
1820 1.7 gwr b_tbl = get_b_table();
1821 1.7 gwr
1822 1.1 gwr /* Point the parent A table descriptor to this new B table. */
1823 1.7 gwr a_dte->addr.raw = mmu_vtop(b_tbl->bt_dtbl);
1824 1.7 gwr a_dte->attr.raw = MMU_LONG_DTE_LU | MMU_DT_SHORT;
1825 1.7 gwr a_tbl->at_ecnt++; /* Update parent's valid entry count */
1826 1.7 gwr
1827 1.1 gwr /* Create the necessary back references to the parent table */
1828 1.1 gwr b_tbl->bt_parent = a_tbl;
1829 1.1 gwr b_tbl->bt_pidx = a_idx;
1830 1.7 gwr
1831 1.7 gwr /*
1832 1.7 gwr * If this table is to be wired, make sure the parent A table
1833 1.1 gwr * wired count is updated to reflect that it has another wired
1834 1.1 gwr * entry.
1835 1.1 gwr */
1836 1.1 gwr if (wired)
1837 1.1 gwr a_tbl->at_wcnt++;
1838 1.1 gwr else if (llevel == NONE)
1839 1.1 gwr llevel = NEWB;
1840 1.1 gwr }
1841 1.1 gwr
1842 1.7 gwr /*
1843 1.7 gwr * Step 3 - Walk into the C table, if there is no valid C table,
1844 1.1 gwr * allocate one.
1845 1.1 gwr */
1846 1.1 gwr
1847 1.1 gwr b_idx = MMU_TIB(va); /* Calculate the TIB of the VA */
1848 1.1 gwr b_dte = &b_tbl->bt_dtbl[b_idx]; /* Retrieve descriptor from table */
1849 1.1 gwr if (MMU_VALID_DT(*b_dte)) { /* Is the descriptor valid? */
1850 1.7 gwr /* The descriptor is valid. Use the C table it points to. */
1851 1.1 gwr /**************************************
1852 1.1 gwr * c_idx *
1853 1.1 gwr * | v *
1854 1.1 gwr * \- b_tbl -> +-+-+-+-+-+-+-+-+-+-+- *
1855 1.1 gwr * | | | | | | | | | | | *
1856 1.1 gwr * +-+-+-+-+-+-+-+-+-+-+- *
1857 1.1 gwr * | *
1858 1.1 gwr * \- c_tbl -> +-+-- *
1859 1.1 gwr * | | | *
1860 1.1 gwr * +-+-- *
1861 1.1 gwr **************************************/
1862 1.7 gwr c_pte = mmu_ptov(MMU_PTE_PA(*b_dte));
1863 1.1 gwr c_tbl = mmuC2tmgr(c_pte);
1864 1.7 gwr
1865 1.7 gwr /* If mapping is wired and table is not */
1866 1.1 gwr if (wired && !c_tbl->ct_wcnt) {
1867 1.1 gwr TAILQ_REMOVE(&c_pool, c_tbl, ct_link);
1868 1.1 gwr b_tbl->bt_wcnt++;
1869 1.1 gwr }
1870 1.1 gwr } else {
1871 1.7 gwr /* The descriptor is invalid. Allocate a new C table. */
1872 1.7 gwr c_tbl = get_c_table();
1873 1.7 gwr
1874 1.1 gwr /* Point the parent B table descriptor to this new C table. */
1875 1.7 gwr b_dte->attr.raw = mmu_vtop(c_tbl->ct_dtbl);
1876 1.7 gwr b_dte->attr.raw |= MMU_DT_SHORT;
1877 1.7 gwr b_tbl->bt_ecnt++; /* Update parent's valid entry count */
1878 1.7 gwr
1879 1.1 gwr /* Create the necessary back references to the parent table */
1880 1.1 gwr c_tbl->ct_parent = b_tbl;
1881 1.1 gwr c_tbl->ct_pidx = b_idx;
1882 1.7 gwr
1883 1.7 gwr /*
1884 1.7 gwr * If this table is to be wired, make sure the parent B table
1885 1.1 gwr * wired count is updated to reflect that it has another wired
1886 1.1 gwr * entry.
1887 1.1 gwr */
1888 1.1 gwr if (wired)
1889 1.1 gwr b_tbl->bt_wcnt++;
1890 1.1 gwr else if (llevel == NONE)
1891 1.1 gwr llevel = NEWC;
1892 1.1 gwr }
1893 1.1 gwr
1894 1.7 gwr /*
1895 1.7 gwr * Step 4 - Deposit a page descriptor (PTE) into the appropriate
1896 1.1 gwr * slot of the C table, describing the PA to which the VA is mapped.
1897 1.1 gwr */
1898 1.1 gwr
1899 1.1 gwr pte_idx = MMU_TIC(va);
1900 1.1 gwr c_pte = &c_tbl->ct_dtbl[pte_idx];
1901 1.1 gwr if (MMU_VALID_DT(*c_pte)) { /* Is the entry currently valid? */
1902 1.7 gwr /*
1903 1.7 gwr * The PTE is currently valid. This particular call
1904 1.1 gwr * is just a synonym for one (or more) of the following
1905 1.1 gwr * operations:
1906 1.7 gwr * change protection of a page
1907 1.1 gwr * change wiring status of a page
1908 1.1 gwr * remove the mapping of a page
1909 1.7 gwr *
1910 1.7 gwr * XXX - Semi critical: This code should unwire the PTE
1911 1.7 gwr * and, possibly, associated parent tables if this is a
1912 1.7 gwr * change wiring operation. Currently it does not.
1913 1.7 gwr *
1914 1.7 gwr * This may be ok if pmap_change_wiring() is the only
1915 1.7 gwr * interface used to UNWIRE a page.
1916 1.1 gwr */
1917 1.7 gwr
1918 1.7 gwr /* First check if this is a wiring operation. */
1919 1.7 gwr if (wired && (c_pte->attr.raw & MMU_SHORT_PTE_WIRED)) {
1920 1.7 gwr /*
1921 1.7 gwr * The PTE is already wired. To prevent it from being
1922 1.7 gwr * counted as a new wiring operation, reset the 'wired'
1923 1.7 gwr * variable.
1924 1.7 gwr */
1925 1.7 gwr wired = FALSE;
1926 1.7 gwr }
1927 1.7 gwr
1928 1.1 gwr /* Is the new address the same as the old? */
1929 1.1 gwr if (MMU_PTE_PA(*c_pte) == pa) {
1930 1.7 gwr /*
1931 1.7 gwr * Yes, mark that it does not need to be reinserted
1932 1.7 gwr * into the PV list.
1933 1.7 gwr */
1934 1.7 gwr insert = FALSE;
1935 1.7 gwr
1936 1.7 gwr /*
1937 1.7 gwr * Clear all but the modified, referenced and wired
1938 1.7 gwr * bits on the PTE.
1939 1.7 gwr */
1940 1.7 gwr c_pte->attr.raw &= (MMU_SHORT_PTE_M
1941 1.7 gwr | MMU_SHORT_PTE_USED | MMU_SHORT_PTE_WIRED);
1942 1.1 gwr } else {
1943 1.1 gwr /* No, remove the old entry */
1944 1.1 gwr pmap_remove_pte(c_pte);
1945 1.7 gwr insert = TRUE;
1946 1.1 gwr }
1947 1.8 gwr
1948 1.8 gwr /*
1949 1.8 gwr * TLB flush is only necessary if modifying current map.
1950 1.8 gwr * However, in pmap_enter(), the pmap almost always IS
1951 1.8 gwr * the current pmap, so don't even bother to check.
1952 1.8 gwr */
1953 1.8 gwr TBIS(va);
1954 1.1 gwr } else {
1955 1.7 gwr /*
1956 1.7 gwr * The PTE is invalid. Increment the valid entry count in
1957 1.8 gwr * the C table manager to reflect the addition of a new entry.
1958 1.7 gwr */
1959 1.1 gwr c_tbl->ct_ecnt++;
1960 1.8 gwr
1961 1.8 gwr /* XXX - temporarily make sure the PTE is cleared. */
1962 1.8 gwr c_pte->attr.raw = 0;
1963 1.1 gwr
1964 1.7 gwr /* It will also need to be inserted into the PV list. */
1965 1.7 gwr insert = TRUE;
1966 1.7 gwr }
1967 1.7 gwr
1968 1.7 gwr /*
1969 1.7 gwr * If page is changing from unwired to wired status, set an unused bit
1970 1.7 gwr * within the PTE to indicate that it is wired. Also increment the
1971 1.7 gwr * wired entry count in the C table manager.
1972 1.7 gwr */
1973 1.7 gwr if (wired) {
1974 1.1 gwr c_pte->attr.raw |= MMU_SHORT_PTE_WIRED;
1975 1.7 gwr c_tbl->ct_wcnt++;
1976 1.1 gwr }
1977 1.1 gwr
1978 1.7 gwr /*
1979 1.7 gwr * Map the page, being careful to preserve modify/reference/wired
1980 1.7 gwr * bits. At this point it is assumed that the PTE either has no bits
1981 1.7 gwr * set, or if there are set bits, they are only modified, reference or
1982 1.7 gwr * wired bits. If not, the following statement will cause erratic
1983 1.7 gwr * behavior.
1984 1.7 gwr */
1985 1.8 gwr #ifdef PMAP_DEBUG
1986 1.7 gwr if (c_pte->attr.raw & ~(MMU_SHORT_PTE_M |
1987 1.7 gwr MMU_SHORT_PTE_USED | MMU_SHORT_PTE_WIRED)) {
1988 1.7 gwr printf("pmap_enter: junk left in PTE at %p\n", c_pte);
1989 1.7 gwr Debugger();
1990 1.7 gwr }
1991 1.7 gwr #endif
1992 1.7 gwr c_pte->attr.raw |= ((u_long) pa | MMU_DT_PAGE);
1993 1.7 gwr
1994 1.7 gwr /*
1995 1.7 gwr * If the mapping should be read-only, set the write protect
1996 1.7 gwr * bit in the PTE.
1997 1.7 gwr */
1998 1.7 gwr if (!(prot & VM_PROT_WRITE))
1999 1.7 gwr c_pte->attr.raw |= MMU_SHORT_PTE_WP;
2000 1.7 gwr
2001 1.7 gwr /*
2002 1.7 gwr * If the mapping should be cache inhibited (indicated by the flag
2003 1.7 gwr * bits found on the lower order of the physical address.)
2004 1.7 gwr * mark the PTE as a cache inhibited page.
2005 1.7 gwr */
2006 1.7 gwr if (flags & PMAP_NC)
2007 1.7 gwr c_pte->attr.raw |= MMU_SHORT_PTE_CI;
2008 1.7 gwr
2009 1.7 gwr /*
2010 1.7 gwr * If the physical address being mapped is managed by the PV
2011 1.7 gwr * system then link the pte into the list of pages mapped to that
2012 1.7 gwr * address.
2013 1.7 gwr */
2014 1.7 gwr if (insert && managed) {
2015 1.7 gwr pv = pa2pv(pa);
2016 1.7 gwr nidx = pteidx(c_pte);
2017 1.7 gwr
2018 1.7 gwr s = splimp();
2019 1.7 gwr pvebase[nidx].pve_next = pv->pv_idx;
2020 1.7 gwr pv->pv_idx = nidx;
2021 1.7 gwr splx(s);
2022 1.7 gwr }
2023 1.1 gwr
2024 1.1 gwr /* Move any allocated tables back into the active pool. */
2025 1.1 gwr
2026 1.1 gwr switch (llevel) {
2027 1.1 gwr case NEWA:
2028 1.1 gwr TAILQ_INSERT_TAIL(&a_pool, a_tbl, at_link);
2029 1.1 gwr /* FALLTHROUGH */
2030 1.1 gwr case NEWB:
2031 1.1 gwr TAILQ_INSERT_TAIL(&b_pool, b_tbl, bt_link);
2032 1.1 gwr /* FALLTHROUGH */
2033 1.1 gwr case NEWC:
2034 1.1 gwr TAILQ_INSERT_TAIL(&c_pool, c_tbl, ct_link);
2035 1.1 gwr /* FALLTHROUGH */
2036 1.1 gwr default:
2037 1.1 gwr break;
2038 1.1 gwr }
2039 1.1 gwr }
2040 1.1 gwr
2041 1.1 gwr /* pmap_enter_kernel INTERNAL
2042 1.1 gwr **
2043 1.1 gwr * Map the given virtual address to the given physical address within the
2044 1.1 gwr * kernel address space. This function exists because the kernel map does
2045 1.1 gwr * not do dynamic table allocation. It consists of a contiguous array of ptes
2046 1.1 gwr * and can be edited directly without the need to walk through any tables.
2047 1.1 gwr *
2048 1.1 gwr * XXX: "Danger, Will Robinson!"
2049 1.1 gwr * Note that the kernel should never take a fault on any page
2050 1.1 gwr * between [ KERNBASE .. virtual_avail ] and this is checked in
2051 1.1 gwr * trap.c for kernel-mode MMU faults. This means that mappings
2052 1.1 gwr * created in that range must be implicily wired. -gwr
2053 1.1 gwr */
2054 1.1 gwr void
2055 1.1 gwr pmap_enter_kernel(va, pa, prot)
2056 1.1 gwr vm_offset_t va;
2057 1.1 gwr vm_offset_t pa;
2058 1.1 gwr vm_prot_t prot;
2059 1.1 gwr {
2060 1.7 gwr boolean_t was_valid, insert;
2061 1.7 gwr u_short pte_idx, pv_idx;
2062 1.7 gwr int s, flags;
2063 1.1 gwr mmu_short_pte_t *pte;
2064 1.7 gwr pv_t *pv;
2065 1.7 gwr vm_offset_t old_pa;
2066 1.7 gwr
2067 1.7 gwr flags = (pa & ~MMU_PAGE_MASK);
2068 1.7 gwr pa &= MMU_PAGE_MASK;
2069 1.7 gwr
2070 1.7 gwr /*
2071 1.7 gwr * Calculate the index of the PTE being modified.
2072 1.7 gwr */
2073 1.16 gwr pte_idx = (u_long) _btop(va - KERNBASE);
2074 1.1 gwr
2075 1.22 jeremy /* This array is traditionally named "Sysmap" */
2076 1.7 gwr pte = &kernCbase[pte_idx];
2077 1.7 gwr
2078 1.7 gwr s = splimp();
2079 1.7 gwr if (MMU_VALID_DT(*pte)) {
2080 1.1 gwr was_valid = TRUE;
2081 1.7 gwr /*
2082 1.7 gwr * If the PTE is already mapped to an address and it differs
2083 1.7 gwr * from the address requested, unlink it from the PV list.
2084 1.7 gwr *
2085 1.8 gwr * This only applies to mappings within virtual_avail
2086 1.7 gwr * and VM_MAX_KERNEL_ADDRESS. All others are not requests
2087 1.7 gwr * from the VM system and should not be part of the PV system.
2088 1.7 gwr */
2089 1.8 gwr if ((va >= virtual_avail) && (va < VM_MAX_KERNEL_ADDRESS)) {
2090 1.7 gwr old_pa = MMU_PTE_PA(*pte);
2091 1.7 gwr if (pa != old_pa) {
2092 1.7 gwr if (is_managed(old_pa)) {
2093 1.7 gwr /* XXX - Make this into a function call? */
2094 1.7 gwr pv = pa2pv(old_pa);
2095 1.7 gwr pv_idx = pv->pv_idx;
2096 1.7 gwr if (pv_idx == pte_idx) {
2097 1.7 gwr pv->pv_idx = pvebase[pte_idx].pve_next;
2098 1.7 gwr } else {
2099 1.7 gwr while (pvebase[pv_idx].pve_next != pte_idx)
2100 1.7 gwr pv_idx = pvebase[pv_idx].pve_next;
2101 1.7 gwr pvebase[pv_idx].pve_next =
2102 1.7 gwr pvebase[pte_idx].pve_next;
2103 1.7 gwr }
2104 1.7 gwr /* Save modified/reference bits */
2105 1.7 gwr pv->pv_flags |= (u_short) pte->attr.raw;
2106 1.7 gwr }
2107 1.7 gwr if (is_managed(pa))
2108 1.7 gwr insert = TRUE;
2109 1.7 gwr else
2110 1.7 gwr insert = FALSE;
2111 1.8 gwr /*
2112 1.8 gwr * Clear out any old bits in the PTE.
2113 1.8 gwr */
2114 1.8 gwr pte->attr.raw = MMU_DT_INVALID;
2115 1.7 gwr } else {
2116 1.7 gwr /*
2117 1.7 gwr * Old PA and new PA are the same. No need to relink
2118 1.7 gwr * the mapping within the PV list.
2119 1.7 gwr */
2120 1.7 gwr insert = FALSE;
2121 1.8 gwr
2122 1.8 gwr /*
2123 1.8 gwr * Save any mod/ref bits on the PTE.
2124 1.8 gwr */
2125 1.8 gwr pte->attr.raw &= (MMU_SHORT_PTE_USED|MMU_SHORT_PTE_M);
2126 1.7 gwr }
2127 1.7 gwr } else {
2128 1.7 gwr /*
2129 1.8 gwr * If the VA lies below virtual_avail or beyond
2130 1.8 gwr * VM_MAX_KERNEL_ADDRESS, it is not a request by the VM
2131 1.8 gwr * system and hence does not need to be linked into the PV
2132 1.8 gwr * system.
2133 1.7 gwr */
2134 1.7 gwr insert = FALSE;
2135 1.8 gwr pte->attr.raw = MMU_DT_INVALID;
2136 1.7 gwr }
2137 1.7 gwr } else {
2138 1.8 gwr pte->attr.raw = MMU_DT_INVALID;
2139 1.7 gwr was_valid = FALSE;
2140 1.8 gwr if ((va >= virtual_avail) && (va < VM_MAX_KERNEL_ADDRESS)) {
2141 1.7 gwr if (is_managed(pa))
2142 1.7 gwr insert = TRUE;
2143 1.7 gwr else
2144 1.7 gwr insert = FALSE;
2145 1.8 gwr } else
2146 1.7 gwr insert = FALSE;
2147 1.7 gwr }
2148 1.7 gwr
2149 1.7 gwr /*
2150 1.8 gwr * Map the page. Being careful to preserve modified/referenced bits
2151 1.8 gwr * on the PTE.
2152 1.7 gwr */
2153 1.7 gwr pte->attr.raw |= (pa | MMU_DT_PAGE);
2154 1.1 gwr
2155 1.1 gwr if (!(prot & VM_PROT_WRITE)) /* If access should be read-only */
2156 1.1 gwr pte->attr.raw |= MMU_SHORT_PTE_WP;
2157 1.7 gwr if (flags & PMAP_NC)
2158 1.1 gwr pte->attr.raw |= MMU_SHORT_PTE_CI;
2159 1.8 gwr if (was_valid)
2160 1.7 gwr TBIS(va);
2161 1.1 gwr
2162 1.7 gwr /*
2163 1.7 gwr * Insert the PTE into the PV system, if need be.
2164 1.7 gwr */
2165 1.7 gwr if (insert) {
2166 1.7 gwr pv = pa2pv(pa);
2167 1.7 gwr pvebase[pte_idx].pve_next = pv->pv_idx;
2168 1.7 gwr pv->pv_idx = pte_idx;
2169 1.7 gwr }
2170 1.7 gwr splx(s);
2171 1.7 gwr
2172 1.1 gwr }
2173 1.1 gwr
2174 1.1 gwr /* pmap_protect INTERFACE
2175 1.1 gwr **
2176 1.7 gwr * Apply the given protection to the given virtual address range within
2177 1.1 gwr * the given map.
2178 1.1 gwr *
2179 1.1 gwr * It is ok for the protection applied to be stronger than what is
2180 1.1 gwr * specified. We use this to our advantage when the given map has no
2181 1.7 gwr * mapping for the virtual address. By skipping a page when this
2182 1.1 gwr * is discovered, we are effectively applying a protection of VM_PROT_NONE,
2183 1.1 gwr * and therefore do not need to map the page just to apply a protection
2184 1.1 gwr * code. Only pmap_enter() needs to create new mappings if they do not exist.
2185 1.7 gwr *
2186 1.7 gwr * XXX - This function could be speeded up by using pmap_stroll() for inital
2187 1.7 gwr * setup, and then manual scrolling in the for() loop.
2188 1.1 gwr */
2189 1.1 gwr void
2190 1.7 gwr pmap_protect(pmap, startva, endva, prot)
2191 1.1 gwr pmap_t pmap;
2192 1.7 gwr vm_offset_t startva, endva;
2193 1.1 gwr vm_prot_t prot;
2194 1.1 gwr {
2195 1.7 gwr boolean_t iscurpmap;
2196 1.1 gwr int a_idx, b_idx, c_idx;
2197 1.1 gwr a_tmgr_t *a_tbl;
2198 1.1 gwr b_tmgr_t *b_tbl;
2199 1.1 gwr c_tmgr_t *c_tbl;
2200 1.1 gwr mmu_short_pte_t *pte;
2201 1.1 gwr
2202 1.1 gwr if (pmap == NULL)
2203 1.1 gwr return;
2204 1.1 gwr if (pmap == pmap_kernel()) {
2205 1.7 gwr pmap_protect_kernel(startva, endva, prot);
2206 1.1 gwr return;
2207 1.1 gwr }
2208 1.1 gwr
2209 1.11 jeremy /*
2210 1.12 jeremy * In this particular pmap implementation, there are only three
2211 1.12 jeremy * types of memory protection: 'all' (read/write/execute),
2212 1.12 jeremy * 'read-only' (read/execute) and 'none' (no mapping.)
2213 1.12 jeremy * It is not possible for us to treat 'executable' as a separate
2214 1.12 jeremy * protection type. Therefore, protection requests that seek to
2215 1.12 jeremy * remove execute permission while retaining read or write, and those
2216 1.12 jeremy * that make little sense (write-only for example) are ignored.
2217 1.11 jeremy */
2218 1.12 jeremy switch (prot) {
2219 1.12 jeremy case VM_PROT_NONE:
2220 1.12 jeremy /*
2221 1.12 jeremy * A request to apply the protection code of
2222 1.12 jeremy * 'VM_PROT_NONE' is a synonym for pmap_remove().
2223 1.12 jeremy */
2224 1.12 jeremy pmap_remove(pmap, startva, endva);
2225 1.12 jeremy return;
2226 1.12 jeremy case VM_PROT_EXECUTE:
2227 1.12 jeremy case VM_PROT_READ:
2228 1.12 jeremy case VM_PROT_READ|VM_PROT_EXECUTE:
2229 1.12 jeremy /* continue */
2230 1.12 jeremy break;
2231 1.12 jeremy case VM_PROT_WRITE:
2232 1.12 jeremy case VM_PROT_WRITE|VM_PROT_READ:
2233 1.12 jeremy case VM_PROT_WRITE|VM_PROT_EXECUTE:
2234 1.12 jeremy case VM_PROT_ALL:
2235 1.12 jeremy /* None of these should happen in a sane system. */
2236 1.12 jeremy return;
2237 1.11 jeremy }
2238 1.11 jeremy
2239 1.11 jeremy /*
2240 1.11 jeremy * If the pmap has no A table, it has no mappings and therefore
2241 1.11 jeremy * there is nothing to protect.
2242 1.11 jeremy */
2243 1.11 jeremy if ((a_tbl = pmap->pm_a_tmgr) == NULL)
2244 1.11 jeremy return;
2245 1.11 jeremy
2246 1.11 jeremy a_idx = MMU_TIA(startva);
2247 1.11 jeremy b_idx = MMU_TIB(startva);
2248 1.11 jeremy c_idx = MMU_TIC(startva);
2249 1.11 jeremy b_tbl = (b_tmgr_t *) c_tbl = NULL;
2250 1.11 jeremy
2251 1.7 gwr iscurpmap = (pmap == current_pmap());
2252 1.11 jeremy while (startva < endva) {
2253 1.11 jeremy if (b_tbl || MMU_VALID_DT(a_tbl->at_dtbl[a_idx])) {
2254 1.11 jeremy if (b_tbl == NULL) {
2255 1.11 jeremy b_tbl = (b_tmgr_t *) a_tbl->at_dtbl[a_idx].addr.raw;
2256 1.11 jeremy b_tbl = mmu_ptov((vm_offset_t) b_tbl);
2257 1.11 jeremy b_tbl = mmuB2tmgr((mmu_short_dte_t *) b_tbl);
2258 1.11 jeremy }
2259 1.11 jeremy if (c_tbl || MMU_VALID_DT(b_tbl->bt_dtbl[b_idx])) {
2260 1.11 jeremy if (c_tbl == NULL) {
2261 1.11 jeremy c_tbl = (c_tmgr_t *) MMU_DTE_PA(b_tbl->bt_dtbl[b_idx]);
2262 1.11 jeremy c_tbl = mmu_ptov((vm_offset_t) c_tbl);
2263 1.11 jeremy c_tbl = mmuC2tmgr((mmu_short_pte_t *) c_tbl);
2264 1.11 jeremy }
2265 1.11 jeremy if (MMU_VALID_DT(c_tbl->ct_dtbl[c_idx])) {
2266 1.11 jeremy pte = &c_tbl->ct_dtbl[c_idx];
2267 1.12 jeremy /* make the mapping read-only */
2268 1.12 jeremy pte->attr.raw |= MMU_SHORT_PTE_WP;
2269 1.11 jeremy /*
2270 1.11 jeremy * If we just modified the current address space,
2271 1.11 jeremy * flush any translations for the modified page from
2272 1.11 jeremy * the translation cache and any data from it in the
2273 1.11 jeremy * data cache.
2274 1.11 jeremy */
2275 1.11 jeremy if (iscurpmap)
2276 1.11 jeremy TBIS(startva);
2277 1.11 jeremy }
2278 1.11 jeremy startva += NBPG;
2279 1.1 gwr
2280 1.11 jeremy if (++c_idx >= MMU_C_TBL_SIZE) { /* exceeded C table? */
2281 1.11 jeremy c_tbl = NULL;
2282 1.11 jeremy c_idx = 0;
2283 1.11 jeremy if (++b_idx >= MMU_B_TBL_SIZE) { /* exceeded B table? */
2284 1.11 jeremy b_tbl = NULL;
2285 1.11 jeremy b_idx = 0;
2286 1.11 jeremy }
2287 1.11 jeremy }
2288 1.11 jeremy } else { /* C table wasn't valid */
2289 1.11 jeremy c_tbl = NULL;
2290 1.11 jeremy c_idx = 0;
2291 1.11 jeremy startva += MMU_TIB_RANGE;
2292 1.11 jeremy if (++b_idx >= MMU_B_TBL_SIZE) { /* exceeded B table? */
2293 1.11 jeremy b_tbl = NULL;
2294 1.11 jeremy b_idx = 0;
2295 1.11 jeremy }
2296 1.11 jeremy } /* C table */
2297 1.11 jeremy } else { /* B table wasn't valid */
2298 1.11 jeremy b_tbl = NULL;
2299 1.11 jeremy b_idx = 0;
2300 1.11 jeremy startva += MMU_TIA_RANGE;
2301 1.11 jeremy a_idx++;
2302 1.11 jeremy } /* B table */
2303 1.1 gwr }
2304 1.1 gwr }
2305 1.1 gwr
2306 1.1 gwr /* pmap_protect_kernel INTERNAL
2307 1.1 gwr **
2308 1.7 gwr * Apply the given protection code to a kernel address range.
2309 1.1 gwr */
2310 1.1 gwr void
2311 1.7 gwr pmap_protect_kernel(startva, endva, prot)
2312 1.7 gwr vm_offset_t startva, endva;
2313 1.1 gwr vm_prot_t prot;
2314 1.1 gwr {
2315 1.7 gwr vm_offset_t va;
2316 1.1 gwr mmu_short_pte_t *pte;
2317 1.1 gwr
2318 1.16 gwr pte = &kernCbase[(unsigned long) _btop(startva - KERNBASE)];
2319 1.7 gwr for (va = startva; va < endva; va += NBPG, pte++) {
2320 1.7 gwr if (MMU_VALID_DT(*pte)) {
2321 1.7 gwr switch (prot) {
2322 1.7 gwr case VM_PROT_ALL:
2323 1.7 gwr break;
2324 1.7 gwr case VM_PROT_EXECUTE:
2325 1.7 gwr case VM_PROT_READ:
2326 1.7 gwr case VM_PROT_READ|VM_PROT_EXECUTE:
2327 1.7 gwr pte->attr.raw |= MMU_SHORT_PTE_WP;
2328 1.7 gwr break;
2329 1.7 gwr case VM_PROT_NONE:
2330 1.7 gwr /* this is an alias for 'pmap_remove_kernel' */
2331 1.7 gwr pmap_remove_pte(pte);
2332 1.7 gwr break;
2333 1.7 gwr default:
2334 1.7 gwr break;
2335 1.7 gwr }
2336 1.7 gwr /*
2337 1.7 gwr * since this is the kernel, immediately flush any cached
2338 1.7 gwr * descriptors for this address.
2339 1.7 gwr */
2340 1.7 gwr TBIS(va);
2341 1.1 gwr }
2342 1.1 gwr }
2343 1.1 gwr }
2344 1.1 gwr
2345 1.1 gwr /* pmap_change_wiring INTERFACE
2346 1.1 gwr **
2347 1.1 gwr * Changes the wiring of the specified page.
2348 1.1 gwr *
2349 1.1 gwr * This function is called from vm_fault.c to unwire
2350 1.1 gwr * a mapping. It really should be called 'pmap_unwire'
2351 1.1 gwr * because it is never asked to do anything but remove
2352 1.1 gwr * wirings.
2353 1.1 gwr */
2354 1.1 gwr void
2355 1.1 gwr pmap_change_wiring(pmap, va, wire)
2356 1.1 gwr pmap_t pmap;
2357 1.1 gwr vm_offset_t va;
2358 1.1 gwr boolean_t wire;
2359 1.1 gwr {
2360 1.1 gwr int a_idx, b_idx, c_idx;
2361 1.1 gwr a_tmgr_t *a_tbl;
2362 1.1 gwr b_tmgr_t *b_tbl;
2363 1.1 gwr c_tmgr_t *c_tbl;
2364 1.1 gwr mmu_short_pte_t *pte;
2365 1.1 gwr
2366 1.1 gwr /* Kernel mappings always remain wired. */
2367 1.1 gwr if (pmap == pmap_kernel())
2368 1.1 gwr return;
2369 1.1 gwr
2370 1.1 gwr #ifdef PMAP_DEBUG
2371 1.1 gwr if (wire == TRUE)
2372 1.1 gwr panic("pmap_change_wiring: wire requested.");
2373 1.1 gwr #endif
2374 1.1 gwr
2375 1.7 gwr /*
2376 1.7 gwr * Walk through the tables. If the walk terminates without
2377 1.1 gwr * a valid PTE then the address wasn't wired in the first place.
2378 1.1 gwr * Return immediately.
2379 1.1 gwr */
2380 1.1 gwr if (pmap_stroll(pmap, va, &a_tbl, &b_tbl, &c_tbl, &pte, &a_idx,
2381 1.1 gwr &b_idx, &c_idx) == FALSE)
2382 1.1 gwr return;
2383 1.1 gwr
2384 1.1 gwr
2385 1.1 gwr /* Is the PTE wired? If not, return. */
2386 1.1 gwr if (!(pte->attr.raw & MMU_SHORT_PTE_WIRED))
2387 1.1 gwr return;
2388 1.1 gwr
2389 1.1 gwr /* Remove the wiring bit. */
2390 1.1 gwr pte->attr.raw &= ~(MMU_SHORT_PTE_WIRED);
2391 1.1 gwr
2392 1.7 gwr /*
2393 1.7 gwr * Decrement the wired entry count in the C table.
2394 1.1 gwr * If it reaches zero the following things happen:
2395 1.1 gwr * 1. The table no longer has any wired entries and is considered
2396 1.1 gwr * unwired.
2397 1.1 gwr * 2. It is placed on the available queue.
2398 1.1 gwr * 3. The parent table's wired entry count is decremented.
2399 1.1 gwr * 4. If it reaches zero, this process repeats at step 1 and
2400 1.1 gwr * stops at after reaching the A table.
2401 1.1 gwr */
2402 1.7 gwr if (--c_tbl->ct_wcnt == 0) {
2403 1.1 gwr TAILQ_INSERT_TAIL(&c_pool, c_tbl, ct_link);
2404 1.7 gwr if (--b_tbl->bt_wcnt == 0) {
2405 1.1 gwr TAILQ_INSERT_TAIL(&b_pool, b_tbl, bt_link);
2406 1.7 gwr if (--a_tbl->at_wcnt == 0) {
2407 1.1 gwr TAILQ_INSERT_TAIL(&a_pool, a_tbl, at_link);
2408 1.1 gwr }
2409 1.1 gwr }
2410 1.1 gwr }
2411 1.1 gwr }
2412 1.1 gwr
2413 1.1 gwr /* pmap_pageable INTERFACE
2414 1.1 gwr **
2415 1.1 gwr * Make the specified range of addresses within the given pmap,
2416 1.1 gwr * 'pageable' or 'not-pageable'. A pageable page must not cause
2417 1.1 gwr * any faults when referenced. A non-pageable page may.
2418 1.1 gwr *
2419 1.1 gwr * This routine is only advisory. The VM system will call pmap_enter()
2420 1.1 gwr * to wire or unwire pages that are going to be made pageable before calling
2421 1.1 gwr * this function. By the time this routine is called, everything that needs
2422 1.1 gwr * to be done has already been done.
2423 1.1 gwr */
2424 1.1 gwr void
2425 1.1 gwr pmap_pageable(pmap, start, end, pageable)
2426 1.1 gwr pmap_t pmap;
2427 1.1 gwr vm_offset_t start, end;
2428 1.1 gwr boolean_t pageable;
2429 1.1 gwr {
2430 1.1 gwr /* not implemented. */
2431 1.1 gwr }
2432 1.1 gwr
2433 1.1 gwr /* pmap_copy INTERFACE
2434 1.1 gwr **
2435 1.1 gwr * Copy the mappings of a range of addresses in one pmap, into
2436 1.1 gwr * the destination address of another.
2437 1.1 gwr *
2438 1.1 gwr * This routine is advisory. Should we one day decide that MMU tables
2439 1.1 gwr * may be shared by more than one pmap, this function should be used to
2440 1.1 gwr * link them together. Until that day however, we do nothing.
2441 1.1 gwr */
2442 1.1 gwr void
2443 1.1 gwr pmap_copy(pmap_a, pmap_b, dst, len, src)
2444 1.1 gwr pmap_t pmap_a, pmap_b;
2445 1.1 gwr vm_offset_t dst;
2446 1.1 gwr vm_size_t len;
2447 1.1 gwr vm_offset_t src;
2448 1.1 gwr {
2449 1.1 gwr /* not implemented. */
2450 1.1 gwr }
2451 1.1 gwr
2452 1.1 gwr /* pmap_copy_page INTERFACE
2453 1.1 gwr **
2454 1.1 gwr * Copy the contents of one physical page into another.
2455 1.1 gwr *
2456 1.7 gwr * This function makes use of two virtual pages allocated in pmap_bootstrap()
2457 1.7 gwr * to map the two specified physical pages into the kernel address space. It
2458 1.7 gwr * then uses bcopy() to copy one into the other.
2459 1.7 gwr *
2460 1.7 gwr * Note: We could use the transparent translation registers to make the
2461 1.7 gwr * mappings. If we do so, be sure to disable interrupts before using them.
2462 1.1 gwr */
2463 1.1 gwr void
2464 1.1 gwr pmap_copy_page(src, dst)
2465 1.1 gwr vm_offset_t src, dst;
2466 1.1 gwr {
2467 1.23 jeremy int s;
2468 1.23 jeremy vm_offset_t oldsrc, olddst;
2469 1.1 gwr
2470 1.23 jeremy s = splimp();
2471 1.23 jeremy oldsrc = olddst = 0; /*XXXgcc*/
2472 1.23 jeremy if (tmp_vpage0_inuse++) {
2473 1.23 jeremy oldsrc = pmap_extract_kernel(tmp_vpages[0]);
2474 1.23 jeremy }
2475 1.23 jeremy if (tmp_vpage1_inuse++) {
2476 1.23 jeremy olddst = pmap_extract_kernel(tmp_vpages[1]);
2477 1.23 jeremy }
2478 1.23 jeremy splx(s);
2479 1.23 jeremy
2480 1.23 jeremy /* Map pages as non-cacheable to avoid cache polution? */
2481 1.1 gwr pmap_enter_kernel(tmp_vpages[0], src, VM_PROT_READ);
2482 1.1 gwr pmap_enter_kernel(tmp_vpages[1], dst, VM_PROT_READ|VM_PROT_WRITE);
2483 1.7 gwr copypage((char *) tmp_vpages[0], (char *) tmp_vpages[1]);
2484 1.7 gwr
2485 1.23 jeremy s = splimp();
2486 1.23 jeremy if (--tmp_vpage0_inuse) {
2487 1.23 jeremy pmap_enter_kernel(tmp_vpages[0], oldsrc, VM_PROT_READ);
2488 1.23 jeremy }
2489 1.23 jeremy if (--tmp_vpage1_inuse) {
2490 1.23 jeremy pmap_enter_kernel(tmp_vpages[1], olddst,
2491 1.23 jeremy VM_PROT_READ|VM_PROT_WRITE);
2492 1.23 jeremy }
2493 1.23 jeremy splx(s);
2494 1.1 gwr }
2495 1.1 gwr
2496 1.1 gwr /* pmap_zero_page INTERFACE
2497 1.1 gwr **
2498 1.1 gwr * Zero the contents of the specified physical page.
2499 1.1 gwr *
2500 1.7 gwr * Uses one of the virtual pages allocated in pmap_boostrap()
2501 1.1 gwr * to map the specified page into the kernel address space. Then uses
2502 1.1 gwr * bzero() to zero out the page.
2503 1.1 gwr */
2504 1.1 gwr void
2505 1.1 gwr pmap_zero_page(pa)
2506 1.1 gwr vm_offset_t pa;
2507 1.1 gwr {
2508 1.23 jeremy int s;
2509 1.23 jeremy vm_offset_t oldpa;
2510 1.23 jeremy
2511 1.23 jeremy s = splimp();
2512 1.23 jeremy oldpa = 0; /*XXXgcc*/
2513 1.23 jeremy if (tmp_vpage1_inuse++) {
2514 1.23 jeremy oldpa = pmap_extract_kernel(tmp_vpages[1]);
2515 1.23 jeremy }
2516 1.23 jeremy splx(s);
2517 1.1 gwr
2518 1.23 jeremy pmap_enter_kernel(tmp_vpages[1], pa, VM_PROT_READ|VM_PROT_WRITE);
2519 1.23 jeremy zeropage((char *) tmp_vpages[1]);
2520 1.1 gwr
2521 1.23 jeremy s = splimp();
2522 1.23 jeremy if (--tmp_vpage1_inuse) {
2523 1.23 jeremy pmap_enter_kernel(tmp_vpages[1], oldpa,
2524 1.23 jeremy VM_PROT_READ|VM_PROT_WRITE);
2525 1.23 jeremy }
2526 1.23 jeremy splx(s);
2527 1.1 gwr }
2528 1.1 gwr
2529 1.1 gwr /* pmap_collect INTERFACE
2530 1.1 gwr **
2531 1.7 gwr * Called from the VM system when we are about to swap out
2532 1.7 gwr * the process using this pmap. This should give up any
2533 1.7 gwr * resources held here, including all its MMU tables.
2534 1.1 gwr */
2535 1.1 gwr void
2536 1.1 gwr pmap_collect(pmap)
2537 1.1 gwr pmap_t pmap;
2538 1.1 gwr {
2539 1.7 gwr /* XXX - todo... */
2540 1.1 gwr }
2541 1.1 gwr
2542 1.1 gwr /* pmap_create INTERFACE
2543 1.1 gwr **
2544 1.1 gwr * Create and return a pmap structure.
2545 1.1 gwr */
2546 1.1 gwr pmap_t
2547 1.1 gwr pmap_create(size)
2548 1.1 gwr vm_size_t size;
2549 1.1 gwr {
2550 1.1 gwr pmap_t pmap;
2551 1.1 gwr
2552 1.1 gwr if (size)
2553 1.1 gwr return NULL;
2554 1.1 gwr
2555 1.1 gwr pmap = (pmap_t) malloc(sizeof(struct pmap), M_VMPMAP, M_WAITOK);
2556 1.1 gwr pmap_pinit(pmap);
2557 1.1 gwr
2558 1.1 gwr return pmap;
2559 1.1 gwr }
2560 1.1 gwr
2561 1.1 gwr /* pmap_pinit INTERNAL
2562 1.1 gwr **
2563 1.1 gwr * Initialize a pmap structure.
2564 1.1 gwr */
2565 1.1 gwr void
2566 1.1 gwr pmap_pinit(pmap)
2567 1.1 gwr pmap_t pmap;
2568 1.1 gwr {
2569 1.1 gwr bzero(pmap, sizeof(struct pmap));
2570 1.7 gwr pmap->pm_a_tmgr = NULL;
2571 1.7 gwr pmap->pm_a_phys = kernAphys;
2572 1.1 gwr }
2573 1.1 gwr
2574 1.1 gwr /* pmap_release INTERFACE
2575 1.1 gwr **
2576 1.1 gwr * Release any resources held by the given pmap.
2577 1.1 gwr *
2578 1.1 gwr * This is the reverse analog to pmap_pinit. It does not
2579 1.1 gwr * necessarily mean for the pmap structure to be deallocated,
2580 1.1 gwr * as in pmap_destroy.
2581 1.1 gwr */
2582 1.1 gwr void
2583 1.1 gwr pmap_release(pmap)
2584 1.1 gwr pmap_t pmap;
2585 1.1 gwr {
2586 1.7 gwr /*
2587 1.7 gwr * As long as the pmap contains no mappings,
2588 1.1 gwr * which always should be the case whenever
2589 1.1 gwr * this function is called, there really should
2590 1.1 gwr * be nothing to do.
2591 1.1 gwr */
2592 1.1 gwr #ifdef PMAP_DEBUG
2593 1.1 gwr if (pmap == NULL)
2594 1.1 gwr return;
2595 1.1 gwr if (pmap == pmap_kernel())
2596 1.9 gwr panic("pmap_release: kernel pmap");
2597 1.1 gwr #endif
2598 1.9 gwr /*
2599 1.9 gwr * XXX - If this pmap has an A table, give it back.
2600 1.9 gwr * The pmap SHOULD be empty by now, and pmap_remove
2601 1.9 gwr * should have already given back the A table...
2602 1.9 gwr * However, I see: pmap->pm_a_tmgr->at_ecnt == 1
2603 1.9 gwr * at this point, which means some mapping was not
2604 1.9 gwr * removed when it should have been. -gwr
2605 1.9 gwr */
2606 1.7 gwr if (pmap->pm_a_tmgr != NULL) {
2607 1.9 gwr /* First make sure we are not using it! */
2608 1.9 gwr if (kernel_crp.rp_addr == pmap->pm_a_phys) {
2609 1.9 gwr kernel_crp.rp_addr = kernAphys;
2610 1.9 gwr loadcrp(&kernel_crp);
2611 1.9 gwr }
2612 1.13 gwr #ifdef PMAP_DEBUG /* XXX - todo! */
2613 1.13 gwr /* XXX - Now complain... */
2614 1.13 gwr printf("pmap_release: still have table\n");
2615 1.13 gwr Debugger();
2616 1.13 gwr #endif
2617 1.7 gwr free_a_table(pmap->pm_a_tmgr, TRUE);
2618 1.7 gwr pmap->pm_a_tmgr = NULL;
2619 1.7 gwr pmap->pm_a_phys = kernAphys;
2620 1.7 gwr }
2621 1.1 gwr }
2622 1.1 gwr
2623 1.1 gwr /* pmap_reference INTERFACE
2624 1.1 gwr **
2625 1.1 gwr * Increment the reference count of a pmap.
2626 1.1 gwr */
2627 1.1 gwr void
2628 1.1 gwr pmap_reference(pmap)
2629 1.1 gwr pmap_t pmap;
2630 1.1 gwr {
2631 1.1 gwr if (pmap == NULL)
2632 1.1 gwr return;
2633 1.1 gwr
2634 1.1 gwr /* pmap_lock(pmap); */
2635 1.1 gwr pmap->pm_refcount++;
2636 1.1 gwr /* pmap_unlock(pmap); */
2637 1.1 gwr }
2638 1.1 gwr
2639 1.1 gwr /* pmap_dereference INTERNAL
2640 1.1 gwr **
2641 1.1 gwr * Decrease the reference count on the given pmap
2642 1.1 gwr * by one and return the current count.
2643 1.1 gwr */
2644 1.1 gwr int
2645 1.1 gwr pmap_dereference(pmap)
2646 1.1 gwr pmap_t pmap;
2647 1.1 gwr {
2648 1.1 gwr int rtn;
2649 1.1 gwr
2650 1.1 gwr if (pmap == NULL)
2651 1.1 gwr return 0;
2652 1.1 gwr
2653 1.1 gwr /* pmap_lock(pmap); */
2654 1.1 gwr rtn = --pmap->pm_refcount;
2655 1.1 gwr /* pmap_unlock(pmap); */
2656 1.1 gwr
2657 1.1 gwr return rtn;
2658 1.1 gwr }
2659 1.1 gwr
2660 1.1 gwr /* pmap_destroy INTERFACE
2661 1.1 gwr **
2662 1.1 gwr * Decrement a pmap's reference count and delete
2663 1.1 gwr * the pmap if it becomes zero. Will be called
2664 1.1 gwr * only after all mappings have been removed.
2665 1.1 gwr */
2666 1.1 gwr void
2667 1.1 gwr pmap_destroy(pmap)
2668 1.1 gwr pmap_t pmap;
2669 1.1 gwr {
2670 1.1 gwr if (pmap == NULL)
2671 1.1 gwr return;
2672 1.1 gwr if (pmap == &kernel_pmap)
2673 1.1 gwr panic("pmap_destroy: kernel_pmap!");
2674 1.1 gwr if (pmap_dereference(pmap) == 0) {
2675 1.1 gwr pmap_release(pmap);
2676 1.1 gwr free(pmap, M_VMPMAP);
2677 1.1 gwr }
2678 1.1 gwr }
2679 1.1 gwr
2680 1.1 gwr /* pmap_is_referenced INTERFACE
2681 1.1 gwr **
2682 1.1 gwr * Determine if the given physical page has been
2683 1.1 gwr * referenced (read from [or written to.])
2684 1.1 gwr */
2685 1.1 gwr boolean_t
2686 1.1 gwr pmap_is_referenced(pa)
2687 1.1 gwr vm_offset_t pa;
2688 1.1 gwr {
2689 1.1 gwr pv_t *pv;
2690 1.7 gwr int idx, s;
2691 1.1 gwr
2692 1.1 gwr if (!pv_initialized)
2693 1.1 gwr return FALSE;
2694 1.7 gwr /* XXX - this may be unecessary. */
2695 1.1 gwr if (!is_managed(pa))
2696 1.1 gwr return FALSE;
2697 1.1 gwr
2698 1.1 gwr pv = pa2pv(pa);
2699 1.7 gwr /*
2700 1.7 gwr * Check the flags on the pv head. If they are set,
2701 1.1 gwr * return immediately. Otherwise a search must be done.
2702 1.7 gwr */
2703 1.1 gwr if (pv->pv_flags & PV_FLAGS_USED)
2704 1.1 gwr return TRUE;
2705 1.7 gwr else {
2706 1.7 gwr s = splimp();
2707 1.7 gwr /*
2708 1.7 gwr * Search through all pv elements pointing
2709 1.1 gwr * to this page and query their reference bits
2710 1.1 gwr */
2711 1.7 gwr for (idx = pv->pv_idx; idx != PVE_EOL; idx =
2712 1.7 gwr pvebase[idx].pve_next)
2713 1.7 gwr if (MMU_PTE_USED(kernCbase[idx])) {
2714 1.7 gwr splx(s);
2715 1.1 gwr return TRUE;
2716 1.7 gwr }
2717 1.7 gwr splx(s);
2718 1.7 gwr }
2719 1.1 gwr
2720 1.1 gwr return FALSE;
2721 1.1 gwr }
2722 1.1 gwr
2723 1.1 gwr /* pmap_is_modified INTERFACE
2724 1.1 gwr **
2725 1.1 gwr * Determine if the given physical page has been
2726 1.1 gwr * modified (written to.)
2727 1.1 gwr */
2728 1.1 gwr boolean_t
2729 1.1 gwr pmap_is_modified(pa)
2730 1.1 gwr vm_offset_t pa;
2731 1.1 gwr {
2732 1.1 gwr pv_t *pv;
2733 1.7 gwr int idx, s;
2734 1.1 gwr
2735 1.1 gwr if (!pv_initialized)
2736 1.1 gwr return FALSE;
2737 1.7 gwr /* XXX - this may be unecessary. */
2738 1.1 gwr if (!is_managed(pa))
2739 1.1 gwr return FALSE;
2740 1.1 gwr
2741 1.1 gwr /* see comments in pmap_is_referenced() */
2742 1.1 gwr pv = pa2pv(pa);
2743 1.7 gwr if (pv->pv_flags & PV_FLAGS_MDFY) {
2744 1.1 gwr return TRUE;
2745 1.7 gwr } else {
2746 1.7 gwr s = splimp();
2747 1.7 gwr for (idx = pv->pv_idx; idx != PVE_EOL; idx =
2748 1.7 gwr pvebase[idx].pve_next)
2749 1.7 gwr if (MMU_PTE_MODIFIED(kernCbase[idx])) {
2750 1.7 gwr splx(s);
2751 1.1 gwr return TRUE;
2752 1.7 gwr }
2753 1.7 gwr splx(s);
2754 1.7 gwr }
2755 1.7 gwr
2756 1.1 gwr return FALSE;
2757 1.1 gwr }
2758 1.1 gwr
2759 1.1 gwr /* pmap_page_protect INTERFACE
2760 1.1 gwr **
2761 1.1 gwr * Applies the given protection to all mappings to the given
2762 1.1 gwr * physical page.
2763 1.1 gwr */
2764 1.1 gwr void
2765 1.1 gwr pmap_page_protect(pa, prot)
2766 1.1 gwr vm_offset_t pa;
2767 1.1 gwr vm_prot_t prot;
2768 1.1 gwr {
2769 1.1 gwr pv_t *pv;
2770 1.7 gwr int idx, s;
2771 1.8 gwr vm_offset_t va;
2772 1.1 gwr struct mmu_short_pte_struct *pte;
2773 1.8 gwr c_tmgr_t *c_tbl;
2774 1.8 gwr pmap_t pmap, curpmap;
2775 1.1 gwr
2776 1.1 gwr if (!is_managed(pa))
2777 1.1 gwr return;
2778 1.1 gwr
2779 1.8 gwr curpmap = current_pmap();
2780 1.1 gwr pv = pa2pv(pa);
2781 1.7 gwr s = splimp();
2782 1.7 gwr for (idx = pv->pv_idx; idx != PVE_EOL; idx = pvebase[idx].pve_next) {
2783 1.7 gwr pte = &kernCbase[idx];
2784 1.1 gwr switch (prot) {
2785 1.1 gwr case VM_PROT_ALL:
2786 1.1 gwr /* do nothing */
2787 1.1 gwr break;
2788 1.7 gwr case VM_PROT_EXECUTE:
2789 1.1 gwr case VM_PROT_READ:
2790 1.1 gwr case VM_PROT_READ|VM_PROT_EXECUTE:
2791 1.1 gwr pte->attr.raw |= MMU_SHORT_PTE_WP;
2792 1.8 gwr
2793 1.8 gwr /*
2794 1.8 gwr * Determine the virtual address mapped by
2795 1.8 gwr * the PTE and flush ATC entries if necessary.
2796 1.8 gwr */
2797 1.8 gwr va = pmap_get_pteinfo(idx, &pmap, &c_tbl);
2798 1.8 gwr if (pmap == curpmap || pmap == pmap_kernel())
2799 1.8 gwr TBIS(va);
2800 1.1 gwr break;
2801 1.1 gwr case VM_PROT_NONE:
2802 1.7 gwr /* Save the mod/ref bits. */
2803 1.7 gwr pv->pv_flags |= pte->attr.raw;
2804 1.7 gwr /* Invalidate the PTE. */
2805 1.7 gwr pte->attr.raw = MMU_DT_INVALID;
2806 1.8 gwr
2807 1.8 gwr /*
2808 1.8 gwr * Update table counts. And flush ATC entries
2809 1.8 gwr * if necessary.
2810 1.8 gwr */
2811 1.8 gwr va = pmap_get_pteinfo(idx, &pmap, &c_tbl);
2812 1.8 gwr
2813 1.8 gwr /*
2814 1.8 gwr * If the PTE belongs to the kernel map,
2815 1.8 gwr * be sure to flush the page it maps.
2816 1.8 gwr */
2817 1.8 gwr if (pmap == pmap_kernel()) {
2818 1.8 gwr TBIS(va);
2819 1.8 gwr } else {
2820 1.8 gwr /*
2821 1.8 gwr * The PTE belongs to a user map.
2822 1.8 gwr * update the entry count in the C
2823 1.8 gwr * table to which it belongs and flush
2824 1.8 gwr * the ATC if the mapping belongs to
2825 1.8 gwr * the current pmap.
2826 1.8 gwr */
2827 1.8 gwr c_tbl->ct_ecnt--;
2828 1.8 gwr if (pmap == curpmap)
2829 1.8 gwr TBIS(va);
2830 1.8 gwr }
2831 1.1 gwr break;
2832 1.1 gwr default:
2833 1.1 gwr break;
2834 1.1 gwr }
2835 1.1 gwr }
2836 1.8 gwr
2837 1.8 gwr /*
2838 1.8 gwr * If the protection code indicates that all mappings to the page
2839 1.8 gwr * be removed, truncate the PV list to zero entries.
2840 1.8 gwr */
2841 1.7 gwr if (prot == VM_PROT_NONE)
2842 1.7 gwr pv->pv_idx = PVE_EOL;
2843 1.7 gwr splx(s);
2844 1.1 gwr }
2845 1.1 gwr
2846 1.7 gwr /* pmap_get_pteinfo INTERNAL
2847 1.1 gwr **
2848 1.7 gwr * Called internally to find the pmap and virtual address within that
2849 1.8 gwr * map to which the pte at the given index maps. Also includes the PTE's C
2850 1.8 gwr * table manager.
2851 1.1 gwr *
2852 1.7 gwr * Returns the pmap in the argument provided, and the virtual address
2853 1.7 gwr * by return value.
2854 1.1 gwr */
2855 1.1 gwr vm_offset_t
2856 1.8 gwr pmap_get_pteinfo(idx, pmap, tbl)
2857 1.8 gwr u_int idx;
2858 1.7 gwr pmap_t *pmap;
2859 1.7 gwr c_tmgr_t **tbl;
2860 1.1 gwr {
2861 1.1 gwr a_tmgr_t *a_tbl;
2862 1.1 gwr b_tmgr_t *b_tbl;
2863 1.1 gwr c_tmgr_t *c_tbl;
2864 1.1 gwr vm_offset_t va = 0;
2865 1.1 gwr
2866 1.7 gwr /*
2867 1.7 gwr * Determine if the PTE is a kernel PTE or a user PTE.
2868 1.1 gwr */
2869 1.8 gwr if (idx >= NUM_KERN_PTES) {
2870 1.7 gwr /*
2871 1.7 gwr * The PTE belongs to a user mapping.
2872 1.8 gwr * Find the virtual address by decoding table indices.
2873 1.7 gwr * Each successive decode will reveal the address from
2874 1.7 gwr * least to most significant bit fashion.
2875 1.7 gwr *
2876 1.7 gwr * 31 0
2877 1.7 gwr * +-------------------------------+
2878 1.8 gwr * |AAAAAAABBBBBBCCCCCC............|
2879 1.7 gwr * +-------------------------------+
2880 1.7 gwr */
2881 1.8 gwr /* XXX: c_tbl = mmuC2tmgr(pte); */
2882 1.8 gwr /* XXX: Would like an inline for this to validate idx... */
2883 1.8 gwr c_tbl = &Ctmgrbase[(idx - NUM_KERN_PTES) / MMU_C_TBL_SIZE];
2884 1.7 gwr b_tbl = c_tbl->ct_parent;
2885 1.8 gwr a_tbl = b_tbl->bt_parent;
2886 1.8 gwr *pmap = a_tbl->at_parent;
2887 1.8 gwr *tbl = c_tbl;
2888 1.7 gwr
2889 1.8 gwr /* Start with the 'C' bits, then add B and A... */
2890 1.8 gwr va |= ((idx % MMU_C_TBL_SIZE) << MMU_TIC_SHIFT);
2891 1.7 gwr va |= (c_tbl->ct_pidx << MMU_TIB_SHIFT);
2892 1.7 gwr va |= (b_tbl->bt_pidx << MMU_TIA_SHIFT);
2893 1.7 gwr } else {
2894 1.7 gwr /*
2895 1.7 gwr * The PTE belongs to the kernel map.
2896 1.7 gwr */
2897 1.8 gwr *pmap = pmap_kernel();
2898 1.8 gwr
2899 1.16 gwr va = _ptob(idx);
2900 1.7 gwr va += KERNBASE;
2901 1.7 gwr }
2902 1.7 gwr
2903 1.1 gwr return va;
2904 1.1 gwr }
2905 1.1 gwr
2906 1.1 gwr /* pmap_clear_modify INTERFACE
2907 1.1 gwr **
2908 1.1 gwr * Clear the modification bit on the page at the specified
2909 1.1 gwr * physical address.
2910 1.1 gwr *
2911 1.1 gwr */
2912 1.1 gwr void
2913 1.1 gwr pmap_clear_modify(pa)
2914 1.1 gwr vm_offset_t pa;
2915 1.1 gwr {
2916 1.19 jeremy if (!is_managed(pa))
2917 1.19 jeremy return;
2918 1.1 gwr pmap_clear_pv(pa, PV_FLAGS_MDFY);
2919 1.1 gwr }
2920 1.1 gwr
2921 1.1 gwr /* pmap_clear_reference INTERFACE
2922 1.1 gwr **
2923 1.1 gwr * Clear the referenced bit on the page at the specified
2924 1.1 gwr * physical address.
2925 1.1 gwr */
2926 1.1 gwr void
2927 1.1 gwr pmap_clear_reference(pa)
2928 1.1 gwr vm_offset_t pa;
2929 1.1 gwr {
2930 1.19 jeremy if (!is_managed(pa))
2931 1.19 jeremy return;
2932 1.1 gwr pmap_clear_pv(pa, PV_FLAGS_USED);
2933 1.1 gwr }
2934 1.1 gwr
2935 1.1 gwr /* pmap_clear_pv INTERNAL
2936 1.1 gwr **
2937 1.1 gwr * Clears the specified flag from the specified physical address.
2938 1.1 gwr * (Used by pmap_clear_modify() and pmap_clear_reference().)
2939 1.1 gwr *
2940 1.1 gwr * Flag is one of:
2941 1.1 gwr * PV_FLAGS_MDFY - Page modified bit.
2942 1.1 gwr * PV_FLAGS_USED - Page used (referenced) bit.
2943 1.1 gwr *
2944 1.1 gwr * This routine must not only clear the flag on the pv list
2945 1.1 gwr * head. It must also clear the bit on every pte in the pv
2946 1.1 gwr * list associated with the address.
2947 1.1 gwr */
2948 1.1 gwr void
2949 1.1 gwr pmap_clear_pv(pa, flag)
2950 1.1 gwr vm_offset_t pa;
2951 1.1 gwr int flag;
2952 1.1 gwr {
2953 1.1 gwr pv_t *pv;
2954 1.7 gwr int idx, s;
2955 1.7 gwr vm_offset_t va;
2956 1.7 gwr pmap_t pmap;
2957 1.1 gwr mmu_short_pte_t *pte;
2958 1.7 gwr c_tmgr_t *c_tbl;
2959 1.1 gwr
2960 1.1 gwr pv = pa2pv(pa);
2961 1.7 gwr
2962 1.7 gwr s = splimp();
2963 1.1 gwr pv->pv_flags &= ~(flag);
2964 1.7 gwr for (idx = pv->pv_idx; idx != PVE_EOL; idx = pvebase[idx].pve_next) {
2965 1.7 gwr pte = &kernCbase[idx];
2966 1.1 gwr pte->attr.raw &= ~(flag);
2967 1.7 gwr /*
2968 1.7 gwr * The MC68030 MMU will not set the modified or
2969 1.7 gwr * referenced bits on any MMU tables for which it has
2970 1.7 gwr * a cached descriptor with its modify bit set. To insure
2971 1.7 gwr * that it will modify these bits on the PTE during the next
2972 1.7 gwr * time it is written to or read from, we must flush it from
2973 1.7 gwr * the ATC.
2974 1.7 gwr *
2975 1.7 gwr * Ordinarily it is only necessary to flush the descriptor
2976 1.7 gwr * if it is used in the current address space. But since I
2977 1.7 gwr * am not sure that there will always be a notion of
2978 1.7 gwr * 'the current address space' when this function is called,
2979 1.7 gwr * I will skip the test and always flush the address. It
2980 1.7 gwr * does no harm.
2981 1.7 gwr */
2982 1.8 gwr va = pmap_get_pteinfo(idx, &pmap, &c_tbl);
2983 1.7 gwr TBIS(va);
2984 1.1 gwr }
2985 1.7 gwr splx(s);
2986 1.1 gwr }
2987 1.1 gwr
2988 1.1 gwr /* pmap_extract INTERFACE
2989 1.1 gwr **
2990 1.1 gwr * Return the physical address mapped by the virtual address
2991 1.8 gwr * in the specified pmap or 0 if it is not known.
2992 1.1 gwr *
2993 1.1 gwr * Note: this function should also apply an exclusive lock
2994 1.1 gwr * on the pmap system during its duration.
2995 1.1 gwr */
2996 1.1 gwr vm_offset_t
2997 1.1 gwr pmap_extract(pmap, va)
2998 1.1 gwr pmap_t pmap;
2999 1.1 gwr vm_offset_t va;
3000 1.1 gwr {
3001 1.1 gwr int a_idx, b_idx, pte_idx;
3002 1.1 gwr a_tmgr_t *a_tbl;
3003 1.1 gwr b_tmgr_t *b_tbl;
3004 1.1 gwr c_tmgr_t *c_tbl;
3005 1.1 gwr mmu_short_pte_t *c_pte;
3006 1.1 gwr
3007 1.1 gwr if (pmap == pmap_kernel())
3008 1.1 gwr return pmap_extract_kernel(va);
3009 1.1 gwr if (pmap == NULL)
3010 1.1 gwr return 0;
3011 1.1 gwr
3012 1.1 gwr if (pmap_stroll(pmap, va, &a_tbl, &b_tbl, &c_tbl,
3013 1.7 gwr &c_pte, &a_idx, &b_idx, &pte_idx) == FALSE)
3014 1.1 gwr return 0;
3015 1.1 gwr
3016 1.7 gwr if (!MMU_VALID_DT(*c_pte))
3017 1.1 gwr return 0;
3018 1.7 gwr
3019 1.8 gwr return (MMU_PTE_PA(*c_pte));
3020 1.1 gwr }
3021 1.1 gwr
3022 1.1 gwr /* pmap_extract_kernel INTERNAL
3023 1.1 gwr **
3024 1.8 gwr * Extract a translation from the kernel address space.
3025 1.1 gwr */
3026 1.1 gwr vm_offset_t
3027 1.1 gwr pmap_extract_kernel(va)
3028 1.1 gwr vm_offset_t va;
3029 1.1 gwr {
3030 1.1 gwr mmu_short_pte_t *pte;
3031 1.1 gwr
3032 1.16 gwr pte = &kernCbase[(u_int) _btop(va - KERNBASE)];
3033 1.1 gwr return MMU_PTE_PA(*pte);
3034 1.1 gwr }
3035 1.1 gwr
3036 1.1 gwr /* pmap_remove_kernel INTERNAL
3037 1.1 gwr **
3038 1.1 gwr * Remove the mapping of a range of virtual addresses from the kernel map.
3039 1.9 gwr * The arguments are already page-aligned.
3040 1.1 gwr */
3041 1.1 gwr void
3042 1.9 gwr pmap_remove_kernel(sva, eva)
3043 1.9 gwr vm_offset_t sva;
3044 1.9 gwr vm_offset_t eva;
3045 1.1 gwr {
3046 1.9 gwr int idx, eidx;
3047 1.9 gwr
3048 1.9 gwr #ifdef PMAP_DEBUG
3049 1.9 gwr if ((sva & PGOFSET) || (eva & PGOFSET))
3050 1.9 gwr panic("pmap_remove_kernel: alignment");
3051 1.9 gwr #endif
3052 1.1 gwr
3053 1.16 gwr idx = _btop(sva - KERNBASE);
3054 1.16 gwr eidx = _btop(eva - KERNBASE);
3055 1.9 gwr
3056 1.9 gwr while (idx < eidx)
3057 1.9 gwr pmap_remove_pte(&kernCbase[idx++]);
3058 1.7 gwr /* Always flush the ATC when maniplating the kernel address space. */
3059 1.8 gwr TBIAS();
3060 1.1 gwr }
3061 1.1 gwr
3062 1.1 gwr /* pmap_remove INTERFACE
3063 1.1 gwr **
3064 1.1 gwr * Remove the mapping of a range of virtual addresses from the given pmap.
3065 1.7 gwr *
3066 1.7 gwr * If the range contains any wired entries, this function will probably create
3067 1.7 gwr * disaster.
3068 1.1 gwr */
3069 1.1 gwr void
3070 1.1 gwr pmap_remove(pmap, start, end)
3071 1.1 gwr pmap_t pmap;
3072 1.1 gwr vm_offset_t start;
3073 1.1 gwr vm_offset_t end;
3074 1.1 gwr {
3075 1.7 gwr
3076 1.1 gwr if (pmap == pmap_kernel()) {
3077 1.1 gwr pmap_remove_kernel(start, end);
3078 1.1 gwr return;
3079 1.1 gwr }
3080 1.1 gwr
3081 1.7 gwr /*
3082 1.7 gwr * XXX - Temporary(?) statement to prevent panic caused
3083 1.7 gwr * by vm_alloc_with_pager() handing us a software map (ie NULL)
3084 1.7 gwr * to remove because it couldn't get backing store.
3085 1.7 gwr * (I guess.)
3086 1.7 gwr */
3087 1.7 gwr if (pmap == NULL)
3088 1.7 gwr return;
3089 1.7 gwr
3090 1.7 gwr /*
3091 1.7 gwr * If the pmap doesn't have an A table of its own, it has no mappings
3092 1.7 gwr * that can be removed.
3093 1.1 gwr */
3094 1.7 gwr if (pmap->pm_a_tmgr == NULL)
3095 1.7 gwr return;
3096 1.7 gwr
3097 1.7 gwr /*
3098 1.7 gwr * Remove the specified range from the pmap. If the function
3099 1.7 gwr * returns true, the operation removed all the valid mappings
3100 1.7 gwr * in the pmap and freed its A table. If this happened to the
3101 1.7 gwr * currently loaded pmap, the MMU root pointer must be reloaded
3102 1.7 gwr * with the default 'kernel' map.
3103 1.7 gwr */
3104 1.7 gwr if (pmap_remove_a(pmap->pm_a_tmgr, start, end)) {
3105 1.9 gwr if (kernel_crp.rp_addr == pmap->pm_a_phys) {
3106 1.9 gwr kernel_crp.rp_addr = kernAphys;
3107 1.9 gwr loadcrp(&kernel_crp);
3108 1.9 gwr /* will do TLB flush below */
3109 1.9 gwr }
3110 1.7 gwr pmap->pm_a_tmgr = NULL;
3111 1.7 gwr pmap->pm_a_phys = kernAphys;
3112 1.1 gwr }
3113 1.9 gwr
3114 1.9 gwr /*
3115 1.9 gwr * If we just modified the current address space,
3116 1.9 gwr * make sure to flush the MMU cache.
3117 1.9 gwr *
3118 1.9 gwr * XXX - this could be an unecessarily large flush.
3119 1.9 gwr * XXX - Could decide, based on the size of the VA range
3120 1.9 gwr * to be removed, whether to flush "by pages" or "all".
3121 1.9 gwr */
3122 1.9 gwr if (pmap == current_pmap())
3123 1.9 gwr TBIAU();
3124 1.1 gwr }
3125 1.1 gwr
3126 1.1 gwr /* pmap_remove_a INTERNAL
3127 1.1 gwr **
3128 1.1 gwr * This is function number one in a set of three that removes a range
3129 1.1 gwr * of memory in the most efficient manner by removing the highest possible
3130 1.1 gwr * tables from the memory space. This particular function attempts to remove
3131 1.1 gwr * as many B tables as it can, delegating the remaining fragmented ranges to
3132 1.1 gwr * pmap_remove_b().
3133 1.1 gwr *
3134 1.7 gwr * If the removal operation results in an empty A table, the function returns
3135 1.7 gwr * TRUE.
3136 1.7 gwr *
3137 1.1 gwr * It's ugly but will do for now.
3138 1.1 gwr */
3139 1.7 gwr boolean_t
3140 1.1 gwr pmap_remove_a(a_tbl, start, end)
3141 1.1 gwr a_tmgr_t *a_tbl;
3142 1.1 gwr vm_offset_t start;
3143 1.1 gwr vm_offset_t end;
3144 1.1 gwr {
3145 1.7 gwr boolean_t empty;
3146 1.1 gwr int idx;
3147 1.7 gwr vm_offset_t nstart, nend;
3148 1.1 gwr b_tmgr_t *b_tbl;
3149 1.1 gwr mmu_long_dte_t *a_dte;
3150 1.1 gwr mmu_short_dte_t *b_dte;
3151 1.8 gwr
3152 1.7 gwr /*
3153 1.7 gwr * The following code works with what I call a 'granularity
3154 1.7 gwr * reduction algorithim'. A range of addresses will always have
3155 1.7 gwr * the following properties, which are classified according to
3156 1.7 gwr * how the range relates to the size of the current granularity
3157 1.7 gwr * - an A table entry:
3158 1.7 gwr *
3159 1.7 gwr * 1 2 3 4
3160 1.7 gwr * -+---+---+---+---+---+---+---+-
3161 1.7 gwr * -+---+---+---+---+---+---+---+-
3162 1.7 gwr *
3163 1.7 gwr * A range will always start on a granularity boundary, illustrated
3164 1.7 gwr * by '+' signs in the table above, or it will start at some point
3165 1.7 gwr * inbetween a granularity boundary, as illustrated by point 1.
3166 1.7 gwr * The first step in removing a range of addresses is to remove the
3167 1.7 gwr * range between 1 and 2, the nearest granularity boundary. This
3168 1.7 gwr * job is handled by the section of code governed by the
3169 1.7 gwr * 'if (start < nstart)' statement.
3170 1.7 gwr *
3171 1.7 gwr * A range will always encompass zero or more intergral granules,
3172 1.7 gwr * illustrated by points 2 and 3. Integral granules are easy to
3173 1.7 gwr * remove. The removal of these granules is the second step, and
3174 1.7 gwr * is handled by the code block 'if (nstart < nend)'.
3175 1.7 gwr *
3176 1.7 gwr * Lastly, a range will always end on a granularity boundary,
3177 1.7 gwr * ill. by point 3, or it will fall just beyond one, ill. by point
3178 1.7 gwr * 4. The last step involves removing this range and is handled by
3179 1.7 gwr * the code block 'if (nend < end)'.
3180 1.7 gwr */
3181 1.1 gwr nstart = MMU_ROUND_UP_A(start);
3182 1.1 gwr nend = MMU_ROUND_A(end);
3183 1.1 gwr
3184 1.1 gwr if (start < nstart) {
3185 1.7 gwr /*
3186 1.7 gwr * This block is executed if the range starts between
3187 1.7 gwr * a granularity boundary.
3188 1.7 gwr *
3189 1.7 gwr * First find the DTE which is responsible for mapping
3190 1.7 gwr * the start of the range.
3191 1.7 gwr */
3192 1.1 gwr idx = MMU_TIA(start);
3193 1.1 gwr a_dte = &a_tbl->at_dtbl[idx];
3194 1.7 gwr
3195 1.7 gwr /*
3196 1.7 gwr * If the DTE is valid then delegate the removal of the sub
3197 1.7 gwr * range to pmap_remove_b(), which can remove addresses at
3198 1.7 gwr * a finer granularity.
3199 1.7 gwr */
3200 1.1 gwr if (MMU_VALID_DT(*a_dte)) {
3201 1.7 gwr b_dte = mmu_ptov(a_dte->addr.raw);
3202 1.1 gwr b_tbl = mmuB2tmgr(b_dte);
3203 1.7 gwr
3204 1.7 gwr /*
3205 1.7 gwr * The sub range to be removed starts at the start
3206 1.7 gwr * of the full range we were asked to remove, and ends
3207 1.7 gwr * at the greater of:
3208 1.7 gwr * 1. The end of the full range, -or-
3209 1.7 gwr * 2. The end of the full range, rounded down to the
3210 1.7 gwr * nearest granularity boundary.
3211 1.7 gwr */
3212 1.7 gwr if (end < nstart)
3213 1.8 gwr empty = pmap_remove_b(b_tbl, start, end);
3214 1.7 gwr else
3215 1.8 gwr empty = pmap_remove_b(b_tbl, start, nstart);
3216 1.7 gwr
3217 1.7 gwr /*
3218 1.7 gwr * If the removal resulted in an empty B table,
3219 1.7 gwr * invalidate the DTE that points to it and decrement
3220 1.7 gwr * the valid entry count of the A table.
3221 1.7 gwr */
3222 1.7 gwr if (empty) {
3223 1.7 gwr a_dte->attr.raw = MMU_DT_INVALID;
3224 1.7 gwr a_tbl->at_ecnt--;
3225 1.1 gwr }
3226 1.1 gwr }
3227 1.7 gwr /*
3228 1.7 gwr * If the DTE is invalid, the address range is already non-
3229 1.7 gwr * existant and can simply be skipped.
3230 1.7 gwr */
3231 1.1 gwr }
3232 1.1 gwr if (nstart < nend) {
3233 1.7 gwr /*
3234 1.8 gwr * This block is executed if the range spans a whole number
3235 1.7 gwr * multiple of granules (A table entries.)
3236 1.7 gwr *
3237 1.7 gwr * First find the DTE which is responsible for mapping
3238 1.7 gwr * the start of the first granule involved.
3239 1.7 gwr */
3240 1.1 gwr idx = MMU_TIA(nstart);
3241 1.1 gwr a_dte = &a_tbl->at_dtbl[idx];
3242 1.7 gwr
3243 1.7 gwr /*
3244 1.7 gwr * Remove entire sub-granules (B tables) one at a time,
3245 1.7 gwr * until reaching the end of the range.
3246 1.7 gwr */
3247 1.7 gwr for (; nstart < nend; a_dte++, nstart += MMU_TIA_RANGE)
3248 1.1 gwr if (MMU_VALID_DT(*a_dte)) {
3249 1.7 gwr /*
3250 1.7 gwr * Find the B table manager for the
3251 1.7 gwr * entry and free it.
3252 1.7 gwr */
3253 1.7 gwr b_dte = mmu_ptov(a_dte->addr.raw);
3254 1.1 gwr b_tbl = mmuB2tmgr(b_dte);
3255 1.7 gwr free_b_table(b_tbl, TRUE);
3256 1.7 gwr
3257 1.7 gwr /*
3258 1.7 gwr * Invalidate the DTE that points to the
3259 1.7 gwr * B table and decrement the valid entry
3260 1.7 gwr * count of the A table.
3261 1.7 gwr */
3262 1.1 gwr a_dte->attr.raw = MMU_DT_INVALID;
3263 1.1 gwr a_tbl->at_ecnt--;
3264 1.1 gwr }
3265 1.1 gwr }
3266 1.1 gwr if (nend < end) {
3267 1.7 gwr /*
3268 1.7 gwr * This block is executed if the range ends beyond a
3269 1.7 gwr * granularity boundary.
3270 1.7 gwr *
3271 1.7 gwr * First find the DTE which is responsible for mapping
3272 1.7 gwr * the start of the nearest (rounded down) granularity
3273 1.7 gwr * boundary.
3274 1.7 gwr */
3275 1.1 gwr idx = MMU_TIA(nend);
3276 1.1 gwr a_dte = &a_tbl->at_dtbl[idx];
3277 1.7 gwr
3278 1.7 gwr /*
3279 1.7 gwr * If the DTE is valid then delegate the removal of the sub
3280 1.7 gwr * range to pmap_remove_b(), which can remove addresses at
3281 1.7 gwr * a finer granularity.
3282 1.7 gwr */
3283 1.1 gwr if (MMU_VALID_DT(*a_dte)) {
3284 1.7 gwr /*
3285 1.7 gwr * Find the B table manager for the entry
3286 1.7 gwr * and hand it to pmap_remove_b() along with
3287 1.7 gwr * the sub range.
3288 1.7 gwr */
3289 1.7 gwr b_dte = mmu_ptov(a_dte->addr.raw);
3290 1.1 gwr b_tbl = mmuB2tmgr(b_dte);
3291 1.7 gwr
3292 1.8 gwr empty = pmap_remove_b(b_tbl, nend, end);
3293 1.7 gwr
3294 1.7 gwr /*
3295 1.7 gwr * If the removal resulted in an empty B table,
3296 1.7 gwr * invalidate the DTE that points to it and decrement
3297 1.7 gwr * the valid entry count of the A table.
3298 1.7 gwr */
3299 1.7 gwr if (empty) {
3300 1.7 gwr a_dte->attr.raw = MMU_DT_INVALID;
3301 1.7 gwr a_tbl->at_ecnt--;
3302 1.7 gwr }
3303 1.1 gwr }
3304 1.1 gwr }
3305 1.7 gwr
3306 1.7 gwr /*
3307 1.7 gwr * If there are no more entries in the A table, release it
3308 1.7 gwr * back to the available pool and return TRUE.
3309 1.7 gwr */
3310 1.7 gwr if (a_tbl->at_ecnt == 0) {
3311 1.7 gwr a_tbl->at_parent = NULL;
3312 1.7 gwr TAILQ_REMOVE(&a_pool, a_tbl, at_link);
3313 1.7 gwr TAILQ_INSERT_HEAD(&a_pool, a_tbl, at_link);
3314 1.7 gwr empty = TRUE;
3315 1.7 gwr } else {
3316 1.7 gwr empty = FALSE;
3317 1.7 gwr }
3318 1.7 gwr
3319 1.7 gwr return empty;
3320 1.1 gwr }
3321 1.1 gwr
3322 1.1 gwr /* pmap_remove_b INTERNAL
3323 1.1 gwr **
3324 1.1 gwr * Remove a range of addresses from an address space, trying to remove entire
3325 1.1 gwr * C tables if possible.
3326 1.7 gwr *
3327 1.7 gwr * If the operation results in an empty B table, the function returns TRUE.
3328 1.1 gwr */
3329 1.7 gwr boolean_t
3330 1.8 gwr pmap_remove_b(b_tbl, start, end)
3331 1.1 gwr b_tmgr_t *b_tbl;
3332 1.1 gwr vm_offset_t start;
3333 1.1 gwr vm_offset_t end;
3334 1.1 gwr {
3335 1.7 gwr boolean_t empty;
3336 1.1 gwr int idx;
3337 1.1 gwr vm_offset_t nstart, nend, rstart;
3338 1.1 gwr c_tmgr_t *c_tbl;
3339 1.1 gwr mmu_short_dte_t *b_dte;
3340 1.1 gwr mmu_short_pte_t *c_dte;
3341 1.1 gwr
3342 1.1 gwr
3343 1.1 gwr nstart = MMU_ROUND_UP_B(start);
3344 1.1 gwr nend = MMU_ROUND_B(end);
3345 1.1 gwr
3346 1.1 gwr if (start < nstart) {
3347 1.1 gwr idx = MMU_TIB(start);
3348 1.1 gwr b_dte = &b_tbl->bt_dtbl[idx];
3349 1.1 gwr if (MMU_VALID_DT(*b_dte)) {
3350 1.7 gwr c_dte = mmu_ptov(MMU_DTE_PA(*b_dte));
3351 1.1 gwr c_tbl = mmuC2tmgr(c_dte);
3352 1.7 gwr if (end < nstart)
3353 1.8 gwr empty = pmap_remove_c(c_tbl, start, end);
3354 1.7 gwr else
3355 1.8 gwr empty = pmap_remove_c(c_tbl, start, nstart);
3356 1.7 gwr if (empty) {
3357 1.7 gwr b_dte->attr.raw = MMU_DT_INVALID;
3358 1.7 gwr b_tbl->bt_ecnt--;
3359 1.1 gwr }
3360 1.1 gwr }
3361 1.1 gwr }
3362 1.1 gwr if (nstart < nend) {
3363 1.1 gwr idx = MMU_TIB(nstart);
3364 1.1 gwr b_dte = &b_tbl->bt_dtbl[idx];
3365 1.1 gwr rstart = nstart;
3366 1.1 gwr while (rstart < nend) {
3367 1.1 gwr if (MMU_VALID_DT(*b_dte)) {
3368 1.7 gwr c_dte = mmu_ptov(MMU_DTE_PA(*b_dte));
3369 1.1 gwr c_tbl = mmuC2tmgr(c_dte);
3370 1.7 gwr free_c_table(c_tbl, TRUE);
3371 1.1 gwr b_dte->attr.raw = MMU_DT_INVALID;
3372 1.1 gwr b_tbl->bt_ecnt--;
3373 1.1 gwr }
3374 1.1 gwr b_dte++;
3375 1.1 gwr rstart += MMU_TIB_RANGE;
3376 1.1 gwr }
3377 1.1 gwr }
3378 1.1 gwr if (nend < end) {
3379 1.1 gwr idx = MMU_TIB(nend);
3380 1.1 gwr b_dte = &b_tbl->bt_dtbl[idx];
3381 1.1 gwr if (MMU_VALID_DT(*b_dte)) {
3382 1.7 gwr c_dte = mmu_ptov(MMU_DTE_PA(*b_dte));
3383 1.1 gwr c_tbl = mmuC2tmgr(c_dte);
3384 1.8 gwr empty = pmap_remove_c(c_tbl, nend, end);
3385 1.7 gwr if (empty) {
3386 1.7 gwr b_dte->attr.raw = MMU_DT_INVALID;
3387 1.7 gwr b_tbl->bt_ecnt--;
3388 1.7 gwr }
3389 1.1 gwr }
3390 1.1 gwr }
3391 1.7 gwr
3392 1.7 gwr if (b_tbl->bt_ecnt == 0) {
3393 1.7 gwr b_tbl->bt_parent = NULL;
3394 1.7 gwr TAILQ_REMOVE(&b_pool, b_tbl, bt_link);
3395 1.7 gwr TAILQ_INSERT_HEAD(&b_pool, b_tbl, bt_link);
3396 1.7 gwr empty = TRUE;
3397 1.7 gwr } else {
3398 1.7 gwr empty = FALSE;
3399 1.7 gwr }
3400 1.7 gwr
3401 1.7 gwr return empty;
3402 1.1 gwr }
3403 1.1 gwr
3404 1.1 gwr /* pmap_remove_c INTERNAL
3405 1.1 gwr **
3406 1.1 gwr * Remove a range of addresses from the given C table.
3407 1.1 gwr */
3408 1.7 gwr boolean_t
3409 1.8 gwr pmap_remove_c(c_tbl, start, end)
3410 1.1 gwr c_tmgr_t *c_tbl;
3411 1.1 gwr vm_offset_t start;
3412 1.1 gwr vm_offset_t end;
3413 1.1 gwr {
3414 1.7 gwr boolean_t empty;
3415 1.1 gwr int idx;
3416 1.1 gwr mmu_short_pte_t *c_pte;
3417 1.1 gwr
3418 1.1 gwr idx = MMU_TIC(start);
3419 1.1 gwr c_pte = &c_tbl->ct_dtbl[idx];
3420 1.8 gwr for (;start < end; start += MMU_PAGE_SIZE, c_pte++) {
3421 1.7 gwr if (MMU_VALID_DT(*c_pte)) {
3422 1.1 gwr pmap_remove_pte(c_pte);
3423 1.7 gwr c_tbl->ct_ecnt--;
3424 1.7 gwr }
3425 1.1 gwr }
3426 1.7 gwr
3427 1.7 gwr if (c_tbl->ct_ecnt == 0) {
3428 1.7 gwr c_tbl->ct_parent = NULL;
3429 1.7 gwr TAILQ_REMOVE(&c_pool, c_tbl, ct_link);
3430 1.9 gwr TAILQ_INSERT_HEAD(&c_pool, c_tbl, ct_link);
3431 1.9 gwr empty = TRUE;
3432 1.9 gwr } else {
3433 1.9 gwr empty = FALSE;
3434 1.9 gwr }
3435 1.7 gwr
3436 1.9 gwr return empty;
3437 1.1 gwr }
3438 1.1 gwr
3439 1.1 gwr /* is_managed INTERNAL
3440 1.1 gwr **
3441 1.1 gwr * Determine if the given physical address is managed by the PV system.
3442 1.1 gwr * Note that this logic assumes that no one will ask for the status of
3443 1.1 gwr * addresses which lie in-between the memory banks on the 3/80. If they
3444 1.1 gwr * do so, it will falsely report that it is managed.
3445 1.7 gwr *
3446 1.8 gwr * Note: A "managed" address is one that was reported to the VM system as
3447 1.8 gwr * a "usable page" during system startup. As such, the VM system expects the
3448 1.8 gwr * pmap module to keep an accurate track of the useage of those pages.
3449 1.8 gwr * Any page not given to the VM system at startup does not exist (as far as
3450 1.8 gwr * the VM system is concerned) and is therefore "unmanaged." Examples are
3451 1.8 gwr * those pages which belong to the ROM monitor and the memory allocated before
3452 1.8 gwr * the VM system was started.
3453 1.1 gwr */
3454 1.1 gwr boolean_t
3455 1.1 gwr is_managed(pa)
3456 1.1 gwr vm_offset_t pa;
3457 1.1 gwr {
3458 1.1 gwr if (pa >= avail_start && pa < avail_end)
3459 1.1 gwr return TRUE;
3460 1.1 gwr else
3461 1.1 gwr return FALSE;
3462 1.1 gwr }
3463 1.1 gwr
3464 1.1 gwr /* pmap_bootstrap_alloc INTERNAL
3465 1.1 gwr **
3466 1.1 gwr * Used internally for memory allocation at startup when malloc is not
3467 1.1 gwr * available. This code will fail once it crosses the first memory
3468 1.1 gwr * bank boundary on the 3/80. Hopefully by then however, the VM system
3469 1.1 gwr * will be in charge of allocation.
3470 1.1 gwr */
3471 1.1 gwr void *
3472 1.1 gwr pmap_bootstrap_alloc(size)
3473 1.1 gwr int size;
3474 1.1 gwr {
3475 1.1 gwr void *rtn;
3476 1.1 gwr
3477 1.8 gwr #ifdef PMAP_DEBUG
3478 1.7 gwr if (bootstrap_alloc_enabled == FALSE) {
3479 1.7 gwr mon_printf("pmap_bootstrap_alloc: disabled\n");
3480 1.7 gwr sunmon_abort();
3481 1.7 gwr }
3482 1.7 gwr #endif
3483 1.7 gwr
3484 1.1 gwr rtn = (void *) virtual_avail;
3485 1.1 gwr virtual_avail += size;
3486 1.1 gwr
3487 1.8 gwr #ifdef PMAP_DEBUG
3488 1.7 gwr if (virtual_avail > virtual_contig_end) {
3489 1.7 gwr mon_printf("pmap_bootstrap_alloc: out of mem\n");
3490 1.7 gwr sunmon_abort();
3491 1.1 gwr }
3492 1.7 gwr #endif
3493 1.1 gwr
3494 1.1 gwr return rtn;
3495 1.1 gwr }
3496 1.1 gwr
3497 1.1 gwr /* pmap_bootstap_aalign INTERNAL
3498 1.1 gwr **
3499 1.7 gwr * Used to insure that the next call to pmap_bootstrap_alloc() will
3500 1.7 gwr * return a chunk of memory aligned to the specified size.
3501 1.8 gwr *
3502 1.8 gwr * Note: This function will only support alignment sizes that are powers
3503 1.8 gwr * of two.
3504 1.1 gwr */
3505 1.1 gwr void
3506 1.1 gwr pmap_bootstrap_aalign(size)
3507 1.1 gwr int size;
3508 1.1 gwr {
3509 1.7 gwr int off;
3510 1.7 gwr
3511 1.7 gwr off = virtual_avail & (size - 1);
3512 1.7 gwr if (off) {
3513 1.7 gwr (void) pmap_bootstrap_alloc(size - off);
3514 1.1 gwr }
3515 1.1 gwr }
3516 1.7 gwr
3517 1.8 gwr /* pmap_pa_exists
3518 1.8 gwr **
3519 1.8 gwr * Used by the /dev/mem driver to see if a given PA is memory
3520 1.8 gwr * that can be mapped. (The PA is not in a hole.)
3521 1.8 gwr */
3522 1.8 gwr int
3523 1.8 gwr pmap_pa_exists(pa)
3524 1.8 gwr vm_offset_t pa;
3525 1.8 gwr {
3526 1.21 gwr register int i;
3527 1.21 gwr
3528 1.21 gwr for (i = 0; i < SUN3X_NPHYS_RAM_SEGS; i++) {
3529 1.21 gwr if ((pa >= avail_mem[i].pmem_start) &&
3530 1.21 gwr (pa < avail_mem[i].pmem_end))
3531 1.21 gwr return (1);
3532 1.21 gwr if (avail_mem[i].pmem_next == NULL)
3533 1.21 gwr break;
3534 1.21 gwr }
3535 1.8 gwr return (0);
3536 1.8 gwr }
3537 1.8 gwr
3538 1.1 gwr /* pmap_activate INTERFACE
3539 1.1 gwr **
3540 1.7 gwr * This is called by locore.s:cpu_switch when we are switching to a
3541 1.7 gwr * new process. This should load the MMU context for the new proc.
3542 1.7 gwr * XXX - Later, this should be done directly in locore.s
3543 1.1 gwr */
3544 1.1 gwr void
3545 1.7 gwr pmap_activate(pmap)
3546 1.1 gwr pmap_t pmap;
3547 1.1 gwr {
3548 1.7 gwr u_long rootpa;
3549 1.7 gwr
3550 1.7 gwr /* Only do reload/flush if we have to. */
3551 1.7 gwr rootpa = pmap->pm_a_phys;
3552 1.7 gwr if (kernel_crp.rp_addr != rootpa) {
3553 1.7 gwr DPRINT(("pmap_activate(%p)\n", pmap));
3554 1.7 gwr kernel_crp.rp_addr = rootpa;
3555 1.7 gwr loadcrp(&kernel_crp);
3556 1.8 gwr TBIAU();
3557 1.7 gwr }
3558 1.1 gwr }
3559 1.1 gwr
3560 1.1 gwr
3561 1.1 gwr /* pmap_update
3562 1.1 gwr **
3563 1.1 gwr * Apply any delayed changes scheduled for all pmaps immediately.
3564 1.1 gwr *
3565 1.1 gwr * No delayed operations are currently done in this pmap.
3566 1.1 gwr */
3567 1.1 gwr void
3568 1.1 gwr pmap_update()
3569 1.1 gwr {
3570 1.1 gwr /* not implemented. */
3571 1.1 gwr }
3572 1.1 gwr
3573 1.17 gwr /*
3574 1.17 gwr * Fill in the cpu_kcore header for dumpsys()
3575 1.17 gwr * (See machdep.c)
3576 1.17 gwr */
3577 1.17 gwr void
3578 1.17 gwr pmap_set_kcore_hdr(chdr_p)
3579 1.17 gwr cpu_kcore_hdr_t *chdr_p;
3580 1.17 gwr {
3581 1.20 thorpej struct sun3x_kcore_hdr *sh = &chdr_p->un._sun3x;
3582 1.17 gwr u_long spa, len;
3583 1.17 gwr int i;
3584 1.20 thorpej extern char machine[];
3585 1.17 gwr
3586 1.20 thorpej /*
3587 1.20 thorpej * Fill in dispatch information.
3588 1.20 thorpej */
3589 1.20 thorpej strcpy(chdr_p->name, machine);
3590 1.20 thorpej chdr_p->page_size = NBPG;
3591 1.20 thorpej chdr_p->kernbase = KERNBASE;
3592 1.20 thorpej
3593 1.20 thorpej sh->contig_end = virtual_contig_end;
3594 1.20 thorpej sh->kernCbase = (u_long) kernCbase;
3595 1.20 thorpej for (i = 0; i < SUN3X_NPHYS_RAM_SEGS; i++) {
3596 1.17 gwr spa = avail_mem[i].pmem_start;
3597 1.17 gwr spa = _trunc_page(spa);
3598 1.17 gwr len = avail_mem[i].pmem_end - spa;
3599 1.17 gwr len = _round_page(len);
3600 1.20 thorpej sh->ram_segs[i].start = spa;
3601 1.20 thorpej sh->ram_segs[i].size = len;
3602 1.17 gwr }
3603 1.17 gwr }
3604 1.17 gwr
3605 1.17 gwr
3606 1.1 gwr /* pmap_virtual_space INTERFACE
3607 1.1 gwr **
3608 1.1 gwr * Return the current available range of virtual addresses in the
3609 1.1 gwr * arguuments provided. Only really called once.
3610 1.1 gwr */
3611 1.1 gwr void
3612 1.1 gwr pmap_virtual_space(vstart, vend)
3613 1.1 gwr vm_offset_t *vstart, *vend;
3614 1.1 gwr {
3615 1.1 gwr *vstart = virtual_avail;
3616 1.1 gwr *vend = virtual_end;
3617 1.1 gwr }
3618 1.1 gwr
3619 1.1 gwr /* pmap_free_pages INTERFACE
3620 1.1 gwr **
3621 1.1 gwr * Return the number of physical pages still available.
3622 1.1 gwr *
3623 1.1 gwr * This is probably going to be a mess, but it's only called
3624 1.1 gwr * once and it's the only function left that I have to implement!
3625 1.1 gwr */
3626 1.1 gwr u_int
3627 1.1 gwr pmap_free_pages()
3628 1.1 gwr {
3629 1.1 gwr int i;
3630 1.1 gwr u_int left;
3631 1.1 gwr vm_offset_t avail;
3632 1.1 gwr
3633 1.7 gwr avail = avail_next;
3634 1.1 gwr left = 0;
3635 1.1 gwr i = 0;
3636 1.1 gwr while (avail >= avail_mem[i].pmem_end) {
3637 1.1 gwr if (avail_mem[i].pmem_next == NULL)
3638 1.1 gwr return 0;
3639 1.1 gwr i++;
3640 1.1 gwr }
3641 1.20 thorpej while (i < SUN3X_NPHYS_RAM_SEGS) {
3642 1.1 gwr if (avail < avail_mem[i].pmem_start) {
3643 1.1 gwr /* Avail is inside a hole, march it
3644 1.1 gwr * up to the next bank.
3645 1.1 gwr */
3646 1.1 gwr avail = avail_mem[i].pmem_start;
3647 1.1 gwr }
3648 1.16 gwr left += _btop(avail_mem[i].pmem_end - avail);
3649 1.1 gwr if (avail_mem[i].pmem_next == NULL)
3650 1.1 gwr break;
3651 1.1 gwr i++;
3652 1.1 gwr }
3653 1.1 gwr
3654 1.1 gwr return left;
3655 1.1 gwr }
3656 1.1 gwr
3657 1.1 gwr /* pmap_page_index INTERFACE
3658 1.1 gwr **
3659 1.1 gwr * Return the index of the given physical page in a list of useable
3660 1.1 gwr * physical pages in the system. Holes in physical memory may be counted
3661 1.1 gwr * if so desired. As long as pmap_free_pages() and pmap_page_index()
3662 1.1 gwr * agree as to whether holes in memory do or do not count as valid pages,
3663 1.1 gwr * it really doesn't matter. However, if you like to save a little
3664 1.1 gwr * memory, don't count holes as valid pages. This is even more true when
3665 1.1 gwr * the holes are large.
3666 1.1 gwr *
3667 1.8 gwr * We will not count holes as valid pages. We can generate page indices
3668 1.1 gwr * that conform to this by using the memory bank structures initialized
3669 1.1 gwr * in pmap_alloc_pv().
3670 1.1 gwr */
3671 1.1 gwr int
3672 1.1 gwr pmap_page_index(pa)
3673 1.1 gwr vm_offset_t pa;
3674 1.1 gwr {
3675 1.1 gwr struct pmap_physmem_struct *bank = avail_mem;
3676 1.1 gwr
3677 1.7 gwr /* Search for the memory bank with this page. */
3678 1.7 gwr /* XXX - What if it is not physical memory? */
3679 1.1 gwr while (pa > bank->pmem_end)
3680 1.1 gwr bank = bank->pmem_next;
3681 1.1 gwr pa -= bank->pmem_start;
3682 1.1 gwr
3683 1.16 gwr return (bank->pmem_pvbase + _btop(pa));
3684 1.1 gwr }
3685 1.1 gwr
3686 1.1 gwr /* pmap_next_page INTERFACE
3687 1.1 gwr **
3688 1.1 gwr * Place the physical address of the next available page in the
3689 1.1 gwr * argument given. Returns FALSE if there are no more pages left.
3690 1.1 gwr *
3691 1.1 gwr * This function must jump over any holes in physical memory.
3692 1.1 gwr * Once this function is used, any use of pmap_bootstrap_alloc()
3693 1.1 gwr * is a sin. Sinners will be punished with erratic behavior.
3694 1.1 gwr */
3695 1.1 gwr boolean_t
3696 1.1 gwr pmap_next_page(pa)
3697 1.1 gwr vm_offset_t *pa;
3698 1.1 gwr {
3699 1.1 gwr static struct pmap_physmem_struct *curbank = avail_mem;
3700 1.1 gwr
3701 1.8 gwr /* XXX - temporary ROM saving hack. */
3702 1.8 gwr if (avail_next >= avail_end)
3703 1.8 gwr return FALSE;
3704 1.8 gwr
3705 1.7 gwr if (avail_next >= curbank->pmem_end)
3706 1.1 gwr if (curbank->pmem_next == NULL)
3707 1.1 gwr return FALSE;
3708 1.1 gwr else {
3709 1.1 gwr curbank = curbank->pmem_next;
3710 1.7 gwr avail_next = curbank->pmem_start;
3711 1.1 gwr }
3712 1.1 gwr
3713 1.7 gwr *pa = avail_next;
3714 1.7 gwr avail_next += NBPG;
3715 1.1 gwr return TRUE;
3716 1.1 gwr }
3717 1.8 gwr
3718 1.8 gwr /* pmap_count INTERFACE
3719 1.8 gwr **
3720 1.8 gwr * Return the number of resident (valid) pages in the given pmap.
3721 1.8 gwr *
3722 1.8 gwr * Note: If this function is handed the kernel map, it will report
3723 1.8 gwr * that it has no mappings. Hopefully the VM system won't ask for kernel
3724 1.8 gwr * map statistics.
3725 1.8 gwr */
3726 1.8 gwr segsz_t
3727 1.8 gwr pmap_count(pmap, type)
3728 1.8 gwr pmap_t pmap;
3729 1.8 gwr int type;
3730 1.8 gwr {
3731 1.8 gwr u_int count;
3732 1.8 gwr int a_idx, b_idx;
3733 1.8 gwr a_tmgr_t *a_tbl;
3734 1.8 gwr b_tmgr_t *b_tbl;
3735 1.8 gwr c_tmgr_t *c_tbl;
3736 1.8 gwr
3737 1.8 gwr /*
3738 1.8 gwr * If the pmap does not have its own A table manager, it has no
3739 1.8 gwr * valid entires.
3740 1.8 gwr */
3741 1.8 gwr if (pmap->pm_a_tmgr == NULL)
3742 1.8 gwr return 0;
3743 1.8 gwr
3744 1.8 gwr a_tbl = pmap->pm_a_tmgr;
3745 1.8 gwr
3746 1.8 gwr count = 0;
3747 1.8 gwr for (a_idx = 0; a_idx < MMU_TIA(KERNBASE); a_idx++) {
3748 1.8 gwr if (MMU_VALID_DT(a_tbl->at_dtbl[a_idx])) {
3749 1.8 gwr b_tbl = mmuB2tmgr(mmu_ptov(a_tbl->at_dtbl[a_idx].addr.raw));
3750 1.8 gwr for (b_idx = 0; b_idx < MMU_B_TBL_SIZE; b_idx++) {
3751 1.8 gwr if (MMU_VALID_DT(b_tbl->bt_dtbl[b_idx])) {
3752 1.8 gwr c_tbl = mmuC2tmgr(
3753 1.8 gwr mmu_ptov(MMU_DTE_PA(b_tbl->bt_dtbl[b_idx])));
3754 1.8 gwr if (type == 0)
3755 1.8 gwr /*
3756 1.8 gwr * A resident entry count has been requested.
3757 1.8 gwr */
3758 1.8 gwr count += c_tbl->ct_ecnt;
3759 1.8 gwr else
3760 1.8 gwr /*
3761 1.8 gwr * A wired entry count has been requested.
3762 1.8 gwr */
3763 1.8 gwr count += c_tbl->ct_wcnt;
3764 1.8 gwr }
3765 1.8 gwr }
3766 1.8 gwr }
3767 1.8 gwr }
3768 1.8 gwr
3769 1.8 gwr return count;
3770 1.8 gwr }
3771 1.8 gwr
3772 1.1 gwr /************************ SUN3 COMPATIBILITY ROUTINES ********************
3773 1.1 gwr * The following routines are only used by DDB for tricky kernel text *
3774 1.1 gwr * text operations in db_memrw.c. They are provided for sun3 *
3775 1.1 gwr * compatibility. *
3776 1.1 gwr *************************************************************************/
3777 1.1 gwr /* get_pte INTERNAL
3778 1.1 gwr **
3779 1.1 gwr * Return the page descriptor the describes the kernel mapping
3780 1.1 gwr * of the given virtual address.
3781 1.1 gwr */
3782 1.14 gwr extern u_long ptest_addr __P((u_long)); /* XXX: locore.s */
3783 1.13 gwr u_long
3784 1.13 gwr get_pte(va)
3785 1.13 gwr vm_offset_t va;
3786 1.13 gwr {
3787 1.13 gwr u_long pte_pa;
3788 1.13 gwr mmu_short_pte_t *pte;
3789 1.13 gwr
3790 1.13 gwr /* Get the physical address of the PTE */
3791 1.13 gwr pte_pa = ptest_addr(va & ~PGOFSET);
3792 1.13 gwr
3793 1.13 gwr /* Convert to a virtual address... */
3794 1.13 gwr pte = (mmu_short_pte_t *) (KERNBASE + pte_pa);
3795 1.13 gwr
3796 1.13 gwr /* Make sure it is in our level-C tables... */
3797 1.13 gwr if ((pte < kernCbase) ||
3798 1.13 gwr (pte >= &mmuCbase[NUM_USER_PTES]))
3799 1.13 gwr return 0;
3800 1.13 gwr
3801 1.13 gwr /* ... and just return its contents. */
3802 1.13 gwr return (pte->attr.raw);
3803 1.13 gwr }
3804 1.13 gwr
3805 1.1 gwr
3806 1.1 gwr /* set_pte INTERNAL
3807 1.1 gwr **
3808 1.1 gwr * Set the page descriptor that describes the kernel mapping
3809 1.1 gwr * of the given virtual address.
3810 1.1 gwr */
3811 1.1 gwr void
3812 1.1 gwr set_pte(va, pte)
3813 1.1 gwr vm_offset_t va;
3814 1.1 gwr vm_offset_t pte;
3815 1.1 gwr {
3816 1.1 gwr u_long idx;
3817 1.1 gwr
3818 1.7 gwr if (va < KERNBASE)
3819 1.7 gwr return;
3820 1.7 gwr
3821 1.16 gwr idx = (unsigned long) _btop(va - KERNBASE);
3822 1.1 gwr kernCbase[idx].attr.raw = pte;
3823 1.1 gwr }
3824 1.7 gwr
3825 1.8 gwr #ifdef PMAP_DEBUG
3826 1.7 gwr /************************** DEBUGGING ROUTINES **************************
3827 1.7 gwr * The following routines are meant to be an aid to debugging the pmap *
3828 1.7 gwr * system. They are callable from the DDB command line and should be *
3829 1.7 gwr * prepared to be handed unstable or incomplete states of the system. *
3830 1.7 gwr ************************************************************************/
3831 1.7 gwr
3832 1.7 gwr /* pv_list
3833 1.7 gwr **
3834 1.7 gwr * List all pages found on the pv list for the given physical page.
3835 1.8 gwr * To avoid endless loops, the listing will stop at the end of the list
3836 1.7 gwr * or after 'n' entries - whichever comes first.
3837 1.7 gwr */
3838 1.7 gwr void
3839 1.7 gwr pv_list(pa, n)
3840 1.7 gwr vm_offset_t pa;
3841 1.7 gwr int n;
3842 1.7 gwr {
3843 1.7 gwr int idx;
3844 1.7 gwr vm_offset_t va;
3845 1.7 gwr pv_t *pv;
3846 1.7 gwr c_tmgr_t *c_tbl;
3847 1.7 gwr pmap_t pmap;
3848 1.7 gwr
3849 1.7 gwr pv = pa2pv(pa);
3850 1.7 gwr idx = pv->pv_idx;
3851 1.7 gwr
3852 1.7 gwr for (;idx != PVE_EOL && n > 0; idx=pvebase[idx].pve_next, n--) {
3853 1.8 gwr va = pmap_get_pteinfo(idx, &pmap, &c_tbl);
3854 1.7 gwr printf("idx %d, pmap 0x%x, va 0x%x, c_tbl %x\n",
3855 1.7 gwr idx, (u_int) pmap, (u_int) va, (u_int) c_tbl);
3856 1.7 gwr }
3857 1.7 gwr }
3858 1.8 gwr #endif /* PMAP_DEBUG */
3859 1.1 gwr
3860 1.1 gwr #ifdef NOT_YET
3861 1.1 gwr /* and maybe not ever */
3862 1.1 gwr /************************** LOW-LEVEL ROUTINES **************************
3863 1.1 gwr * These routines will eventualy be re-written into assembly and placed *
3864 1.1 gwr * in locore.s. They are here now as stubs so that the pmap module can *
3865 1.1 gwr * be linked as a standalone user program for testing. *
3866 1.1 gwr ************************************************************************/
3867 1.1 gwr /* flush_atc_crp INTERNAL
3868 1.1 gwr **
3869 1.1 gwr * Flush all page descriptors derived from the given CPU Root Pointer
3870 1.1 gwr * (CRP), or 'A' table as it is known here, from the 68851's automatic
3871 1.1 gwr * cache.
3872 1.1 gwr */
3873 1.1 gwr void
3874 1.1 gwr flush_atc_crp(a_tbl)
3875 1.1 gwr {
3876 1.1 gwr mmu_long_rp_t rp;
3877 1.1 gwr
3878 1.1 gwr /* Create a temporary root table pointer that points to the
3879 1.1 gwr * given A table.
3880 1.1 gwr */
3881 1.1 gwr rp.attr.raw = ~MMU_LONG_RP_LU;
3882 1.1 gwr rp.addr.raw = (unsigned int) a_tbl;
3883 1.1 gwr
3884 1.1 gwr mmu_pflushr(&rp);
3885 1.1 gwr /* mmu_pflushr:
3886 1.1 gwr * movel sp(4)@,a0
3887 1.1 gwr * pflushr a0@
3888 1.1 gwr * rts
3889 1.1 gwr */
3890 1.1 gwr }
3891 1.1 gwr #endif /* NOT_YET */
3892