mmu.c 57.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
/*
 * Copyright (C) 2012 - Virtual Open Systems and Columbia University
 * Author: Christoffer Dall <c.dall@virtualopensystems.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License, version 2, as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
 */
18 19 20 21

#include <linux/mman.h>
#include <linux/kvm_host.h>
#include <linux/io.h>
22
#include <linux/hugetlb.h>
23
#include <linux/sched/signal.h>
C
Christoffer Dall 已提交
24
#include <trace/events/kvm.h>
25
#include <asm/pgalloc.h>
26
#include <asm/cacheflush.h>
27 28
#include <asm/kvm_arm.h>
#include <asm/kvm_mmu.h>
C
Christoffer Dall 已提交
29
#include <asm/kvm_mmio.h>
30
#include <asm/kvm_asm.h>
31
#include <asm/kvm_emulate.h>
32
#include <asm/virt.h>
33
#include <asm/system_misc.h>
34 35

#include "trace.h"
36

37
static pgd_t *boot_hyp_pgd;
38
static pgd_t *hyp_pgd;
39
static pgd_t *merged_hyp_pgd;
40 41
static DEFINE_MUTEX(kvm_hyp_pgd_mutex);

42 43 44 45
static unsigned long hyp_idmap_start;
static unsigned long hyp_idmap_end;
static phys_addr_t hyp_idmap_vector;

46 47
static unsigned long io_map_base;

48
#define S2_PGD_SIZE	(PTRS_PER_S2_PGD * sizeof(pgd_t))
49
#define hyp_pgd_order get_order(PTRS_PER_PGD * sizeof(pgd_t))
50

51 52 53 54 55 56
#define KVM_S2PTE_FLAG_IS_IOMAP		(1UL << 0)
#define KVM_S2_FLAG_LOGGING_ACTIVE	(1UL << 1)

static bool memslot_is_logging(struct kvm_memory_slot *memslot)
{
	return memslot->dirty_bitmap && !(memslot->flags & KVM_MEM_READONLY);
57 58 59 60 61 62 63 64 65 66 67
}

/**
 * kvm_flush_remote_tlbs() - flush all VM TLB entries for v7/8
 * @kvm:	pointer to kvm structure.
 *
 * Interface to HYP function to flush all VM TLB entries
 */
void kvm_flush_remote_tlbs(struct kvm *kvm)
{
	kvm_call_hyp(__kvm_tlb_flush_vmid, kvm);
68
}
69

70
static void kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa)
71
{
72
	kvm_call_hyp(__kvm_tlb_flush_vmid_ipa, kvm, ipa);
73 74
}

75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
/*
 * D-Cache management functions. They take the page table entries by
 * value, as they are flushing the cache using the kernel mapping (or
 * kmap on 32bit).
 */
static void kvm_flush_dcache_pte(pte_t pte)
{
	__kvm_flush_dcache_pte(pte);
}

static void kvm_flush_dcache_pmd(pmd_t pmd)
{
	__kvm_flush_dcache_pmd(pmd);
}

static void kvm_flush_dcache_pud(pud_t pud)
{
	__kvm_flush_dcache_pud(pud);
}

95 96 97 98 99
static bool kvm_is_device_pfn(unsigned long pfn)
{
	return !pfn_valid(pfn);
}

100 101 102 103 104 105 106 107 108 109 110
/**
 * stage2_dissolve_pmd() - clear and flush huge PMD entry
 * @kvm:	pointer to kvm structure.
 * @addr:	IPA
 * @pmd:	pmd pointer for IPA
 *
 * Function clears a PMD entry, flushes addr 1st and 2nd stage TLBs. Marks all
 * pages in the range dirty.
 */
static void stage2_dissolve_pmd(struct kvm *kvm, phys_addr_t addr, pmd_t *pmd)
{
111
	if (!pmd_thp_or_huge(*pmd))
112 113 114 115 116 117 118
		return;

	pmd_clear(pmd);
	kvm_tlb_flush_vmid_ipa(kvm, addr);
	put_page(virt_to_page(pmd));
}

119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache,
				  int min, int max)
{
	void *page;

	BUG_ON(max > KVM_NR_MEM_OBJS);
	if (cache->nobjs >= min)
		return 0;
	while (cache->nobjs < max) {
		page = (void *)__get_free_page(PGALLOC_GFP);
		if (!page)
			return -ENOMEM;
		cache->objects[cache->nobjs++] = page;
	}
	return 0;
}

static void mmu_free_memory_cache(struct kvm_mmu_memory_cache *mc)
{
	while (mc->nobjs)
		free_page((unsigned long)mc->objects[--mc->nobjs]);
}

static void *mmu_memory_cache_alloc(struct kvm_mmu_memory_cache *mc)
{
	void *p;

	BUG_ON(!mc || !mc->nobjs);
	p = mc->objects[--mc->nobjs];
	return p;
}

151
static void clear_stage2_pgd_entry(struct kvm *kvm, pgd_t *pgd, phys_addr_t addr)
152
{
153 154
	pud_t *pud_table __maybe_unused = stage2_pud_offset(pgd, 0UL);
	stage2_pgd_clear(pgd);
155
	kvm_tlb_flush_vmid_ipa(kvm, addr);
156
	stage2_pud_free(pud_table);
157
	put_page(virt_to_page(pgd));
158 159
}

160
static void clear_stage2_pud_entry(struct kvm *kvm, pud_t *pud, phys_addr_t addr)
161
{
162 163 164
	pmd_t *pmd_table __maybe_unused = stage2_pmd_offset(pud, 0);
	VM_BUG_ON(stage2_pud_huge(*pud));
	stage2_pud_clear(pud);
165
	kvm_tlb_flush_vmid_ipa(kvm, addr);
166
	stage2_pmd_free(pmd_table);
167 168
	put_page(virt_to_page(pud));
}
169

170
static void clear_stage2_pmd_entry(struct kvm *kvm, pmd_t *pmd, phys_addr_t addr)
171
{
172
	pte_t *pte_table = pte_offset_kernel(pmd, 0);
173
	VM_BUG_ON(pmd_thp_or_huge(*pmd));
174 175 176
	pmd_clear(pmd);
	kvm_tlb_flush_vmid_ipa(kvm, addr);
	pte_free_kernel(NULL, pte_table);
177 178 179
	put_page(virt_to_page(pmd));
}

180 181 182 183 184 185 186 187 188 189 190 191
static inline void kvm_set_pte(pte_t *ptep, pte_t new_pte)
{
	WRITE_ONCE(*ptep, new_pte);
	dsb(ishst);
}

static inline void kvm_set_pmd(pmd_t *pmdp, pmd_t new_pmd)
{
	WRITE_ONCE(*pmdp, new_pmd);
	dsb(ishst);
}

192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208
static inline void kvm_pmd_populate(pmd_t *pmdp, pte_t *ptep)
{
	kvm_set_pmd(pmdp, kvm_mk_pmd(ptep));
}

static inline void kvm_pud_populate(pud_t *pudp, pmd_t *pmdp)
{
	WRITE_ONCE(*pudp, kvm_mk_pud(pmdp));
	dsb(ishst);
}

static inline void kvm_pgd_populate(pgd_t *pgdp, pud_t *pudp)
{
	WRITE_ONCE(*pgdp, kvm_mk_pgd(pudp));
	dsb(ishst);
}

209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
/*
 * Unmapping vs dcache management:
 *
 * If a guest maps certain memory pages as uncached, all writes will
 * bypass the data cache and go directly to RAM.  However, the CPUs
 * can still speculate reads (not writes) and fill cache lines with
 * data.
 *
 * Those cache lines will be *clean* cache lines though, so a
 * clean+invalidate operation is equivalent to an invalidate
 * operation, because no cache lines are marked dirty.
 *
 * Those clean cache lines could be filled prior to an uncached write
 * by the guest, and the cache coherent IO subsystem would therefore
 * end up writing old data to disk.
 *
 * This is why right after unmapping a page/section and invalidating
 * the corresponding TLBs, we call kvm_flush_dcache_p*() to make sure
 * the IO subsystem will never hit in the cache.
228 229 230 231
 *
 * This is all avoided on systems that have ARM64_HAS_STAGE2_FWB, as
 * we then fully enforce cacheability of RAM, no matter what the guest
 * does.
232
 */
233
static void unmap_stage2_ptes(struct kvm *kvm, pmd_t *pmd,
234
		       phys_addr_t addr, phys_addr_t end)
235
{
236 237 238 239 240 241
	phys_addr_t start_addr = addr;
	pte_t *pte, *start_pte;

	start_pte = pte = pte_offset_kernel(pmd, addr);
	do {
		if (!pte_none(*pte)) {
242 243
			pte_t old_pte = *pte;

244 245
			kvm_set_pte(pte, __pte(0));
			kvm_tlb_flush_vmid_ipa(kvm, addr);
246 247

			/* No need to invalidate the cache for device mappings */
248
			if (!kvm_is_device_pfn(pte_pfn(old_pte)))
249 250 251
				kvm_flush_dcache_pte(old_pte);

			put_page(virt_to_page(pte));
252 253 254
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);

255 256
	if (stage2_pte_table_empty(start_pte))
		clear_stage2_pmd_entry(kvm, pmd, start_addr);
257 258
}

259
static void unmap_stage2_pmds(struct kvm *kvm, pud_t *pud,
260
		       phys_addr_t addr, phys_addr_t end)
261
{
262 263
	phys_addr_t next, start_addr = addr;
	pmd_t *pmd, *start_pmd;
264

265
	start_pmd = pmd = stage2_pmd_offset(pud, addr);
266
	do {
267
		next = stage2_pmd_addr_end(addr, end);
268
		if (!pmd_none(*pmd)) {
269
			if (pmd_thp_or_huge(*pmd)) {
270 271
				pmd_t old_pmd = *pmd;

272 273
				pmd_clear(pmd);
				kvm_tlb_flush_vmid_ipa(kvm, addr);
274 275 276

				kvm_flush_dcache_pmd(old_pmd);

277 278
				put_page(virt_to_page(pmd));
			} else {
279
				unmap_stage2_ptes(kvm, pmd, addr, next);
280
			}
281
		}
282
	} while (pmd++, addr = next, addr != end);
283

284 285
	if (stage2_pmd_table_empty(start_pmd))
		clear_stage2_pud_entry(kvm, pud, start_addr);
286
}
287

288
static void unmap_stage2_puds(struct kvm *kvm, pgd_t *pgd,
289 290 291 292
		       phys_addr_t addr, phys_addr_t end)
{
	phys_addr_t next, start_addr = addr;
	pud_t *pud, *start_pud;
293

294
	start_pud = pud = stage2_pud_offset(pgd, addr);
295
	do {
296 297 298
		next = stage2_pud_addr_end(addr, end);
		if (!stage2_pud_none(*pud)) {
			if (stage2_pud_huge(*pud)) {
299 300
				pud_t old_pud = *pud;

301
				stage2_pud_clear(pud);
302
				kvm_tlb_flush_vmid_ipa(kvm, addr);
303
				kvm_flush_dcache_pud(old_pud);
304 305
				put_page(virt_to_page(pud));
			} else {
306
				unmap_stage2_pmds(kvm, pud, addr, next);
307 308
			}
		}
309
	} while (pud++, addr = next, addr != end);
310

311 312
	if (stage2_pud_table_empty(start_pud))
		clear_stage2_pgd_entry(kvm, pgd, start_addr);
313 314
}

315 316 317 318 319 320 321 322 323 324 325 326
/**
 * unmap_stage2_range -- Clear stage2 page table entries to unmap a range
 * @kvm:   The VM pointer
 * @start: The intermediate physical base address of the range to unmap
 * @size:  The size of the area to unmap
 *
 * Clear a range of stage-2 mappings, lowering the various ref-counts.  Must
 * be called while holding mmu_lock (unless for freeing the stage2 pgd before
 * destroying the VM), otherwise another faulting VCPU may come in and mess
 * with things behind our backs.
 */
static void unmap_stage2_range(struct kvm *kvm, phys_addr_t start, u64 size)
327 328 329 330 331
{
	pgd_t *pgd;
	phys_addr_t addr = start, end = start + size;
	phys_addr_t next;

332
	assert_spin_locked(&kvm->mmu_lock);
333 334
	WARN_ON(size & ~PAGE_MASK);

335
	pgd = kvm->arch.pgd + stage2_pgd_index(addr);
336
	do {
337 338 339 340 341 342 343
		/*
		 * Make sure the page table is still active, as another thread
		 * could have possibly freed the page table, while we released
		 * the lock.
		 */
		if (!READ_ONCE(kvm->arch.pgd))
			break;
344 345 346
		next = stage2_pgd_addr_end(addr, end);
		if (!stage2_pgd_none(*pgd))
			unmap_stage2_puds(kvm, pgd, addr, next);
347 348 349 350 351 352
		/*
		 * If the range is too large, release the kvm->mmu_lock
		 * to prevent starvation and lockup detector warnings.
		 */
		if (next != end)
			cond_resched_lock(&kvm->mmu_lock);
353
	} while (pgd++, addr = next, addr != end);
354 355
}

356 357 358 359 360 361 362
static void stage2_flush_ptes(struct kvm *kvm, pmd_t *pmd,
			      phys_addr_t addr, phys_addr_t end)
{
	pte_t *pte;

	pte = pte_offset_kernel(pmd, addr);
	do {
363
		if (!pte_none(*pte) && !kvm_is_device_pfn(pte_pfn(*pte)))
364
			kvm_flush_dcache_pte(*pte);
365 366 367 368 369 370 371 372 373
	} while (pte++, addr += PAGE_SIZE, addr != end);
}

static void stage2_flush_pmds(struct kvm *kvm, pud_t *pud,
			      phys_addr_t addr, phys_addr_t end)
{
	pmd_t *pmd;
	phys_addr_t next;

374
	pmd = stage2_pmd_offset(pud, addr);
375
	do {
376
		next = stage2_pmd_addr_end(addr, end);
377
		if (!pmd_none(*pmd)) {
378
			if (pmd_thp_or_huge(*pmd))
379 380
				kvm_flush_dcache_pmd(*pmd);
			else
381 382 383 384 385 386 387 388 389 390 391
				stage2_flush_ptes(kvm, pmd, addr, next);
		}
	} while (pmd++, addr = next, addr != end);
}

static void stage2_flush_puds(struct kvm *kvm, pgd_t *pgd,
			      phys_addr_t addr, phys_addr_t end)
{
	pud_t *pud;
	phys_addr_t next;

392
	pud = stage2_pud_offset(pgd, addr);
393
	do {
394 395 396
		next = stage2_pud_addr_end(addr, end);
		if (!stage2_pud_none(*pud)) {
			if (stage2_pud_huge(*pud))
397 398
				kvm_flush_dcache_pud(*pud);
			else
399 400 401 402 403 404 405 406 407 408 409 410 411
				stage2_flush_pmds(kvm, pud, addr, next);
		}
	} while (pud++, addr = next, addr != end);
}

static void stage2_flush_memslot(struct kvm *kvm,
				 struct kvm_memory_slot *memslot)
{
	phys_addr_t addr = memslot->base_gfn << PAGE_SHIFT;
	phys_addr_t end = addr + PAGE_SIZE * memslot->npages;
	phys_addr_t next;
	pgd_t *pgd;

412
	pgd = kvm->arch.pgd + stage2_pgd_index(addr);
413
	do {
414
		next = stage2_pgd_addr_end(addr, end);
415 416
		if (!stage2_pgd_none(*pgd))
			stage2_flush_puds(kvm, pgd, addr, next);
417 418 419 420 421 422 423 424 425 426
	} while (pgd++, addr = next, addr != end);
}

/**
 * stage2_flush_vm - Invalidate cache for pages mapped in stage 2
 * @kvm: The struct kvm pointer
 *
 * Go through the stage 2 page tables and invalidate any cache lines
 * backing memory already mapped to the VM.
 */
427
static void stage2_flush_vm(struct kvm *kvm)
428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
{
	struct kvm_memslots *slots;
	struct kvm_memory_slot *memslot;
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
	spin_lock(&kvm->mmu_lock);

	slots = kvm_memslots(kvm);
	kvm_for_each_memslot(memslot, slots)
		stage2_flush_memslot(kvm, memslot);

	spin_unlock(&kvm->mmu_lock);
	srcu_read_unlock(&kvm->srcu, idx);
}

444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519
static void clear_hyp_pgd_entry(pgd_t *pgd)
{
	pud_t *pud_table __maybe_unused = pud_offset(pgd, 0UL);
	pgd_clear(pgd);
	pud_free(NULL, pud_table);
	put_page(virt_to_page(pgd));
}

static void clear_hyp_pud_entry(pud_t *pud)
{
	pmd_t *pmd_table __maybe_unused = pmd_offset(pud, 0);
	VM_BUG_ON(pud_huge(*pud));
	pud_clear(pud);
	pmd_free(NULL, pmd_table);
	put_page(virt_to_page(pud));
}

static void clear_hyp_pmd_entry(pmd_t *pmd)
{
	pte_t *pte_table = pte_offset_kernel(pmd, 0);
	VM_BUG_ON(pmd_thp_or_huge(*pmd));
	pmd_clear(pmd);
	pte_free_kernel(NULL, pte_table);
	put_page(virt_to_page(pmd));
}

static void unmap_hyp_ptes(pmd_t *pmd, phys_addr_t addr, phys_addr_t end)
{
	pte_t *pte, *start_pte;

	start_pte = pte = pte_offset_kernel(pmd, addr);
	do {
		if (!pte_none(*pte)) {
			kvm_set_pte(pte, __pte(0));
			put_page(virt_to_page(pte));
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);

	if (hyp_pte_table_empty(start_pte))
		clear_hyp_pmd_entry(pmd);
}

static void unmap_hyp_pmds(pud_t *pud, phys_addr_t addr, phys_addr_t end)
{
	phys_addr_t next;
	pmd_t *pmd, *start_pmd;

	start_pmd = pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		/* Hyp doesn't use huge pmds */
		if (!pmd_none(*pmd))
			unmap_hyp_ptes(pmd, addr, next);
	} while (pmd++, addr = next, addr != end);

	if (hyp_pmd_table_empty(start_pmd))
		clear_hyp_pud_entry(pud);
}

static void unmap_hyp_puds(pgd_t *pgd, phys_addr_t addr, phys_addr_t end)
{
	phys_addr_t next;
	pud_t *pud, *start_pud;

	start_pud = pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		/* Hyp doesn't use huge puds */
		if (!pud_none(*pud))
			unmap_hyp_pmds(pud, addr, next);
	} while (pud++, addr = next, addr != end);

	if (hyp_pud_table_empty(start_pud))
		clear_hyp_pgd_entry(pgd);
}

520 521 522 523 524 525 526
static unsigned int kvm_pgd_index(unsigned long addr, unsigned int ptrs_per_pgd)
{
	return (addr >> PGDIR_SHIFT) & (ptrs_per_pgd - 1);
}

static void __unmap_hyp_range(pgd_t *pgdp, unsigned long ptrs_per_pgd,
			      phys_addr_t start, u64 size)
527 528 529 530 531 532 533 534 535
{
	pgd_t *pgd;
	phys_addr_t addr = start, end = start + size;
	phys_addr_t next;

	/*
	 * We don't unmap anything from HYP, except at the hyp tear down.
	 * Hence, we don't have to invalidate the TLBs here.
	 */
536
	pgd = pgdp + kvm_pgd_index(addr, ptrs_per_pgd);
537 538 539 540 541 542 543
	do {
		next = pgd_addr_end(addr, end);
		if (!pgd_none(*pgd))
			unmap_hyp_puds(pgd, addr, next);
	} while (pgd++, addr = next, addr != end);
}

544 545 546 547 548 549 550 551 552 553
static void unmap_hyp_range(pgd_t *pgdp, phys_addr_t start, u64 size)
{
	__unmap_hyp_range(pgdp, PTRS_PER_PGD, start, size);
}

static void unmap_hyp_idmap_range(pgd_t *pgdp, phys_addr_t start, u64 size)
{
	__unmap_hyp_range(pgdp, __kvm_idmap_ptrs_per_pgd(), start, size);
}

554
/**
555
 * free_hyp_pgds - free Hyp-mode page tables
556
 *
557 558
 * Assumes hyp_pgd is a page table used strictly in Hyp-mode and
 * therefore contains either mappings in the kernel memory area (above
559
 * PAGE_OFFSET), or device mappings in the idmap range.
560
 *
561 562
 * boot_hyp_pgd should only map the idmap range, and is only used in
 * the extended idmap case.
563
 */
564
void free_hyp_pgds(void)
565
{
566 567
	pgd_t *id_pgd;

568
	mutex_lock(&kvm_hyp_pgd_mutex);
569

570 571 572 573 574 575 576 577 578 579
	id_pgd = boot_hyp_pgd ? boot_hyp_pgd : hyp_pgd;

	if (id_pgd) {
		/* In case we never called hyp_mmu_init() */
		if (!io_map_base)
			io_map_base = hyp_idmap_start;
		unmap_hyp_idmap_range(id_pgd, io_map_base,
				      hyp_idmap_start + PAGE_SIZE - io_map_base);
	}

580 581 582 583 584
	if (boot_hyp_pgd) {
		free_pages((unsigned long)boot_hyp_pgd, hyp_pgd_order);
		boot_hyp_pgd = NULL;
	}

585
	if (hyp_pgd) {
586 587
		unmap_hyp_range(hyp_pgd, kern_hyp_va(PAGE_OFFSET),
				(uintptr_t)high_memory - PAGE_OFFSET);
588

589
		free_pages((unsigned long)hyp_pgd, hyp_pgd_order);
590
		hyp_pgd = NULL;
591
	}
592 593 594 595 596
	if (merged_hyp_pgd) {
		clear_page(merged_hyp_pgd);
		free_page((unsigned long)merged_hyp_pgd);
		merged_hyp_pgd = NULL;
	}
597

598 599 600 601
	mutex_unlock(&kvm_hyp_pgd_mutex);
}

static void create_hyp_pte_mappings(pmd_t *pmd, unsigned long start,
602 603
				    unsigned long end, unsigned long pfn,
				    pgprot_t prot)
604 605 606 607
{
	pte_t *pte;
	unsigned long addr;

608 609
	addr = start;
	do {
610
		pte = pte_offset_kernel(pmd, addr);
611
		kvm_set_pte(pte, kvm_pfn_pte(pfn, prot));
612
		get_page(virt_to_page(pte));
613
		pfn++;
614
	} while (addr += PAGE_SIZE, addr != end);
615 616 617
}

static int create_hyp_pmd_mappings(pud_t *pud, unsigned long start,
618 619
				   unsigned long end, unsigned long pfn,
				   pgprot_t prot)
620 621 622 623 624
{
	pmd_t *pmd;
	pte_t *pte;
	unsigned long addr, next;

625 626
	addr = start;
	do {
627
		pmd = pmd_offset(pud, addr);
628 629 630 631

		BUG_ON(pmd_sect(*pmd));

		if (pmd_none(*pmd)) {
632
			pte = pte_alloc_one_kernel(NULL, addr);
633 634 635 636
			if (!pte) {
				kvm_err("Cannot allocate Hyp pte\n");
				return -ENOMEM;
			}
637
			kvm_pmd_populate(pmd, pte);
638
			get_page(virt_to_page(pmd));
639 640 641 642
		}

		next = pmd_addr_end(addr, end);

643 644
		create_hyp_pte_mappings(pmd, addr, next, pfn, prot);
		pfn += (next - addr) >> PAGE_SHIFT;
645
	} while (addr = next, addr != end);
646 647 648 649

	return 0;
}

650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668
static int create_hyp_pud_mappings(pgd_t *pgd, unsigned long start,
				   unsigned long end, unsigned long pfn,
				   pgprot_t prot)
{
	pud_t *pud;
	pmd_t *pmd;
	unsigned long addr, next;
	int ret;

	addr = start;
	do {
		pud = pud_offset(pgd, addr);

		if (pud_none_or_clear_bad(pud)) {
			pmd = pmd_alloc_one(NULL, addr);
			if (!pmd) {
				kvm_err("Cannot allocate Hyp pmd\n");
				return -ENOMEM;
			}
669
			kvm_pud_populate(pud, pmd);
670 671 672 673 674 675 676 677 678 679 680 681 682
			get_page(virt_to_page(pud));
		}

		next = pud_addr_end(addr, end);
		ret = create_hyp_pmd_mappings(pud, addr, next, pfn, prot);
		if (ret)
			return ret;
		pfn += (next - addr) >> PAGE_SHIFT;
	} while (addr = next, addr != end);

	return 0;
}

683
static int __create_hyp_mappings(pgd_t *pgdp, unsigned long ptrs_per_pgd,
684 685
				 unsigned long start, unsigned long end,
				 unsigned long pfn, pgprot_t prot)
686 687 688 689 690 691 692
{
	pgd_t *pgd;
	pud_t *pud;
	unsigned long addr, next;
	int err = 0;

	mutex_lock(&kvm_hyp_pgd_mutex);
693 694 695
	addr = start & PAGE_MASK;
	end = PAGE_ALIGN(end);
	do {
696
		pgd = pgdp + kvm_pgd_index(addr, ptrs_per_pgd);
697

698 699 700 701
		if (pgd_none(*pgd)) {
			pud = pud_alloc_one(NULL, addr);
			if (!pud) {
				kvm_err("Cannot allocate Hyp pud\n");
702 703 704
				err = -ENOMEM;
				goto out;
			}
705
			kvm_pgd_populate(pgd, pud);
706
			get_page(virt_to_page(pgd));
707 708 709
		}

		next = pgd_addr_end(addr, end);
710
		err = create_hyp_pud_mappings(pgd, addr, next, pfn, prot);
711 712
		if (err)
			goto out;
713
		pfn += (next - addr) >> PAGE_SHIFT;
714
	} while (addr = next, addr != end);
715 716 717 718 719
out:
	mutex_unlock(&kvm_hyp_pgd_mutex);
	return err;
}

720 721 722 723 724 725 726 727 728 729 730
static phys_addr_t kvm_kaddr_to_phys(void *kaddr)
{
	if (!is_vmalloc_addr(kaddr)) {
		BUG_ON(!virt_addr_valid(kaddr));
		return __pa(kaddr);
	} else {
		return page_to_phys(vmalloc_to_page(kaddr)) +
		       offset_in_page(kaddr);
	}
}

731
/**
732
 * create_hyp_mappings - duplicate a kernel virtual address range in Hyp mode
733 734
 * @from:	The virtual kernel start address of the range
 * @to:		The virtual kernel end address of the range (exclusive)
735
 * @prot:	The protection to be applied to this range
736
 *
737 738 739
 * The same virtual address as the kernel virtual address is also used
 * in Hyp-mode mapping (modulo HYP_PAGE_OFFSET) to the same underlying
 * physical pages.
740
 */
741
int create_hyp_mappings(void *from, void *to, pgprot_t prot)
742
{
743 744
	phys_addr_t phys_addr;
	unsigned long virt_addr;
M
Marc Zyngier 已提交
745 746
	unsigned long start = kern_hyp_va((unsigned long)from);
	unsigned long end = kern_hyp_va((unsigned long)to);
747

748 749 750
	if (is_kernel_in_hyp_mode())
		return 0;

751 752
	start = start & PAGE_MASK;
	end = PAGE_ALIGN(end);
753

754 755
	for (virt_addr = start; virt_addr < end; virt_addr += PAGE_SIZE) {
		int err;
756

757
		phys_addr = kvm_kaddr_to_phys(from + virt_addr - start);
758 759
		err = __create_hyp_mappings(hyp_pgd, PTRS_PER_PGD,
					    virt_addr, virt_addr + PAGE_SIZE,
760
					    __phys_to_pfn(phys_addr),
761
					    prot);
762 763 764 765 766
		if (err)
			return err;
	}

	return 0;
767 768
}

769 770
static int __create_hyp_private_mapping(phys_addr_t phys_addr, size_t size,
					unsigned long *haddr, pgprot_t prot)
771
{
772 773 774
	pgd_t *pgd = hyp_pgd;
	unsigned long base;
	int ret = 0;
775

776
	mutex_lock(&kvm_hyp_pgd_mutex);
777

778 779 780 781 782 783 784 785 786 787
	/*
	 * This assumes that we we have enough space below the idmap
	 * page to allocate our VAs. If not, the check below will
	 * kick. A potential alternative would be to detect that
	 * overflow and switch to an allocation above the idmap.
	 *
	 * The allocated size is always a multiple of PAGE_SIZE.
	 */
	size = PAGE_ALIGN(size + offset_in_page(phys_addr));
	base = io_map_base - size;
788

789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808
	/*
	 * Verify that BIT(VA_BITS - 1) hasn't been flipped by
	 * allocating the new area, as it would indicate we've
	 * overflowed the idmap/IO address range.
	 */
	if ((base ^ io_map_base) & BIT(VA_BITS - 1))
		ret = -ENOMEM;
	else
		io_map_base = base;

	mutex_unlock(&kvm_hyp_pgd_mutex);

	if (ret)
		goto out;

	if (__kvm_cpu_uses_extended_idmap())
		pgd = boot_hyp_pgd;

	ret = __create_hyp_mappings(pgd, __kvm_idmap_ptrs_per_pgd(),
				    base, base + size,
809
				    __phys_to_pfn(phys_addr), prot);
810 811 812
	if (ret)
		goto out;

813
	*haddr = base + offset_in_page(phys_addr);
814 815

out:
816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843
	return ret;
}

/**
 * create_hyp_io_mappings - Map IO into both kernel and HYP
 * @phys_addr:	The physical start address which gets mapped
 * @size:	Size of the region being mapped
 * @kaddr:	Kernel VA for this mapping
 * @haddr:	HYP VA for this mapping
 */
int create_hyp_io_mappings(phys_addr_t phys_addr, size_t size,
			   void __iomem **kaddr,
			   void __iomem **haddr)
{
	unsigned long addr;
	int ret;

	*kaddr = ioremap(phys_addr, size);
	if (!*kaddr)
		return -ENOMEM;

	if (is_kernel_in_hyp_mode()) {
		*haddr = *kaddr;
		return 0;
	}

	ret = __create_hyp_private_mapping(phys_addr, size,
					   &addr, PAGE_HYP_DEVICE);
844 845 846
	if (ret) {
		iounmap(*kaddr);
		*kaddr = NULL;
847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872
		*haddr = NULL;
		return ret;
	}

	*haddr = (void __iomem *)addr;
	return 0;
}

/**
 * create_hyp_exec_mappings - Map an executable range into HYP
 * @phys_addr:	The physical start address which gets mapped
 * @size:	Size of the region being mapped
 * @haddr:	HYP VA for this mapping
 */
int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
			     void **haddr)
{
	unsigned long addr;
	int ret;

	BUG_ON(is_kernel_in_hyp_mode());

	ret = __create_hyp_private_mapping(phys_addr, size,
					   &addr, PAGE_HYP_EXEC);
	if (ret) {
		*haddr = NULL;
873 874 875
		return ret;
	}

876
	*haddr = (void *)addr;
877
	return 0;
878 879
}

880 881 882 883
/**
 * kvm_alloc_stage2_pgd - allocate level-1 table for stage-2 translation.
 * @kvm:	The KVM struct pointer for the VM.
 *
884 885 886
 * Allocates only the stage-2 HW PGD level table(s) (can support either full
 * 40-bit input addresses or limited to 32-bit input addresses). Clears the
 * allocated pages.
887 888 889 890 891 892 893 894 895 896 897 898 899
 *
 * Note we don't need locking here as this is only called when the VM is
 * created, which can only be done once.
 */
int kvm_alloc_stage2_pgd(struct kvm *kvm)
{
	pgd_t *pgd;

	if (kvm->arch.pgd != NULL) {
		kvm_err("kvm_arch already initialized?\n");
		return -EINVAL;
	}

900 901 902
	/* Allocate the HW PGD, making sure that each page gets its own refcount */
	pgd = alloc_pages_exact(S2_PGD_SIZE, GFP_KERNEL | __GFP_ZERO);
	if (!pgd)
903 904
		return -ENOMEM;

905 906 907 908
	kvm->arch.pgd = pgd;
	return 0;
}

909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963
static void stage2_unmap_memslot(struct kvm *kvm,
				 struct kvm_memory_slot *memslot)
{
	hva_t hva = memslot->userspace_addr;
	phys_addr_t addr = memslot->base_gfn << PAGE_SHIFT;
	phys_addr_t size = PAGE_SIZE * memslot->npages;
	hva_t reg_end = hva + size;

	/*
	 * A memory region could potentially cover multiple VMAs, and any holes
	 * between them, so iterate over all of them to find out if we should
	 * unmap any of them.
	 *
	 *     +--------------------------------------------+
	 * +---------------+----------------+   +----------------+
	 * |   : VMA 1     |      VMA 2     |   |    VMA 3  :    |
	 * +---------------+----------------+   +----------------+
	 *     |               memory region                |
	 *     +--------------------------------------------+
	 */
	do {
		struct vm_area_struct *vma = find_vma(current->mm, hva);
		hva_t vm_start, vm_end;

		if (!vma || vma->vm_start >= reg_end)
			break;

		/*
		 * Take the intersection of this VMA with the memory region
		 */
		vm_start = max(hva, vma->vm_start);
		vm_end = min(reg_end, vma->vm_end);

		if (!(vma->vm_flags & VM_PFNMAP)) {
			gpa_t gpa = addr + (vm_start - memslot->userspace_addr);
			unmap_stage2_range(kvm, gpa, vm_end - vm_start);
		}
		hva = vm_end;
	} while (hva < reg_end);
}

/**
 * stage2_unmap_vm - Unmap Stage-2 RAM mappings
 * @kvm: The struct kvm pointer
 *
 * Go through the memregions and unmap any reguler RAM
 * backing memory already mapped to the VM.
 */
void stage2_unmap_vm(struct kvm *kvm)
{
	struct kvm_memslots *slots;
	struct kvm_memory_slot *memslot;
	int idx;

	idx = srcu_read_lock(&kvm->srcu);
964
	down_read(&current->mm->mmap_sem);
965 966 967 968 969 970 971
	spin_lock(&kvm->mmu_lock);

	slots = kvm_memslots(kvm);
	kvm_for_each_memslot(memslot, slots)
		stage2_unmap_memslot(kvm, memslot);

	spin_unlock(&kvm->mmu_lock);
972
	up_read(&current->mm->mmap_sem);
973 974 975
	srcu_read_unlock(&kvm->srcu, idx);
}

976 977 978 979 980 981 982 983 984 985
/**
 * kvm_free_stage2_pgd - free all stage-2 tables
 * @kvm:	The KVM struct pointer for the VM.
 *
 * Walks the level-1 page table pointed to by kvm->arch.pgd and frees all
 * underlying level-2 and level-3 tables before freeing the actual level-1 table
 * and setting the struct pointer to NULL.
 */
void kvm_free_stage2_pgd(struct kvm *kvm)
{
986
	void *pgd = NULL;
987

988
	spin_lock(&kvm->mmu_lock);
989 990
	if (kvm->arch.pgd) {
		unmap_stage2_range(kvm, 0, KVM_PHYS_SIZE);
991
		pgd = READ_ONCE(kvm->arch.pgd);
992 993
		kvm->arch.pgd = NULL;
	}
994 995
	spin_unlock(&kvm->mmu_lock);

996
	/* Free the HW pgd, one page at a time */
997 998
	if (pgd)
		free_pages_exact(pgd, S2_PGD_SIZE);
999 1000
}

1001
static pud_t *stage2_get_pud(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
1002
			     phys_addr_t addr)
1003 1004 1005 1006
{
	pgd_t *pgd;
	pud_t *pud;

1007 1008
	pgd = kvm->arch.pgd + stage2_pgd_index(addr);
	if (WARN_ON(stage2_pgd_none(*pgd))) {
1009 1010 1011
		if (!cache)
			return NULL;
		pud = mmu_memory_cache_alloc(cache);
1012
		stage2_pgd_populate(pgd, pud);
1013 1014 1015
		get_page(virt_to_page(pgd));
	}

1016
	return stage2_pud_offset(pgd, addr);
1017 1018 1019 1020 1021 1022 1023 1024 1025
}

static pmd_t *stage2_get_pmd(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
			     phys_addr_t addr)
{
	pud_t *pud;
	pmd_t *pmd;

	pud = stage2_get_pud(kvm, cache, addr);
1026 1027 1028
	if (!pud)
		return NULL;

1029
	if (stage2_pud_none(*pud)) {
1030
		if (!cache)
1031
			return NULL;
1032
		pmd = mmu_memory_cache_alloc(cache);
1033
		stage2_pud_populate(pud, pmd);
1034
		get_page(virt_to_page(pud));
1035 1036
	}

1037
	return stage2_pmd_offset(pud, addr);
1038 1039 1040 1041 1042 1043 1044 1045 1046
}

static int stage2_set_pmd_huge(struct kvm *kvm, struct kvm_mmu_memory_cache
			       *cache, phys_addr_t addr, const pmd_t *new_pmd)
{
	pmd_t *pmd, old_pmd;

	pmd = stage2_get_pmd(kvm, cache, addr);
	VM_BUG_ON(!pmd);
1047

1048
	old_pmd = *pmd;
1049
	if (pmd_present(old_pmd)) {
1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076
		/*
		 * Multiple vcpus faulting on the same PMD entry, can
		 * lead to them sequentially updating the PMD with the
		 * same value. Following the break-before-make
		 * (pmd_clear() followed by tlb_flush()) process can
		 * hinder forward progress due to refaults generated
		 * on missing translations.
		 *
		 * Skip updating the page table if the entry is
		 * unchanged.
		 */
		if (pmd_val(old_pmd) == pmd_val(*new_pmd))
			return 0;

		/*
		 * Mapping in huge pages should only happen through a
		 * fault.  If a page is merged into a transparent huge
		 * page, the individual subpages of that huge page
		 * should be unmapped through MMU notifiers before we
		 * get here.
		 *
		 * Merging of CompoundPages is not supported; they
		 * should become splitting first, unmapped, merged,
		 * and mapped back in on-demand.
		 */
		VM_BUG_ON(pmd_pfn(old_pmd) != pmd_pfn(*new_pmd));

1077
		pmd_clear(pmd);
1078
		kvm_tlb_flush_vmid_ipa(kvm, addr);
1079
	} else {
1080
		get_page(virt_to_page(pmd));
1081 1082 1083
	}

	kvm_set_pmd(pmd, *new_pmd);
1084 1085 1086
	return 0;
}

1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105
static bool stage2_is_exec(struct kvm *kvm, phys_addr_t addr)
{
	pmd_t *pmdp;
	pte_t *ptep;

	pmdp = stage2_get_pmd(kvm, NULL, addr);
	if (!pmdp || pmd_none(*pmdp) || !pmd_present(*pmdp))
		return false;

	if (pmd_thp_or_huge(*pmdp))
		return kvm_s2pmd_exec(pmdp);

	ptep = pte_offset_kernel(pmdp, addr);
	if (!ptep || pte_none(*ptep) || !pte_present(*ptep))
		return false;

	return kvm_s2pte_exec(ptep);
}

1106
static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
1107 1108
			  phys_addr_t addr, const pte_t *new_pte,
			  unsigned long flags)
1109 1110 1111
{
	pmd_t *pmd;
	pte_t *pte, old_pte;
1112 1113 1114 1115
	bool iomap = flags & KVM_S2PTE_FLAG_IS_IOMAP;
	bool logging_active = flags & KVM_S2_FLAG_LOGGING_ACTIVE;

	VM_BUG_ON(logging_active && !cache);
1116

1117
	/* Create stage-2 page table mapping - Levels 0 and 1 */
1118 1119 1120 1121 1122 1123 1124 1125 1126
	pmd = stage2_get_pmd(kvm, cache, addr);
	if (!pmd) {
		/*
		 * Ignore calls from kvm_set_spte_hva for unallocated
		 * address ranges.
		 */
		return 0;
	}

1127 1128 1129 1130 1131 1132 1133
	/*
	 * While dirty page logging - dissolve huge PMD, then continue on to
	 * allocate page.
	 */
	if (logging_active)
		stage2_dissolve_pmd(kvm, addr, pmd);

1134
	/* Create stage-2 page mappings - Level 2 */
1135 1136 1137 1138
	if (pmd_none(*pmd)) {
		if (!cache)
			return 0; /* ignore calls from kvm_set_spte_hva */
		pte = mmu_memory_cache_alloc(cache);
1139
		kvm_pmd_populate(pmd, pte);
1140
		get_page(virt_to_page(pmd));
1141 1142 1143
	}

	pte = pte_offset_kernel(pmd, addr);
1144 1145 1146 1147 1148 1149

	if (iomap && pte_present(*pte))
		return -EFAULT;

	/* Create 2nd stage page table mapping - Level 3 */
	old_pte = *pte;
1150
	if (pte_present(old_pte)) {
1151 1152 1153 1154
		/* Skip page table update if there is no change */
		if (pte_val(old_pte) == pte_val(*new_pte))
			return 0;

1155
		kvm_set_pte(pte, __pte(0));
1156
		kvm_tlb_flush_vmid_ipa(kvm, addr);
1157
	} else {
1158
		get_page(virt_to_page(pte));
1159
	}
1160

1161
	kvm_set_pte(pte, *new_pte);
1162 1163 1164
	return 0;
}

1165 1166 1167 1168 1169 1170 1171
#ifndef __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
static int stage2_ptep_test_and_clear_young(pte_t *pte)
{
	if (pte_young(*pte)) {
		*pte = pte_mkold(*pte);
		return 1;
	}
1172 1173
	return 0;
}
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184
#else
static int stage2_ptep_test_and_clear_young(pte_t *pte)
{
	return __ptep_test_and_clear_young(pte);
}
#endif

static int stage2_pmdp_test_and_clear_young(pmd_t *pmd)
{
	return stage2_ptep_test_and_clear_young((pte_t *)pmd);
}
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194

/**
 * kvm_phys_addr_ioremap - map a device range to guest IPA
 *
 * @kvm:	The KVM pointer
 * @guest_ipa:	The IPA at which to insert the mapping
 * @pa:		The physical address of the device
 * @size:	The size of the mapping
 */
int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
1195
			  phys_addr_t pa, unsigned long size, bool writable)
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
{
	phys_addr_t addr, end;
	int ret = 0;
	unsigned long pfn;
	struct kvm_mmu_memory_cache cache = { 0, };

	end = (guest_ipa + size + PAGE_SIZE - 1) & PAGE_MASK;
	pfn = __phys_to_pfn(pa);

	for (addr = guest_ipa; addr < end; addr += PAGE_SIZE) {
1206
		pte_t pte = kvm_pfn_pte(pfn, PAGE_S2_DEVICE);
1207

1208
		if (writable)
1209
			pte = kvm_s2pte_mkwrite(pte);
1210

1211 1212
		ret = mmu_topup_memory_cache(&cache, KVM_MMU_CACHE_MIN_PAGES,
						KVM_NR_MEM_OBJS);
1213 1214 1215
		if (ret)
			goto out;
		spin_lock(&kvm->mmu_lock);
1216 1217
		ret = stage2_set_pte(kvm, &cache, addr, &pte,
						KVM_S2PTE_FLAG_IS_IOMAP);
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229
		spin_unlock(&kvm->mmu_lock);
		if (ret)
			goto out;

		pfn++;
	}

out:
	mmu_free_memory_cache(&cache);
	return ret;
}

D
Dan Williams 已提交
1230
static bool transparent_hugepage_adjust(kvm_pfn_t *pfnp, phys_addr_t *ipap)
1231
{
D
Dan Williams 已提交
1232
	kvm_pfn_t pfn = *pfnp;
1233
	gfn_t gfn = *ipap >> PAGE_SHIFT;
1234
	struct page *page = pfn_to_page(pfn);
1235

1236 1237 1238 1239 1240 1241
	/*
	 * PageTransCompoungMap() returns true for THP and
	 * hugetlbfs. Make sure the adjustment is done only for THP
	 * pages.
	 */
	if (!PageHuge(page) && PageTransCompoundMap(page)) {
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276
		unsigned long mask;
		/*
		 * The address we faulted on is backed by a transparent huge
		 * page.  However, because we map the compound huge page and
		 * not the individual tail page, we need to transfer the
		 * refcount to the head page.  We have to be careful that the
		 * THP doesn't start to split while we are adjusting the
		 * refcounts.
		 *
		 * We are sure this doesn't happen, because mmu_notifier_retry
		 * was successful and we are holding the mmu_lock, so if this
		 * THP is trying to split, it will be blocked in the mmu
		 * notifier before touching any of the pages, specifically
		 * before being able to call __split_huge_page_refcount().
		 *
		 * We can therefore safely transfer the refcount from PG_tail
		 * to PG_head and switch the pfn from a tail page to the head
		 * page accordingly.
		 */
		mask = PTRS_PER_PMD - 1;
		VM_BUG_ON((gfn & mask) != (pfn & mask));
		if (pfn & mask) {
			*ipap &= PMD_MASK;
			kvm_release_pfn_clean(pfn);
			pfn &= ~mask;
			kvm_get_pfn(pfn);
			*pfnp = pfn;
		}

		return true;
	}

	return false;
}

1277 1278 1279 1280 1281 1282 1283 1284
static bool kvm_is_write_fault(struct kvm_vcpu *vcpu)
{
	if (kvm_vcpu_trap_is_iabt(vcpu))
		return false;

	return kvm_vcpu_dabt_iswrite(vcpu);
}

1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
/**
 * stage2_wp_ptes - write protect PMD range
 * @pmd:	pointer to pmd entry
 * @addr:	range start address
 * @end:	range end address
 */
static void stage2_wp_ptes(pmd_t *pmd, phys_addr_t addr, phys_addr_t end)
{
	pte_t *pte;

	pte = pte_offset_kernel(pmd, addr);
	do {
		if (!pte_none(*pte)) {
			if (!kvm_s2pte_readonly(pte))
				kvm_set_s2pte_readonly(pte);
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);
}

/**
 * stage2_wp_pmds - write protect PUD range
 * @pud:	pointer to pud entry
 * @addr:	range start address
 * @end:	range end address
 */
static void stage2_wp_pmds(pud_t *pud, phys_addr_t addr, phys_addr_t end)
{
	pmd_t *pmd;
	phys_addr_t next;

1315
	pmd = stage2_pmd_offset(pud, addr);
1316 1317

	do {
1318
		next = stage2_pmd_addr_end(addr, end);
1319
		if (!pmd_none(*pmd)) {
1320
			if (pmd_thp_or_huge(*pmd)) {
1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342
				if (!kvm_s2pmd_readonly(pmd))
					kvm_set_s2pmd_readonly(pmd);
			} else {
				stage2_wp_ptes(pmd, addr, next);
			}
		}
	} while (pmd++, addr = next, addr != end);
}

/**
  * stage2_wp_puds - write protect PGD range
  * @pgd:	pointer to pgd entry
  * @addr:	range start address
  * @end:	range end address
  *
  * Process PUD entries, for a huge PUD we cause a panic.
  */
static void  stage2_wp_puds(pgd_t *pgd, phys_addr_t addr, phys_addr_t end)
{
	pud_t *pud;
	phys_addr_t next;

1343
	pud = stage2_pud_offset(pgd, addr);
1344
	do {
1345 1346
		next = stage2_pud_addr_end(addr, end);
		if (!stage2_pud_none(*pud)) {
1347
			/* TODO:PUD not supported, revisit later if supported */
1348
			BUG_ON(stage2_pud_huge(*pud));
1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364
			stage2_wp_pmds(pud, addr, next);
		}
	} while (pud++, addr = next, addr != end);
}

/**
 * stage2_wp_range() - write protect stage2 memory region range
 * @kvm:	The KVM pointer
 * @addr:	Start address of range
 * @end:	End address of range
 */
static void stage2_wp_range(struct kvm *kvm, phys_addr_t addr, phys_addr_t end)
{
	pgd_t *pgd;
	phys_addr_t next;

1365
	pgd = kvm->arch.pgd + stage2_pgd_index(addr);
1366 1367 1368 1369
	do {
		/*
		 * Release kvm_mmu_lock periodically if the memory region is
		 * large. Otherwise, we may see kernel panics with
1370 1371
		 * CONFIG_DETECT_HUNG_TASK, CONFIG_LOCKUP_DETECTOR,
		 * CONFIG_LOCKDEP. Additionally, holding the lock too long
1372 1373 1374
		 * will also starve other vCPUs. We have to also make sure
		 * that the page tables are not freed while we released
		 * the lock.
1375
		 */
1376 1377 1378
		cond_resched_lock(&kvm->mmu_lock);
		if (!READ_ONCE(kvm->arch.pgd))
			break;
1379 1380
		next = stage2_pgd_addr_end(addr, end);
		if (stage2_pgd_present(*pgd))
1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399
			stage2_wp_puds(pgd, addr, next);
	} while (pgd++, addr = next, addr != end);
}

/**
 * kvm_mmu_wp_memory_region() - write protect stage 2 entries for memory slot
 * @kvm:	The KVM pointer
 * @slot:	The memory slot to write protect
 *
 * Called to start logging dirty pages after memory region
 * KVM_MEM_LOG_DIRTY_PAGES operation is called. After this function returns
 * all present PMD and PTEs are write protected in the memory region.
 * Afterwards read of dirty page log can be called.
 *
 * Acquires kvm_mmu_lock. Called with kvm->slots_lock mutex acquired,
 * serializing operations for VM memory regions.
 */
void kvm_mmu_wp_memory_region(struct kvm *kvm, int slot)
{
1400 1401
	struct kvm_memslots *slots = kvm_memslots(kvm);
	struct kvm_memory_slot *memslot = id_to_memslot(slots, slot);
1402 1403 1404 1405 1406 1407 1408 1409
	phys_addr_t start = memslot->base_gfn << PAGE_SHIFT;
	phys_addr_t end = (memslot->base_gfn + memslot->npages) << PAGE_SHIFT;

	spin_lock(&kvm->mmu_lock);
	stage2_wp_range(kvm, start, end);
	spin_unlock(&kvm->mmu_lock);
	kvm_flush_remote_tlbs(kvm);
}
1410 1411

/**
1412
 * kvm_mmu_write_protect_pt_masked() - write protect dirty pages
1413 1414 1415 1416 1417 1418 1419 1420 1421
 * @kvm:	The KVM pointer
 * @slot:	The memory slot associated with mask
 * @gfn_offset:	The gfn offset in memory slot
 * @mask:	The mask of dirty pages at offset 'gfn_offset' in this memory
 *		slot to be write protected
 *
 * Walks bits set in mask write protects the associated pte's. Caller must
 * acquire kvm_mmu_lock.
 */
1422
static void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
1423 1424 1425 1426 1427 1428 1429 1430 1431
		struct kvm_memory_slot *slot,
		gfn_t gfn_offset, unsigned long mask)
{
	phys_addr_t base_gfn = slot->base_gfn + gfn_offset;
	phys_addr_t start = (base_gfn +  __ffs(mask)) << PAGE_SHIFT;
	phys_addr_t end = (base_gfn + __fls(mask) + 1) << PAGE_SHIFT;

	stage2_wp_range(kvm, start, end);
}
1432

1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
/*
 * kvm_arch_mmu_enable_log_dirty_pt_masked - enable dirty logging for selected
 * dirty pages.
 *
 * It calls kvm_mmu_write_protect_pt_masked to write protect selected pages to
 * enable dirty logging for them.
 */
void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
		struct kvm_memory_slot *slot,
		gfn_t gfn_offset, unsigned long mask)
{
	kvm_mmu_write_protect_pt_masked(kvm, slot, gfn_offset, mask);
}

1447
static void clean_dcache_guest_page(kvm_pfn_t pfn, unsigned long size)
1448
{
1449
	__clean_dcache_guest_page(pfn, size);
1450 1451
}

1452
static void invalidate_icache_guest_page(kvm_pfn_t pfn, unsigned long size)
1453
{
1454
	__invalidate_icache_guest_page(pfn, size);
1455 1456
}

1457 1458 1459 1460 1461
static void kvm_send_hwpoison_signal(unsigned long address,
				     struct vm_area_struct *vma)
{
	siginfo_t info;

1462
	clear_siginfo(&info);
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475
	info.si_signo   = SIGBUS;
	info.si_errno   = 0;
	info.si_code    = BUS_MCEERR_AR;
	info.si_addr    = (void __user *)address;

	if (is_vm_hugetlb_page(vma))
		info.si_addr_lsb = huge_page_shift(hstate_vma(vma));
	else
		info.si_addr_lsb = PAGE_SHIFT;

	send_sig_info(SIGBUS, &info, current);
}

1476
static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
1477
			  struct kvm_memory_slot *memslot, unsigned long hva,
1478 1479 1480
			  unsigned long fault_status)
{
	int ret;
1481 1482
	bool write_fault, writable, force_pte = false;
	bool exec_fault, needs_exec;
1483
	unsigned long mmu_seq;
1484 1485
	gfn_t gfn = fault_ipa >> PAGE_SHIFT;
	struct kvm *kvm = vcpu->kvm;
1486
	struct kvm_mmu_memory_cache *memcache = &vcpu->arch.mmu_page_cache;
1487
	struct vm_area_struct *vma;
D
Dan Williams 已提交
1488
	kvm_pfn_t pfn;
1489
	pgprot_t mem_type = PAGE_S2;
1490
	bool logging_active = memslot_is_logging(memslot);
1491
	unsigned long vma_pagesize, flags = 0;
1492

1493
	write_fault = kvm_is_write_fault(vcpu);
1494 1495 1496 1497
	exec_fault = kvm_vcpu_trap_is_iabt(vcpu);
	VM_BUG_ON(write_fault && exec_fault);

	if (fault_status == FSC_PERM && !write_fault && !exec_fault) {
1498 1499 1500 1501
		kvm_err("Unexpected L2 read permission error\n");
		return -EFAULT;
	}

1502 1503 1504
	/* Let's check if we will get back a huge page backed by hugetlbfs */
	down_read(&current->mm->mmap_sem);
	vma = find_vma_intersection(current->mm, hva, hva + 1);
1505 1506 1507 1508 1509 1510
	if (unlikely(!vma)) {
		kvm_err("Failed to find VMA for hva 0x%lx\n", hva);
		up_read(&current->mm->mmap_sem);
		return -EFAULT;
	}

1511 1512
	vma_pagesize = vma_kernel_pagesize(vma);
	if (vma_pagesize == PMD_SIZE && !logging_active) {
1513
		gfn = (fault_ipa & PMD_MASK) >> PAGE_SHIFT;
1514
	} else {
1515 1516 1517 1518 1519 1520
		/*
		 * Fallback to PTE if it's not one of the Stage 2
		 * supported hugepage sizes
		 */
		vma_pagesize = PAGE_SIZE;

1521
		/*
1522 1523 1524 1525 1526 1527 1528
		 * Pages belonging to memslots that don't have the same
		 * alignment for userspace and IPA cannot be mapped using
		 * block descriptors even if the pages belong to a THP for
		 * the process, because the stage-2 block descriptor will
		 * cover more than a single THP and we loose atomicity for
		 * unmapping, updates, and splits of the THP or other pages
		 * in the stage-2 block range.
1529
		 */
1530 1531
		if ((memslot->userspace_addr & ~PMD_MASK) !=
		    ((memslot->base_gfn << PAGE_SHIFT) & ~PMD_MASK))
1532
			force_pte = true;
1533 1534 1535
	}
	up_read(&current->mm->mmap_sem);

1536
	/* We need minimum second+third level pages */
1537 1538
	ret = mmu_topup_memory_cache(memcache, KVM_MMU_CACHE_MIN_PAGES,
				     KVM_NR_MEM_OBJS);
1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
	if (ret)
		return ret;

	mmu_seq = vcpu->kvm->mmu_notifier_seq;
	/*
	 * Ensure the read of mmu_notifier_seq happens before we call
	 * gfn_to_pfn_prot (which calls get_user_pages), so that we don't risk
	 * the page we just got a reference to gets unmapped before we have a
	 * chance to grab the mmu_lock, which ensure that if the page gets
	 * unmapped afterwards, the call to kvm_unmap_hva will take it away
	 * from us again properly. This smp_rmb() interacts with the smp_wmb()
	 * in kvm_mmu_notifier_invalidate_<page|range_end>.
	 */
	smp_rmb();

1554
	pfn = gfn_to_pfn_prot(kvm, gfn, write_fault, &writable);
1555 1556 1557 1558
	if (pfn == KVM_PFN_ERR_HWPOISON) {
		kvm_send_hwpoison_signal(hva, vma);
		return 0;
	}
1559
	if (is_error_noslot_pfn(pfn))
1560 1561
		return -EFAULT;

1562
	if (kvm_is_device_pfn(pfn)) {
1563
		mem_type = PAGE_S2_DEVICE;
1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580
		flags |= KVM_S2PTE_FLAG_IS_IOMAP;
	} else if (logging_active) {
		/*
		 * Faults on pages in a memslot with logging enabled
		 * should not be mapped with huge pages (it introduces churn
		 * and performance degradation), so force a pte mapping.
		 */
		force_pte = true;
		flags |= KVM_S2_FLAG_LOGGING_ACTIVE;

		/*
		 * Only actually map the page as writable if this was a write
		 * fault.
		 */
		if (!write_fault)
			writable = false;
	}
1581

1582 1583
	spin_lock(&kvm->mmu_lock);
	if (mmu_notifier_retry(kvm, mmu_seq))
1584
		goto out_unlock;
1585

1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597
	if (vma_pagesize == PAGE_SIZE && !force_pte) {
		/*
		 * Only PMD_SIZE transparent hugepages(THP) are
		 * currently supported. This code will need to be
		 * updated to support other THP sizes.
		 */
		if (transparent_hugepage_adjust(&pfn, &fault_ipa))
			vma_pagesize = PMD_SIZE;
	}

	if (writable)
		kvm_set_pfn_dirty(pfn);
1598

1599 1600 1601 1602 1603 1604
	if (fault_status != FSC_PERM)
		clean_dcache_guest_page(pfn, vma_pagesize);

	if (exec_fault)
		invalidate_icache_guest_page(pfn, vma_pagesize);

1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615
	/*
	 * If we took an execution fault we have made the
	 * icache/dcache coherent above and should now let the s2
	 * mapping be executable.
	 *
	 * Write faults (!exec_fault && FSC_PERM) are orthogonal to
	 * execute permissions, and we preserve whatever we have.
	 */
	needs_exec = exec_fault ||
		(fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));

1616
	if (vma_pagesize == PMD_SIZE) {
1617 1618 1619 1620
		pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);

		new_pmd = kvm_pmd_mkhuge(new_pmd);

1621
		if (writable)
1622
			new_pmd = kvm_s2pmd_mkwrite(new_pmd);
1623

1624
		if (needs_exec)
1625
			new_pmd = kvm_s2pmd_mkexec(new_pmd);
1626

1627 1628
		ret = stage2_set_pmd_huge(kvm, memcache, fault_ipa, &new_pmd);
	} else {
1629
		pte_t new_pte = kvm_pfn_pte(pfn, mem_type);
1630

1631
		if (writable) {
1632
			new_pte = kvm_s2pte_mkwrite(new_pte);
1633
			mark_page_dirty(kvm, gfn);
1634
		}
1635

1636
		if (needs_exec)
1637
			new_pte = kvm_s2pte_mkexec(new_pte);
1638

1639
		ret = stage2_set_pte(kvm, memcache, fault_ipa, &new_pte, flags);
1640
	}
1641

1642
out_unlock:
1643
	spin_unlock(&kvm->mmu_lock);
1644
	kvm_set_pfn_accessed(pfn);
1645
	kvm_release_pfn_clean(pfn);
1646
	return ret;
1647 1648
}

1649 1650 1651 1652
/*
 * Resolve the access fault by making the page young again.
 * Note that because the faulting entry is guaranteed not to be
 * cached in the TLB, we don't need to invalidate anything.
1653 1654
 * Only the HW Access Flag updates are supported for Stage 2 (no DBM),
 * so there is no need for atomic (pte|pmd)_mkyoung operations.
1655 1656 1657 1658 1659
 */
static void handle_access_fault(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa)
{
	pmd_t *pmd;
	pte_t *pte;
D
Dan Williams 已提交
1660
	kvm_pfn_t pfn;
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
	bool pfn_valid = false;

	trace_kvm_access_fault(fault_ipa);

	spin_lock(&vcpu->kvm->mmu_lock);

	pmd = stage2_get_pmd(vcpu->kvm, NULL, fault_ipa);
	if (!pmd || pmd_none(*pmd))	/* Nothing there */
		goto out;

1671
	if (pmd_thp_or_huge(*pmd)) {	/* THP, HugeTLB */
1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
		*pmd = pmd_mkyoung(*pmd);
		pfn = pmd_pfn(*pmd);
		pfn_valid = true;
		goto out;
	}

	pte = pte_offset_kernel(pmd, fault_ipa);
	if (pte_none(*pte))		/* Nothing there either */
		goto out;

	*pte = pte_mkyoung(*pte);	/* Just a page... */
	pfn = pte_pfn(*pte);
	pfn_valid = true;
out:
	spin_unlock(&vcpu->kvm->mmu_lock);
	if (pfn_valid)
		kvm_set_pfn_accessed(pfn);
}

1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
/**
 * kvm_handle_guest_abort - handles all 2nd stage aborts
 * @vcpu:	the VCPU pointer
 * @run:	the kvm_run structure
 *
 * Any abort that gets to the host is almost guaranteed to be caused by a
 * missing second stage translation table entry, which can mean that either the
 * guest simply needs more memory and we must allocate an appropriate page or it
 * can mean that the guest tried to access I/O memory, which is emulated by user
 * space. The distinction is based on the IPA causing the fault and whether this
 * memory region has been registered as standard RAM by user space.
 */
1703 1704
int kvm_handle_guest_abort(struct kvm_vcpu *vcpu, struct kvm_run *run)
{
1705 1706 1707
	unsigned long fault_status;
	phys_addr_t fault_ipa;
	struct kvm_memory_slot *memslot;
1708 1709
	unsigned long hva;
	bool is_iabt, write_fault, writable;
1710 1711 1712
	gfn_t gfn;
	int ret, idx;

1713 1714 1715
	fault_status = kvm_vcpu_trap_get_fault_type(vcpu);

	fault_ipa = kvm_vcpu_get_fault_ipa(vcpu);
1716
	is_iabt = kvm_vcpu_trap_is_iabt(vcpu);
1717

1718 1719 1720 1721 1722 1723
	/* Synchronous External Abort? */
	if (kvm_vcpu_dabt_isextabt(vcpu)) {
		/*
		 * For RAS the host kernel may handle this abort.
		 * There is no need to pass the error into the guest.
		 */
1724 1725 1726
		if (!handle_guest_sea(fault_ipa, kvm_vcpu_get_hsr(vcpu)))
			return 1;

1727 1728 1729 1730
		if (unlikely(!is_iabt)) {
			kvm_inject_vabt(vcpu);
			return 1;
		}
1731 1732
	}

1733 1734
	trace_kvm_guest_fault(*vcpu_pc(vcpu), kvm_vcpu_get_hsr(vcpu),
			      kvm_vcpu_get_hfar(vcpu), fault_ipa);
1735 1736

	/* Check the stage-2 fault is trans. fault or write fault */
1737 1738
	if (fault_status != FSC_FAULT && fault_status != FSC_PERM &&
	    fault_status != FSC_ACCESS) {
1739 1740 1741 1742
		kvm_err("Unsupported FSC: EC=%#x xFSC=%#lx ESR_EL2=%#lx\n",
			kvm_vcpu_trap_get_class(vcpu),
			(unsigned long)kvm_vcpu_trap_get_fault(vcpu),
			(unsigned long)kvm_vcpu_get_hsr(vcpu));
1743 1744 1745 1746 1747 1748
		return -EFAULT;
	}

	idx = srcu_read_lock(&vcpu->kvm->srcu);

	gfn = fault_ipa >> PAGE_SHIFT;
1749 1750
	memslot = gfn_to_memslot(vcpu->kvm, gfn);
	hva = gfn_to_hva_memslot_prot(memslot, gfn, &writable);
1751
	write_fault = kvm_is_write_fault(vcpu);
1752
	if (kvm_is_error_hva(hva) || (write_fault && !writable)) {
1753 1754
		if (is_iabt) {
			/* Prefetch Abort on I/O address */
1755
			kvm_inject_pabt(vcpu, kvm_vcpu_get_hfar(vcpu));
1756 1757 1758 1759
			ret = 1;
			goto out_unlock;
		}

1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775
		/*
		 * Check for a cache maintenance operation. Since we
		 * ended-up here, we know it is outside of any memory
		 * slot. But we can't find out if that is for a device,
		 * or if the guest is just being stupid. The only thing
		 * we know for sure is that this range cannot be cached.
		 *
		 * So let's assume that the guest is just being
		 * cautious, and skip the instruction.
		 */
		if (kvm_vcpu_dabt_is_cm(vcpu)) {
			kvm_skip_instr(vcpu, kvm_vcpu_trap_il_is32bit(vcpu));
			ret = 1;
			goto out_unlock;
		}

M
Marc Zyngier 已提交
1776 1777 1778 1779 1780 1781 1782
		/*
		 * The IPA is reported as [MAX:12], so we need to
		 * complement it with the bottom 12 bits from the
		 * faulting VA. This is always 12 bits, irrespective
		 * of the page size.
		 */
		fault_ipa |= kvm_vcpu_get_hfar(vcpu) & ((1 << 12) - 1);
C
Christoffer Dall 已提交
1783
		ret = io_mem_abort(vcpu, run, fault_ipa);
1784 1785 1786
		goto out_unlock;
	}

1787 1788 1789
	/* Userspace should not be able to register out-of-bounds IPAs */
	VM_BUG_ON(fault_ipa >= KVM_PHYS_SIZE);

1790 1791 1792 1793 1794 1795
	if (fault_status == FSC_ACCESS) {
		handle_access_fault(vcpu, fault_ipa);
		ret = 1;
		goto out_unlock;
	}

1796
	ret = user_mem_abort(vcpu, fault_ipa, memslot, hva, fault_status);
1797 1798 1799 1800 1801
	if (ret == 0)
		ret = 1;
out_unlock:
	srcu_read_unlock(&vcpu->kvm->srcu, idx);
	return ret;
1802 1803
}

1804 1805 1806 1807
static int handle_hva_to_gpa(struct kvm *kvm,
			     unsigned long start,
			     unsigned long end,
			     int (*handler)(struct kvm *kvm,
1808 1809
					    gpa_t gpa, u64 size,
					    void *data),
1810
			     void *data)
1811 1812 1813
{
	struct kvm_memslots *slots;
	struct kvm_memory_slot *memslot;
1814
	int ret = 0;
1815 1816 1817 1818 1819 1820

	slots = kvm_memslots(kvm);

	/* we only care about the pages that the guest sees */
	kvm_for_each_memslot(memslot, slots) {
		unsigned long hva_start, hva_end;
1821
		gfn_t gpa;
1822 1823 1824 1825 1826 1827 1828

		hva_start = max(start, memslot->userspace_addr);
		hva_end = min(end, memslot->userspace_addr +
					(memslot->npages << PAGE_SHIFT));
		if (hva_start >= hva_end)
			continue;

1829 1830
		gpa = hva_to_gfn_memslot(hva_start, memslot) << PAGE_SHIFT;
		ret |= handler(kvm, gpa, (u64)(hva_end - hva_start), data);
1831
	}
1832 1833

	return ret;
1834 1835
}

1836
static int kvm_unmap_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
1837
{
1838
	unmap_stage2_range(kvm, gpa, size);
1839
	return 0;
1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
}

int kvm_unmap_hva_range(struct kvm *kvm,
			unsigned long start, unsigned long end)
{
	if (!kvm->arch.pgd)
		return 0;

	trace_kvm_unmap_hva_range(start, end);
	handle_hva_to_gpa(kvm, start, end, &kvm_unmap_hva_handler, NULL);
	return 0;
}

1853
static int kvm_set_spte_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
1854 1855 1856
{
	pte_t *pte = (pte_t *)data;

1857
	WARN_ON(size != PAGE_SIZE);
1858 1859 1860 1861 1862 1863 1864 1865
	/*
	 * We can always call stage2_set_pte with KVM_S2PTE_FLAG_LOGGING_ACTIVE
	 * flag clear because MMU notifiers will have unmapped a huge PMD before
	 * calling ->change_pte() (which in turn calls kvm_set_spte_hva()) and
	 * therefore stage2_set_pte() never needs to clear out a huge PMD
	 * through this calling path.
	 */
	stage2_set_pte(kvm, NULL, gpa, pte, 0);
1866
	return 0;
1867 1868 1869 1870 1871 1872
}


void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
{
	unsigned long end = hva + PAGE_SIZE;
1873
	kvm_pfn_t pfn = pte_pfn(pte);
1874 1875 1876 1877 1878 1879
	pte_t stage2_pte;

	if (!kvm->arch.pgd)
		return;

	trace_kvm_set_spte_hva(hva);
1880 1881 1882 1883 1884 1885

	/*
	 * We've moved a page around, probably through CoW, so let's treat it
	 * just like a translation fault and clean the cache to the PoC.
	 */
	clean_dcache_guest_page(pfn, PAGE_SIZE);
1886
	stage2_pte = kvm_pfn_pte(pfn, PAGE_S2);
1887 1888 1889
	handle_hva_to_gpa(kvm, hva, end, &kvm_set_spte_handler, &stage2_pte);
}

1890
static int kvm_age_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
1891 1892 1893 1894
{
	pmd_t *pmd;
	pte_t *pte;

1895
	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE);
1896 1897 1898 1899
	pmd = stage2_get_pmd(kvm, NULL, gpa);
	if (!pmd || pmd_none(*pmd))	/* Nothing there */
		return 0;

1900 1901
	if (pmd_thp_or_huge(*pmd))	/* THP, HugeTLB */
		return stage2_pmdp_test_and_clear_young(pmd);
1902 1903 1904 1905 1906

	pte = pte_offset_kernel(pmd, gpa);
	if (pte_none(*pte))
		return 0;

1907
	return stage2_ptep_test_and_clear_young(pte);
1908 1909
}

1910
static int kvm_test_age_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
1911 1912 1913 1914
{
	pmd_t *pmd;
	pte_t *pte;

1915
	WARN_ON(size != PAGE_SIZE && size != PMD_SIZE);
1916 1917 1918 1919
	pmd = stage2_get_pmd(kvm, NULL, gpa);
	if (!pmd || pmd_none(*pmd))	/* Nothing there */
		return 0;

1920
	if (pmd_thp_or_huge(*pmd))		/* THP, HugeTLB */
1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931
		return pmd_young(*pmd);

	pte = pte_offset_kernel(pmd, gpa);
	if (!pte_none(*pte))		/* Just a page... */
		return pte_young(*pte);

	return 0;
}

int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end)
{
1932 1933
	if (!kvm->arch.pgd)
		return 0;
1934 1935 1936 1937 1938 1939
	trace_kvm_age_hva(start, end);
	return handle_hva_to_gpa(kvm, start, end, kvm_age_hva_handler, NULL);
}

int kvm_test_age_hva(struct kvm *kvm, unsigned long hva)
{
1940 1941
	if (!kvm->arch.pgd)
		return 0;
1942 1943 1944 1945
	trace_kvm_test_age_hva(hva);
	return handle_hva_to_gpa(kvm, hva, hva, kvm_test_age_hva_handler, NULL);
}

1946 1947 1948 1949 1950
void kvm_mmu_free_memory_caches(struct kvm_vcpu *vcpu)
{
	mmu_free_memory_cache(&vcpu->arch.mmu_page_cache);
}

1951 1952
phys_addr_t kvm_mmu_get_httbr(void)
{
1953 1954 1955 1956
	if (__kvm_cpu_uses_extended_idmap())
		return virt_to_phys(merged_hyp_pgd);
	else
		return virt_to_phys(hyp_pgd);
1957 1958
}

1959 1960 1961 1962 1963
phys_addr_t kvm_get_idmap_vector(void)
{
	return hyp_idmap_vector;
}

1964 1965 1966 1967 1968
static int kvm_map_idmap_text(pgd_t *pgd)
{
	int err;

	/* Create the idmap in the boot page tables */
1969
	err = 	__create_hyp_mappings(pgd, __kvm_idmap_ptrs_per_pgd(),
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
				      hyp_idmap_start, hyp_idmap_end,
				      __phys_to_pfn(hyp_idmap_start),
				      PAGE_HYP_EXEC);
	if (err)
		kvm_err("Failed to idmap %lx-%lx\n",
			hyp_idmap_start, hyp_idmap_end);

	return err;
}

1980 1981
int kvm_mmu_init(void)
{
1982 1983
	int err;

1984
	hyp_idmap_start = kvm_virt_to_phys(__hyp_idmap_text_start);
1985
	hyp_idmap_start = ALIGN_DOWN(hyp_idmap_start, PAGE_SIZE);
1986
	hyp_idmap_end = kvm_virt_to_phys(__hyp_idmap_text_end);
1987
	hyp_idmap_end = ALIGN(hyp_idmap_end, PAGE_SIZE);
1988
	hyp_idmap_vector = kvm_virt_to_phys(__kvm_hyp_init);
1989

1990 1991 1992 1993 1994
	/*
	 * We rely on the linker script to ensure at build time that the HYP
	 * init code does not cross a page boundary.
	 */
	BUG_ON((hyp_idmap_start ^ (hyp_idmap_end - 1)) & PAGE_MASK);
1995

1996 1997 1998 1999
	kvm_debug("IDMAP page: %lx\n", hyp_idmap_start);
	kvm_debug("HYP VA range: %lx:%lx\n",
		  kern_hyp_va(PAGE_OFFSET),
		  kern_hyp_va((unsigned long)high_memory - 1));
2000

M
Marc Zyngier 已提交
2001
	if (hyp_idmap_start >= kern_hyp_va(PAGE_OFFSET) &&
2002
	    hyp_idmap_start <  kern_hyp_va((unsigned long)high_memory - 1) &&
2003
	    hyp_idmap_start != (unsigned long)__hyp_idmap_text_start) {
2004 2005 2006 2007 2008 2009 2010 2011 2012
		/*
		 * The idmap page is intersecting with the VA space,
		 * it is not safe to continue further.
		 */
		kvm_err("IDMAP intersecting with HYP VA, unable to continue\n");
		err = -EINVAL;
		goto out;
	}

2013
	hyp_pgd = (pgd_t *)__get_free_pages(GFP_KERNEL | __GFP_ZERO, hyp_pgd_order);
2014
	if (!hyp_pgd) {
2015
		kvm_err("Hyp mode PGD not allocated\n");
2016 2017 2018 2019
		err = -ENOMEM;
		goto out;
	}

2020 2021 2022 2023 2024 2025 2026 2027
	if (__kvm_cpu_uses_extended_idmap()) {
		boot_hyp_pgd = (pgd_t *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
							 hyp_pgd_order);
		if (!boot_hyp_pgd) {
			kvm_err("Hyp boot PGD not allocated\n");
			err = -ENOMEM;
			goto out;
		}
2028

2029 2030 2031
		err = kvm_map_idmap_text(boot_hyp_pgd);
		if (err)
			goto out;
2032

2033 2034 2035 2036 2037 2038 2039
		merged_hyp_pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
		if (!merged_hyp_pgd) {
			kvm_err("Failed to allocate extra HYP pgd\n");
			goto out;
		}
		__kvm_extend_hypmap(boot_hyp_pgd, hyp_pgd, merged_hyp_pgd,
				    hyp_idmap_start);
2040 2041 2042 2043
	} else {
		err = kvm_map_idmap_text(hyp_pgd);
		if (err)
			goto out;
2044 2045
	}

2046
	io_map_base = hyp_idmap_start;
2047
	return 0;
2048
out:
2049
	free_hyp_pgds();
2050
	return err;
2051
}
2052 2053

void kvm_arch_commit_memory_region(struct kvm *kvm,
2054
				   const struct kvm_userspace_memory_region *mem,
2055
				   const struct kvm_memory_slot *old,
2056
				   const struct kvm_memory_slot *new,
2057 2058
				   enum kvm_mr_change change)
{
2059 2060 2061 2062 2063 2064 2065
	/*
	 * At this point memslot has been committed and there is an
	 * allocated dirty_bitmap[], dirty pages will be be tracked while the
	 * memory slot is write protected.
	 */
	if (change != KVM_MR_DELETE && mem->flags & KVM_MEM_LOG_DIRTY_PAGES)
		kvm_mmu_wp_memory_region(kvm, mem->slot);
2066 2067 2068 2069
}

int kvm_arch_prepare_memory_region(struct kvm *kvm,
				   struct kvm_memory_slot *memslot,
2070
				   const struct kvm_userspace_memory_region *mem,
2071 2072
				   enum kvm_mr_change change)
{
2073 2074 2075 2076 2077
	hva_t hva = mem->userspace_addr;
	hva_t reg_end = hva + mem->memory_size;
	bool writable = !(mem->flags & KVM_MEM_READONLY);
	int ret = 0;

2078 2079
	if (change != KVM_MR_CREATE && change != KVM_MR_MOVE &&
			change != KVM_MR_FLAGS_ONLY)
2080 2081
		return 0;

2082 2083 2084 2085 2086 2087 2088 2089
	/*
	 * Prevent userspace from creating a memory region outside of the IPA
	 * space addressable by the KVM guest IPA space.
	 */
	if (memslot->base_gfn + memslot->npages >=
	    (KVM_PHYS_SIZE >> PAGE_SHIFT))
		return -EFAULT;

2090
	down_read(&current->mm->mmap_sem);
2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127
	/*
	 * A memory region could potentially cover multiple VMAs, and any holes
	 * between them, so iterate over all of them to find out if we can map
	 * any of them right now.
	 *
	 *     +--------------------------------------------+
	 * +---------------+----------------+   +----------------+
	 * |   : VMA 1     |      VMA 2     |   |    VMA 3  :    |
	 * +---------------+----------------+   +----------------+
	 *     |               memory region                |
	 *     +--------------------------------------------+
	 */
	do {
		struct vm_area_struct *vma = find_vma(current->mm, hva);
		hva_t vm_start, vm_end;

		if (!vma || vma->vm_start >= reg_end)
			break;

		/*
		 * Mapping a read-only VMA is only allowed if the
		 * memory region is configured as read-only.
		 */
		if (writable && !(vma->vm_flags & VM_WRITE)) {
			ret = -EPERM;
			break;
		}

		/*
		 * Take the intersection of this VMA with the memory region
		 */
		vm_start = max(hva, vma->vm_start);
		vm_end = min(reg_end, vma->vm_end);

		if (vma->vm_flags & VM_PFNMAP) {
			gpa_t gpa = mem->guest_phys_addr +
				    (vm_start - mem->userspace_addr);
2128 2129 2130 2131
			phys_addr_t pa;

			pa = (phys_addr_t)vma->vm_pgoff << PAGE_SHIFT;
			pa += vm_start - vma->vm_start;
2132

2133
			/* IO region dirty page logging not allowed */
2134 2135 2136 2137
			if (memslot->flags & KVM_MEM_LOG_DIRTY_PAGES) {
				ret = -EINVAL;
				goto out;
			}
2138

2139 2140 2141 2142 2143 2144 2145 2146 2147
			ret = kvm_phys_addr_ioremap(kvm, gpa, pa,
						    vm_end - vm_start,
						    writable);
			if (ret)
				break;
		}
		hva = vm_end;
	} while (hva < reg_end);

2148
	if (change == KVM_MR_FLAGS_ONLY)
2149
		goto out;
2150

2151 2152
	spin_lock(&kvm->mmu_lock);
	if (ret)
2153
		unmap_stage2_range(kvm, mem->guest_phys_addr, mem->memory_size);
2154 2155 2156
	else
		stage2_flush_memslot(kvm, memslot);
	spin_unlock(&kvm->mmu_lock);
2157 2158
out:
	up_read(&current->mm->mmap_sem);
2159
	return ret;
2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
}

void kvm_arch_free_memslot(struct kvm *kvm, struct kvm_memory_slot *free,
			   struct kvm_memory_slot *dont)
{
}

int kvm_arch_create_memslot(struct kvm *kvm, struct kvm_memory_slot *slot,
			    unsigned long npages)
{
	return 0;
}

2173
void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen)
2174 2175 2176 2177 2178
{
}

void kvm_arch_flush_shadow_all(struct kvm *kvm)
{
2179
	kvm_free_stage2_pgd(kvm);
2180 2181 2182 2183 2184
}

void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
				   struct kvm_memory_slot *slot)
{
2185 2186 2187 2188 2189 2190
	gpa_t gpa = slot->base_gfn << PAGE_SHIFT;
	phys_addr_t size = slot->npages << PAGE_SHIFT;

	spin_lock(&kvm->mmu_lock);
	unmap_stage2_range(kvm, gpa, size);
	spin_unlock(&kvm->mmu_lock);
2191
}
2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222

/*
 * See note at ARMv7 ARM B1.14.4 (TL;DR: S/W ops are not easily virtualized).
 *
 * Main problems:
 * - S/W ops are local to a CPU (not broadcast)
 * - We have line migration behind our back (speculation)
 * - System caches don't support S/W at all (damn!)
 *
 * In the face of the above, the best we can do is to try and convert
 * S/W ops to VA ops. Because the guest is not allowed to infer the
 * S/W to PA mapping, it can only use S/W to nuke the whole cache,
 * which is a rather good thing for us.
 *
 * Also, it is only used when turning caches on/off ("The expected
 * usage of the cache maintenance instructions that operate by set/way
 * is associated with the cache maintenance instructions associated
 * with the powerdown and powerup of caches, if this is required by
 * the implementation.").
 *
 * We use the following policy:
 *
 * - If we trap a S/W operation, we enable VM trapping to detect
 *   caches being turned on/off, and do a full clean.
 *
 * - We flush the caches on both caches being turned on and off.
 *
 * - Once the caches are enabled, we stop trapping VM ops.
 */
void kvm_set_way_flush(struct kvm_vcpu *vcpu)
{
2223
	unsigned long hcr = *vcpu_hcr(vcpu);
2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237

	/*
	 * If this is the first time we do a S/W operation
	 * (i.e. HCR_TVM not set) flush the whole memory, and set the
	 * VM trapping.
	 *
	 * Otherwise, rely on the VM trapping to wait for the MMU +
	 * Caches to be turned off. At that point, we'll be able to
	 * clean the caches again.
	 */
	if (!(hcr & HCR_TVM)) {
		trace_kvm_set_way_flush(*vcpu_pc(vcpu),
					vcpu_has_cache_enabled(vcpu));
		stage2_flush_vm(vcpu->kvm);
2238
		*vcpu_hcr(vcpu) = hcr | HCR_TVM;
2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255
	}
}

void kvm_toggle_cache(struct kvm_vcpu *vcpu, bool was_enabled)
{
	bool now_enabled = vcpu_has_cache_enabled(vcpu);

	/*
	 * If switching the MMU+caches on, need to invalidate the caches.
	 * If switching it off, need to clean the caches.
	 * Clean + invalidate does the trick always.
	 */
	if (now_enabled != was_enabled)
		stage2_flush_vm(vcpu->kvm);

	/* Caches are now on, stop trapping VM ops (until a S/W op) */
	if (now_enabled)
2256
		*vcpu_hcr(vcpu) &= ~HCR_TVM;
2257 2258 2259

	trace_kvm_toggle_cache(*vcpu_pc(vcpu), was_enabled, now_enabled);
}