source: mainline/kernel/generic/src/mm/as.c@ 6785b88b

lfn serial ticket/834-toolchain-update topic/msim-upgrade topic/simplify-dev-export
Last change on this file since 6785b88b was 6785b88b, checked in by Jiri Svoboda <jiri@…>, 7 years ago

Fix indentation.

  • Property mode set to 100644
File size: 57.2 KB
RevLine 
[20d50a1]1/*
[0321109]2 * Copyright (c) 2010 Jakub Jermar
[88cc71c0]3 * Copyright (c) 2018 Jiri Svoboda
[20d50a1]4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 *
10 * - Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * - Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * - The name of the author may not be used to endorse or promote products
16 * derived from this software without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
19 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
20 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
21 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
22 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
23 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
24 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
25 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
26 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
27 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
28 */
29
[174156fd]30/** @addtogroup kernel_generic_mm
[b45c443]31 * @{
32 */
33
[9179d0a]34/**
[b45c443]35 * @file
[da1bafb]36 * @brief Address space related functions.
[9179d0a]37 *
[20d50a1]38 * This file contains address space manipulation functions.
39 * Roughly speaking, this is a higher-level client of
40 * Virtual Address Translation (VAT) subsystem.
[9179d0a]41 *
42 * Functionality provided by this file allows one to
[cc73a8a1]43 * create address spaces and create, resize and share
[9179d0a]44 * address space areas.
45 *
46 * @see page.c
47 *
[20d50a1]48 */
49
50#include <mm/as.h>
[ef67bab]51#include <arch/mm/as.h>
[20d50a1]52#include <mm/page.h>
53#include <mm/frame.h>
[085d973]54#include <mm/slab.h>
[20d50a1]55#include <mm/tlb.h>
56#include <arch/mm/page.h>
57#include <genarch/mm/page_pt.h>
[2802767]58#include <genarch/mm/page_ht.h>
[4512d7e]59#include <mm/asid.h>
[20d50a1]60#include <arch/mm/asid.h>
[31d8e10]61#include <preemption.h>
[20d50a1]62#include <synch/spinlock.h>
[1068f6a]63#include <synch/mutex.h>
[5c9a08b]64#include <adt/list.h>
[252127e]65#include <adt/btree.h>
[df0103f7]66#include <proc/task.h>
[e3c762cd]67#include <proc/thread.h>
[20d50a1]68#include <arch/asm.h>
[df0103f7]69#include <panic.h>
[63e27ef]70#include <assert.h>
[bab75df6]71#include <stdio.h>
[44a7ee5]72#include <mem.h>
[5a7d9d1]73#include <macros.h>
[0b37882]74#include <bitops.h>
[20d50a1]75#include <arch.h>
[df0103f7]76#include <errno.h>
77#include <config.h>
[25bf215]78#include <align.h>
[d99c1d2]79#include <typedefs.h>
[e3c762cd]80#include <syscall/copy.h>
81#include <arch/interrupt.h>
[1dbc43f]82#include <interrupt.h>
[20d50a1]83
[cc73a8a1]84/**
85 * Each architecture decides what functions will be used to carry out
86 * address space operations such as creating or locking page tables.
87 */
[ef67bab]88as_operations_t *as_operations = NULL;
[20d50a1]89
[fc47885]90/** Slab for as_t objects.
[da1bafb]91 *
[57da95c]92 */
[82d515e9]93static slab_cache_t *as_cache;
[57da95c]94
[fc47885]95/** ASID subsystem lock.
96 *
97 * This lock protects:
[55b77d9]98 * - inactive_as_with_asid_list
[879585a3]99 * - as->asid for each as of the as_t type
100 * - asids_allocated counter
[da1bafb]101 *
[6f4495f5]102 */
[879585a3]103SPINLOCK_INITIALIZE(asidlock);
[7e4e532]104
105/**
[fc47885]106 * Inactive address spaces (on all processors)
107 * that have valid ASID.
[7e4e532]108 */
[55b77d9]109LIST_INITIALIZE(inactive_as_with_asid_list);
[7e4e532]110
[071a8ae6]111/** Kernel address space. */
112as_t *AS_KERNEL = NULL;
113
[88cc71c0]114static void *as_areas_getkey(odlink_t *);
115static int as_areas_cmp(void *, void *);
116
[b7fd2a0]117NO_TRACE static errno_t as_constructor(void *obj, unsigned int flags)
[29b2bbf]118{
119 as_t *as = (as_t *) obj;
[a35b458]120
[29b2bbf]121 link_initialize(&as->inactive_as_with_asid_link);
[7f341820]122 mutex_initialize(&as->lock, MUTEX_PASSIVE);
[a35b458]123
[fc47885]124 return as_constructor_arch(as, flags);
[29b2bbf]125}
126
[7a0359b]127NO_TRACE static size_t as_destructor(void *obj)
[29b2bbf]128{
[fc47885]129 return as_destructor_arch((as_t *) obj);
[29b2bbf]130}
131
[ef67bab]132/** Initialize address space subsystem. */
133void as_init(void)
134{
135 as_arch_init();
[a35b458]136
[82d515e9]137 as_cache = slab_cache_create("as_t", sizeof(as_t), 0,
[6f4495f5]138 as_constructor, as_destructor, SLAB_CACHE_MAGDEFERRED);
[a35b458]139
[8e1ea655]140 AS_KERNEL = as_create(FLAG_AS_KERNEL);
[125e944]141 if (!AS_KERNEL)
[f651e80]142 panic("Cannot create kernel address space.");
[a35b458]143
[fc47885]144 /*
145 * Make sure the kernel address space
[76fca31]146 * reference count never drops to zero.
147 */
[6193351]148 as_hold(AS_KERNEL);
[ef67bab]149}
150
[071a8ae6]151/** Create address space.
152 *
[da1bafb]153 * @param flags Flags that influence the way in wich the address
154 * space is created.
155 *
[071a8ae6]156 */
[da1bafb]157as_t *as_create(unsigned int flags)
[20d50a1]158{
[82d515e9]159 as_t *as = (as_t *) slab_alloc(as_cache, 0);
[29b2bbf]160 (void) as_create_arch(as, 0);
[a35b458]161
[88cc71c0]162 odict_initialize(&as->as_areas, as_areas_getkey, as_areas_cmp);
[a35b458]163
[bb68433]164 if (flags & FLAG_AS_KERNEL)
165 as->asid = ASID_KERNEL;
166 else
167 as->asid = ASID_INVALID;
[a35b458]168
[78de83de]169 refcount_init(&as->refcount);
[47800e0]170 as->cpu_refcount = 0;
[a35b458]171
[b3f8fb7]172#ifdef AS_PAGE_TABLE
[80bcaed]173 as->genarch.page_table = page_table_create(flags);
[b3f8fb7]174#else
175 page_table_create(flags);
176#endif
[a35b458]177
[20d50a1]178 return as;
179}
180
[482826d]181/** Destroy adress space.
182 *
[6f4495f5]183 * When there are no tasks referencing this address space (i.e. its refcount is
184 * zero), the address space can be destroyed.
[31d8e10]185 *
186 * We know that we don't hold any spinlock.
[6745592]187 *
[da1bafb]188 * @param as Address space to be destroyed.
189 *
[482826d]190 */
191void as_destroy(as_t *as)
[5be1923]192{
[31d8e10]193 DEADLOCK_PROBE_INIT(p_asidlock);
[a35b458]194
[63e27ef]195 assert(as != AS);
[78de83de]196 assert(refcount_unique(&as->refcount));
[a35b458]197
[482826d]198 /*
[663bb537]199 * Since there is no reference to this address space, it is safe not to
200 * lock its mutex.
[482826d]201 */
[a35b458]202
[31d8e10]203 /*
204 * We need to avoid deadlock between TLB shootdown and asidlock.
205 * We therefore try to take asid conditionally and if we don't succeed,
206 * we enable interrupts and try again. This is done while preemption is
207 * disabled to prevent nested context switches. We also depend on the
208 * fact that so far no spinlocks are held.
209 */
210 preemption_disable();
[da1bafb]211 ipl_t ipl = interrupts_read();
[a35b458]212
[31d8e10]213retry:
214 interrupts_disable();
215 if (!spinlock_trylock(&asidlock)) {
216 interrupts_enable();
217 DEADLOCK_PROBE(p_asidlock, DEADLOCK_THRESHOLD);
218 goto retry;
219 }
[a35b458]220
[da1bafb]221 /* Interrupts disabled, enable preemption */
222 preemption_enable();
[a35b458]223
[da1bafb]224 if ((as->asid != ASID_INVALID) && (as != AS_KERNEL)) {
[1624aae]225 if (as->cpu_refcount == 0)
[31e8ddd]226 list_remove(&as->inactive_as_with_asid_link);
[a35b458]227
[482826d]228 asid_put(as->asid);
229 }
[a35b458]230
[879585a3]231 spinlock_unlock(&asidlock);
[fdaad75d]232 interrupts_restore(ipl);
[a35b458]233
[482826d]234 /*
235 * Destroy address space areas of the address space.
[88cc71c0]236 * Need to start from the beginning each time since we are destroying
237 * the areas.
[da1bafb]238 */
[88cc71c0]239 as_area_t *area = as_area_first(as);
240 while (area != NULL) {
241 /*
242 * XXX We already have as_area_t, but as_area_destroy will
243 * have to search for it. This could be made faster.
244 */
245 as_area_destroy(as, area->base);
246 area = as_area_first(as);
[482826d]247 }
[a35b458]248
[88cc71c0]249 odict_finalize(&as->as_areas);
[a35b458]250
[b3f8fb7]251#ifdef AS_PAGE_TABLE
[80bcaed]252 page_table_destroy(as->genarch.page_table);
[b3f8fb7]253#else
254 page_table_destroy(NULL);
255#endif
[a35b458]256
[82d515e9]257 slab_free(as_cache, as);
[5be1923]258}
259
[0321109]260/** Hold a reference to an address space.
261 *
[fc47885]262 * Holding a reference to an address space prevents destruction
263 * of that address space.
[0321109]264 *
[da1bafb]265 * @param as Address space to be held.
266 *
[0321109]267 */
[7a0359b]268NO_TRACE void as_hold(as_t *as)
[0321109]269{
[78de83de]270 refcount_up(&as->refcount);
[0321109]271}
272
273/** Release a reference to an address space.
274 *
[fc47885]275 * The last one to release a reference to an address space
276 * destroys the address space.
[0321109]277 *
[78de83de]278 * @param as Address space to be released.
[da1bafb]279 *
[0321109]280 */
[7a0359b]281NO_TRACE void as_release(as_t *as)
[0321109]282{
[78de83de]283 if (refcount_down(&as->refcount))
[0321109]284 as_destroy(as);
285}
286
[88cc71c0]287/** Return first address space area.
288 *
289 * @param as Address space
290 * @return First area in @a as (i.e. area with the lowest base address)
291 * or @c NULL if there is none
292 */
293as_area_t *as_area_first(as_t *as)
294{
295 odlink_t *odlink = odict_first(&as->as_areas);
296 if (odlink == NULL)
297 return NULL;
298
299 return odict_get_instance(odlink, as_area_t, las_areas);
300}
301
302/** Return next address space area.
303 *
304 * @param cur Current area
305 * @return Next area in the same address space or @c NULL if @a cur is the
306 * last area.
307 */
308as_area_t *as_area_next(as_area_t *cur)
309{
310 odlink_t *odlink = odict_next(&cur->las_areas, &cur->as->as_areas);
311 if (odlink == NULL)
312 return NULL;
313
314 return odict_get_instance(odlink, as_area_t, las_areas);
315}
316
317/** Determine if area with specified parameters would conflict with
318 * a specific existing address space area.
319 *
320 * @param addr Starting virtual address of the area being tested.
321 * @param count Number of pages in the area being tested.
322 * @param guarded True if the area being tested is protected by guard pages.
323 * @param area Area against which we are testing.
324 *
325 * @return True if the two areas conflict, false otherwise.
326 */
327NO_TRACE static bool area_is_conflicting(uintptr_t addr,
328 size_t count, bool guarded, as_area_t *area)
329{
330 assert((addr % PAGE_SIZE) == 0);
331
332 size_t gsize = P2SZ(count);
[6785b88b]333 size_t agsize = P2SZ(area->pages);
[cd1ecf11]334
335 /*
336 * A guarded area has one guard page before, one page after.
337 * What we do here is: if either area is guarded, we add
338 * PAGE_SIZE to the size of both areas. That guarantees
339 * they will be spaced at least one page apart.
340 */
341 if (guarded || (area->flags & AS_AREA_GUARD) != 0) {
342 /* Add guard page size unless area is at the end of VA domain */
343 if (!overflows(addr, P2SZ(count)))
344 gsize += PAGE_SIZE;
345
346 /* Add guard page size unless area is at the end of VA domain */
347 if (!overflows(area->base, P2SZ(area->pages)))
348 agsize += PAGE_SIZE;
349 }
[88cc71c0]350
351 return overlaps(addr, gsize, area->base, agsize);
352
353}
354
[e3ee9b9]355/** Check area conflicts with other areas.
356 *
[35a3d950]357 * @param as Address space.
358 * @param addr Starting virtual address of the area being tested.
359 * @param count Number of pages in the area being tested.
360 * @param guarded True if the area being tested is protected by guard pages.
[88cc71c0]361 * @param avoid Do not touch this area. I.e. this area is not considered
362 * as presenting a conflict.
[e3ee9b9]363 *
364 * @return True if there is no conflict, false otherwise.
365 *
366 */
[0b37882]367NO_TRACE static bool check_area_conflicts(as_t *as, uintptr_t addr,
[35a3d950]368 size_t count, bool guarded, as_area_t *avoid)
[e3ee9b9]369{
[63e27ef]370 assert((addr % PAGE_SIZE) == 0);
371 assert(mutex_locked(&as->lock));
[94795812]372
373 /*
374 * If the addition of the supposed area address and size overflows,
375 * report conflict.
376 */
377 if (overflows_into_positive(addr, P2SZ(count)))
378 return false;
[a35b458]379
[e3ee9b9]380 /*
381 * We don't want any area to have conflicts with NULL page.
382 */
[b6f3e7e]383 if (overlaps(addr, P2SZ(count), (uintptr_t) NULL, PAGE_SIZE))
[e3ee9b9]384 return false;
[35a3d950]385
[e3ee9b9]386 /*
[88cc71c0]387 * To determine if we overlap with another area, we just need
388 * to look at overlap with the last area with base address <=
389 * to ours and on the first area with base address > than ours.
390 *
391 * First find last area with <= base address.
[e3ee9b9]392 */
[88cc71c0]393 odlink_t *odlink = odict_find_leq(&as->as_areas, &addr, NULL);
394 if (odlink != NULL) {
395 as_area_t *area = odict_get_instance(odlink, as_area_t,
396 las_areas);
[a35b458]397
[0b37882]398 if (area != avoid) {
399 mutex_lock(&area->lock);
[88cc71c0]400 if (area_is_conflicting(addr, count, guarded, area)) {
[0b37882]401 mutex_unlock(&area->lock);
402 return false;
403 }
[a35b458]404
[e3ee9b9]405 mutex_unlock(&area->lock);
406 }
[88cc71c0]407
408 /* Next area */
409 odlink = odict_next(odlink, &as->as_areas);
[e3ee9b9]410 }
[a35b458]411
[d9d0088]412 /*
413 * Next area, if any, is the first with base > than our base address.
414 * If there was no area with <= base, we need to look at the first area.
415 */
416 if (odlink == NULL)
417 odlink = odict_first(&as->as_areas);
418
[88cc71c0]419 if (odlink != NULL) {
420 as_area_t *area = odict_get_instance(odlink, as_area_t,
421 las_areas);
[a35b458]422
[0b37882]423 if (area != avoid) {
424 mutex_lock(&area->lock);
[88cc71c0]425 if (area_is_conflicting(addr, count, guarded, area)) {
[0b37882]426 mutex_unlock(&area->lock);
427 return false;
428 }
[a35b458]429
[e3ee9b9]430 mutex_unlock(&area->lock);
431 }
432 }
[a35b458]433
[e3ee9b9]434 /*
435 * So far, the area does not conflict with other areas.
[57355a40]436 * Check if it is contained in the user address space.
[e3ee9b9]437 */
438 if (!KERNEL_ADDRESS_SPACE_SHADOWED) {
[57355a40]439 return iswithin(USER_ADDRESS_SPACE_START,
440 (USER_ADDRESS_SPACE_END - USER_ADDRESS_SPACE_START) + 1,
441 addr, P2SZ(count));
[e3ee9b9]442 }
[a35b458]443
[e3ee9b9]444 return true;
445}
446
[fbcdeb8]447/** Return pointer to unmapped address space area
448 *
449 * The address space must be already locked when calling
450 * this function.
451 *
[35a3d950]452 * @param as Address space.
453 * @param bound Lowest address bound.
454 * @param size Requested size of the allocation.
455 * @param guarded True if the allocation must be protected by guard pages.
[fbcdeb8]456 *
457 * @return Address of the beginning of unmapped address space area.
458 * @return -1 if no suitable address space area was found.
459 *
460 */
461NO_TRACE static uintptr_t as_get_unmapped_area(as_t *as, uintptr_t bound,
[35a3d950]462 size_t size, bool guarded)
[fbcdeb8]463{
[63e27ef]464 assert(mutex_locked(&as->lock));
[a35b458]465
[fbcdeb8]466 if (size == 0)
467 return (uintptr_t) -1;
[a35b458]468
[fbcdeb8]469 /*
470 * Make sure we allocate from page-aligned
471 * address. Check for possible overflow in
472 * each step.
473 */
[a35b458]474
[fbcdeb8]475 size_t pages = SIZE2FRAMES(size);
[a35b458]476
[fbcdeb8]477 /*
478 * Find the lowest unmapped address aligned on the size
479 * boundary, not smaller than bound and of the required size.
480 */
[a35b458]481
[fbcdeb8]482 /* First check the bound address itself */
483 uintptr_t addr = ALIGN_UP(bound, PAGE_SIZE);
[35a3d950]484 if (addr >= bound) {
485 if (guarded) {
[7c3fb9b]486 /*
487 * Leave an unmapped page between the lower
[35a3d950]488 * bound and the area's start address.
489 */
490 addr += P2SZ(1);
491 }
492
493 if (check_area_conflicts(as, addr, pages, guarded, NULL))
494 return addr;
495 }
[a35b458]496
[fbcdeb8]497 /* Eventually check the addresses behind each area */
[88cc71c0]498 as_area_t *area = as_area_first(as);
499 while (area != NULL) {
500 mutex_lock(&area->lock);
[a35b458]501
[d9d0088]502 addr = area->base + P2SZ(area->pages);
[a35b458]503
[88cc71c0]504 if (guarded || area->flags & AS_AREA_GUARD) {
505 /*
506 * We must leave an unmapped page
507 * between the two areas.
508 */
509 addr += P2SZ(1);
510 }
[35a3d950]511
[88cc71c0]512 bool avail =
513 ((addr >= bound) && (addr >= area->base) &&
514 (check_area_conflicts(as, addr, pages, guarded, area)));
[35a3d950]515
[88cc71c0]516 mutex_unlock(&area->lock);
[a35b458]517
[88cc71c0]518 if (avail)
519 return addr;
[a35b458]520
[88cc71c0]521 area = as_area_next(area);
[fbcdeb8]522 }
[a35b458]523
[fbcdeb8]524 /* No suitable address space area found */
525 return (uintptr_t) -1;
526}
527
[83b6ba9f]528/** Remove reference to address space area share info.
529 *
530 * If the reference count drops to 0, the sh_info is deallocated.
531 *
532 * @param sh_info Pointer to address space area share info.
533 *
534 */
535NO_TRACE static void sh_info_remove_reference(share_info_t *sh_info)
536{
537 bool dealloc = false;
[a35b458]538
[83b6ba9f]539 mutex_lock(&sh_info->lock);
[63e27ef]540 assert(sh_info->refcount);
[a35b458]541
[83b6ba9f]542 if (--sh_info->refcount == 0) {
543 dealloc = true;
[a35b458]544
[83b6ba9f]545 /*
546 * Now walk carefully the pagemap B+tree and free/remove
547 * reference from all frames found there.
548 */
549 list_foreach(sh_info->pagemap.leaf_list, leaf_link,
550 btree_node_t, node) {
551 btree_key_t i;
[a35b458]552
[83b6ba9f]553 for (i = 0; i < node->keys; i++)
554 frame_free((uintptr_t) node->value[i], 1);
555 }
[a35b458]556
[83b6ba9f]557 }
558 mutex_unlock(&sh_info->lock);
[a35b458]559
[83b6ba9f]560 if (dealloc) {
561 if (sh_info->backend && sh_info->backend->destroy_shared_data) {
562 sh_info->backend->destroy_shared_data(
563 sh_info->backend_shared_data);
564 }
565 btree_destroy(&sh_info->pagemap);
566 free(sh_info);
567 }
568}
569
[20d50a1]570/** Create address space area of common attributes.
571 *
572 * The created address space area is added to the target address space.
573 *
[da1bafb]574 * @param as Target address space.
575 * @param flags Flags of the area memory.
576 * @param size Size of area.
577 * @param attrs Attributes of the area.
578 * @param backend Address space area backend. NULL if no backend is used.
[826599a2]579 * @param backend_data NULL or a pointer to custom backend data.
[fbcdeb8]580 * @param base Starting virtual address of the area.
[f2c3fed]581 * If set to AS_AREA_ANY, a suitable mappable area is
582 * found.
583 * @param bound Lowest address bound if base is set to AS_AREA_ANY.
[fbcdeb8]584 * Otherwise ignored.
[da1bafb]585 *
586 * @return Address space area on success or NULL on failure.
[20d50a1]587 *
588 */
[da1bafb]589as_area_t *as_area_create(as_t *as, unsigned int flags, size_t size,
[fbcdeb8]590 unsigned int attrs, mem_backend_t *backend,
591 mem_backend_data_t *backend_data, uintptr_t *base, uintptr_t bound)
[20d50a1]592{
[f2c3fed]593 if ((*base != (uintptr_t) AS_AREA_ANY) && !IS_ALIGNED(*base, PAGE_SIZE))
[37e7d2b9]594 return NULL;
[a35b458]595
[0b37882]596 if (size == 0)
[dbbeb26]597 return NULL;
[0941e9ae]598
[0b37882]599 size_t pages = SIZE2FRAMES(size);
[a35b458]600
[37e7d2b9]601 /* Writeable executable areas are not supported. */
602 if ((flags & AS_AREA_EXEC) && (flags & AS_AREA_WRITE))
603 return NULL;
[35a3d950]604
605 bool const guarded = flags & AS_AREA_GUARD;
[a35b458]606
[1068f6a]607 mutex_lock(&as->lock);
[a35b458]608
[f2c3fed]609 if (*base == (uintptr_t) AS_AREA_ANY) {
[35a3d950]610 *base = as_get_unmapped_area(as, bound, size, guarded);
[fbcdeb8]611 if (*base == (uintptr_t) -1) {
612 mutex_unlock(&as->lock);
613 return NULL;
614 }
615 }
[35a3d950]616
[83b6ba9f]617 if (overflows_into_positive(*base, size)) {
618 mutex_unlock(&as->lock);
[0941e9ae]619 return NULL;
[83b6ba9f]620 }
[0941e9ae]621
[35a3d950]622 if (!check_area_conflicts(as, *base, pages, guarded, NULL)) {
[1068f6a]623 mutex_unlock(&as->lock);
[37e7d2b9]624 return NULL;
625 }
[a35b458]626
[11b285d]627 as_area_t *area = (as_area_t *) malloc(sizeof(as_area_t));
[7473807]628 if (!area) {
629 mutex_unlock(&as->lock);
630 return NULL;
631 }
[a35b458]632
[da1bafb]633 mutex_initialize(&area->lock, MUTEX_PASSIVE);
[a35b458]634
[da1bafb]635 area->as = as;
[88cc71c0]636 odlink_initialize(&area->las_areas);
[da1bafb]637 area->flags = flags;
638 area->attributes = attrs;
[0b37882]639 area->pages = pages;
[fc47885]640 area->resident = 0;
[fbcdeb8]641 area->base = *base;
[da1bafb]642 area->backend = backend;
[83b6ba9f]643 area->sh_info = NULL;
[a35b458]644
[0ee077ee]645 if (backend_data)
[da1bafb]646 area->backend_data = *backend_data;
[0ee077ee]647 else
[da1bafb]648 memsetb(&area->backend_data, sizeof(area->backend_data), 0);
[83b6ba9f]649
650 share_info_t *si = NULL;
651
652 /*
[ae7d03c]653 * Create the sharing info structure.
654 * We do this in advance for every new area, even if it is not going
655 * to be shared.
656 */
[83b6ba9f]657 if (!(attrs & AS_AREA_ATTR_PARTIAL)) {
[11b285d]658 si = (share_info_t *) malloc(sizeof(share_info_t));
[7473807]659 if (!si) {
660 free(area);
661 mutex_unlock(&as->lock);
662 return NULL;
663 }
[83b6ba9f]664 mutex_initialize(&si->lock, MUTEX_PASSIVE);
665 si->refcount = 1;
666 si->shared = false;
667 si->backend_shared_data = NULL;
668 si->backend = backend;
669 btree_create(&si->pagemap);
670
671 area->sh_info = si;
[a35b458]672
[83b6ba9f]673 if (area->backend && area->backend->create_shared_data) {
674 if (!area->backend->create_shared_data(area)) {
675 free(area);
676 mutex_unlock(&as->lock);
677 sh_info_remove_reference(si);
678 return NULL;
679 }
680 }
681 }
682
[e394b736]683 if (area->backend && area->backend->create) {
684 if (!area->backend->create(area)) {
685 free(area);
686 mutex_unlock(&as->lock);
[83b6ba9f]687 if (!(attrs & AS_AREA_ATTR_PARTIAL))
688 sh_info_remove_reference(si);
[e394b736]689 return NULL;
690 }
691 }
[83b6ba9f]692
[da1bafb]693 btree_create(&area->used_space);
[88cc71c0]694 odict_insert(&area->las_areas, &as->as_areas, NULL);
[a35b458]695
[1068f6a]696 mutex_unlock(&as->lock);
[a35b458]697
[da1bafb]698 return area;
[20d50a1]699}
700
[e3ee9b9]701/** Find address space area and lock it.
702 *
703 * @param as Address space.
704 * @param va Virtual address.
705 *
706 * @return Locked address space area containing va on success or
707 * NULL on failure.
708 *
709 */
[7a0359b]710NO_TRACE static as_area_t *find_area_and_lock(as_t *as, uintptr_t va)
[e3ee9b9]711{
[63e27ef]712 assert(mutex_locked(&as->lock));
[a35b458]713
[88cc71c0]714 odlink_t *odlink = odict_find_leq(&as->as_areas, &va, NULL);
715 if (odlink == NULL)
716 return NULL;
[a35b458]717
[88cc71c0]718 as_area_t *area = odict_get_instance(odlink, as_area_t, las_areas);
719 mutex_lock(&area->lock);
[a35b458]720
[88cc71c0]721 assert(area->base <= va);
[a35b458]722
[88cc71c0]723 if (va <= area->base + (P2SZ(area->pages) - 1))
724 return area;
[a35b458]725
[88cc71c0]726 mutex_unlock(&area->lock);
[e3ee9b9]727 return NULL;
728}
729
[df0103f7]730/** Find address space area and change it.
731 *
[da1bafb]732 * @param as Address space.
733 * @param address Virtual address belonging to the area to be changed.
734 * Must be page-aligned.
735 * @param size New size of the virtual memory block starting at
736 * address.
737 * @param flags Flags influencing the remap operation. Currently unused.
738 *
739 * @return Zero on success or a value from @ref errno.h otherwise.
[df0103f7]740 *
[da1bafb]741 */
[b7fd2a0]742errno_t as_area_resize(as_t *as, uintptr_t address, size_t size, unsigned int flags)
[df0103f7]743{
[59fb782]744 if (!IS_ALIGNED(address, PAGE_SIZE))
745 return EINVAL;
746
[1068f6a]747 mutex_lock(&as->lock);
[a35b458]748
[df0103f7]749 /*
750 * Locate the area.
751 */
[da1bafb]752 as_area_t *area = find_area_and_lock(as, address);
[df0103f7]753 if (!area) {
[1068f6a]754 mutex_unlock(&as->lock);
[7242a78e]755 return ENOENT;
[df0103f7]756 }
[01029fc]757
758 if (!area->backend->is_resizable(area)) {
[df0103f7]759 /*
[01029fc]760 * The backend does not support resizing for this area.
[df0103f7]761 */
[1068f6a]762 mutex_unlock(&area->lock);
763 mutex_unlock(&as->lock);
[7242a78e]764 return ENOTSUP;
[df0103f7]765 }
[a35b458]766
[83b6ba9f]767 mutex_lock(&area->sh_info->lock);
768 if (area->sh_info->shared) {
[8182031]769 /*
[da1bafb]770 * Remapping of shared address space areas
[8182031]771 * is not supported.
772 */
[83b6ba9f]773 mutex_unlock(&area->sh_info->lock);
[8182031]774 mutex_unlock(&area->lock);
775 mutex_unlock(&as->lock);
776 return ENOTSUP;
777 }
[83b6ba9f]778 mutex_unlock(&area->sh_info->lock);
[a35b458]779
[da1bafb]780 size_t pages = SIZE2FRAMES((address - area->base) + size);
[df0103f7]781 if (!pages) {
782 /*
783 * Zero size address space areas are not allowed.
784 */
[1068f6a]785 mutex_unlock(&area->lock);
786 mutex_unlock(&as->lock);
[7242a78e]787 return EPERM;
[df0103f7]788 }
[a35b458]789
[df0103f7]790 if (pages < area->pages) {
[b6f3e7e]791 uintptr_t start_free = area->base + P2SZ(pages);
[a35b458]792
[df0103f7]793 /*
794 * Shrinking the area.
795 * No need to check for overlaps.
796 */
[a35b458]797
[c964521]798 page_table_lock(as, false);
[a35b458]799
[56789125]800 /*
801 * Remove frames belonging to used space starting from
802 * the highest addresses downwards until an overlap with
803 * the resized address space area is found. Note that this
804 * is also the right way to remove part of the used_space
805 * B+tree leaf list.
[da1bafb]806 */
807 bool cond = true;
808 while (cond) {
[63e27ef]809 assert(!list_empty(&area->used_space.leaf_list));
[a35b458]810
[da1bafb]811 btree_node_t *node =
[55b77d9]812 list_get_instance(list_last(&area->used_space.leaf_list),
[6f4495f5]813 btree_node_t, leaf_link);
[a35b458]814
[c81132d]815 if ((cond = (node->keys != 0))) {
[da1bafb]816 uintptr_t ptr = node->key[node->keys - 1];
[315b7e3]817 size_t node_size =
[98000fb]818 (size_t) node->value[node->keys - 1];
[da1bafb]819 size_t i = 0;
[a35b458]820
[315b7e3]821 if (overlaps(ptr, P2SZ(node_size), area->base,
[b6f3e7e]822 P2SZ(pages))) {
[a35b458]823
[315b7e3]824 if (ptr + P2SZ(node_size) <= start_free) {
[56789125]825 /*
[6f4495f5]826 * The whole interval fits
827 * completely in the resized
828 * address space area.
[56789125]829 */
830 break;
831 }
[a35b458]832
[56789125]833 /*
[6f4495f5]834 * Part of the interval corresponding
835 * to b and c overlaps with the resized
836 * address space area.
[56789125]837 */
[a35b458]838
[da1bafb]839 /* We are almost done */
840 cond = false;
841 i = (start_free - ptr) >> PAGE_WIDTH;
[6745592]842 if (!used_space_remove(area, start_free,
[315b7e3]843 node_size - i))
[da1bafb]844 panic("Cannot remove used space.");
[56789125]845 } else {
846 /*
[6f4495f5]847 * The interval of used space can be
848 * completely removed.
[56789125]849 */
[315b7e3]850 if (!used_space_remove(area, ptr, node_size))
[da1bafb]851 panic("Cannot remove used space.");
[56789125]852 }
[a35b458]853
[d67dfdc]854 /*
855 * Start TLB shootdown sequence.
856 *
857 * The sequence is rather short and can be
858 * repeated multiple times. The reason is that
859 * we don't want to have used_space_remove()
860 * inside the sequence as it may use a blocking
861 * memory allocation for its B+tree. Blocking
862 * while holding the tlblock spinlock is
863 * forbidden and would hit a kernel assertion.
864 */
865
866 ipl_t ipl = tlb_shootdown_start(TLB_INVL_PAGES,
867 as->asid, area->base + P2SZ(pages),
868 area->pages - pages);
[a35b458]869
[315b7e3]870 for (; i < node_size; i++) {
[38dc82d]871 pte_t pte;
872 bool found = page_mapping_find(as,
873 ptr + P2SZ(i), false, &pte);
[a35b458]874
[63e27ef]875 assert(found);
876 assert(PTE_VALID(&pte));
877 assert(PTE_PRESENT(&pte));
[a35b458]878
[da1bafb]879 if ((area->backend) &&
880 (area->backend->frame_free)) {
[0ee077ee]881 area->backend->frame_free(area,
[b6f3e7e]882 ptr + P2SZ(i),
[38dc82d]883 PTE_GET_FRAME(&pte));
[8182031]884 }
[a35b458]885
[b6f3e7e]886 page_mapping_remove(as, ptr + P2SZ(i));
[56789125]887 }
[a35b458]888
[d67dfdc]889 /*
890 * Finish TLB shootdown sequence.
891 */
[a35b458]892
[d67dfdc]893 tlb_invalidate_pages(as->asid,
894 area->base + P2SZ(pages),
895 area->pages - pages);
[a35b458]896
[d67dfdc]897 /*
898 * Invalidate software translation caches
899 * (e.g. TSB on sparc64, PHT on ppc32).
900 */
901 as_invalidate_translation_cache(as,
902 area->base + P2SZ(pages),
903 area->pages - pages);
904 tlb_shootdown_finalize(ipl);
905 }
906 }
[da1bafb]907 page_table_unlock(as, false);
[df0103f7]908 } else {
909 /*
910 * Growing the area.
[0941e9ae]911 */
912
[94795812]913 if (overflows_into_positive(address, P2SZ(pages)))
[0941e9ae]914 return EINVAL;
915
916 /*
[df0103f7]917 * Check for overlaps with other address space areas.
918 */
[35a3d950]919 bool const guarded = area->flags & AS_AREA_GUARD;
920 if (!check_area_conflicts(as, address, pages, guarded, area)) {
[1068f6a]921 mutex_unlock(&area->lock);
[da1bafb]922 mutex_unlock(&as->lock);
[7242a78e]923 return EADDRNOTAVAIL;
[df0103f7]924 }
[da1bafb]925 }
[a35b458]926
[e394b736]927 if (area->backend && area->backend->resize) {
928 if (!area->backend->resize(area, pages)) {
929 mutex_unlock(&area->lock);
930 mutex_unlock(&as->lock);
931 return ENOMEM;
932 }
933 }
[a35b458]934
[df0103f7]935 area->pages = pages;
[a35b458]936
[1068f6a]937 mutex_unlock(&area->lock);
938 mutex_unlock(&as->lock);
[a35b458]939
[7242a78e]940 return 0;
941}
942
943/** Destroy address space area.
944 *
[da1bafb]945 * @param as Address space.
946 * @param address Address within the area to be deleted.
947 *
948 * @return Zero on success or a value from @ref errno.h on failure.
[7242a78e]949 *
950 */
[b7fd2a0]951errno_t as_area_destroy(as_t *as, uintptr_t address)
[7242a78e]952{
[1068f6a]953 mutex_lock(&as->lock);
[a35b458]954
[da1bafb]955 as_area_t *area = find_area_and_lock(as, address);
[7242a78e]956 if (!area) {
[1068f6a]957 mutex_unlock(&as->lock);
[7242a78e]958 return ENOENT;
959 }
[e394b736]960
961 if (area->backend && area->backend->destroy)
962 area->backend->destroy(area);
[a35b458]963
[c964521]964 page_table_lock(as, false);
[a35b458]965
[5552d60]966 /*
967 * Start TLB shootdown sequence.
968 */
[402eda5]969 ipl_t ipl = tlb_shootdown_start(TLB_INVL_PAGES, as->asid, area->base,
970 area->pages);
[a35b458]971
[567807b1]972 /*
973 * Visit only the pages mapped by used_space B+tree.
974 */
[feeac0d]975 list_foreach(area->used_space.leaf_list, leaf_link, btree_node_t,
976 node) {
[da1bafb]977 btree_key_t i;
[a35b458]978
[f8d069e8]979 for (i = 0; i < node->keys; i++) {
[da1bafb]980 uintptr_t ptr = node->key[i];
981 size_t size;
[a35b458]982
[da1bafb]983 for (size = 0; size < (size_t) node->value[i]; size++) {
[38dc82d]984 pte_t pte;
985 bool found = page_mapping_find(as,
[3bacee1]986 ptr + P2SZ(size), false, &pte);
[a35b458]987
[63e27ef]988 assert(found);
989 assert(PTE_VALID(&pte));
990 assert(PTE_PRESENT(&pte));
[a35b458]991
[da1bafb]992 if ((area->backend) &&
993 (area->backend->frame_free)) {
994 area->backend->frame_free(area,
[b6f3e7e]995 ptr + P2SZ(size),
[38dc82d]996 PTE_GET_FRAME(&pte));
[56789125]997 }
[a35b458]998
[b6f3e7e]999 page_mapping_remove(as, ptr + P2SZ(size));
[7242a78e]1000 }
1001 }
1002 }
[a35b458]1003
[7242a78e]1004 /*
[5552d60]1005 * Finish TLB shootdown sequence.
[7242a78e]1006 */
[a35b458]1007
[f1d1f5d3]1008 tlb_invalidate_pages(as->asid, area->base, area->pages);
[a35b458]1009
[f1d1f5d3]1010 /*
[eef1b031]1011 * Invalidate potential software translation caches
1012 * (e.g. TSB on sparc64, PHT on ppc32).
[f1d1f5d3]1013 */
1014 as_invalidate_translation_cache(as, area->base, area->pages);
[402eda5]1015 tlb_shootdown_finalize(ipl);
[a35b458]1016
[c964521]1017 page_table_unlock(as, false);
[a35b458]1018
[5552d60]1019 btree_destroy(&area->used_space);
[a35b458]1020
[8d4f2ae]1021 area->attributes |= AS_AREA_ATTR_PARTIAL;
[a35b458]1022
[83b6ba9f]1023 sh_info_remove_reference(area->sh_info);
[a35b458]1024
[1068f6a]1025 mutex_unlock(&area->lock);
[a35b458]1026
[7242a78e]1027 /*
1028 * Remove the empty area from address space.
1029 */
[88cc71c0]1030 odict_remove(&area->las_areas);
[a35b458]1031
[8d4f2ae]1032 free(area);
[a35b458]1033
[f1d1f5d3]1034 mutex_unlock(&as->lock);
[7242a78e]1035 return 0;
[df0103f7]1036}
1037
[8d6bc2d5]1038/** Share address space area with another or the same address space.
[df0103f7]1039 *
[0ee077ee]1040 * Address space area mapping is shared with a new address space area.
1041 * If the source address space area has not been shared so far,
1042 * a new sh_info is created. The new address space area simply gets the
1043 * sh_info of the source area. The process of duplicating the
1044 * mapping is done through the backend share function.
[da1bafb]1045 *
1046 * @param src_as Pointer to source address space.
1047 * @param src_base Base address of the source address space area.
1048 * @param acc_size Expected size of the source area.
1049 * @param dst_as Pointer to destination address space.
[fd4d8c0]1050 * @param dst_flags_mask Destination address space area flags mask.
[fbcdeb8]1051 * @param dst_base Target base address. If set to -1,
1052 * a suitable mappable area is found.
1053 * @param bound Lowest address bound if dst_base is set to -1.
1054 * Otherwise ignored.
[df0103f7]1055 *
[da1bafb]1056 * @return Zero on success.
1057 * @return ENOENT if there is no such task or such address space.
1058 * @return EPERM if there was a problem in accepting the area.
1059 * @return ENOMEM if there was a problem in allocating destination
1060 * address space area.
1061 * @return ENOTSUP if the address space area backend does not support
1062 * sharing.
1063 *
[df0103f7]1064 */
[b7fd2a0]1065errno_t as_area_share(as_t *src_as, uintptr_t src_base, size_t acc_size,
[fbcdeb8]1066 as_t *dst_as, unsigned int dst_flags_mask, uintptr_t *dst_base,
1067 uintptr_t bound)
[df0103f7]1068{
[1068f6a]1069 mutex_lock(&src_as->lock);
[da1bafb]1070 as_area_t *src_area = find_area_and_lock(src_as, src_base);
[a9e8b39]1071 if (!src_area) {
[6fa476f7]1072 /*
1073 * Could not find the source address space area.
1074 */
[1068f6a]1075 mutex_unlock(&src_as->lock);
[6fa476f7]1076 return ENOENT;
1077 }
[a35b458]1078
[01029fc]1079 if (!src_area->backend->is_shareable(src_area)) {
[8d6bc2d5]1080 /*
[01029fc]1081 * The backend does not permit sharing of this area.
[8d6bc2d5]1082 */
1083 mutex_unlock(&src_area->lock);
1084 mutex_unlock(&src_as->lock);
1085 return ENOTSUP;
1086 }
[a35b458]1087
[b6f3e7e]1088 size_t src_size = P2SZ(src_area->pages);
[da1bafb]1089 unsigned int src_flags = src_area->flags;
1090 mem_backend_t *src_backend = src_area->backend;
1091 mem_backend_data_t src_backend_data = src_area->backend_data;
[a35b458]1092
[1ec1fd8]1093 /* Share the cacheable flag from the original mapping */
1094 if (src_flags & AS_AREA_CACHEABLE)
1095 dst_flags_mask |= AS_AREA_CACHEABLE;
[a35b458]1096
[da1bafb]1097 if ((src_size != acc_size) ||
1098 ((src_flags & dst_flags_mask) != dst_flags_mask)) {
[8d6bc2d5]1099 mutex_unlock(&src_area->lock);
1100 mutex_unlock(&src_as->lock);
[df0103f7]1101 return EPERM;
1102 }
[a35b458]1103
[8d6bc2d5]1104 /*
1105 * Now we are committed to sharing the area.
[8440473]1106 * First, prepare the area for sharing.
[8d6bc2d5]1107 * Then it will be safe to unlock it.
1108 */
[da1bafb]1109 share_info_t *sh_info = src_area->sh_info;
[a35b458]1110
[83b6ba9f]1111 mutex_lock(&sh_info->lock);
1112 sh_info->refcount++;
1113 bool shared = sh_info->shared;
1114 sh_info->shared = true;
1115 mutex_unlock(&sh_info->lock);
1116
1117 if (!shared) {
[c0697c4c]1118 /*
1119 * Call the backend to setup sharing.
[83b6ba9f]1120 * This only happens once for each sh_info.
[c0697c4c]1121 */
1122 src_area->backend->share(src_area);
[8d6bc2d5]1123 }
[a35b458]1124
[8d6bc2d5]1125 mutex_unlock(&src_area->lock);
1126 mutex_unlock(&src_as->lock);
[a35b458]1127
[df0103f7]1128 /*
[a9e8b39]1129 * Create copy of the source address space area.
1130 * The destination area is created with AS_AREA_ATTR_PARTIAL
1131 * attribute set which prevents race condition with
1132 * preliminary as_page_fault() calls.
[fd4d8c0]1133 * The flags of the source area are masked against dst_flags_mask
1134 * to support sharing in less privileged mode.
[df0103f7]1135 */
[fbcdeb8]1136 as_area_t *dst_area = as_area_create(dst_as, dst_flags_mask,
1137 src_size, AS_AREA_ATTR_PARTIAL, src_backend,
1138 &src_backend_data, dst_base, bound);
[a9e8b39]1139 if (!dst_area) {
[df0103f7]1140 /*
1141 * Destination address space area could not be created.
1142 */
[8d6bc2d5]1143 sh_info_remove_reference(sh_info);
[a35b458]1144
[df0103f7]1145 return ENOMEM;
1146 }
[a35b458]1147
[a9e8b39]1148 /*
1149 * Now the destination address space area has been
1150 * fully initialized. Clear the AS_AREA_ATTR_PARTIAL
[8d6bc2d5]1151 * attribute and set the sh_info.
[da1bafb]1152 */
1153 mutex_lock(&dst_as->lock);
[1068f6a]1154 mutex_lock(&dst_area->lock);
[a9e8b39]1155 dst_area->attributes &= ~AS_AREA_ATTR_PARTIAL;
[8d6bc2d5]1156 dst_area->sh_info = sh_info;
[1068f6a]1157 mutex_unlock(&dst_area->lock);
[da1bafb]1158 mutex_unlock(&dst_as->lock);
[a35b458]1159
[df0103f7]1160 return 0;
1161}
1162
[fb84455]1163/** Check access mode for address space area.
1164 *
[da1bafb]1165 * @param area Address space area.
1166 * @param access Access mode.
1167 *
1168 * @return False if access violates area's permissions, true
1169 * otherwise.
[fb84455]1170 *
1171 */
[97bdb4a]1172NO_TRACE bool as_area_check_access(as_area_t *area, pf_access_t access)
[fb84455]1173{
[63e27ef]1174 assert(mutex_locked(&area->lock));
[a35b458]1175
[fb84455]1176 int flagmap[] = {
1177 [PF_ACCESS_READ] = AS_AREA_READ,
1178 [PF_ACCESS_WRITE] = AS_AREA_WRITE,
1179 [PF_ACCESS_EXEC] = AS_AREA_EXEC
1180 };
[a35b458]1181
[fb84455]1182 if (!(area->flags & flagmap[access]))
1183 return false;
[a35b458]1184
[fb84455]1185 return true;
1186}
1187
[e3ee9b9]1188/** Convert address space area flags to page flags.
1189 *
1190 * @param aflags Flags of some address space area.
1191 *
1192 * @return Flags to be passed to page_mapping_insert().
1193 *
1194 */
[7a0359b]1195NO_TRACE static unsigned int area_flags_to_page_flags(unsigned int aflags)
[e3ee9b9]1196{
1197 unsigned int flags = PAGE_USER | PAGE_PRESENT;
[a35b458]1198
[e3ee9b9]1199 if (aflags & AS_AREA_READ)
1200 flags |= PAGE_READ;
[a35b458]1201
[e3ee9b9]1202 if (aflags & AS_AREA_WRITE)
1203 flags |= PAGE_WRITE;
[a35b458]1204
[e3ee9b9]1205 if (aflags & AS_AREA_EXEC)
1206 flags |= PAGE_EXEC;
[a35b458]1207
[e3ee9b9]1208 if (aflags & AS_AREA_CACHEABLE)
1209 flags |= PAGE_CACHEABLE;
[a35b458]1210
[e3ee9b9]1211 return flags;
1212}
1213
[6745592]1214/** Change adress space area flags.
[c98e6ee]1215 *
1216 * The idea is to have the same data, but with a different access mode.
1217 * This is needed e.g. for writing code into memory and then executing it.
1218 * In order for this to work properly, this may copy the data
1219 * into private anonymous memory (unless it's already there).
1220 *
[76fca31]1221 * @param as Address space.
1222 * @param flags Flags of the area memory.
1223 * @param address Address within the area to be changed.
1224 *
1225 * @return Zero on success or a value from @ref errno.h on failure.
[c98e6ee]1226 *
1227 */
[b7fd2a0]1228errno_t as_area_change_flags(as_t *as, unsigned int flags, uintptr_t address)
[c98e6ee]1229{
1230 /* Flags for the new memory mapping */
[da1bafb]1231 unsigned int page_flags = area_flags_to_page_flags(flags);
[a35b458]1232
[c98e6ee]1233 mutex_lock(&as->lock);
[a35b458]1234
[da1bafb]1235 as_area_t *area = find_area_and_lock(as, address);
[c98e6ee]1236 if (!area) {
1237 mutex_unlock(&as->lock);
1238 return ENOENT;
1239 }
[a35b458]1240
[83b6ba9f]1241 if (area->backend != &anon_backend) {
[c98e6ee]1242 /* Copying non-anonymous memory not supported yet */
1243 mutex_unlock(&area->lock);
1244 mutex_unlock(&as->lock);
1245 return ENOTSUP;
1246 }
[83b6ba9f]1247
1248 mutex_lock(&area->sh_info->lock);
1249 if (area->sh_info->shared) {
1250 /* Copying shared areas not supported yet */
1251 mutex_unlock(&area->sh_info->lock);
1252 mutex_unlock(&area->lock);
1253 mutex_unlock(&as->lock);
1254 return ENOTSUP;
1255 }
1256 mutex_unlock(&area->sh_info->lock);
[a35b458]1257
[c98e6ee]1258 /*
1259 * Compute total number of used pages in the used_space B+tree
1260 */
[da1bafb]1261 size_t used_pages = 0;
[a35b458]1262
[feeac0d]1263 list_foreach(area->used_space.leaf_list, leaf_link, btree_node_t,
1264 node) {
[da1bafb]1265 btree_key_t i;
[a35b458]1266
[da1bafb]1267 for (i = 0; i < node->keys; i++)
[98000fb]1268 used_pages += (size_t) node->value[i];
[c98e6ee]1269 }
[a35b458]1270
[c98e6ee]1271 /* An array for storing frame numbers */
[11b285d]1272 uintptr_t *old_frame = malloc(used_pages * sizeof(uintptr_t));
[7473807]1273 if (!old_frame) {
1274 mutex_unlock(&area->lock);
1275 mutex_unlock(&as->lock);
1276 return ENOMEM;
1277 }
[a35b458]1278
[c964521]1279 page_table_lock(as, false);
[a35b458]1280
[c98e6ee]1281 /*
1282 * Start TLB shootdown sequence.
1283 */
[402eda5]1284 ipl_t ipl = tlb_shootdown_start(TLB_INVL_PAGES, as->asid, area->base,
1285 area->pages);
[a35b458]1286
[c98e6ee]1287 /*
1288 * Remove used pages from page tables and remember their frame
1289 * numbers.
1290 */
[da1bafb]1291 size_t frame_idx = 0;
[a35b458]1292
[feeac0d]1293 list_foreach(area->used_space.leaf_list, leaf_link, btree_node_t,
1294 node) {
[da1bafb]1295 btree_key_t i;
[a35b458]1296
[c98e6ee]1297 for (i = 0; i < node->keys; i++) {
[da1bafb]1298 uintptr_t ptr = node->key[i];
1299 size_t size;
[a35b458]1300
[da1bafb]1301 for (size = 0; size < (size_t) node->value[i]; size++) {
[38dc82d]1302 pte_t pte;
1303 bool found = page_mapping_find(as,
1304 ptr + P2SZ(size), false, &pte);
[a35b458]1305
[63e27ef]1306 assert(found);
1307 assert(PTE_VALID(&pte));
1308 assert(PTE_PRESENT(&pte));
[a35b458]1309
[38dc82d]1310 old_frame[frame_idx++] = PTE_GET_FRAME(&pte);
[a35b458]1311
[c98e6ee]1312 /* Remove old mapping */
[b6f3e7e]1313 page_mapping_remove(as, ptr + P2SZ(size));
[c98e6ee]1314 }
1315 }
1316 }
[a35b458]1317
[c98e6ee]1318 /*
1319 * Finish TLB shootdown sequence.
1320 */
[a35b458]1321
[c98e6ee]1322 tlb_invalidate_pages(as->asid, area->base, area->pages);
[a35b458]1323
[c98e6ee]1324 /*
[eef1b031]1325 * Invalidate potential software translation caches
1326 * (e.g. TSB on sparc64, PHT on ppc32).
[c98e6ee]1327 */
1328 as_invalidate_translation_cache(as, area->base, area->pages);
[402eda5]1329 tlb_shootdown_finalize(ipl);
[a35b458]1330
[c964521]1331 page_table_unlock(as, false);
[a35b458]1332
[ae7f6fb]1333 /*
1334 * Set the new flags.
1335 */
1336 area->flags = flags;
[a35b458]1337
[c98e6ee]1338 /*
1339 * Map pages back in with new flags. This step is kept separate
[6745592]1340 * so that the memory area could not be accesed with both the old and
1341 * the new flags at once.
[c98e6ee]1342 */
1343 frame_idx = 0;
[a35b458]1344
[feeac0d]1345 list_foreach(area->used_space.leaf_list, leaf_link, btree_node_t,
1346 node) {
[da1bafb]1347 btree_key_t i;
[a35b458]1348
[c98e6ee]1349 for (i = 0; i < node->keys; i++) {
[da1bafb]1350 uintptr_t ptr = node->key[i];
1351 size_t size;
[a35b458]1352
[da1bafb]1353 for (size = 0; size < (size_t) node->value[i]; size++) {
[c98e6ee]1354 page_table_lock(as, false);
[a35b458]1355
[c98e6ee]1356 /* Insert the new mapping */
[b6f3e7e]1357 page_mapping_insert(as, ptr + P2SZ(size),
[c98e6ee]1358 old_frame[frame_idx++], page_flags);
[a35b458]1359
[c98e6ee]1360 page_table_unlock(as, false);
1361 }
1362 }
1363 }
[a35b458]1364
[c98e6ee]1365 free(old_frame);
[a35b458]1366
[c98e6ee]1367 mutex_unlock(&area->lock);
1368 mutex_unlock(&as->lock);
[a35b458]1369
[c98e6ee]1370 return 0;
1371}
1372
[20d50a1]1373/** Handle page fault within the current address space.
1374 *
[6745592]1375 * This is the high-level page fault handler. It decides whether the page fault
1376 * can be resolved by any backend and if so, it invokes the backend to resolve
1377 * the page fault.
[8182031]1378 *
[20d50a1]1379 * Interrupts are assumed disabled.
1380 *
[59fb782]1381 * @param address Faulting address.
1382 * @param access Access mode that caused the page fault (i.e.
1383 * read/write/exec).
1384 * @param istate Pointer to the interrupted state.
[da1bafb]1385 *
1386 * @return AS_PF_FAULT on page fault.
1387 * @return AS_PF_OK on success.
1388 * @return AS_PF_DEFER if the fault was caused by copy_to_uspace()
1389 * or copy_from_uspace().
[20d50a1]1390 *
1391 */
[59fb782]1392int as_page_fault(uintptr_t address, pf_access_t access, istate_t *istate)
[20d50a1]1393{
[59fb782]1394 uintptr_t page = ALIGN_DOWN(address, PAGE_SIZE);
[908bb96]1395 int rc = AS_PF_FAULT;
1396
[1068f6a]1397 if (!THREAD)
[1dbc43f]1398 goto page_fault;
[a35b458]1399
[7af8c0e]1400 if (!AS)
[1dbc43f]1401 goto page_fault;
[a35b458]1402
[1068f6a]1403 mutex_lock(&AS->lock);
[da1bafb]1404 as_area_t *area = find_area_and_lock(AS, page);
[20d50a1]1405 if (!area) {
1406 /*
1407 * No area contained mapping for 'page'.
1408 * Signal page fault to low-level handler.
1409 */
[1068f6a]1410 mutex_unlock(&AS->lock);
[e3c762cd]1411 goto page_fault;
[20d50a1]1412 }
[a35b458]1413
[a9e8b39]1414 if (area->attributes & AS_AREA_ATTR_PARTIAL) {
1415 /*
1416 * The address space area is not fully initialized.
1417 * Avoid possible race by returning error.
1418 */
[1068f6a]1419 mutex_unlock(&area->lock);
1420 mutex_unlock(&AS->lock);
[da1bafb]1421 goto page_fault;
[a9e8b39]1422 }
[a35b458]1423
[da1bafb]1424 if ((!area->backend) || (!area->backend->page_fault)) {
[8182031]1425 /*
1426 * The address space area is not backed by any backend
1427 * or the backend cannot handle page faults.
1428 */
1429 mutex_unlock(&area->lock);
1430 mutex_unlock(&AS->lock);
[da1bafb]1431 goto page_fault;
[8182031]1432 }
[a35b458]1433
[2299914]1434 page_table_lock(AS, false);
[a35b458]1435
[2299914]1436 /*
[6745592]1437 * To avoid race condition between two page faults on the same address,
1438 * we need to make sure the mapping has not been already inserted.
[2299914]1439 */
[38dc82d]1440 pte_t pte;
1441 bool found = page_mapping_find(AS, page, false, &pte);
[560b81c]1442 if (found && PTE_PRESENT(&pte)) {
1443 if (((access == PF_ACCESS_READ) && PTE_READABLE(&pte)) ||
1444 (access == PF_ACCESS_WRITE && PTE_WRITABLE(&pte)) ||
1445 (access == PF_ACCESS_EXEC && PTE_EXECUTABLE(&pte))) {
1446 page_table_unlock(AS, false);
1447 mutex_unlock(&area->lock);
1448 mutex_unlock(&AS->lock);
1449 return AS_PF_OK;
[2299914]1450 }
1451 }
[a35b458]1452
[20d50a1]1453 /*
[8182031]1454 * Resort to the backend page fault handler.
[20d50a1]1455 */
[908bb96]1456 rc = area->backend->page_fault(area, page, access);
1457 if (rc != AS_PF_OK) {
[8182031]1458 page_table_unlock(AS, false);
1459 mutex_unlock(&area->lock);
1460 mutex_unlock(&AS->lock);
1461 goto page_fault;
1462 }
[a35b458]1463
[8182031]1464 page_table_unlock(AS, false);
[1068f6a]1465 mutex_unlock(&area->lock);
1466 mutex_unlock(&AS->lock);
[e3c762cd]1467 return AS_PF_OK;
[a35b458]1468
[e3c762cd]1469page_fault:
1470 if (THREAD->in_copy_from_uspace) {
1471 THREAD->in_copy_from_uspace = false;
[6f4495f5]1472 istate_set_retaddr(istate,
1473 (uintptr_t) &memcpy_from_uspace_failover_address);
[e3c762cd]1474 } else if (THREAD->in_copy_to_uspace) {
1475 THREAD->in_copy_to_uspace = false;
[6f4495f5]1476 istate_set_retaddr(istate,
1477 (uintptr_t) &memcpy_to_uspace_failover_address);
[908bb96]1478 } else if (rc == AS_PF_SILENT) {
1479 printf("Killing task %" PRIu64 " due to a "
1480 "failed late reservation request.\n", TASK->taskid);
1481 task_kill_self(true);
[e3c762cd]1482 } else {
[59fb782]1483 fault_if_from_uspace(istate, "Page fault: %p.", (void *) address);
1484 panic_memtrap(istate, access, address, NULL);
[e3c762cd]1485 }
[a35b458]1486
[e3c762cd]1487 return AS_PF_DEFER;
[20d50a1]1488}
1489
[7e4e532]1490/** Switch address spaces.
[1068f6a]1491 *
1492 * Note that this function cannot sleep as it is essentially a part of
[879585a3]1493 * scheduling. Sleeping here would lead to deadlock on wakeup. Another
1494 * thing which is forbidden in this context is locking the address space.
[20d50a1]1495 *
[7250d2c]1496 * When this function is entered, no spinlocks may be held.
[31d8e10]1497 *
[da1bafb]1498 * @param old Old address space or NULL.
1499 * @param new New address space.
1500 *
[20d50a1]1501 */
[80bcaed]1502void as_switch(as_t *old_as, as_t *new_as)
[20d50a1]1503{
[31d8e10]1504 DEADLOCK_PROBE_INIT(p_asidlock);
1505 preemption_disable();
[a35b458]1506
[31d8e10]1507retry:
1508 (void) interrupts_disable();
1509 if (!spinlock_trylock(&asidlock)) {
[da1bafb]1510 /*
[31d8e10]1511 * Avoid deadlock with TLB shootdown.
1512 * We can enable interrupts here because
1513 * preemption is disabled. We should not be
1514 * holding any other lock.
1515 */
1516 (void) interrupts_enable();
1517 DEADLOCK_PROBE(p_asidlock, DEADLOCK_THRESHOLD);
1518 goto retry;
1519 }
1520 preemption_enable();
[a35b458]1521
[7e4e532]1522 /*
1523 * First, take care of the old address space.
[da1bafb]1524 */
[80bcaed]1525 if (old_as) {
[63e27ef]1526 assert(old_as->cpu_refcount);
[a35b458]1527
[da1bafb]1528 if ((--old_as->cpu_refcount == 0) && (old_as != AS_KERNEL)) {
[7e4e532]1529 /*
1530 * The old address space is no longer active on
1531 * any processor. It can be appended to the
1532 * list of inactive address spaces with assigned
1533 * ASID.
1534 */
[63e27ef]1535 assert(old_as->asid != ASID_INVALID);
[a35b458]1536
[2057572]1537 list_append(&old_as->inactive_as_with_asid_link,
[55b77d9]1538 &inactive_as_with_asid_list);
[7e4e532]1539 }
[a35b458]1540
[57da95c]1541 /*
1542 * Perform architecture-specific tasks when the address space
1543 * is being removed from the CPU.
1544 */
[80bcaed]1545 as_deinstall_arch(old_as);
[7e4e532]1546 }
[a35b458]1547
[7e4e532]1548 /*
1549 * Second, prepare the new address space.
1550 */
[80bcaed]1551 if ((new_as->cpu_refcount++ == 0) && (new_as != AS_KERNEL)) {
[879585a3]1552 if (new_as->asid != ASID_INVALID)
[80bcaed]1553 list_remove(&new_as->inactive_as_with_asid_link);
[879585a3]1554 else
1555 new_as->asid = asid_get();
[7e4e532]1556 }
[a35b458]1557
[80bcaed]1558#ifdef AS_PAGE_TABLE
1559 SET_PTL0_ADDRESS(new_as->genarch.page_table);
1560#endif
[a35b458]1561
[20d50a1]1562 /*
1563 * Perform architecture-specific steps.
[4512d7e]1564 * (e.g. write ASID to hardware register etc.)
[20d50a1]1565 */
[80bcaed]1566 as_install_arch(new_as);
[a35b458]1567
[879585a3]1568 spinlock_unlock(&asidlock);
[a35b458]1569
[80bcaed]1570 AS = new_as;
[20d50a1]1571}
[6a3c9a7]1572
[df0103f7]1573/** Compute flags for virtual address translation subsytem.
1574 *
[da1bafb]1575 * @param area Address space area.
1576 *
1577 * @return Flags to be used in page_mapping_insert().
[df0103f7]1578 *
1579 */
[97bdb4a]1580NO_TRACE unsigned int as_area_get_flags(as_area_t *area)
[df0103f7]1581{
[63e27ef]1582 assert(mutex_locked(&area->lock));
[a35b458]1583
[da1bafb]1584 return area_flags_to_page_flags(area->flags);
[df0103f7]1585}
1586
[88cc71c0]1587/** Get key function for the @c as_t.as_areas ordered dictionary.
1588 *
1589 * @param odlink Link
1590 * @return Pointer to task ID cast as 'void *'
1591 */
1592static void *as_areas_getkey(odlink_t *odlink)
1593{
1594 as_area_t *area = odict_get_instance(odlink, as_area_t, las_areas);
1595 return (void *) &area->base;
1596}
1597
1598/** Key comparison function for the @c as_t.as_areas ordered dictionary.
1599 *
1600 * @param a Pointer to area A base
1601 * @param b Pointer to area B base
1602 * @return -1, 0, 1 iff base of A is lower than, equal to, higher than B
1603 */
1604static int as_areas_cmp(void *a, void *b)
1605{
1606 uintptr_t base_a = *(uintptr_t *)a;
1607 uintptr_t base_b = *(uintptr_t *)b;
1608
1609 if (base_a < base_b)
1610 return -1;
1611 else if (base_a == base_b)
1612 return 0;
1613 else
1614 return +1;
1615}
1616
[ef67bab]1617/** Create page table.
1618 *
[6745592]1619 * Depending on architecture, create either address space private or global page
1620 * table.
[ef67bab]1621 *
[da1bafb]1622 * @param flags Flags saying whether the page table is for the kernel
1623 * address space.
1624 *
1625 * @return First entry of the page table.
[ef67bab]1626 *
1627 */
[97bdb4a]1628NO_TRACE pte_t *page_table_create(unsigned int flags)
[ef67bab]1629{
[63e27ef]1630 assert(as_operations);
1631 assert(as_operations->page_table_create);
[a35b458]1632
[bd1deed]1633 return as_operations->page_table_create(flags);
[ef67bab]1634}
[d3e7ff4]1635
[482826d]1636/** Destroy page table.
1637 *
1638 * Destroy page table in architecture specific way.
1639 *
[da1bafb]1640 * @param page_table Physical address of PTL0.
1641 *
[482826d]1642 */
[97bdb4a]1643NO_TRACE void page_table_destroy(pte_t *page_table)
[482826d]1644{
[63e27ef]1645 assert(as_operations);
1646 assert(as_operations->page_table_destroy);
[a35b458]1647
[bd1deed]1648 as_operations->page_table_destroy(page_table);
[482826d]1649}
1650
[2299914]1651/** Lock page table.
1652 *
1653 * This function should be called before any page_mapping_insert(),
1654 * page_mapping_remove() and page_mapping_find().
[da1bafb]1655 *
[2299914]1656 * Locking order is such that address space areas must be locked
1657 * prior to this call. Address space can be locked prior to this
1658 * call in which case the lock argument is false.
1659 *
[da1bafb]1660 * @param as Address space.
1661 * @param lock If false, do not attempt to lock as->lock.
1662 *
[2299914]1663 */
[97bdb4a]1664NO_TRACE void page_table_lock(as_t *as, bool lock)
[2299914]1665{
[63e27ef]1666 assert(as_operations);
1667 assert(as_operations->page_table_lock);
[a35b458]1668
[2299914]1669 as_operations->page_table_lock(as, lock);
1670}
1671
1672/** Unlock page table.
1673 *
[da1bafb]1674 * @param as Address space.
1675 * @param unlock If false, do not attempt to unlock as->lock.
1676 *
[2299914]1677 */
[97bdb4a]1678NO_TRACE void page_table_unlock(as_t *as, bool unlock)
[2299914]1679{
[63e27ef]1680 assert(as_operations);
1681 assert(as_operations->page_table_unlock);
[a35b458]1682
[2299914]1683 as_operations->page_table_unlock(as, unlock);
1684}
1685
[ada559c]1686/** Test whether page tables are locked.
1687 *
[e3ee9b9]1688 * @param as Address space where the page tables belong.
[ada559c]1689 *
[e3ee9b9]1690 * @return True if the page tables belonging to the address soace
1691 * are locked, otherwise false.
[ada559c]1692 */
[97bdb4a]1693NO_TRACE bool page_table_locked(as_t *as)
[ada559c]1694{
[63e27ef]1695 assert(as_operations);
1696 assert(as_operations->page_table_locked);
[ada559c]1697
1698 return as_operations->page_table_locked(as);
1699}
1700
[b878df3]1701/** Return size of the address space area with given base.
1702 *
[1d432f9]1703 * @param base Arbitrary address inside the address space area.
[da1bafb]1704 *
1705 * @return Size of the address space area in bytes or zero if it
1706 * does not exist.
[b878df3]1707 *
1708 */
1709size_t as_area_get_size(uintptr_t base)
[7c23af9]1710{
1711 size_t size;
[a35b458]1712
[1d432f9]1713 page_table_lock(AS, true);
[da1bafb]1714 as_area_t *src_area = find_area_and_lock(AS, base);
[a35b458]1715
[6745592]1716 if (src_area) {
[b6f3e7e]1717 size = P2SZ(src_area->pages);
[1068f6a]1718 mutex_unlock(&src_area->lock);
[da1bafb]1719 } else
[7c23af9]1720 size = 0;
[a35b458]1721
[1d432f9]1722 page_table_unlock(AS, true);
[7c23af9]1723 return size;
1724}
1725
[25bf215]1726/** Mark portion of address space area as used.
1727 *
1728 * The address space area must be already locked.
1729 *
[da1bafb]1730 * @param area Address space area.
1731 * @param page First page to be marked.
1732 * @param count Number of page to be marked.
1733 *
[fc47885]1734 * @return False on failure or true on success.
[25bf215]1735 *
1736 */
[fc47885]1737bool used_space_insert(as_area_t *area, uintptr_t page, size_t count)
[25bf215]1738{
[63e27ef]1739 assert(mutex_locked(&area->lock));
1740 assert(IS_ALIGNED(page, PAGE_SIZE));
1741 assert(count);
[a35b458]1742
[566da7f8]1743 btree_node_t *leaf = NULL;
[da1bafb]1744 size_t pages = (size_t) btree_search(&area->used_space, page, &leaf);
[25bf215]1745 if (pages) {
1746 /*
1747 * We hit the beginning of some used space.
1748 */
[fc47885]1749 return false;
[25bf215]1750 }
[566da7f8]1751
[63e27ef]1752 assert(leaf != NULL);
[a35b458]1753
[a6cb8cb]1754 if (!leaf->keys) {
[da1bafb]1755 btree_insert(&area->used_space, page, (void *) count, leaf);
[fc47885]1756 goto success;
[a6cb8cb]1757 }
[a35b458]1758
[da1bafb]1759 btree_node_t *node = btree_leaf_node_left_neighbour(&area->used_space, leaf);
[25bf215]1760 if (node) {
[6f4495f5]1761 uintptr_t left_pg = node->key[node->keys - 1];
1762 uintptr_t right_pg = leaf->key[0];
[98000fb]1763 size_t left_cnt = (size_t) node->value[node->keys - 1];
1764 size_t right_cnt = (size_t) leaf->value[0];
[a35b458]1765
[25bf215]1766 /*
1767 * Examine the possibility that the interval fits
1768 * somewhere between the rightmost interval of
1769 * the left neigbour and the first interval of the leaf.
1770 */
[a35b458]1771
[25bf215]1772 if (page >= right_pg) {
1773 /* Do nothing. */
[b6f3e7e]1774 } else if (overlaps(page, P2SZ(count), left_pg,
1775 P2SZ(left_cnt))) {
[25bf215]1776 /* The interval intersects with the left interval. */
[fc47885]1777 return false;
[b6f3e7e]1778 } else if (overlaps(page, P2SZ(count), right_pg,
1779 P2SZ(right_cnt))) {
[25bf215]1780 /* The interval intersects with the right interval. */
[fc47885]1781 return false;
[b6f3e7e]1782 } else if ((page == left_pg + P2SZ(left_cnt)) &&
1783 (page + P2SZ(count) == right_pg)) {
[6f4495f5]1784 /*
1785 * The interval can be added by merging the two already
1786 * present intervals.
1787 */
[56789125]1788 node->value[node->keys - 1] += count + right_cnt;
[da1bafb]1789 btree_remove(&area->used_space, right_pg, leaf);
[fc47885]1790 goto success;
[b6f3e7e]1791 } else if (page == left_pg + P2SZ(left_cnt)) {
[da1bafb]1792 /*
[6f4495f5]1793 * The interval can be added by simply growing the left
1794 * interval.
1795 */
[56789125]1796 node->value[node->keys - 1] += count;
[fc47885]1797 goto success;
[b6f3e7e]1798 } else if (page + P2SZ(count) == right_pg) {
[25bf215]1799 /*
[6f4495f5]1800 * The interval can be addded by simply moving base of
1801 * the right interval down and increasing its size
1802 * accordingly.
[25bf215]1803 */
[56789125]1804 leaf->value[0] += count;
[25bf215]1805 leaf->key[0] = page;
[fc47885]1806 goto success;
[25bf215]1807 } else {
1808 /*
1809 * The interval is between both neigbouring intervals,
1810 * but cannot be merged with any of them.
1811 */
[da1bafb]1812 btree_insert(&area->used_space, page, (void *) count,
[6f4495f5]1813 leaf);
[fc47885]1814 goto success;
[25bf215]1815 }
1816 } else if (page < leaf->key[0]) {
[7f1c620]1817 uintptr_t right_pg = leaf->key[0];
[98000fb]1818 size_t right_cnt = (size_t) leaf->value[0];
[a35b458]1819
[25bf215]1820 /*
[6f4495f5]1821 * Investigate the border case in which the left neighbour does
1822 * not exist but the interval fits from the left.
[25bf215]1823 */
[a35b458]1824
[b6f3e7e]1825 if (overlaps(page, P2SZ(count), right_pg, P2SZ(right_cnt))) {
[25bf215]1826 /* The interval intersects with the right interval. */
[fc47885]1827 return false;
[b6f3e7e]1828 } else if (page + P2SZ(count) == right_pg) {
[25bf215]1829 /*
[6f4495f5]1830 * The interval can be added by moving the base of the
1831 * right interval down and increasing its size
1832 * accordingly.
[25bf215]1833 */
1834 leaf->key[0] = page;
[56789125]1835 leaf->value[0] += count;
[fc47885]1836 goto success;
[25bf215]1837 } else {
1838 /*
1839 * The interval doesn't adjoin with the right interval.
1840 * It must be added individually.
1841 */
[da1bafb]1842 btree_insert(&area->used_space, page, (void *) count,
[6f4495f5]1843 leaf);
[fc47885]1844 goto success;
[25bf215]1845 }
1846 }
[a35b458]1847
[da1bafb]1848 node = btree_leaf_node_right_neighbour(&area->used_space, leaf);
[25bf215]1849 if (node) {
[6f4495f5]1850 uintptr_t left_pg = leaf->key[leaf->keys - 1];
1851 uintptr_t right_pg = node->key[0];
[98000fb]1852 size_t left_cnt = (size_t) leaf->value[leaf->keys - 1];
1853 size_t right_cnt = (size_t) node->value[0];
[a35b458]1854
[25bf215]1855 /*
1856 * Examine the possibility that the interval fits
1857 * somewhere between the leftmost interval of
1858 * the right neigbour and the last interval of the leaf.
1859 */
[a35b458]1860
[25bf215]1861 if (page < left_pg) {
1862 /* Do nothing. */
[b6f3e7e]1863 } else if (overlaps(page, P2SZ(count), left_pg,
1864 P2SZ(left_cnt))) {
[25bf215]1865 /* The interval intersects with the left interval. */
[fc47885]1866 return false;
[b6f3e7e]1867 } else if (overlaps(page, P2SZ(count), right_pg,
1868 P2SZ(right_cnt))) {
[25bf215]1869 /* The interval intersects with the right interval. */
[fc47885]1870 return false;
[b6f3e7e]1871 } else if ((page == left_pg + P2SZ(left_cnt)) &&
1872 (page + P2SZ(count) == right_pg)) {
[6f4495f5]1873 /*
1874 * The interval can be added by merging the two already
1875 * present intervals.
[da1bafb]1876 */
[56789125]1877 leaf->value[leaf->keys - 1] += count + right_cnt;
[da1bafb]1878 btree_remove(&area->used_space, right_pg, node);
[fc47885]1879 goto success;
[b6f3e7e]1880 } else if (page == left_pg + P2SZ(left_cnt)) {
[6f4495f5]1881 /*
1882 * The interval can be added by simply growing the left
1883 * interval.
[da1bafb]1884 */
[fc47885]1885 leaf->value[leaf->keys - 1] += count;
1886 goto success;
[b6f3e7e]1887 } else if (page + P2SZ(count) == right_pg) {
[25bf215]1888 /*
[6f4495f5]1889 * The interval can be addded by simply moving base of
1890 * the right interval down and increasing its size
1891 * accordingly.
[25bf215]1892 */
[56789125]1893 node->value[0] += count;
[25bf215]1894 node->key[0] = page;
[fc47885]1895 goto success;
[25bf215]1896 } else {
1897 /*
1898 * The interval is between both neigbouring intervals,
1899 * but cannot be merged with any of them.
1900 */
[da1bafb]1901 btree_insert(&area->used_space, page, (void *) count,
[6f4495f5]1902 leaf);
[fc47885]1903 goto success;
[25bf215]1904 }
1905 } else if (page >= leaf->key[leaf->keys - 1]) {
[7f1c620]1906 uintptr_t left_pg = leaf->key[leaf->keys - 1];
[98000fb]1907 size_t left_cnt = (size_t) leaf->value[leaf->keys - 1];
[a35b458]1908
[25bf215]1909 /*
[6f4495f5]1910 * Investigate the border case in which the right neighbour
1911 * does not exist but the interval fits from the right.
[25bf215]1912 */
[a35b458]1913
[b6f3e7e]1914 if (overlaps(page, P2SZ(count), left_pg, P2SZ(left_cnt))) {
[56789125]1915 /* The interval intersects with the left interval. */
[fc47885]1916 return false;
[b6f3e7e]1917 } else if (left_pg + P2SZ(left_cnt) == page) {
[6f4495f5]1918 /*
1919 * The interval can be added by growing the left
1920 * interval.
1921 */
[56789125]1922 leaf->value[leaf->keys - 1] += count;
[fc47885]1923 goto success;
[25bf215]1924 } else {
1925 /*
1926 * The interval doesn't adjoin with the left interval.
1927 * It must be added individually.
1928 */
[da1bafb]1929 btree_insert(&area->used_space, page, (void *) count,
[6f4495f5]1930 leaf);
[fc47885]1931 goto success;
[25bf215]1932 }
1933 }
[a35b458]1934
[25bf215]1935 /*
[6f4495f5]1936 * Note that if the algorithm made it thus far, the interval can fit
1937 * only between two other intervals of the leaf. The two border cases
1938 * were already resolved.
[25bf215]1939 */
[da1bafb]1940 btree_key_t i;
[25bf215]1941 for (i = 1; i < leaf->keys; i++) {
1942 if (page < leaf->key[i]) {
[6f4495f5]1943 uintptr_t left_pg = leaf->key[i - 1];
1944 uintptr_t right_pg = leaf->key[i];
[98000fb]1945 size_t left_cnt = (size_t) leaf->value[i - 1];
1946 size_t right_cnt = (size_t) leaf->value[i];
[a35b458]1947
[25bf215]1948 /*
1949 * The interval fits between left_pg and right_pg.
1950 */
[a35b458]1951
[b6f3e7e]1952 if (overlaps(page, P2SZ(count), left_pg,
1953 P2SZ(left_cnt))) {
[6f4495f5]1954 /*
1955 * The interval intersects with the left
1956 * interval.
1957 */
[fc47885]1958 return false;
[b6f3e7e]1959 } else if (overlaps(page, P2SZ(count), right_pg,
1960 P2SZ(right_cnt))) {
[6f4495f5]1961 /*
1962 * The interval intersects with the right
1963 * interval.
1964 */
[fc47885]1965 return false;
[b6f3e7e]1966 } else if ((page == left_pg + P2SZ(left_cnt)) &&
1967 (page + P2SZ(count) == right_pg)) {
[6f4495f5]1968 /*
1969 * The interval can be added by merging the two
1970 * already present intervals.
1971 */
[56789125]1972 leaf->value[i - 1] += count + right_cnt;
[da1bafb]1973 btree_remove(&area->used_space, right_pg, leaf);
[fc47885]1974 goto success;
[b6f3e7e]1975 } else if (page == left_pg + P2SZ(left_cnt)) {
[6f4495f5]1976 /*
1977 * The interval can be added by simply growing
1978 * the left interval.
1979 */
[56789125]1980 leaf->value[i - 1] += count;
[fc47885]1981 goto success;
[b6f3e7e]1982 } else if (page + P2SZ(count) == right_pg) {
[25bf215]1983 /*
[da1bafb]1984 * The interval can be addded by simply moving
[6f4495f5]1985 * base of the right interval down and
1986 * increasing its size accordingly.
[da1bafb]1987 */
[56789125]1988 leaf->value[i] += count;
[25bf215]1989 leaf->key[i] = page;
[fc47885]1990 goto success;
[25bf215]1991 } else {
1992 /*
[6f4495f5]1993 * The interval is between both neigbouring
1994 * intervals, but cannot be merged with any of
1995 * them.
[25bf215]1996 */
[da1bafb]1997 btree_insert(&area->used_space, page,
[6f4495f5]1998 (void *) count, leaf);
[fc47885]1999 goto success;
[25bf215]2000 }
2001 }
2002 }
[a35b458]2003
[7e752b2]2004 panic("Inconsistency detected while adding %zu pages of used "
2005 "space at %p.", count, (void *) page);
[a35b458]2006
[fc47885]2007success:
2008 area->resident += count;
2009 return true;
[25bf215]2010}
2011
2012/** Mark portion of address space area as unused.
2013 *
2014 * The address space area must be already locked.
2015 *
[da1bafb]2016 * @param area Address space area.
2017 * @param page First page to be marked.
2018 * @param count Number of page to be marked.
2019 *
[fc47885]2020 * @return False on failure or true on success.
[25bf215]2021 *
2022 */
[fc47885]2023bool used_space_remove(as_area_t *area, uintptr_t page, size_t count)
[25bf215]2024{
[63e27ef]2025 assert(mutex_locked(&area->lock));
2026 assert(IS_ALIGNED(page, PAGE_SIZE));
2027 assert(count);
[a35b458]2028
[da1bafb]2029 btree_node_t *leaf;
2030 size_t pages = (size_t) btree_search(&area->used_space, page, &leaf);
[25bf215]2031 if (pages) {
2032 /*
2033 * We are lucky, page is the beginning of some interval.
2034 */
2035 if (count > pages) {
[fc47885]2036 return false;
[25bf215]2037 } else if (count == pages) {
[da1bafb]2038 btree_remove(&area->used_space, page, leaf);
[fc47885]2039 goto success;
[25bf215]2040 } else {
2041 /*
2042 * Find the respective interval.
2043 * Decrease its size and relocate its start address.
2044 */
[da1bafb]2045 btree_key_t i;
[25bf215]2046 for (i = 0; i < leaf->keys; i++) {
2047 if (leaf->key[i] == page) {
[b6f3e7e]2048 leaf->key[i] += P2SZ(count);
[56789125]2049 leaf->value[i] -= count;
[fc47885]2050 goto success;
[25bf215]2051 }
2052 }
[a35b458]2053
[25bf215]2054 goto error;
2055 }
2056 }
[a35b458]2057
[b6f3e7e]2058 btree_node_t *node = btree_leaf_node_left_neighbour(&area->used_space,
2059 leaf);
[da1bafb]2060 if ((node) && (page < leaf->key[0])) {
[7f1c620]2061 uintptr_t left_pg = node->key[node->keys - 1];
[98000fb]2062 size_t left_cnt = (size_t) node->value[node->keys - 1];
[a35b458]2063
[b6f3e7e]2064 if (overlaps(left_pg, P2SZ(left_cnt), page, P2SZ(count))) {
2065 if (page + P2SZ(count) == left_pg + P2SZ(left_cnt)) {
[25bf215]2066 /*
[6f4495f5]2067 * The interval is contained in the rightmost
2068 * interval of the left neighbour and can be
2069 * removed by updating the size of the bigger
2070 * interval.
[25bf215]2071 */
[56789125]2072 node->value[node->keys - 1] -= count;
[fc47885]2073 goto success;
[b6f3e7e]2074 } else if (page + P2SZ(count) <
2075 left_pg + P2SZ(left_cnt)) {
2076 size_t new_cnt;
2077
[25bf215]2078 /*
[6f4495f5]2079 * The interval is contained in the rightmost
2080 * interval of the left neighbour but its
2081 * removal requires both updating the size of
2082 * the original interval and also inserting a
2083 * new interval.
[25bf215]2084 */
[b6f3e7e]2085 new_cnt = ((left_pg + P2SZ(left_cnt)) -
2086 (page + P2SZ(count))) >> PAGE_WIDTH;
[56789125]2087 node->value[node->keys - 1] -= count + new_cnt;
[da1bafb]2088 btree_insert(&area->used_space, page +
[b6f3e7e]2089 P2SZ(count), (void *) new_cnt, leaf);
[fc47885]2090 goto success;
[25bf215]2091 }
2092 }
[a35b458]2093
[fc47885]2094 return false;
[da1bafb]2095 } else if (page < leaf->key[0])
[fc47885]2096 return false;
[a35b458]2097
[25bf215]2098 if (page > leaf->key[leaf->keys - 1]) {
[7f1c620]2099 uintptr_t left_pg = leaf->key[leaf->keys - 1];
[98000fb]2100 size_t left_cnt = (size_t) leaf->value[leaf->keys - 1];
[a35b458]2101
[b6f3e7e]2102 if (overlaps(left_pg, P2SZ(left_cnt), page, P2SZ(count))) {
2103 if (page + P2SZ(count) == left_pg + P2SZ(left_cnt)) {
[25bf215]2104 /*
[6f4495f5]2105 * The interval is contained in the rightmost
2106 * interval of the leaf and can be removed by
2107 * updating the size of the bigger interval.
[25bf215]2108 */
[56789125]2109 leaf->value[leaf->keys - 1] -= count;
[fc47885]2110 goto success;
[b6f3e7e]2111 } else if (page + P2SZ(count) < left_pg +
2112 P2SZ(left_cnt)) {
2113 size_t new_cnt;
2114
[25bf215]2115 /*
[6f4495f5]2116 * The interval is contained in the rightmost
2117 * interval of the leaf but its removal
2118 * requires both updating the size of the
2119 * original interval and also inserting a new
2120 * interval.
[25bf215]2121 */
[b6f3e7e]2122 new_cnt = ((left_pg + P2SZ(left_cnt)) -
2123 (page + P2SZ(count))) >> PAGE_WIDTH;
[56789125]2124 leaf->value[leaf->keys - 1] -= count + new_cnt;
[da1bafb]2125 btree_insert(&area->used_space, page +
[b6f3e7e]2126 P2SZ(count), (void *) new_cnt, leaf);
[fc47885]2127 goto success;
[25bf215]2128 }
2129 }
[a35b458]2130
[fc47885]2131 return false;
[da1bafb]2132 }
[a35b458]2133
[25bf215]2134 /*
2135 * The border cases have been already resolved.
[fc47885]2136 * Now the interval can be only between intervals of the leaf.
[25bf215]2137 */
[da1bafb]2138 btree_key_t i;
[25bf215]2139 for (i = 1; i < leaf->keys - 1; i++) {
2140 if (page < leaf->key[i]) {
[7f1c620]2141 uintptr_t left_pg = leaf->key[i - 1];
[98000fb]2142 size_t left_cnt = (size_t) leaf->value[i - 1];
[a35b458]2143
[25bf215]2144 /*
[6f4495f5]2145 * Now the interval is between intervals corresponding
2146 * to (i - 1) and i.
[25bf215]2147 */
[b6f3e7e]2148 if (overlaps(left_pg, P2SZ(left_cnt), page,
2149 P2SZ(count))) {
2150 if (page + P2SZ(count) ==
2151 left_pg + P2SZ(left_cnt)) {
[25bf215]2152 /*
[6f4495f5]2153 * The interval is contained in the
2154 * interval (i - 1) of the leaf and can
2155 * be removed by updating the size of
2156 * the bigger interval.
[25bf215]2157 */
[56789125]2158 leaf->value[i - 1] -= count;
[fc47885]2159 goto success;
[b6f3e7e]2160 } else if (page + P2SZ(count) <
2161 left_pg + P2SZ(left_cnt)) {
2162 size_t new_cnt;
2163
[25bf215]2164 /*
[6f4495f5]2165 * The interval is contained in the
2166 * interval (i - 1) of the leaf but its
2167 * removal requires both updating the
2168 * size of the original interval and
[25bf215]2169 * also inserting a new interval.
2170 */
[b6f3e7e]2171 new_cnt = ((left_pg + P2SZ(left_cnt)) -
2172 (page + P2SZ(count))) >>
[6f4495f5]2173 PAGE_WIDTH;
[56789125]2174 leaf->value[i - 1] -= count + new_cnt;
[da1bafb]2175 btree_insert(&area->used_space, page +
[b6f3e7e]2176 P2SZ(count), (void *) new_cnt,
[6f4495f5]2177 leaf);
[fc47885]2178 goto success;
[25bf215]2179 }
2180 }
[a35b458]2181
[fc47885]2182 return false;
[25bf215]2183 }
2184 }
[a35b458]2185
[25bf215]2186error:
[7e752b2]2187 panic("Inconsistency detected while removing %zu pages of used "
2188 "space from %p.", count, (void *) page);
[a35b458]2189
[fc47885]2190success:
2191 area->resident -= count;
2192 return true;
[25bf215]2193}
2194
[df0103f7]2195/*
2196 * Address space related syscalls.
2197 */
2198
[fbcdeb8]2199sysarg_t sys_as_area_create(uintptr_t base, size_t size, unsigned int flags,
[ae6021d]2200 uintptr_t bound, as_area_pager_info_t *pager_info)
[df0103f7]2201{
[fbcdeb8]2202 uintptr_t virt = base;
[75b139f]2203 mem_backend_t *backend;
2204 mem_backend_data_t backend_data;
2205
[ae6021d]2206 if (pager_info == AS_AREA_UNPAGED)
[75b139f]2207 backend = &anon_backend;
2208 else {
2209 backend = &user_backend;
[ae6021d]2210 if (copy_from_uspace(&backend_data.pager_info, pager_info,
[3bacee1]2211 sizeof(as_area_pager_info_t)) != EOK) {
[ae6021d]2212 return (sysarg_t) AS_MAP_FAILED;
2213 }
[75b139f]2214 }
[c4c2406]2215 as_area_t *area = as_area_create(AS, flags, size,
[75b139f]2216 AS_AREA_ATTR_NONE, backend, &backend_data, &virt, bound);
[fbcdeb8]2217 if (area == NULL)
[f2c3fed]2218 return (sysarg_t) AS_MAP_FAILED;
[a35b458]2219
[fbcdeb8]2220 return (sysarg_t) virt;
[df0103f7]2221}
2222
[b7fd2a0]2223sys_errno_t sys_as_area_resize(uintptr_t address, size_t size, unsigned int flags)
[df0103f7]2224{
[b7fd2a0]2225 return (sys_errno_t) as_area_resize(AS, address, size, 0);
[7242a78e]2226}
2227
[b7fd2a0]2228sys_errno_t sys_as_area_change_flags(uintptr_t address, unsigned int flags)
[c98e6ee]2229{
[b7fd2a0]2230 return (sys_errno_t) as_area_change_flags(AS, flags, address);
[c98e6ee]2231}
2232
[b7fd2a0]2233sys_errno_t sys_as_area_destroy(uintptr_t address)
[7242a78e]2234{
[b7fd2a0]2235 return (sys_errno_t) as_area_destroy(AS, address);
[df0103f7]2236}
[b45c443]2237
[336db295]2238/** Get list of adress space areas.
2239 *
[da1bafb]2240 * @param as Address space.
2241 * @param obuf Place to save pointer to returned buffer.
2242 * @param osize Place to save size of returned buffer.
2243 *
[336db295]2244 */
2245void as_get_area_info(as_t *as, as_area_info_t **obuf, size_t *osize)
2246{
2247 mutex_lock(&as->lock);
[a35b458]2248
[88cc71c0]2249 /* Count number of areas. */
2250 size_t area_cnt = odict_count(&as->as_areas);
[a35b458]2251
[da1bafb]2252 size_t isize = area_cnt * sizeof(as_area_info_t);
[11b285d]2253 as_area_info_t *info = nfmalloc(isize);
[a35b458]2254
[88cc71c0]2255 /* Record area data. */
[a35b458]2256
[da1bafb]2257 size_t area_idx = 0;
[a35b458]2258
[88cc71c0]2259 as_area_t *area = as_area_first(as);
2260 while (area != NULL) {
2261 assert(area_idx < area_cnt);
2262 mutex_lock(&area->lock);
[a35b458]2263
[88cc71c0]2264 info[area_idx].start_addr = area->base;
2265 info[area_idx].size = P2SZ(area->pages);
2266 info[area_idx].flags = area->flags;
2267 ++area_idx;
[a35b458]2268
[88cc71c0]2269 mutex_unlock(&area->lock);
2270 area = as_area_next(area);
[336db295]2271 }
[a35b458]2272
[336db295]2273 mutex_unlock(&as->lock);
[a35b458]2274
[336db295]2275 *obuf = info;
2276 *osize = isize;
2277}
2278
[64c2ad5]2279/** Print out information about address space.
2280 *
[da1bafb]2281 * @param as Address space.
2282 *
[64c2ad5]2283 */
2284void as_print(as_t *as)
2285{
2286 mutex_lock(&as->lock);
[a35b458]2287
[0b37882]2288 /* Print out info about address space areas */
[88cc71c0]2289 as_area_t *area = as_area_first(as);
2290 while (area != NULL) {
2291 mutex_lock(&area->lock);
2292 printf("as_area: %p, base=%p, pages=%zu"
2293 " (%p - %p)\n", area, (void *) area->base,
2294 area->pages, (void *) area->base,
2295 (void *) (area->base + P2SZ(area->pages)));
2296 mutex_unlock(&area->lock);
[a35b458]2297
[88cc71c0]2298 area = as_area_next(area);
[64c2ad5]2299 }
[a35b458]2300
[64c2ad5]2301 mutex_unlock(&as->lock);
2302}
2303
[cc73a8a1]2304/** @}
[b45c443]2305 */
Note: See TracBrowser for help on using the repository browser.