source: mainline/kernel/generic/src/mm/as.c@ e32e092

lfn serial ticket/834-toolchain-update topic/msim-upgrade topic/simplify-dev-export
Last change on this file since e32e092 was e32e092, checked in by Jiri Svoboda <jirik.svoboda@…>, 17 years ago

Declare arguments for memstr.h operations as pointers instead of uintptr_t.

  • Property mode set to 100644
File size: 45.6 KB
RevLine 
[20d50a1]1/*
[df4ed85]2 * Copyright (c) 2001-2006 Jakub Jermar
[20d50a1]3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 *
9 * - Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * - Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * - The name of the author may not be used to endorse or promote products
15 * derived from this software without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 */
28
[cc73a8a1]29/** @addtogroup genericmm
[b45c443]30 * @{
31 */
32
[9179d0a]33/**
[b45c443]34 * @file
[9179d0a]35 * @brief Address space related functions.
36 *
[20d50a1]37 * This file contains address space manipulation functions.
38 * Roughly speaking, this is a higher-level client of
39 * Virtual Address Translation (VAT) subsystem.
[9179d0a]40 *
41 * Functionality provided by this file allows one to
[cc73a8a1]42 * create address spaces and create, resize and share
[9179d0a]43 * address space areas.
44 *
45 * @see page.c
46 *
[20d50a1]47 */
48
49#include <mm/as.h>
[ef67bab]50#include <arch/mm/as.h>
[20d50a1]51#include <mm/page.h>
52#include <mm/frame.h>
[085d973]53#include <mm/slab.h>
[20d50a1]54#include <mm/tlb.h>
55#include <arch/mm/page.h>
56#include <genarch/mm/page_pt.h>
[2802767]57#include <genarch/mm/page_ht.h>
[4512d7e]58#include <mm/asid.h>
[20d50a1]59#include <arch/mm/asid.h>
[31d8e10]60#include <preemption.h>
[20d50a1]61#include <synch/spinlock.h>
[1068f6a]62#include <synch/mutex.h>
[5c9a08b]63#include <adt/list.h>
[252127e]64#include <adt/btree.h>
[df0103f7]65#include <proc/task.h>
[e3c762cd]66#include <proc/thread.h>
[20d50a1]67#include <arch/asm.h>
[df0103f7]68#include <panic.h>
[20d50a1]69#include <debug.h>
[df0103f7]70#include <print.h>
[20d50a1]71#include <memstr.h>
[5a7d9d1]72#include <macros.h>
[20d50a1]73#include <arch.h>
[df0103f7]74#include <errno.h>
75#include <config.h>
[25bf215]76#include <align.h>
[df0103f7]77#include <arch/types.h>
[e3c762cd]78#include <syscall/copy.h>
79#include <arch/interrupt.h>
[20d50a1]80
[92778f2]81#ifdef CONFIG_VIRT_IDX_DCACHE
82#include <arch/mm/cache.h>
83#endif /* CONFIG_VIRT_IDX_DCACHE */
84
[bd1deed]85#ifndef __OBJC__
[cc73a8a1]86/**
87 * Each architecture decides what functions will be used to carry out
88 * address space operations such as creating or locking page tables.
89 */
[ef67bab]90as_operations_t *as_operations = NULL;
[20d50a1]91
[57da95c]92/**
93 * Slab for as_t objects.
94 */
95static slab_cache_t *as_slab;
[c993e45]96#endif
[57da95c]97
[6f4495f5]98/**
[879585a3]99 * This lock serializes access to the ASID subsystem.
100 * It protects:
101 * - inactive_as_with_asid_head list
102 * - as->asid for each as of the as_t type
103 * - asids_allocated counter
[6f4495f5]104 */
[879585a3]105SPINLOCK_INITIALIZE(asidlock);
[7e4e532]106
107/**
108 * This list contains address spaces that are not active on any
109 * processor and that have valid ASID.
110 */
111LIST_INITIALIZE(inactive_as_with_asid_head);
112
[071a8ae6]113/** Kernel address space. */
114as_t *AS_KERNEL = NULL;
115
[df0103f7]116static int area_flags_to_page_flags(int aflags);
[7f1c620]117static as_area_t *find_area_and_lock(as_t *as, uintptr_t va);
[6f4495f5]118static bool check_area_conflicts(as_t *as, uintptr_t va, size_t size,
119 as_area_t *avoid_area);
[8182031]120static void sh_info_remove_reference(share_info_t *sh_info);
[20d50a1]121
[c993e45]122#ifndef __OBJC__
[29b2bbf]123static int as_constructor(void *obj, int flags)
124{
125 as_t *as = (as_t *) obj;
126 int rc;
127
128 link_initialize(&as->inactive_as_with_asid_link);
129 mutex_initialize(&as->lock);
130
131 rc = as_constructor_arch(as, flags);
132
133 return rc;
134}
135
136static int as_destructor(void *obj)
137{
138 as_t *as = (as_t *) obj;
139
140 return as_destructor_arch(as);
141}
[c993e45]142#endif
[29b2bbf]143
[ef67bab]144/** Initialize address space subsystem. */
145void as_init(void)
146{
147 as_arch_init();
[c993e45]148
149#ifndef __OBJC__
[29b2bbf]150 as_slab = slab_cache_create("as_slab", sizeof(as_t), 0,
[6f4495f5]151 as_constructor, as_destructor, SLAB_CACHE_MAGDEFERRED);
[c993e45]152#endif
[57da95c]153
[8e1ea655]154 AS_KERNEL = as_create(FLAG_AS_KERNEL);
[125e944]155 if (!AS_KERNEL)
156 panic("can't create kernel address space\n");
157
[ef67bab]158}
159
[071a8ae6]160/** Create address space.
161 *
162 * @param flags Flags that influence way in wich the address space is created.
163 */
[ef67bab]164as_t *as_create(int flags)
[20d50a1]165{
166 as_t *as;
167
[c993e45]168#ifdef __OBJC__
169 as = [as_t new];
170 link_initialize(&as->inactive_as_with_asid_link);
171 mutex_initialize(&as->lock);
172 (void) as_constructor_arch(as, flags);
173#else
[57da95c]174 as = (as_t *) slab_alloc(as_slab, 0);
[c993e45]175#endif
[29b2bbf]176 (void) as_create_arch(as, 0);
177
[252127e]178 btree_create(&as->as_area_btree);
[bb68433]179
180 if (flags & FLAG_AS_KERNEL)
181 as->asid = ASID_KERNEL;
182 else
183 as->asid = ASID_INVALID;
184
[31d8e10]185 atomic_set(&as->refcount, 0);
[47800e0]186 as->cpu_refcount = 0;
[b3f8fb7]187#ifdef AS_PAGE_TABLE
[80bcaed]188 as->genarch.page_table = page_table_create(flags);
[b3f8fb7]189#else
190 page_table_create(flags);
191#endif
[20d50a1]192
193 return as;
194}
195
[482826d]196/** Destroy adress space.
197 *
[6f4495f5]198 * When there are no tasks referencing this address space (i.e. its refcount is
199 * zero), the address space can be destroyed.
[31d8e10]200 *
201 * We know that we don't hold any spinlock.
[482826d]202 */
203void as_destroy(as_t *as)
[5be1923]204{
[482826d]205 ipl_t ipl;
[6f9a9bc]206 bool cond;
[31d8e10]207 DEADLOCK_PROBE_INIT(p_asidlock);
[482826d]208
[31d8e10]209 ASSERT(atomic_get(&as->refcount) == 0);
[482826d]210
211 /*
212 * Since there is no reference to this area,
213 * it is safe not to lock its mutex.
214 */
[879585a3]215
[31d8e10]216 /*
217 * We need to avoid deadlock between TLB shootdown and asidlock.
218 * We therefore try to take asid conditionally and if we don't succeed,
219 * we enable interrupts and try again. This is done while preemption is
220 * disabled to prevent nested context switches. We also depend on the
221 * fact that so far no spinlocks are held.
222 */
223 preemption_disable();
224 ipl = interrupts_read();
225retry:
226 interrupts_disable();
227 if (!spinlock_trylock(&asidlock)) {
228 interrupts_enable();
229 DEADLOCK_PROBE(p_asidlock, DEADLOCK_THRESHOLD);
230 goto retry;
231 }
232 preemption_enable(); /* Interrupts disabled, enable preemption */
[31e8ddd]233 if (as->asid != ASID_INVALID && as != AS_KERNEL) {
[6f9a9bc]234 if (as != AS && as->cpu_refcount == 0)
[31e8ddd]235 list_remove(&as->inactive_as_with_asid_link);
[482826d]236 asid_put(as->asid);
237 }
[879585a3]238 spinlock_unlock(&asidlock);
[482826d]239
240 /*
241 * Destroy address space areas of the address space.
[8440473]242 * The B+tree must be walked carefully because it is
[6f9a9bc]243 * also being destroyed.
[482826d]244 */
[6f9a9bc]245 for (cond = true; cond; ) {
[482826d]246 btree_node_t *node;
[6f9a9bc]247
248 ASSERT(!list_empty(&as->as_area_btree.leaf_head));
[6f4495f5]249 node = list_get_instance(as->as_area_btree.leaf_head.next,
250 btree_node_t, leaf_link);
[6f9a9bc]251
252 if ((cond = node->keys)) {
253 as_area_destroy(as, node->key[0]);
254 }
[482826d]255 }
[f8d069e8]256
[152b2b0]257 btree_destroy(&as->as_area_btree);
[b3f8fb7]258#ifdef AS_PAGE_TABLE
[80bcaed]259 page_table_destroy(as->genarch.page_table);
[b3f8fb7]260#else
261 page_table_destroy(NULL);
262#endif
[5be1923]263
[482826d]264 interrupts_restore(ipl);
[c993e45]265
266#ifdef __OBJC__
267 [as free];
268#else
[57da95c]269 slab_free(as_slab, as);
[c993e45]270#endif
[5be1923]271}
272
[20d50a1]273/** Create address space area of common attributes.
274 *
275 * The created address space area is added to the target address space.
276 *
277 * @param as Target address space.
[a9e8b39]278 * @param flags Flags of the area memory.
[37e7d2b9]279 * @param size Size of area.
[20d50a1]280 * @param base Base address of area.
[a9e8b39]281 * @param attrs Attributes of the area.
[8182031]282 * @param backend Address space area backend. NULL if no backend is used.
283 * @param backend_data NULL or a pointer to an array holding two void *.
[20d50a1]284 *
285 * @return Address space area on success or NULL on failure.
286 */
[c738d65]287as_area_t *
288as_area_create(as_t *as, int flags, size_t size, uintptr_t base, int attrs,
[0ee077ee]289 mem_backend_t *backend, mem_backend_data_t *backend_data)
[20d50a1]290{
291 ipl_t ipl;
292 as_area_t *a;
293
294 if (base % PAGE_SIZE)
[37e7d2b9]295 return NULL;
296
[dbbeb26]297 if (!size)
298 return NULL;
299
[37e7d2b9]300 /* Writeable executable areas are not supported. */
301 if ((flags & AS_AREA_EXEC) && (flags & AS_AREA_WRITE))
302 return NULL;
[20d50a1]303
304 ipl = interrupts_disable();
[1068f6a]305 mutex_lock(&as->lock);
[20d50a1]306
[37e7d2b9]307 if (!check_area_conflicts(as, base, size, NULL)) {
[1068f6a]308 mutex_unlock(&as->lock);
[37e7d2b9]309 interrupts_restore(ipl);
310 return NULL;
311 }
[20d50a1]312
[bb68433]313 a = (as_area_t *) malloc(sizeof(as_area_t), 0);
314
[1068f6a]315 mutex_initialize(&a->lock);
[bb68433]316
[0ee077ee]317 a->as = as;
[c23502d]318 a->flags = flags;
[a9e8b39]319 a->attributes = attrs;
[37e7d2b9]320 a->pages = SIZE2FRAMES(size);
[bb68433]321 a->base = base;
[8182031]322 a->sh_info = NULL;
323 a->backend = backend;
[0ee077ee]324 if (backend_data)
325 a->backend_data = *backend_data;
326 else
[e32e092]327 memsetb(&a->backend_data, sizeof(a->backend_data), 0);
[0ee077ee]328
[25bf215]329 btree_create(&a->used_space);
[bb68433]330
[252127e]331 btree_insert(&as->as_area_btree, base, (void *) a, NULL);
[20d50a1]332
[1068f6a]333 mutex_unlock(&as->lock);
[20d50a1]334 interrupts_restore(ipl);
[f9425006]335
[20d50a1]336 return a;
337}
338
[df0103f7]339/** Find address space area and change it.
340 *
341 * @param as Address space.
[6f4495f5]342 * @param address Virtual address belonging to the area to be changed. Must be
343 * page-aligned.
[df0103f7]344 * @param size New size of the virtual memory block starting at address.
345 * @param flags Flags influencing the remap operation. Currently unused.
346 *
[7242a78e]347 * @return Zero on success or a value from @ref errno.h otherwise.
[df0103f7]348 */
[7f1c620]349int as_area_resize(as_t *as, uintptr_t address, size_t size, int flags)
[df0103f7]350{
[7242a78e]351 as_area_t *area;
[df0103f7]352 ipl_t ipl;
353 size_t pages;
354
355 ipl = interrupts_disable();
[1068f6a]356 mutex_lock(&as->lock);
[df0103f7]357
358 /*
359 * Locate the area.
360 */
361 area = find_area_and_lock(as, address);
362 if (!area) {
[1068f6a]363 mutex_unlock(&as->lock);
[df0103f7]364 interrupts_restore(ipl);
[7242a78e]365 return ENOENT;
[df0103f7]366 }
367
[0ee077ee]368 if (area->backend == &phys_backend) {
[df0103f7]369 /*
370 * Remapping of address space areas associated
371 * with memory mapped devices is not supported.
372 */
[1068f6a]373 mutex_unlock(&area->lock);
374 mutex_unlock(&as->lock);
[df0103f7]375 interrupts_restore(ipl);
[7242a78e]376 return ENOTSUP;
[df0103f7]377 }
[8182031]378 if (area->sh_info) {
379 /*
380 * Remapping of shared address space areas
381 * is not supported.
382 */
383 mutex_unlock(&area->lock);
384 mutex_unlock(&as->lock);
385 interrupts_restore(ipl);
386 return ENOTSUP;
387 }
[df0103f7]388
389 pages = SIZE2FRAMES((address - area->base) + size);
390 if (!pages) {
391 /*
392 * Zero size address space areas are not allowed.
393 */
[1068f6a]394 mutex_unlock(&area->lock);
395 mutex_unlock(&as->lock);
[df0103f7]396 interrupts_restore(ipl);
[7242a78e]397 return EPERM;
[df0103f7]398 }
399
400 if (pages < area->pages) {
[56789125]401 bool cond;
[7f1c620]402 uintptr_t start_free = area->base + pages*PAGE_SIZE;
[df0103f7]403
404 /*
405 * Shrinking the area.
406 * No need to check for overlaps.
407 */
408
[5552d60]409 /*
410 * Start TLB shootdown sequence.
411 */
[6f4495f5]412 tlb_shootdown_start(TLB_INVL_PAGES, AS->asid, area->base +
413 pages * PAGE_SIZE, area->pages - pages);
[5552d60]414
[56789125]415 /*
416 * Remove frames belonging to used space starting from
417 * the highest addresses downwards until an overlap with
418 * the resized address space area is found. Note that this
419 * is also the right way to remove part of the used_space
420 * B+tree leaf list.
421 */
422 for (cond = true; cond;) {
423 btree_node_t *node;
424
425 ASSERT(!list_empty(&area->used_space.leaf_head));
[6f4495f5]426 node =
427 list_get_instance(area->used_space.leaf_head.prev,
428 btree_node_t, leaf_link);
[56789125]429 if ((cond = (bool) node->keys)) {
[7f1c620]430 uintptr_t b = node->key[node->keys - 1];
[6f4495f5]431 count_t c =
432 (count_t) node->value[node->keys - 1];
[6c441cf8]433 unsigned int i = 0;
[56789125]434
[6f4495f5]435 if (overlaps(b, c * PAGE_SIZE, area->base,
[4638401]436 pages * PAGE_SIZE)) {
[56789125]437
[6f4495f5]438 if (b + c * PAGE_SIZE <= start_free) {
[56789125]439 /*
[6f4495f5]440 * The whole interval fits
441 * completely in the resized
442 * address space area.
[56789125]443 */
444 break;
445 }
446
447 /*
[6f4495f5]448 * Part of the interval corresponding
449 * to b and c overlaps with the resized
450 * address space area.
[56789125]451 */
452
453 cond = false; /* we are almost done */
454 i = (start_free - b) >> PAGE_WIDTH;
[2b8b0ca]455 if (!used_space_remove(area, start_free, c - i))
456 panic("Could not remove used space.\n");
[56789125]457 } else {
458 /*
[6f4495f5]459 * The interval of used space can be
460 * completely removed.
[56789125]461 */
462 if (!used_space_remove(area, b, c))
[2b8b0ca]463 panic("Could not remove used space.\n");
[56789125]464 }
465
466 for (; i < c; i++) {
467 pte_t *pte;
468
469 page_table_lock(as, false);
[6f4495f5]470 pte = page_mapping_find(as, b +
471 i * PAGE_SIZE);
472 ASSERT(pte && PTE_VALID(pte) &&
473 PTE_PRESENT(pte));
474 if (area->backend &&
475 area->backend->frame_free) {
[0ee077ee]476 area->backend->frame_free(area,
[6f4495f5]477 b + i * PAGE_SIZE,
478 PTE_GET_FRAME(pte));
[8182031]479 }
[6f4495f5]480 page_mapping_remove(as, b +
481 i * PAGE_SIZE);
[56789125]482 page_table_unlock(as, false);
483 }
[df0103f7]484 }
485 }
[5552d60]486
[df0103f7]487 /*
[5552d60]488 * Finish TLB shootdown sequence.
[df0103f7]489 */
[31d8e10]490
[6f4495f5]491 tlb_invalidate_pages(as->asid, area->base + pages * PAGE_SIZE,
492 area->pages - pages);
[f1d1f5d3]493 /*
494 * Invalidate software translation caches (e.g. TSB on sparc64).
495 */
[6f4495f5]496 as_invalidate_translation_cache(as, area->base +
497 pages * PAGE_SIZE, area->pages - pages);
[31d8e10]498 tlb_shootdown_finalize();
499
[df0103f7]500 } else {
501 /*
502 * Growing the area.
503 * Check for overlaps with other address space areas.
504 */
[6f4495f5]505 if (!check_area_conflicts(as, address, pages * PAGE_SIZE,
506 area)) {
[1068f6a]507 mutex_unlock(&area->lock);
508 mutex_unlock(&as->lock);
[df0103f7]509 interrupts_restore(ipl);
[7242a78e]510 return EADDRNOTAVAIL;
[df0103f7]511 }
512 }
513
514 area->pages = pages;
515
[1068f6a]516 mutex_unlock(&area->lock);
517 mutex_unlock(&as->lock);
[df0103f7]518 interrupts_restore(ipl);
519
[7242a78e]520 return 0;
521}
522
523/** Destroy address space area.
524 *
525 * @param as Address space.
526 * @param address Address withing the area to be deleted.
527 *
528 * @return Zero on success or a value from @ref errno.h on failure.
529 */
[7f1c620]530int as_area_destroy(as_t *as, uintptr_t address)
[7242a78e]531{
532 as_area_t *area;
[7f1c620]533 uintptr_t base;
[f8d069e8]534 link_t *cur;
[7242a78e]535 ipl_t ipl;
536
537 ipl = interrupts_disable();
[1068f6a]538 mutex_lock(&as->lock);
[7242a78e]539
540 area = find_area_and_lock(as, address);
541 if (!area) {
[1068f6a]542 mutex_unlock(&as->lock);
[7242a78e]543 interrupts_restore(ipl);
544 return ENOENT;
545 }
546
[56789125]547 base = area->base;
548
[5552d60]549 /*
550 * Start TLB shootdown sequence.
551 */
[f1d1f5d3]552 tlb_shootdown_start(TLB_INVL_PAGES, as->asid, area->base, area->pages);
[5552d60]553
[567807b1]554 /*
555 * Visit only the pages mapped by used_space B+tree.
556 */
[6f4495f5]557 for (cur = area->used_space.leaf_head.next;
558 cur != &area->used_space.leaf_head; cur = cur->next) {
[567807b1]559 btree_node_t *node;
[6c441cf8]560 unsigned int i;
[56789125]561
[f8d069e8]562 node = list_get_instance(cur, btree_node_t, leaf_link);
563 for (i = 0; i < node->keys; i++) {
[7f1c620]564 uintptr_t b = node->key[i];
[f8d069e8]565 count_t j;
[567807b1]566 pte_t *pte;
[56789125]567
[f8d069e8]568 for (j = 0; j < (count_t) node->value[i]; j++) {
[567807b1]569 page_table_lock(as, false);
[6f4495f5]570 pte = page_mapping_find(as, b + j * PAGE_SIZE);
571 ASSERT(pte && PTE_VALID(pte) &&
572 PTE_PRESENT(pte));
573 if (area->backend &&
574 area->backend->frame_free) {
575 area->backend->frame_free(area, b +
[4638401]576 j * PAGE_SIZE, PTE_GET_FRAME(pte));
[56789125]577 }
[6f4495f5]578 page_mapping_remove(as, b + j * PAGE_SIZE);
[567807b1]579 page_table_unlock(as, false);
[7242a78e]580 }
581 }
582 }
[56789125]583
[7242a78e]584 /*
[5552d60]585 * Finish TLB shootdown sequence.
[7242a78e]586 */
[31d8e10]587
[f1d1f5d3]588 tlb_invalidate_pages(as->asid, area->base, area->pages);
589 /*
[6f4495f5]590 * Invalidate potential software translation caches (e.g. TSB on
591 * sparc64).
[f1d1f5d3]592 */
593 as_invalidate_translation_cache(as, area->base, area->pages);
[31d8e10]594 tlb_shootdown_finalize();
[f1d1f5d3]595
[5552d60]596 btree_destroy(&area->used_space);
[7242a78e]597
[8d4f2ae]598 area->attributes |= AS_AREA_ATTR_PARTIAL;
[8182031]599
600 if (area->sh_info)
601 sh_info_remove_reference(area->sh_info);
602
[1068f6a]603 mutex_unlock(&area->lock);
[7242a78e]604
605 /*
606 * Remove the empty area from address space.
607 */
[f1d1f5d3]608 btree_remove(&as->as_area_btree, base, NULL);
[7242a78e]609
[8d4f2ae]610 free(area);
611
[f1d1f5d3]612 mutex_unlock(&as->lock);
[7242a78e]613 interrupts_restore(ipl);
614 return 0;
[df0103f7]615}
616
[8d6bc2d5]617/** Share address space area with another or the same address space.
[df0103f7]618 *
[0ee077ee]619 * Address space area mapping is shared with a new address space area.
620 * If the source address space area has not been shared so far,
621 * a new sh_info is created. The new address space area simply gets the
622 * sh_info of the source area. The process of duplicating the
623 * mapping is done through the backend share function.
[8d6bc2d5]624 *
[fd4d8c0]625 * @param src_as Pointer to source address space.
[a9e8b39]626 * @param src_base Base address of the source address space area.
[fd4d8c0]627 * @param acc_size Expected size of the source area.
[46fc2f9]628 * @param dst_as Pointer to destination address space.
[fd4d8c0]629 * @param dst_base Target base address.
630 * @param dst_flags_mask Destination address space area flags mask.
[df0103f7]631 *
[d0485c6]632 * @return Zero on success or ENOENT if there is no such task or if there is no
633 * such address space area, EPERM if there was a problem in accepting the area
634 * or ENOMEM if there was a problem in allocating destination address space
635 * area. ENOTSUP is returned if the address space area backend does not support
[2057572]636 * sharing.
[df0103f7]637 */
[7f1c620]638int as_area_share(as_t *src_as, uintptr_t src_base, size_t acc_size,
[c0697c4c]639 as_t *dst_as, uintptr_t dst_base, int dst_flags_mask)
[df0103f7]640{
641 ipl_t ipl;
[a9e8b39]642 int src_flags;
643 size_t src_size;
644 as_area_t *src_area, *dst_area;
[8d6bc2d5]645 share_info_t *sh_info;
[0ee077ee]646 mem_backend_t *src_backend;
647 mem_backend_data_t src_backend_data;
[d6e5cbc]648
[7c23af9]649 ipl = interrupts_disable();
[1068f6a]650 mutex_lock(&src_as->lock);
[7c23af9]651 src_area = find_area_and_lock(src_as, src_base);
[a9e8b39]652 if (!src_area) {
[6fa476f7]653 /*
654 * Could not find the source address space area.
655 */
[1068f6a]656 mutex_unlock(&src_as->lock);
[6fa476f7]657 interrupts_restore(ipl);
658 return ENOENT;
659 }
[d0485c6]660
[0ee077ee]661 if (!src_area->backend || !src_area->backend->share) {
[8d6bc2d5]662 /*
[f47fd19]663 * There is no backend or the backend does not
[0ee077ee]664 * know how to share the area.
[8d6bc2d5]665 */
666 mutex_unlock(&src_area->lock);
667 mutex_unlock(&src_as->lock);
668 interrupts_restore(ipl);
669 return ENOTSUP;
670 }
671
[a9e8b39]672 src_size = src_area->pages * PAGE_SIZE;
673 src_flags = src_area->flags;
[0ee077ee]674 src_backend = src_area->backend;
675 src_backend_data = src_area->backend_data;
[1ec1fd8]676
677 /* Share the cacheable flag from the original mapping */
678 if (src_flags & AS_AREA_CACHEABLE)
679 dst_flags_mask |= AS_AREA_CACHEABLE;
680
[6f4495f5]681 if (src_size != acc_size ||
682 (src_flags & dst_flags_mask) != dst_flags_mask) {
[8d6bc2d5]683 mutex_unlock(&src_area->lock);
684 mutex_unlock(&src_as->lock);
[df0103f7]685 interrupts_restore(ipl);
686 return EPERM;
687 }
[8d6bc2d5]688
689 /*
690 * Now we are committed to sharing the area.
[8440473]691 * First, prepare the area for sharing.
[8d6bc2d5]692 * Then it will be safe to unlock it.
693 */
694 sh_info = src_area->sh_info;
695 if (!sh_info) {
696 sh_info = (share_info_t *) malloc(sizeof(share_info_t), 0);
697 mutex_initialize(&sh_info->lock);
698 sh_info->refcount = 2;
699 btree_create(&sh_info->pagemap);
700 src_area->sh_info = sh_info;
[c0697c4c]701 /*
702 * Call the backend to setup sharing.
703 */
704 src_area->backend->share(src_area);
[8d6bc2d5]705 } else {
706 mutex_lock(&sh_info->lock);
707 sh_info->refcount++;
708 mutex_unlock(&sh_info->lock);
709 }
710
711 mutex_unlock(&src_area->lock);
712 mutex_unlock(&src_as->lock);
713
[df0103f7]714 /*
[a9e8b39]715 * Create copy of the source address space area.
716 * The destination area is created with AS_AREA_ATTR_PARTIAL
717 * attribute set which prevents race condition with
718 * preliminary as_page_fault() calls.
[fd4d8c0]719 * The flags of the source area are masked against dst_flags_mask
720 * to support sharing in less privileged mode.
[df0103f7]721 */
[76d7305]722 dst_area = as_area_create(dst_as, dst_flags_mask, src_size, dst_base,
[6f4495f5]723 AS_AREA_ATTR_PARTIAL, src_backend, &src_backend_data);
[a9e8b39]724 if (!dst_area) {
[df0103f7]725 /*
726 * Destination address space area could not be created.
727 */
[8d6bc2d5]728 sh_info_remove_reference(sh_info);
729
[df0103f7]730 interrupts_restore(ipl);
731 return ENOMEM;
732 }
[92778f2]733
[a9e8b39]734 /*
735 * Now the destination address space area has been
736 * fully initialized. Clear the AS_AREA_ATTR_PARTIAL
[8d6bc2d5]737 * attribute and set the sh_info.
[a9e8b39]738 */
[92778f2]739 mutex_lock(&dst_as->lock);
[1068f6a]740 mutex_lock(&dst_area->lock);
[a9e8b39]741 dst_area->attributes &= ~AS_AREA_ATTR_PARTIAL;
[8d6bc2d5]742 dst_area->sh_info = sh_info;
[1068f6a]743 mutex_unlock(&dst_area->lock);
[92778f2]744 mutex_unlock(&dst_as->lock);
745
[df0103f7]746 interrupts_restore(ipl);
747
748 return 0;
749}
750
[fb84455]751/** Check access mode for address space area.
752 *
753 * The address space area must be locked prior to this call.
754 *
755 * @param area Address space area.
756 * @param access Access mode.
757 *
758 * @return False if access violates area's permissions, true otherwise.
759 */
760bool as_area_check_access(as_area_t *area, pf_access_t access)
761{
762 int flagmap[] = {
763 [PF_ACCESS_READ] = AS_AREA_READ,
764 [PF_ACCESS_WRITE] = AS_AREA_WRITE,
765 [PF_ACCESS_EXEC] = AS_AREA_EXEC
766 };
767
768 if (!(area->flags & flagmap[access]))
769 return false;
770
771 return true;
772}
773
[20d50a1]774/** Handle page fault within the current address space.
775 *
[8182031]776 * This is the high-level page fault handler. It decides
777 * whether the page fault can be resolved by any backend
778 * and if so, it invokes the backend to resolve the page
779 * fault.
780 *
[20d50a1]781 * Interrupts are assumed disabled.
782 *
783 * @param page Faulting page.
[567807b1]784 * @param access Access mode that caused the fault (i.e. read/write/exec).
[e3c762cd]785 * @param istate Pointer to interrupted state.
[20d50a1]786 *
[8182031]787 * @return AS_PF_FAULT on page fault, AS_PF_OK on success or AS_PF_DEFER if the
788 * fault was caused by copy_to_uspace() or copy_from_uspace().
[20d50a1]789 */
[7f1c620]790int as_page_fault(uintptr_t page, pf_access_t access, istate_t *istate)
[20d50a1]791{
[2299914]792 pte_t *pte;
[d3e7ff4]793 as_area_t *area;
[20d50a1]794
[1068f6a]795 if (!THREAD)
[8182031]796 return AS_PF_FAULT;
[1068f6a]797
[20d50a1]798 ASSERT(AS);
[2299914]799
[1068f6a]800 mutex_lock(&AS->lock);
[d3e7ff4]801 area = find_area_and_lock(AS, page);
[20d50a1]802 if (!area) {
803 /*
804 * No area contained mapping for 'page'.
805 * Signal page fault to low-level handler.
806 */
[1068f6a]807 mutex_unlock(&AS->lock);
[e3c762cd]808 goto page_fault;
[20d50a1]809 }
810
[a9e8b39]811 if (area->attributes & AS_AREA_ATTR_PARTIAL) {
812 /*
813 * The address space area is not fully initialized.
814 * Avoid possible race by returning error.
815 */
[1068f6a]816 mutex_unlock(&area->lock);
817 mutex_unlock(&AS->lock);
[e3c762cd]818 goto page_fault;
[a9e8b39]819 }
820
[0ee077ee]821 if (!area->backend || !area->backend->page_fault) {
[8182031]822 /*
823 * The address space area is not backed by any backend
824 * or the backend cannot handle page faults.
825 */
826 mutex_unlock(&area->lock);
827 mutex_unlock(&AS->lock);
828 goto page_fault;
829 }
[1ace9ea]830
[2299914]831 page_table_lock(AS, false);
832
833 /*
834 * To avoid race condition between two page faults
835 * on the same address, we need to make sure
836 * the mapping has not been already inserted.
837 */
838 if ((pte = page_mapping_find(AS, page))) {
839 if (PTE_PRESENT(pte)) {
[fb84455]840 if (((access == PF_ACCESS_READ) && PTE_READABLE(pte)) ||
[6f4495f5]841 (access == PF_ACCESS_WRITE && PTE_WRITABLE(pte)) ||
842 (access == PF_ACCESS_EXEC && PTE_EXECUTABLE(pte))) {
[fb84455]843 page_table_unlock(AS, false);
844 mutex_unlock(&area->lock);
845 mutex_unlock(&AS->lock);
846 return AS_PF_OK;
847 }
[2299914]848 }
849 }
[20d50a1]850
851 /*
[8182031]852 * Resort to the backend page fault handler.
[20d50a1]853 */
[0ee077ee]854 if (area->backend->page_fault(area, page, access) != AS_PF_OK) {
[8182031]855 page_table_unlock(AS, false);
856 mutex_unlock(&area->lock);
857 mutex_unlock(&AS->lock);
858 goto page_fault;
859 }
[20d50a1]860
[8182031]861 page_table_unlock(AS, false);
[1068f6a]862 mutex_unlock(&area->lock);
863 mutex_unlock(&AS->lock);
[e3c762cd]864 return AS_PF_OK;
865
866page_fault:
867 if (THREAD->in_copy_from_uspace) {
868 THREAD->in_copy_from_uspace = false;
[6f4495f5]869 istate_set_retaddr(istate,
870 (uintptr_t) &memcpy_from_uspace_failover_address);
[e3c762cd]871 } else if (THREAD->in_copy_to_uspace) {
872 THREAD->in_copy_to_uspace = false;
[6f4495f5]873 istate_set_retaddr(istate,
874 (uintptr_t) &memcpy_to_uspace_failover_address);
[e3c762cd]875 } else {
876 return AS_PF_FAULT;
877 }
878
879 return AS_PF_DEFER;
[20d50a1]880}
881
[7e4e532]882/** Switch address spaces.
[1068f6a]883 *
884 * Note that this function cannot sleep as it is essentially a part of
[879585a3]885 * scheduling. Sleeping here would lead to deadlock on wakeup. Another
886 * thing which is forbidden in this context is locking the address space.
[20d50a1]887 *
[31d8e10]888 * When this function is enetered, no spinlocks may be held.
889 *
[7e4e532]890 * @param old Old address space or NULL.
891 * @param new New address space.
[20d50a1]892 */
[80bcaed]893void as_switch(as_t *old_as, as_t *new_as)
[20d50a1]894{
[31d8e10]895 DEADLOCK_PROBE_INIT(p_asidlock);
896 preemption_disable();
897retry:
898 (void) interrupts_disable();
899 if (!spinlock_trylock(&asidlock)) {
900 /*
901 * Avoid deadlock with TLB shootdown.
902 * We can enable interrupts here because
903 * preemption is disabled. We should not be
904 * holding any other lock.
905 */
906 (void) interrupts_enable();
907 DEADLOCK_PROBE(p_asidlock, DEADLOCK_THRESHOLD);
908 goto retry;
909 }
910 preemption_enable();
[7e4e532]911
912 /*
913 * First, take care of the old address space.
914 */
[80bcaed]915 if (old_as) {
916 ASSERT(old_as->cpu_refcount);
917 if((--old_as->cpu_refcount == 0) && (old_as != AS_KERNEL)) {
[7e4e532]918 /*
919 * The old address space is no longer active on
920 * any processor. It can be appended to the
921 * list of inactive address spaces with assigned
922 * ASID.
923 */
[2057572]924 ASSERT(old_as->asid != ASID_INVALID);
925 list_append(&old_as->inactive_as_with_asid_link,
926 &inactive_as_with_asid_head);
[7e4e532]927 }
[57da95c]928
929 /*
930 * Perform architecture-specific tasks when the address space
931 * is being removed from the CPU.
932 */
[80bcaed]933 as_deinstall_arch(old_as);
[7e4e532]934 }
935
936 /*
937 * Second, prepare the new address space.
938 */
[80bcaed]939 if ((new_as->cpu_refcount++ == 0) && (new_as != AS_KERNEL)) {
[879585a3]940 if (new_as->asid != ASID_INVALID)
[80bcaed]941 list_remove(&new_as->inactive_as_with_asid_link);
[879585a3]942 else
943 new_as->asid = asid_get();
[7e4e532]944 }
[80bcaed]945#ifdef AS_PAGE_TABLE
946 SET_PTL0_ADDRESS(new_as->genarch.page_table);
947#endif
[7e4e532]948
[20d50a1]949 /*
950 * Perform architecture-specific steps.
[4512d7e]951 * (e.g. write ASID to hardware register etc.)
[20d50a1]952 */
[80bcaed]953 as_install_arch(new_as);
[879585a3]954
955 spinlock_unlock(&asidlock);
[20d50a1]956
[80bcaed]957 AS = new_as;
[20d50a1]958}
[6a3c9a7]959
[df0103f7]960/** Convert address space area flags to page flags.
[6a3c9a7]961 *
[df0103f7]962 * @param aflags Flags of some address space area.
[6a3c9a7]963 *
[df0103f7]964 * @return Flags to be passed to page_mapping_insert().
[6a3c9a7]965 */
[df0103f7]966int area_flags_to_page_flags(int aflags)
[6a3c9a7]967{
968 int flags;
969
[9a8d91b]970 flags = PAGE_USER | PAGE_PRESENT;
[c23502d]971
[df0103f7]972 if (aflags & AS_AREA_READ)
[c23502d]973 flags |= PAGE_READ;
974
[df0103f7]975 if (aflags & AS_AREA_WRITE)
[c23502d]976 flags |= PAGE_WRITE;
977
[df0103f7]978 if (aflags & AS_AREA_EXEC)
[c23502d]979 flags |= PAGE_EXEC;
[6a3c9a7]980
[0ee077ee]981 if (aflags & AS_AREA_CACHEABLE)
[9a8d91b]982 flags |= PAGE_CACHEABLE;
983
[6a3c9a7]984 return flags;
985}
[ef67bab]986
[df0103f7]987/** Compute flags for virtual address translation subsytem.
988 *
989 * The address space area must be locked.
990 * Interrupts must be disabled.
991 *
992 * @param a Address space area.
993 *
994 * @return Flags to be used in page_mapping_insert().
995 */
[8182031]996int as_area_get_flags(as_area_t *a)
[df0103f7]997{
998 return area_flags_to_page_flags(a->flags);
999}
1000
[ef67bab]1001/** Create page table.
1002 *
1003 * Depending on architecture, create either address space
1004 * private or global page table.
1005 *
1006 * @param flags Flags saying whether the page table is for kernel address space.
1007 *
1008 * @return First entry of the page table.
1009 */
1010pte_t *page_table_create(int flags)
1011{
[bd1deed]1012#ifdef __OBJC__
1013 return [as_t page_table_create: flags];
1014#else
1015 ASSERT(as_operations);
1016 ASSERT(as_operations->page_table_create);
1017
1018 return as_operations->page_table_create(flags);
1019#endif
[ef67bab]1020}
[d3e7ff4]1021
[482826d]1022/** Destroy page table.
1023 *
1024 * Destroy page table in architecture specific way.
1025 *
1026 * @param page_table Physical address of PTL0.
1027 */
1028void page_table_destroy(pte_t *page_table)
1029{
[bd1deed]1030#ifdef __OBJC__
1031 return [as_t page_table_destroy: page_table];
1032#else
1033 ASSERT(as_operations);
1034 ASSERT(as_operations->page_table_destroy);
1035
1036 as_operations->page_table_destroy(page_table);
1037#endif
[482826d]1038}
1039
[2299914]1040/** Lock page table.
1041 *
1042 * This function should be called before any page_mapping_insert(),
1043 * page_mapping_remove() and page_mapping_find().
1044 *
1045 * Locking order is such that address space areas must be locked
1046 * prior to this call. Address space can be locked prior to this
1047 * call in which case the lock argument is false.
1048 *
1049 * @param as Address space.
[9179d0a]1050 * @param lock If false, do not attempt to lock as->lock.
[2299914]1051 */
1052void page_table_lock(as_t *as, bool lock)
1053{
[bd1deed]1054#ifdef __OBJC__
1055 [as page_table_lock: lock];
1056#else
[2299914]1057 ASSERT(as_operations);
1058 ASSERT(as_operations->page_table_lock);
[bd1deed]1059
[2299914]1060 as_operations->page_table_lock(as, lock);
[bd1deed]1061#endif
[2299914]1062}
1063
1064/** Unlock page table.
1065 *
1066 * @param as Address space.
[9179d0a]1067 * @param unlock If false, do not attempt to unlock as->lock.
[2299914]1068 */
1069void page_table_unlock(as_t *as, bool unlock)
1070{
[bd1deed]1071#ifdef __OBJC__
1072 [as page_table_unlock: unlock];
1073#else
[2299914]1074 ASSERT(as_operations);
1075 ASSERT(as_operations->page_table_unlock);
[bd1deed]1076
[2299914]1077 as_operations->page_table_unlock(as, unlock);
[bd1deed]1078#endif
[2299914]1079}
1080
[d3e7ff4]1081
1082/** Find address space area and lock it.
1083 *
1084 * The address space must be locked and interrupts must be disabled.
1085 *
1086 * @param as Address space.
1087 * @param va Virtual address.
1088 *
[6f4495f5]1089 * @return Locked address space area containing va on success or NULL on
1090 * failure.
[d3e7ff4]1091 */
[7f1c620]1092as_area_t *find_area_and_lock(as_t *as, uintptr_t va)
[d3e7ff4]1093{
1094 as_area_t *a;
[252127e]1095 btree_node_t *leaf, *lnode;
[6c441cf8]1096 unsigned int i;
[252127e]1097
1098 a = (as_area_t *) btree_search(&as->as_area_btree, va, &leaf);
1099 if (a) {
1100 /* va is the base address of an address space area */
[1068f6a]1101 mutex_lock(&a->lock);
[252127e]1102 return a;
1103 }
[d3e7ff4]1104
[252127e]1105 /*
[c47912f]1106 * Search the leaf node and the righmost record of its left neighbour
[252127e]1107 * to find out whether this is a miss or va belongs to an address
1108 * space area found there.
1109 */
1110
1111 /* First, search the leaf node itself. */
1112 for (i = 0; i < leaf->keys; i++) {
1113 a = (as_area_t *) leaf->value[i];
[1068f6a]1114 mutex_lock(&a->lock);
[252127e]1115 if ((a->base <= va) && (va < a->base + a->pages * PAGE_SIZE)) {
1116 return a;
1117 }
[1068f6a]1118 mutex_unlock(&a->lock);
[252127e]1119 }
[d3e7ff4]1120
[252127e]1121 /*
[c47912f]1122 * Second, locate the left neighbour and test its last record.
[b26db0c]1123 * Because of its position in the B+tree, it must have base < va.
[252127e]1124 */
[6f4495f5]1125 lnode = btree_leaf_node_left_neighbour(&as->as_area_btree, leaf);
1126 if (lnode) {
[252127e]1127 a = (as_area_t *) lnode->value[lnode->keys - 1];
[1068f6a]1128 mutex_lock(&a->lock);
[252127e]1129 if (va < a->base + a->pages * PAGE_SIZE) {
[37e7d2b9]1130 return a;
[252127e]1131 }
[1068f6a]1132 mutex_unlock(&a->lock);
[d3e7ff4]1133 }
1134
1135 return NULL;
1136}
[37e7d2b9]1137
1138/** Check area conflicts with other areas.
1139 *
1140 * The address space must be locked and interrupts must be disabled.
1141 *
1142 * @param as Address space.
1143 * @param va Starting virtual address of the area being tested.
1144 * @param size Size of the area being tested.
1145 * @param avoid_area Do not touch this area.
1146 *
1147 * @return True if there is no conflict, false otherwise.
1148 */
[6f4495f5]1149bool check_area_conflicts(as_t *as, uintptr_t va, size_t size,
1150 as_area_t *avoid_area)
[37e7d2b9]1151{
1152 as_area_t *a;
[252127e]1153 btree_node_t *leaf, *node;
[6c441cf8]1154 unsigned int i;
[37e7d2b9]1155
[5a7d9d1]1156 /*
1157 * We don't want any area to have conflicts with NULL page.
1158 */
1159 if (overlaps(va, size, NULL, PAGE_SIZE))
1160 return false;
1161
[252127e]1162 /*
1163 * The leaf node is found in O(log n), where n is proportional to
1164 * the number of address space areas belonging to as.
1165 * The check for conflicts is then attempted on the rightmost
[c47912f]1166 * record in the left neighbour, the leftmost record in the right
1167 * neighbour and all records in the leaf node itself.
[252127e]1168 */
1169
1170 if ((a = (as_area_t *) btree_search(&as->as_area_btree, va, &leaf))) {
1171 if (a != avoid_area)
1172 return false;
1173 }
1174
1175 /* First, check the two border cases. */
[c47912f]1176 if ((node = btree_leaf_node_left_neighbour(&as->as_area_btree, leaf))) {
[252127e]1177 a = (as_area_t *) node->value[node->keys - 1];
[1068f6a]1178 mutex_lock(&a->lock);
[252127e]1179 if (overlaps(va, size, a->base, a->pages * PAGE_SIZE)) {
[1068f6a]1180 mutex_unlock(&a->lock);
[252127e]1181 return false;
1182 }
[1068f6a]1183 mutex_unlock(&a->lock);
[252127e]1184 }
[6f4495f5]1185 node = btree_leaf_node_right_neighbour(&as->as_area_btree, leaf);
1186 if (node) {
[252127e]1187 a = (as_area_t *) node->value[0];
[1068f6a]1188 mutex_lock(&a->lock);
[252127e]1189 if (overlaps(va, size, a->base, a->pages * PAGE_SIZE)) {
[1068f6a]1190 mutex_unlock(&a->lock);
[252127e]1191 return false;
1192 }
[1068f6a]1193 mutex_unlock(&a->lock);
[252127e]1194 }
1195
1196 /* Second, check the leaf node. */
1197 for (i = 0; i < leaf->keys; i++) {
1198 a = (as_area_t *) leaf->value[i];
[37e7d2b9]1199
1200 if (a == avoid_area)
1201 continue;
[252127e]1202
[1068f6a]1203 mutex_lock(&a->lock);
[252127e]1204 if (overlaps(va, size, a->base, a->pages * PAGE_SIZE)) {
[1068f6a]1205 mutex_unlock(&a->lock);
[252127e]1206 return false;
1207 }
[1068f6a]1208 mutex_unlock(&a->lock);
[5a7d9d1]1209 }
[37e7d2b9]1210
[5a7d9d1]1211 /*
1212 * So far, the area does not conflict with other areas.
1213 * Check if it doesn't conflict with kernel address space.
1214 */
1215 if (!KERNEL_ADDRESS_SPACE_SHADOWED) {
1216 return !overlaps(va, size,
[6f4495f5]1217 KERNEL_ADDRESS_SPACE_START,
1218 KERNEL_ADDRESS_SPACE_END - KERNEL_ADDRESS_SPACE_START);
[37e7d2b9]1219 }
1220
1221 return true;
1222}
[df0103f7]1223
[b878df3]1224/** Return size of the address space area with given base.
1225 *
1226 * @param base Arbitrary address insede the address space area.
1227 *
1228 * @return Size of the address space area in bytes or zero if it
1229 * does not exist.
1230 */
1231size_t as_area_get_size(uintptr_t base)
[7c23af9]1232{
1233 ipl_t ipl;
1234 as_area_t *src_area;
1235 size_t size;
1236
1237 ipl = interrupts_disable();
1238 src_area = find_area_and_lock(AS, base);
1239 if (src_area){
1240 size = src_area->pages * PAGE_SIZE;
[1068f6a]1241 mutex_unlock(&src_area->lock);
[7c23af9]1242 } else {
1243 size = 0;
1244 }
1245 interrupts_restore(ipl);
1246 return size;
1247}
1248
[25bf215]1249/** Mark portion of address space area as used.
1250 *
1251 * The address space area must be already locked.
1252 *
1253 * @param a Address space area.
1254 * @param page First page to be marked.
1255 * @param count Number of page to be marked.
1256 *
1257 * @return 0 on failure and 1 on success.
1258 */
[7f1c620]1259int used_space_insert(as_area_t *a, uintptr_t page, count_t count)
[25bf215]1260{
1261 btree_node_t *leaf, *node;
1262 count_t pages;
[6c441cf8]1263 unsigned int i;
[25bf215]1264
1265 ASSERT(page == ALIGN_DOWN(page, PAGE_SIZE));
1266 ASSERT(count);
1267
1268 pages = (count_t) btree_search(&a->used_space, page, &leaf);
1269 if (pages) {
1270 /*
1271 * We hit the beginning of some used space.
1272 */
1273 return 0;
1274 }
1275
[a6cb8cb]1276 if (!leaf->keys) {
1277 btree_insert(&a->used_space, page, (void *) count, leaf);
1278 return 1;
1279 }
1280
[25bf215]1281 node = btree_leaf_node_left_neighbour(&a->used_space, leaf);
1282 if (node) {
[6f4495f5]1283 uintptr_t left_pg = node->key[node->keys - 1];
1284 uintptr_t right_pg = leaf->key[0];
1285 count_t left_cnt = (count_t) node->value[node->keys - 1];
1286 count_t right_cnt = (count_t) leaf->value[0];
[25bf215]1287
1288 /*
1289 * Examine the possibility that the interval fits
1290 * somewhere between the rightmost interval of
1291 * the left neigbour and the first interval of the leaf.
1292 */
1293
1294 if (page >= right_pg) {
1295 /* Do nothing. */
[6f4495f5]1296 } else if (overlaps(page, count * PAGE_SIZE, left_pg,
1297 left_cnt * PAGE_SIZE)) {
[25bf215]1298 /* The interval intersects with the left interval. */
1299 return 0;
[6f4495f5]1300 } else if (overlaps(page, count * PAGE_SIZE, right_pg,
1301 right_cnt * PAGE_SIZE)) {
[25bf215]1302 /* The interval intersects with the right interval. */
1303 return 0;
[6f4495f5]1304 } else if ((page == left_pg + left_cnt * PAGE_SIZE) &&
1305 (page + count * PAGE_SIZE == right_pg)) {
1306 /*
1307 * The interval can be added by merging the two already
1308 * present intervals.
1309 */
[56789125]1310 node->value[node->keys - 1] += count + right_cnt;
[25bf215]1311 btree_remove(&a->used_space, right_pg, leaf);
1312 return 1;
[6f4495f5]1313 } else if (page == left_pg + left_cnt * PAGE_SIZE) {
1314 /*
1315 * The interval can be added by simply growing the left
1316 * interval.
1317 */
[56789125]1318 node->value[node->keys - 1] += count;
[25bf215]1319 return 1;
[6f4495f5]1320 } else if (page + count * PAGE_SIZE == right_pg) {
[25bf215]1321 /*
[6f4495f5]1322 * The interval can be addded by simply moving base of
1323 * the right interval down and increasing its size
1324 * accordingly.
[25bf215]1325 */
[56789125]1326 leaf->value[0] += count;
[25bf215]1327 leaf->key[0] = page;
1328 return 1;
1329 } else {
1330 /*
1331 * The interval is between both neigbouring intervals,
1332 * but cannot be merged with any of them.
1333 */
[6f4495f5]1334 btree_insert(&a->used_space, page, (void *) count,
1335 leaf);
[25bf215]1336 return 1;
1337 }
1338 } else if (page < leaf->key[0]) {
[7f1c620]1339 uintptr_t right_pg = leaf->key[0];
[25bf215]1340 count_t right_cnt = (count_t) leaf->value[0];
1341
1342 /*
[6f4495f5]1343 * Investigate the border case in which the left neighbour does
1344 * not exist but the interval fits from the left.
[25bf215]1345 */
1346
[6f4495f5]1347 if (overlaps(page, count * PAGE_SIZE, right_pg,
1348 right_cnt * PAGE_SIZE)) {
[25bf215]1349 /* The interval intersects with the right interval. */
1350 return 0;
[6f4495f5]1351 } else if (page + count * PAGE_SIZE == right_pg) {
[25bf215]1352 /*
[6f4495f5]1353 * The interval can be added by moving the base of the
1354 * right interval down and increasing its size
1355 * accordingly.
[25bf215]1356 */
1357 leaf->key[0] = page;
[56789125]1358 leaf->value[0] += count;
[25bf215]1359 return 1;
1360 } else {
1361 /*
1362 * The interval doesn't adjoin with the right interval.
1363 * It must be added individually.
1364 */
[6f4495f5]1365 btree_insert(&a->used_space, page, (void *) count,
1366 leaf);
[25bf215]1367 return 1;
1368 }
1369 }
1370
1371 node = btree_leaf_node_right_neighbour(&a->used_space, leaf);
1372 if (node) {
[6f4495f5]1373 uintptr_t left_pg = leaf->key[leaf->keys - 1];
1374 uintptr_t right_pg = node->key[0];
1375 count_t left_cnt = (count_t) leaf->value[leaf->keys - 1];
1376 count_t right_cnt = (count_t) node->value[0];
[25bf215]1377
1378 /*
1379 * Examine the possibility that the interval fits
1380 * somewhere between the leftmost interval of
1381 * the right neigbour and the last interval of the leaf.
1382 */
1383
1384 if (page < left_pg) {
1385 /* Do nothing. */
[6f4495f5]1386 } else if (overlaps(page, count * PAGE_SIZE, left_pg,
1387 left_cnt * PAGE_SIZE)) {
[25bf215]1388 /* The interval intersects with the left interval. */
1389 return 0;
[6f4495f5]1390 } else if (overlaps(page, count * PAGE_SIZE, right_pg,
1391 right_cnt * PAGE_SIZE)) {
[25bf215]1392 /* The interval intersects with the right interval. */
1393 return 0;
[6f4495f5]1394 } else if ((page == left_pg + left_cnt * PAGE_SIZE) &&
1395 (page + count * PAGE_SIZE == right_pg)) {
1396 /*
1397 * The interval can be added by merging the two already
1398 * present intervals.
1399 * */
[56789125]1400 leaf->value[leaf->keys - 1] += count + right_cnt;
[25bf215]1401 btree_remove(&a->used_space, right_pg, node);
1402 return 1;
[6f4495f5]1403 } else if (page == left_pg + left_cnt * PAGE_SIZE) {
1404 /*
1405 * The interval can be added by simply growing the left
1406 * interval.
1407 * */
[56789125]1408 leaf->value[leaf->keys - 1] += count;
[25bf215]1409 return 1;
[6f4495f5]1410 } else if (page + count * PAGE_SIZE == right_pg) {
[25bf215]1411 /*
[6f4495f5]1412 * The interval can be addded by simply moving base of
1413 * the right interval down and increasing its size
1414 * accordingly.
[25bf215]1415 */
[56789125]1416 node->value[0] += count;
[25bf215]1417 node->key[0] = page;
1418 return 1;
1419 } else {
1420 /*
1421 * The interval is between both neigbouring intervals,
1422 * but cannot be merged with any of them.
1423 */
[6f4495f5]1424 btree_insert(&a->used_space, page, (void *) count,
1425 leaf);
[25bf215]1426 return 1;
1427 }
1428 } else if (page >= leaf->key[leaf->keys - 1]) {
[7f1c620]1429 uintptr_t left_pg = leaf->key[leaf->keys - 1];
[25bf215]1430 count_t left_cnt = (count_t) leaf->value[leaf->keys - 1];
1431
1432 /*
[6f4495f5]1433 * Investigate the border case in which the right neighbour
1434 * does not exist but the interval fits from the right.
[25bf215]1435 */
1436
[6f4495f5]1437 if (overlaps(page, count * PAGE_SIZE, left_pg,
1438 left_cnt * PAGE_SIZE)) {
[56789125]1439 /* The interval intersects with the left interval. */
[25bf215]1440 return 0;
[6f4495f5]1441 } else if (left_pg + left_cnt * PAGE_SIZE == page) {
1442 /*
1443 * The interval can be added by growing the left
1444 * interval.
1445 */
[56789125]1446 leaf->value[leaf->keys - 1] += count;
[25bf215]1447 return 1;
1448 } else {
1449 /*
1450 * The interval doesn't adjoin with the left interval.
1451 * It must be added individually.
1452 */
[6f4495f5]1453 btree_insert(&a->used_space, page, (void *) count,
1454 leaf);
[25bf215]1455 return 1;
1456 }
1457 }
1458
1459 /*
[6f4495f5]1460 * Note that if the algorithm made it thus far, the interval can fit
1461 * only between two other intervals of the leaf. The two border cases
1462 * were already resolved.
[25bf215]1463 */
1464 for (i = 1; i < leaf->keys; i++) {
1465 if (page < leaf->key[i]) {
[6f4495f5]1466 uintptr_t left_pg = leaf->key[i - 1];
1467 uintptr_t right_pg = leaf->key[i];
1468 count_t left_cnt = (count_t) leaf->value[i - 1];
1469 count_t right_cnt = (count_t) leaf->value[i];
[25bf215]1470
1471 /*
1472 * The interval fits between left_pg and right_pg.
1473 */
1474
[6f4495f5]1475 if (overlaps(page, count * PAGE_SIZE, left_pg,
1476 left_cnt * PAGE_SIZE)) {
1477 /*
1478 * The interval intersects with the left
1479 * interval.
1480 */
[25bf215]1481 return 0;
[6f4495f5]1482 } else if (overlaps(page, count * PAGE_SIZE, right_pg,
1483 right_cnt * PAGE_SIZE)) {
1484 /*
1485 * The interval intersects with the right
1486 * interval.
1487 */
[25bf215]1488 return 0;
[6f4495f5]1489 } else if ((page == left_pg + left_cnt * PAGE_SIZE) &&
1490 (page + count * PAGE_SIZE == right_pg)) {
1491 /*
1492 * The interval can be added by merging the two
1493 * already present intervals.
1494 */
[56789125]1495 leaf->value[i - 1] += count + right_cnt;
[25bf215]1496 btree_remove(&a->used_space, right_pg, leaf);
1497 return 1;
[6f4495f5]1498 } else if (page == left_pg + left_cnt * PAGE_SIZE) {
1499 /*
1500 * The interval can be added by simply growing
1501 * the left interval.
1502 */
[56789125]1503 leaf->value[i - 1] += count;
[25bf215]1504 return 1;
[6f4495f5]1505 } else if (page + count * PAGE_SIZE == right_pg) {
[25bf215]1506 /*
[6f4495f5]1507 * The interval can be addded by simply moving
1508 * base of the right interval down and
1509 * increasing its size accordingly.
[25bf215]1510 */
[56789125]1511 leaf->value[i] += count;
[25bf215]1512 leaf->key[i] = page;
1513 return 1;
1514 } else {
1515 /*
[6f4495f5]1516 * The interval is between both neigbouring
1517 * intervals, but cannot be merged with any of
1518 * them.
[25bf215]1519 */
[6f4495f5]1520 btree_insert(&a->used_space, page,
1521 (void *) count, leaf);
[25bf215]1522 return 1;
1523 }
1524 }
1525 }
1526
[2b8b0ca]1527 panic("Inconsistency detected while adding %" PRIc " pages of used space at "
[6f4495f5]1528 "%p.\n", count, page);
[25bf215]1529}
1530
1531/** Mark portion of address space area as unused.
1532 *
1533 * The address space area must be already locked.
1534 *
1535 * @param a Address space area.
1536 * @param page First page to be marked.
1537 * @param count Number of page to be marked.
1538 *
1539 * @return 0 on failure and 1 on success.
1540 */
[7f1c620]1541int used_space_remove(as_area_t *a, uintptr_t page, count_t count)
[25bf215]1542{
1543 btree_node_t *leaf, *node;
1544 count_t pages;
[6c441cf8]1545 unsigned int i;
[25bf215]1546
1547 ASSERT(page == ALIGN_DOWN(page, PAGE_SIZE));
1548 ASSERT(count);
1549
1550 pages = (count_t) btree_search(&a->used_space, page, &leaf);
1551 if (pages) {
1552 /*
1553 * We are lucky, page is the beginning of some interval.
1554 */
1555 if (count > pages) {
1556 return 0;
1557 } else if (count == pages) {
1558 btree_remove(&a->used_space, page, leaf);
[56789125]1559 return 1;
[25bf215]1560 } else {
1561 /*
1562 * Find the respective interval.
1563 * Decrease its size and relocate its start address.
1564 */
1565 for (i = 0; i < leaf->keys; i++) {
1566 if (leaf->key[i] == page) {
[6f4495f5]1567 leaf->key[i] += count * PAGE_SIZE;
[56789125]1568 leaf->value[i] -= count;
[25bf215]1569 return 1;
1570 }
1571 }
1572 goto error;
1573 }
1574 }
1575
1576 node = btree_leaf_node_left_neighbour(&a->used_space, leaf);
1577 if (node && page < leaf->key[0]) {
[7f1c620]1578 uintptr_t left_pg = node->key[node->keys - 1];
[25bf215]1579 count_t left_cnt = (count_t) node->value[node->keys - 1];
1580
[6f4495f5]1581 if (overlaps(left_pg, left_cnt * PAGE_SIZE, page,
1582 count * PAGE_SIZE)) {
1583 if (page + count * PAGE_SIZE ==
1584 left_pg + left_cnt * PAGE_SIZE) {
[25bf215]1585 /*
[6f4495f5]1586 * The interval is contained in the rightmost
1587 * interval of the left neighbour and can be
1588 * removed by updating the size of the bigger
1589 * interval.
[25bf215]1590 */
[56789125]1591 node->value[node->keys - 1] -= count;
[25bf215]1592 return 1;
[6f4495f5]1593 } else if (page + count * PAGE_SIZE <
1594 left_pg + left_cnt*PAGE_SIZE) {
[56789125]1595 count_t new_cnt;
[25bf215]1596
1597 /*
[6f4495f5]1598 * The interval is contained in the rightmost
1599 * interval of the left neighbour but its
1600 * removal requires both updating the size of
1601 * the original interval and also inserting a
1602 * new interval.
[25bf215]1603 */
[6f4495f5]1604 new_cnt = ((left_pg + left_cnt * PAGE_SIZE) -
1605 (page + count*PAGE_SIZE)) >> PAGE_WIDTH;
[56789125]1606 node->value[node->keys - 1] -= count + new_cnt;
[6f4495f5]1607 btree_insert(&a->used_space, page +
1608 count * PAGE_SIZE, (void *) new_cnt, leaf);
[25bf215]1609 return 1;
1610 }
1611 }
1612 return 0;
1613 } else if (page < leaf->key[0]) {
1614 return 0;
1615 }
1616
1617 if (page > leaf->key[leaf->keys - 1]) {
[7f1c620]1618 uintptr_t left_pg = leaf->key[leaf->keys - 1];
[25bf215]1619 count_t left_cnt = (count_t) leaf->value[leaf->keys - 1];
1620
[6f4495f5]1621 if (overlaps(left_pg, left_cnt * PAGE_SIZE, page,
1622 count * PAGE_SIZE)) {
1623 if (page + count * PAGE_SIZE ==
1624 left_pg + left_cnt * PAGE_SIZE) {
[25bf215]1625 /*
[6f4495f5]1626 * The interval is contained in the rightmost
1627 * interval of the leaf and can be removed by
1628 * updating the size of the bigger interval.
[25bf215]1629 */
[56789125]1630 leaf->value[leaf->keys - 1] -= count;
[25bf215]1631 return 1;
[6f4495f5]1632 } else if (page + count * PAGE_SIZE < left_pg +
1633 left_cnt * PAGE_SIZE) {
[56789125]1634 count_t new_cnt;
[25bf215]1635
1636 /*
[6f4495f5]1637 * The interval is contained in the rightmost
1638 * interval of the leaf but its removal
1639 * requires both updating the size of the
1640 * original interval and also inserting a new
1641 * interval.
[25bf215]1642 */
[6f4495f5]1643 new_cnt = ((left_pg + left_cnt * PAGE_SIZE) -
1644 (page + count * PAGE_SIZE)) >> PAGE_WIDTH;
[56789125]1645 leaf->value[leaf->keys - 1] -= count + new_cnt;
[6f4495f5]1646 btree_insert(&a->used_space, page +
1647 count * PAGE_SIZE, (void *) new_cnt, leaf);
[25bf215]1648 return 1;
1649 }
1650 }
1651 return 0;
1652 }
1653
1654 /*
1655 * The border cases have been already resolved.
1656 * Now the interval can be only between intervals of the leaf.
1657 */
1658 for (i = 1; i < leaf->keys - 1; i++) {
1659 if (page < leaf->key[i]) {
[7f1c620]1660 uintptr_t left_pg = leaf->key[i - 1];
[25bf215]1661 count_t left_cnt = (count_t) leaf->value[i - 1];
1662
1663 /*
[6f4495f5]1664 * Now the interval is between intervals corresponding
1665 * to (i - 1) and i.
[25bf215]1666 */
[6f4495f5]1667 if (overlaps(left_pg, left_cnt * PAGE_SIZE, page,
1668 count * PAGE_SIZE)) {
1669 if (page + count * PAGE_SIZE ==
1670 left_pg + left_cnt*PAGE_SIZE) {
[25bf215]1671 /*
[6f4495f5]1672 * The interval is contained in the
1673 * interval (i - 1) of the leaf and can
1674 * be removed by updating the size of
1675 * the bigger interval.
[25bf215]1676 */
[56789125]1677 leaf->value[i - 1] -= count;
[25bf215]1678 return 1;
[6f4495f5]1679 } else if (page + count * PAGE_SIZE <
1680 left_pg + left_cnt * PAGE_SIZE) {
[56789125]1681 count_t new_cnt;
[25bf215]1682
1683 /*
[6f4495f5]1684 * The interval is contained in the
1685 * interval (i - 1) of the leaf but its
1686 * removal requires both updating the
1687 * size of the original interval and
[25bf215]1688 * also inserting a new interval.
1689 */
[6f4495f5]1690 new_cnt = ((left_pg +
1691 left_cnt * PAGE_SIZE) -
1692 (page + count * PAGE_SIZE)) >>
1693 PAGE_WIDTH;
[56789125]1694 leaf->value[i - 1] -= count + new_cnt;
[6f4495f5]1695 btree_insert(&a->used_space, page +
1696 count * PAGE_SIZE, (void *) new_cnt,
1697 leaf);
[25bf215]1698 return 1;
1699 }
1700 }
1701 return 0;
1702 }
1703 }
1704
1705error:
[2b8b0ca]1706 panic("Inconsistency detected while removing %" PRIc " pages of used space "
[6f4495f5]1707 "from %p.\n", count, page);
[25bf215]1708}
1709
[8182031]1710/** Remove reference to address space area share info.
1711 *
1712 * If the reference count drops to 0, the sh_info is deallocated.
1713 *
1714 * @param sh_info Pointer to address space area share info.
1715 */
1716void sh_info_remove_reference(share_info_t *sh_info)
1717{
1718 bool dealloc = false;
1719
1720 mutex_lock(&sh_info->lock);
1721 ASSERT(sh_info->refcount);
1722 if (--sh_info->refcount == 0) {
1723 dealloc = true;
[f8d069e8]1724 link_t *cur;
[8182031]1725
1726 /*
1727 * Now walk carefully the pagemap B+tree and free/remove
1728 * reference from all frames found there.
1729 */
[6f4495f5]1730 for (cur = sh_info->pagemap.leaf_head.next;
1731 cur != &sh_info->pagemap.leaf_head; cur = cur->next) {
[8182031]1732 btree_node_t *node;
[6c441cf8]1733 unsigned int i;
[8182031]1734
[f8d069e8]1735 node = list_get_instance(cur, btree_node_t, leaf_link);
1736 for (i = 0; i < node->keys; i++)
[7f1c620]1737 frame_free((uintptr_t) node->value[i]);
[8182031]1738 }
1739
1740 }
1741 mutex_unlock(&sh_info->lock);
1742
1743 if (dealloc) {
1744 btree_destroy(&sh_info->pagemap);
1745 free(sh_info);
1746 }
1747}
1748
[df0103f7]1749/*
1750 * Address space related syscalls.
1751 */
1752
1753/** Wrapper for as_area_create(). */
[7f1c620]1754unative_t sys_as_area_create(uintptr_t address, size_t size, int flags)
[df0103f7]1755{
[6f4495f5]1756 if (as_area_create(AS, flags | AS_AREA_CACHEABLE, size, address,
1757 AS_AREA_ATTR_NONE, &anon_backend, NULL))
[7f1c620]1758 return (unative_t) address;
[df0103f7]1759 else
[7f1c620]1760 return (unative_t) -1;
[df0103f7]1761}
1762
[c6e314a]1763/** Wrapper for as_area_resize(). */
[7f1c620]1764unative_t sys_as_area_resize(uintptr_t address, size_t size, int flags)
[df0103f7]1765{
[7f1c620]1766 return (unative_t) as_area_resize(AS, address, size, 0);
[7242a78e]1767}
1768
[c6e314a]1769/** Wrapper for as_area_destroy(). */
[7f1c620]1770unative_t sys_as_area_destroy(uintptr_t address)
[7242a78e]1771{
[7f1c620]1772 return (unative_t) as_area_destroy(AS, address);
[df0103f7]1773}
[b45c443]1774
[64c2ad5]1775/** Print out information about address space.
1776 *
1777 * @param as Address space.
1778 */
1779void as_print(as_t *as)
1780{
1781 ipl_t ipl;
1782
1783 ipl = interrupts_disable();
1784 mutex_lock(&as->lock);
1785
1786 /* print out info about address space areas */
1787 link_t *cur;
[6f4495f5]1788 for (cur = as->as_area_btree.leaf_head.next;
1789 cur != &as->as_area_btree.leaf_head; cur = cur->next) {
1790 btree_node_t *node;
1791
1792 node = list_get_instance(cur, btree_node_t, leaf_link);
[64c2ad5]1793
[6c441cf8]1794 unsigned int i;
[64c2ad5]1795 for (i = 0; i < node->keys; i++) {
[7ba7c6d]1796 as_area_t *area = node->value[i];
[64c2ad5]1797
1798 mutex_lock(&area->lock);
[2b8b0ca]1799 printf("as_area: %p, base=%p, pages=%" PRIc " (%p - %p)\n",
[6f4495f5]1800 area, area->base, area->pages, area->base,
[2b8b0ca]1801 area->base + FRAMES2SIZE(area->pages));
[64c2ad5]1802 mutex_unlock(&area->lock);
1803 }
1804 }
1805
1806 mutex_unlock(&as->lock);
1807 interrupts_restore(ipl);
1808}
1809
[cc73a8a1]1810/** @}
[b45c443]1811 */
Note: See TracBrowser for help on using the repository browser.