kvm_mmu.h 10.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 * 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.
 */

#ifndef __ARM_KVM_MMU_H__
#define __ARM_KVM_MMU_H__

22 23
#include <asm/memory.h>
#include <asm/page.h>
24

25 26 27 28
/*
 * We directly use the kernel VA for the HYP, as we can directly share
 * the mapping (HTTBR "covers" TTBR1).
 */
M
Marc Zyngier 已提交
29
#define kern_hyp_va(kva)	(kva)
30

31 32 33 34 35 36 37
/* Contrary to arm64, there is no need to generate a PC-relative address */
#define hyp_symbol_addr(s)						\
	({								\
		typeof(s) *addr = &(s);					\
		addr;							\
	})

38 39
#ifndef __ASSEMBLY__

40
#include <linux/highmem.h>
41
#include <asm/cacheflush.h>
42
#include <asm/cputype.h>
43
#include <asm/kvm_arm.h>
44
#include <asm/kvm_hyp.h>
45
#include <asm/pgalloc.h>
46
#include <asm/stage2_pgtable.h>
47

48 49 50
/* Ensure compatibility with arm64 */
#define VA_BITS			32

51 52 53 54 55 56 57
#define kvm_phys_shift(kvm)		KVM_PHYS_SHIFT
#define kvm_phys_size(kvm)		(1ULL << kvm_phys_shift(kvm))
#define kvm_phys_mask(kvm)		(kvm_phys_size(kvm) - 1ULL)
#define kvm_vttbr_baddr_mask(kvm)	VTTBR_BADDR_MASK

#define stage2_pgd_size(kvm)		(PTRS_PER_S2_PGD * sizeof(pgd_t))

58
int create_hyp_mappings(void *from, void *to, pgprot_t prot);
59
int create_hyp_io_mappings(phys_addr_t phys_addr, size_t size,
60 61
			   void __iomem **kaddr,
			   void __iomem **haddr);
62 63
int create_hyp_exec_mappings(phys_addr_t phys_addr, size_t size,
			     void **haddr);
64
void free_hyp_pgds(void);
65

66
void stage2_unmap_vm(struct kvm *kvm);
67 68 69
int kvm_alloc_stage2_pgd(struct kvm *kvm);
void kvm_free_stage2_pgd(struct kvm *kvm);
int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
70
			  phys_addr_t pa, unsigned long size, bool writable);
71 72 73 74 75

int kvm_handle_guest_abort(struct kvm_vcpu *vcpu, struct kvm_run *run);

void kvm_mmu_free_memory_caches(struct kvm_vcpu *vcpu);

76
phys_addr_t kvm_mmu_get_httbr(void);
77
phys_addr_t kvm_get_idmap_vector(void);
78 79
int kvm_mmu_init(void);
void kvm_clear_hyp_idmap(void);
80 81 82 83

#define kvm_mk_pmd(ptep)	__pmd(__pa(ptep) | PMD_TYPE_TABLE)
#define kvm_mk_pud(pmdp)	__pud(__pa(pmdp) | PMD_TYPE_TABLE)
#define kvm_mk_pgd(pudp)	({ BUILD_BUG(); 0; })
84

85 86
#define kvm_pfn_pte(pfn, prot)	pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot)	pfn_pmd(pfn, prot)
87
#define kvm_pfn_pud(pfn, prot)	(__pud(0))
88

89 90 91
#define kvm_pud_pfn(pud)	({ WARN_ON(1); 0; })


92
#define kvm_pmd_mkhuge(pmd)	pmd_mkhuge(pmd)
93 94
/* No support for pud hugepages */
#define kvm_pud_mkhuge(pud)	( {WARN_ON(1); pud; })
95

96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
/*
 * The following kvm_*pud*() functions are provided strictly to allow
 * sharing code with arm64. They should never be called in practice.
 */
static inline void kvm_set_s2pud_readonly(pud_t *pud)
{
	WARN_ON(1);
}

static inline bool kvm_s2pud_readonly(pud_t *pud)
{
	WARN_ON(1);
	return false;
}

111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
static inline void kvm_set_pud(pud_t *pud, pud_t new_pud)
{
	WARN_ON(1);
}

static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
{
	WARN_ON(1);
	return pud;
}

static inline pud_t kvm_s2pud_mkexec(pud_t pud)
{
	WARN_ON(1);
	return pud;
}

128 129 130 131 132 133
static inline bool kvm_s2pud_exec(pud_t *pud)
{
	WARN_ON(1);
	return false;
}

134 135 136 137 138 139
static inline pud_t kvm_s2pud_mkyoung(pud_t pud)
{
	BUG();
	return pud;
}

140 141 142 143 144 145
static inline bool kvm_s2pud_young(pud_t pud)
{
	WARN_ON(1);
	return false;
}

146
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
147
{
148 149
	pte_val(pte) |= L_PTE_S2_RDWR;
	return pte;
150 151
}

152
static inline pmd_t kvm_s2pmd_mkwrite(pmd_t pmd)
153
{
154 155
	pmd_val(pmd) |= L_PMD_S2_RDWR;
	return pmd;
156 157
}

158 159 160 161 162 163 164 165 166 167 168 169
static inline pte_t kvm_s2pte_mkexec(pte_t pte)
{
	pte_val(pte) &= ~L_PTE_XN;
	return pte;
}

static inline pmd_t kvm_s2pmd_mkexec(pmd_t pmd)
{
	pmd_val(pmd) &= ~PMD_SECT_XN;
	return pmd;
}

170 171 172 173 174 175 176 177 178 179
static inline void kvm_set_s2pte_readonly(pte_t *pte)
{
	pte_val(*pte) = (pte_val(*pte) & ~L_PTE_S2_RDWR) | L_PTE_S2_RDONLY;
}

static inline bool kvm_s2pte_readonly(pte_t *pte)
{
	return (pte_val(*pte) & L_PTE_S2_RDWR) == L_PTE_S2_RDONLY;
}

180 181 182 183 184
static inline bool kvm_s2pte_exec(pte_t *pte)
{
	return !(pte_val(*pte) & L_PTE_XN);
}

185 186 187 188 189 190 191 192 193 194
static inline void kvm_set_s2pmd_readonly(pmd_t *pmd)
{
	pmd_val(*pmd) = (pmd_val(*pmd) & ~L_PMD_S2_RDWR) | L_PMD_S2_RDONLY;
}

static inline bool kvm_s2pmd_readonly(pmd_t *pmd)
{
	return (pmd_val(*pmd) & L_PMD_S2_RDWR) == L_PMD_S2_RDONLY;
}

195 196 197 198 199
static inline bool kvm_s2pmd_exec(pmd_t *pmd)
{
	return !(pmd_val(*pmd) & PMD_SECT_XN);
}

200 201 202 203 204 205
static inline bool kvm_page_empty(void *ptr)
{
	struct page *ptr_page = virt_to_page(ptr);
	return page_count(ptr_page) == 1;
}

206 207
#define kvm_pte_table_empty(kvm, ptep) kvm_page_empty(ptep)
#define kvm_pmd_table_empty(kvm, pmdp) kvm_page_empty(pmdp)
208
#define kvm_pud_table_empty(kvm, pudp) false
209

210 211 212
#define hyp_pte_table_empty(ptep) kvm_page_empty(ptep)
#define hyp_pmd_table_empty(pmdp) kvm_page_empty(pmdp)
#define hyp_pud_table_empty(pudp) false
213

214 215
struct kvm;

216 217 218 219
#define kvm_flush_dcache_to_poc(a,l)	__cpuc_flush_dcache_area((a), (l))

static inline bool vcpu_has_cache_enabled(struct kvm_vcpu *vcpu)
{
220
	return (vcpu_cp15(vcpu, c1_SCTLR) & 0b101) == 0b101;
221 222
}

223
static inline void __clean_dcache_guest_page(kvm_pfn_t pfn, unsigned long size)
224 225
{
	/*
226
	 * Clean the dcache to the Point of Coherency.
227 228 229 230 231
	 *
	 * We need to do this through a kernel mapping (using the
	 * user-space mapping has proved to be the wrong
	 * solution). For that, we need to kmap one page at a time,
	 * and iterate over the range.
232
	 */
233

234
	VM_BUG_ON(size & ~PAGE_MASK);
235 236 237 238

	while (size) {
		void *va = kmap_atomic_pfn(pfn);

239
		kvm_flush_dcache_to_poc(va, PAGE_SIZE);
240 241 242 243 244 245

		size -= PAGE_SIZE;
		pfn++;

		kunmap_atomic(va);
	}
246
}
247

248
static inline void __invalidate_icache_guest_page(kvm_pfn_t pfn,
249
						  unsigned long size)
250
{
251 252
	u32 iclsz;

253 254 255 256 257 258 259 260 261 262 263 264
	/*
	 * If we are going to insert an instruction page and the icache is
	 * either VIPT or PIPT, there is a potential problem where the host
	 * (or another VM) may have used the same page as this guest, and we
	 * read incorrect data from the icache.  If we're using a PIPT cache,
	 * we can invalidate just that page, but if we are using a VIPT cache
	 * we need to invalidate the entire icache - damn shame - as written
	 * in the ARM ARM (DDI 0406C.b - Page B3-1393).
	 *
	 * VIVT caches are tagged using both the ASID and the VMID and doesn't
	 * need any kind of flushing (DDI 0406C.b - Page B3-1392).
	 */
265

266
	VM_BUG_ON(size & ~PAGE_MASK);
267

268 269 270 271
	if (icache_is_vivt_asid_tagged())
		return;

	if (!icache_is_pipt()) {
272 273
		/* any kind of VIPT cache */
		__flush_icache_all();
274 275 276
		return;
	}

277 278 279 280 281 282 283 284 285
	/*
	 * CTR IminLine contains Log2 of the number of words in the
	 * cache line, so we can get the number of words as
	 * 2 << (IminLine - 1).  To get the number of bytes, we
	 * multiply by 4 (the number of bytes in a 32-bit word), and
	 * get 4 << (IminLine).
	 */
	iclsz = 4 << (read_cpuid(CPUID_CACHETYPE) & 0xf);

286 287
	while (size) {
		void *va = kmap_atomic_pfn(pfn);
288 289
		void *end = va + PAGE_SIZE;
		void *addr = va;
290

291 292 293 294
		do {
			write_sysreg(addr, ICIMVAU);
			addr += iclsz;
		} while (addr < end);
295

296 297
		dsb(ishst);
		isb();
298 299 300 301 302 303 304

		size -= PAGE_SIZE;
		pfn++;

		kunmap_atomic(va);
	}

305 306 307 308 309
	/* Check if we need to invalidate the BTB */
	if ((read_cpuid_ext(CPUID_EXT_MMFR1) >> 28) != 4) {
		write_sysreg(0, BPIALLIS);
		dsb(ishst);
		isb();
310 311 312
	}
}

313 314 315 316 317 318 319 320 321 322 323 324
static inline void __kvm_flush_dcache_pte(pte_t pte)
{
	void *va = kmap_atomic(pte_page(pte));

	kvm_flush_dcache_to_poc(va, PAGE_SIZE);

	kunmap_atomic(va);
}

static inline void __kvm_flush_dcache_pmd(pmd_t pmd)
{
	unsigned long size = PMD_SIZE;
D
Dan Williams 已提交
325
	kvm_pfn_t pfn = pmd_pfn(pmd);
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342

	while (size) {
		void *va = kmap_atomic_pfn(pfn);

		kvm_flush_dcache_to_poc(va, PAGE_SIZE);

		pfn++;
		size -= PAGE_SIZE;

		kunmap_atomic(va);
	}
}

static inline void __kvm_flush_dcache_pud(pud_t pud)
{
}

343
#define kvm_virt_to_phys(x)		virt_to_idmap((unsigned long)(x))
344

345 346
void kvm_set_way_flush(struct kvm_vcpu *vcpu);
void kvm_toggle_cache(struct kvm_vcpu *vcpu, bool was_enabled);
347

348 349 350 351 352
static inline bool __kvm_cpu_uses_extended_idmap(void)
{
	return false;
}

353 354 355 356 357
static inline unsigned long __kvm_idmap_ptrs_per_pgd(void)
{
	return PTRS_PER_PGD;
}

358 359 360 361 362
static inline void __kvm_extend_hypmap(pgd_t *boot_hyp_pgd,
				       pgd_t *hyp_pgd,
				       pgd_t *merged_hyp_pgd,
				       unsigned long hyp_idmap_start) { }

363 364 365 366 367
static inline unsigned int kvm_get_vmid_bits(void)
{
	return 8;
}

368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
/*
 * We are not in the kvm->srcu critical section most of the time, so we take
 * the SRCU read lock here. Since we copy the data from the user page, we
 * can immediately drop the lock again.
 */
static inline int kvm_read_guest_lock(struct kvm *kvm,
				      gpa_t gpa, void *data, unsigned long len)
{
	int srcu_idx = srcu_read_lock(&kvm->srcu);
	int ret = kvm_read_guest(kvm, gpa, data, len);

	srcu_read_unlock(&kvm->srcu, srcu_idx);

	return ret;
}

384 385
static inline void *kvm_get_hyp_vector(void)
{
386 387 388 389 390 391 392 393 394
	switch(read_cpuid_part()) {
#ifdef CONFIG_HARDEN_BRANCH_PREDICTOR
	case ARM_CPU_PART_CORTEX_A12:
	case ARM_CPU_PART_CORTEX_A17:
	{
		extern char __kvm_hyp_vector_bp_inv[];
		return kvm_ksym_ref(__kvm_hyp_vector_bp_inv);
	}

395
	case ARM_CPU_PART_BRAHMA_B15:
396 397 398 399 400
	case ARM_CPU_PART_CORTEX_A15:
	{
		extern char __kvm_hyp_vector_ic_inv[];
		return kvm_ksym_ref(__kvm_hyp_vector_ic_inv);
	}
401 402 403 404 405 406 407
#endif
	default:
	{
		extern char __kvm_hyp_vector[];
		return kvm_ksym_ref(__kvm_hyp_vector);
	}
	}
408 409 410 411 412 413 414
}

static inline int kvm_map_vectors(void)
{
	return 0;
}

415 416 417 418 419
static inline int hyp_map_aux_data(void)
{
	return 0;
}

420 421
#define kvm_phys_to_vttbr(addr)		(addr)

422 423
static inline void kvm_set_ipa_limit(void) {}

424
static __always_inline u64 kvm_get_vttbr(struct kvm *kvm)
425
{
426 427 428 429 430 431
	struct kvm_vmid *vmid = &kvm->arch.vmid;
	u64 vmid_field, baddr;

	baddr = kvm->arch.pgd_phys;
	vmid_field = (u64)vmid->vmid << VTTBR_VMID_SHIFT;
	return kvm_phys_to_vttbr(baddr) | vmid_field;
432 433
}

434 435
#endif	/* !__ASSEMBLY__ */

436
#endif /* __ARM_KVM_MMU_H__ */