e500_tlb.c 33.1 KB
Newer Older
1
/*
2
 * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. All rights reserved.
3 4
 *
 * Author: Yu Liu, yu.liu@freescale.com
S
Scott Wood 已提交
5
 *         Scott Wood, scottwood@freescale.com
6
 *         Ashish Kalra, ashish.kalra@freescale.com
S
Scott Wood 已提交
7
 *         Varun Sethi, varun.sethi@freescale.com
8 9 10 11 12 13 14 15 16 17
 *
 * Description:
 * This file is based on arch/powerpc/kvm/44x_tlb.c,
 * by Hollis Blanchard <hollisb@us.ibm.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.
 */

18
#include <linux/kernel.h>
19
#include <linux/types.h>
20
#include <linux/slab.h>
21 22 23 24
#include <linux/string.h>
#include <linux/kvm.h>
#include <linux/kvm_host.h>
#include <linux/highmem.h>
S
Scott Wood 已提交
25 26 27 28 29
#include <linux/log2.h>
#include <linux/uaccess.h>
#include <linux/sched.h>
#include <linux/rwsem.h>
#include <linux/vmalloc.h>
30
#include <linux/hugetlb.h>
31 32
#include <asm/kvm_ppc.h>

33
#include "e500.h"
34
#include "trace.h"
35
#include "timing.h"
36

37
#define to_htlb1_esel(esel) (host_tlb_params[1].entries - (esel) - 1)
38

39
static struct kvmppc_e500_tlb_params host_tlb_params[E500_TLB_NUM];
40

41
static inline unsigned int gtlb0_get_next_victim(
42 43 44 45
		struct kvmppc_vcpu_e500 *vcpu_e500)
{
	unsigned int victim;

L
Liu Yu 已提交
46
	victim = vcpu_e500->gtlb_nv[0]++;
S
Scott Wood 已提交
47
	if (unlikely(vcpu_e500->gtlb_nv[0] >= vcpu_e500->gtlb_params[0].ways))
L
Liu Yu 已提交
48
		vcpu_e500->gtlb_nv[0] = 0;
49 50 51 52 53 54

	return victim;
}

static inline unsigned int tlb1_max_shadow_size(void)
{
S
Scott Wood 已提交
55
	/* reserve one entry for magic page */
56
	return host_tlb_params[1].entries - tlbcam_index - 1;
57 58
}

S
Scott Wood 已提交
59
static inline int tlbe_is_writable(struct kvm_book3e_206_tlb_entry *tlbe)
60
{
S
Scott Wood 已提交
61
	return tlbe->mas7_3 & (MAS3_SW|MAS3_UW);
62 63 64 65 66 67 68
}

static inline u32 e500_shadow_mas3_attrib(u32 mas3, int usermode)
{
	/* Mask off reserved bits. */
	mas3 &= MAS3_ATTRIB_MASK;

S
Scott Wood 已提交
69
#ifndef CONFIG_KVM_BOOKE_HV
70 71 72 73 74 75 76
	if (!usermode) {
		/* Guest is in supervisor mode,
		 * so we need to translate guest
		 * supervisor permissions into user permissions. */
		mas3 &= ~E500_TLB_USER_PERM_MASK;
		mas3 |= (mas3 & E500_TLB_SUPER_PERM_MASK) << 1;
	}
S
Scott Wood 已提交
77 78 79
	mas3 |= E500_TLB_SUPER_PERM_MASK;
#endif
	return mas3;
80 81 82 83
}

static inline u32 e500_shadow_mas2_attrib(u32 mas2, int usermode)
{
84 85 86
#ifdef CONFIG_SMP
	return (mas2 & MAS2_ATTRIB_MASK) | MAS2_M;
#else
87
	return mas2 & MAS2_ATTRIB_MASK;
88
#endif
89 90 91 92 93
}

/*
 * writing shadow tlb entry to host TLB
 */
S
Scott Wood 已提交
94 95
static inline void __write_host_tlbe(struct kvm_book3e_206_tlb_entry *stlbe,
				     uint32_t mas0)
96
{
97 98 99 100
	unsigned long flags;

	local_irq_save(flags);
	mtspr(SPRN_MAS0, mas0);
101
	mtspr(SPRN_MAS1, stlbe->mas1);
S
Scott Wood 已提交
102 103 104
	mtspr(SPRN_MAS2, (unsigned long)stlbe->mas2);
	mtspr(SPRN_MAS3, (u32)stlbe->mas7_3);
	mtspr(SPRN_MAS7, (u32)(stlbe->mas7_3 >> 32));
S
Scott Wood 已提交
105 106 107
#ifdef CONFIG_KVM_BOOKE_HV
	mtspr(SPRN_MAS8, stlbe->mas8);
#endif
108
	asm volatile("isync; tlbwe" : : : "memory");
S
Scott Wood 已提交
109 110 111 112 113 114

#ifdef CONFIG_KVM_BOOKE_HV
	/* Must clear mas8 for other host tlbwe's */
	mtspr(SPRN_MAS8, 0);
	isync();
#endif
115
	local_irq_restore(flags);
L
Liu Yu 已提交
116 117 118

	trace_kvm_booke206_stlb_write(mas0, stlbe->mas8, stlbe->mas1,
	                              stlbe->mas2, stlbe->mas7_3);
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
/*
 * Acquire a mas0 with victim hint, as if we just took a TLB miss.
 *
 * We don't care about the address we're searching for, other than that it's
 * in the right set and is not present in the TLB.  Using a zero PID and a
 * userspace address means we don't have to set and then restore MAS5, or
 * calculate a proper MAS6 value.
 */
static u32 get_host_mas0(unsigned long eaddr)
{
	unsigned long flags;
	u32 mas0;

	local_irq_save(flags);
	mtspr(SPRN_MAS6, 0);
	asm volatile("tlbsx 0, %0" : : "b" (eaddr & ~CONFIG_PAGE_OFFSET));
	mas0 = mfspr(SPRN_MAS0);
	local_irq_restore(flags);

	return mas0;
}

/* sesel is for tlb1 only */
144
static inline void write_host_tlbe(struct kvmppc_vcpu_e500 *vcpu_e500,
145
		int tlbsel, int sesel, struct kvm_book3e_206_tlb_entry *stlbe)
146
{
147 148
	u32 mas0;

149
	if (tlbsel == 0) {
150 151
		mas0 = get_host_mas0(stlbe->mas2);
		__write_host_tlbe(stlbe, mas0);
152
	} else {
153 154
		__write_host_tlbe(stlbe,
				  MAS0_TLBSEL(1) |
155
				  MAS0_ESEL(to_htlb1_esel(sesel)));
156 157 158
	}
}

159
#ifdef CONFIG_KVM_E500
S
Scott Wood 已提交
160 161
void kvmppc_map_magic(struct kvm_vcpu *vcpu)
{
L
Liu Yu 已提交
162
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
S
Scott Wood 已提交
163
	struct kvm_book3e_206_tlb_entry magic;
S
Scott Wood 已提交
164
	ulong shared_page = ((ulong)vcpu->arch.shared) & PAGE_MASK;
L
Liu Yu 已提交
165
	unsigned int stid;
S
Scott Wood 已提交
166 167 168 169 170
	pfn_t pfn;

	pfn = (pfn_t)virt_to_phys((void *)shared_page) >> PAGE_SHIFT;
	get_page(pfn_to_page(pfn));

L
Liu Yu 已提交
171 172 173 174
	preempt_disable();
	stid = kvmppc_e500_get_sid(vcpu_e500, 0, 0, 0, 0);

	magic.mas1 = MAS1_VALID | MAS1_TS | MAS1_TID(stid) |
S
Scott Wood 已提交
175 176
		     MAS1_TSIZE(BOOK3E_PAGESZ_4K);
	magic.mas2 = vcpu->arch.magic_page_ea | MAS2_M;
S
Scott Wood 已提交
177 178
	magic.mas7_3 = ((u64)pfn << PAGE_SHIFT) |
		       MAS3_SW | MAS3_SR | MAS3_UW | MAS3_UR;
L
Liu Yu 已提交
179
	magic.mas8 = 0;
S
Scott Wood 已提交
180 181

	__write_host_tlbe(&magic, MAS0_TLBSEL(1) | MAS0_ESEL(tlbcam_index));
L
Liu Yu 已提交
182
	preempt_enable();
S
Scott Wood 已提交
183
}
184
#endif
L
Liu Yu 已提交
185

186 187
static void inval_gtlbe_on_host(struct kvmppc_vcpu_e500 *vcpu_e500,
				int tlbsel, int esel)
L
Liu Yu 已提交
188
{
S
Scott Wood 已提交
189 190
	struct kvm_book3e_206_tlb_entry *gtlbe =
		get_entry(vcpu_e500, tlbsel, esel);
L
Liu Yu 已提交
191

192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
	if (tlbsel == 1 &&
	    vcpu_e500->gtlb_priv[1][esel].ref.flags & E500_TLB_BITMAP) {
		u64 tmp = vcpu_e500->g2h_tlb1_map[esel];
		int hw_tlb_indx;
		unsigned long flags;

		local_irq_save(flags);
		while (tmp) {
			hw_tlb_indx = __ilog2_u64(tmp & -tmp);
			mtspr(SPRN_MAS0,
			      MAS0_TLBSEL(1) |
			      MAS0_ESEL(to_htlb1_esel(hw_tlb_indx)));
			mtspr(SPRN_MAS1, 0);
			asm volatile("tlbwe");
			vcpu_e500->h2g_tlb1_rmap[hw_tlb_indx] = 0;
			tmp &= tmp - 1;
		}
		mb();
		vcpu_e500->g2h_tlb1_map[esel] = 0;
		vcpu_e500->gtlb_priv[1][esel].ref.flags &= ~E500_TLB_BITMAP;
		local_irq_restore(flags);

214
		return;
L
Liu Yu 已提交
215 216
	}

217 218
	/* Guest tlbe is backed by at most one host tlbe per shadow pid. */
	kvmppc_e500_tlbil_one(vcpu_e500, gtlbe);
219 220
}

221 222 223 224 225 226 227 228 229 230 231 232
static int tlb0_set_base(gva_t addr, int sets, int ways)
{
	int set_base;

	set_base = (addr >> PAGE_SHIFT) & (sets - 1);
	set_base *= ways;

	return set_base;
}

static int gtlb0_set_base(struct kvmppc_vcpu_e500 *vcpu_e500, gva_t addr)
{
S
Scott Wood 已提交
233 234
	return tlb0_set_base(addr, vcpu_e500->gtlb_params[0].sets,
			     vcpu_e500->gtlb_params[0].ways);
235 236
}

237
static unsigned int get_tlb_esel(struct kvm_vcpu *vcpu, int tlbsel)
238
{
239 240
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
	int esel = get_tlb_esel_bit(vcpu);
241 242

	if (tlbsel == 0) {
S
Scott Wood 已提交
243
		esel &= vcpu_e500->gtlb_params[0].ways - 1;
244
		esel += gtlb0_set_base(vcpu_e500, vcpu->arch.shared->mas2);
245
	} else {
S
Scott Wood 已提交
246
		esel &= vcpu_e500->gtlb_params[tlbsel].entries - 1;
247 248 249 250 251
	}

	return esel;
}

252 253 254 255
/* Search the guest TLB for a matching entry. */
static int kvmppc_e500_tlb_index(struct kvmppc_vcpu_e500 *vcpu_e500,
		gva_t eaddr, int tlbsel, unsigned int pid, int as)
{
S
Scott Wood 已提交
256 257
	int size = vcpu_e500->gtlb_params[tlbsel].entries;
	unsigned int set_base, offset;
258 259
	int i;

260
	if (tlbsel == 0) {
261
		set_base = gtlb0_set_base(vcpu_e500, eaddr);
S
Scott Wood 已提交
262
		size = vcpu_e500->gtlb_params[0].ways;
263 264 265 266
	} else {
		set_base = 0;
	}

S
Scott Wood 已提交
267 268
	offset = vcpu_e500->gtlb_offset[tlbsel];

269
	for (i = 0; i < size; i++) {
S
Scott Wood 已提交
270 271
		struct kvm_book3e_206_tlb_entry *tlbe =
			&vcpu_e500->gtlb_arch[offset + set_base + i];
272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
		unsigned int tid;

		if (eaddr < get_tlb_eaddr(tlbe))
			continue;

		if (eaddr > get_tlb_end(tlbe))
			continue;

		tid = get_tlb_tid(tlbe);
		if (tid && (tid != pid))
			continue;

		if (!get_tlb_v(tlbe))
			continue;

		if (get_tlb_ts(tlbe) != as && as != -1)
			continue;

290
		return set_base + i;
291 292 293 294 295
	}

	return -1;
}

296
static inline void kvmppc_e500_ref_setup(struct tlbe_ref *ref,
S
Scott Wood 已提交
297
					 struct kvm_book3e_206_tlb_entry *gtlbe,
298
					 pfn_t pfn)
299
{
300 301
	ref->pfn = pfn;
	ref->flags = E500_TLB_VALID;
302

L
Liu Yu 已提交
303
	if (tlbe_is_writable(gtlbe))
304
		ref->flags |= E500_TLB_DIRTY;
305 306
}

307
static inline void kvmppc_e500_ref_release(struct tlbe_ref *ref)
308
{
309 310 311
	if (ref->flags & E500_TLB_VALID) {
		if (ref->flags & E500_TLB_DIRTY)
			kvm_release_pfn_dirty(ref->pfn);
L
Liu Yu 已提交
312
		else
313 314 315 316 317 318
			kvm_release_pfn_clean(ref->pfn);

		ref->flags = 0;
	}
}

319 320 321 322 323 324 325 326 327 328
static void clear_tlb1_bitmap(struct kvmppc_vcpu_e500 *vcpu_e500)
{
	if (vcpu_e500->g2h_tlb1_map)
		memset(vcpu_e500->g2h_tlb1_map,
		       sizeof(u64) * vcpu_e500->gtlb_params[1].entries, 0);
	if (vcpu_e500->h2g_tlb1_rmap)
		memset(vcpu_e500->h2g_tlb1_rmap,
		       sizeof(unsigned int) * host_tlb_params[1].entries, 0);
}

329 330 331 332
static void clear_tlb_privs(struct kvmppc_vcpu_e500 *vcpu_e500)
{
	int tlbsel = 0;
	int i;
333

S
Scott Wood 已提交
334
	for (i = 0; i < vcpu_e500->gtlb_params[tlbsel].entries; i++) {
335 336 337
		struct tlbe_ref *ref =
			&vcpu_e500->gtlb_priv[tlbsel][i].ref;
		kvmppc_e500_ref_release(ref);
L
Liu Yu 已提交
338
	}
339 340
}

341 342 343 344 345
static void clear_tlb_refs(struct kvmppc_vcpu_e500 *vcpu_e500)
{
	int stlbsel = 1;
	int i;

346
	kvmppc_e500_tlbil_all(vcpu_e500);
S
Scott Wood 已提交
347

348 349 350 351 352 353 354 355 356
	for (i = 0; i < host_tlb_params[stlbsel].entries; i++) {
		struct tlbe_ref *ref =
			&vcpu_e500->tlb_refs[stlbsel][i];
		kvmppc_e500_ref_release(ref);
	}

	clear_tlb_privs(vcpu_e500);
}

357 358 359 360
static inline void kvmppc_e500_deliver_tlb_miss(struct kvm_vcpu *vcpu,
		unsigned int eaddr, int as)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
361
	unsigned int victim, tsized;
362 363
	int tlbsel;

364
	/* since we only have two TLBs, only lower bit is used. */
365
	tlbsel = (vcpu->arch.shared->mas4 >> 28) & 0x1;
366
	victim = (tlbsel == 0) ? gtlb0_get_next_victim(vcpu_e500) : 0;
367
	tsized = (vcpu->arch.shared->mas4 >> 7) & 0x1f;
368

369
	vcpu->arch.shared->mas0 = MAS0_TLBSEL(tlbsel) | MAS0_ESEL(victim)
L
Liu Yu 已提交
370
		| MAS0_NV(vcpu_e500->gtlb_nv[tlbsel]);
371
	vcpu->arch.shared->mas1 = MAS1_VALID | (as ? MAS1_TS : 0)
372
		| MAS1_TID(get_tlbmiss_tid(vcpu))
373
		| MAS1_TSIZE(tsized);
374 375 376 377
	vcpu->arch.shared->mas2 = (eaddr & MAS2_EPN)
		| (vcpu->arch.shared->mas4 & MAS2_ATTRIB_MASK);
	vcpu->arch.shared->mas7_3 &= MAS3_U0 | MAS3_U1 | MAS3_U2 | MAS3_U3;
	vcpu->arch.shared->mas6 = (vcpu->arch.shared->mas6 & MAS6_SPID1)
378 379 380 381
		| (get_cur_pid(vcpu) << 16)
		| (as ? MAS6_SAS : 0);
}

382
/* TID must be supplied by the caller */
S
Scott Wood 已提交
383
static inline void kvmppc_e500_setup_stlbe(
384
	struct kvm_vcpu *vcpu,
S
Scott Wood 已提交
385 386 387
	struct kvm_book3e_206_tlb_entry *gtlbe,
	int tsize, struct tlbe_ref *ref, u64 gvaddr,
	struct kvm_book3e_206_tlb_entry *stlbe)
L
Liu Yu 已提交
388
{
389
	pfn_t pfn = ref->pfn;
390
	u32 pr = vcpu->arch.shared->msr & MSR_PR;
391 392

	BUG_ON(!(ref->flags & E500_TLB_VALID));
L
Liu Yu 已提交
393

394 395 396 397 398 399
	/* Force IPROT=0 for all guest mappings. */
	stlbe->mas1 = MAS1_TSIZE(tsize) | get_tlb_sts(gtlbe) | MAS1_VALID;
	stlbe->mas2 = (gvaddr & MAS2_EPN) |
		      e500_shadow_mas2_attrib(gtlbe->mas2, pr);
	stlbe->mas7_3 = ((u64)pfn << PAGE_SHIFT) |
			e500_shadow_mas3_attrib(gtlbe->mas7_3, pr);
S
Scott Wood 已提交
400 401 402 403

#ifdef CONFIG_KVM_BOOKE_HV
	stlbe->mas8 = MAS8_TGS | vcpu->kvm->arch.lpid;
#endif
L
Liu Yu 已提交
404 405
}

406
static inline void kvmppc_e500_shadow_map(struct kvmppc_vcpu_e500 *vcpu_e500,
S
Scott Wood 已提交
407
	u64 gvaddr, gfn_t gfn, struct kvm_book3e_206_tlb_entry *gtlbe,
408
	int tlbsel, struct kvm_book3e_206_tlb_entry *stlbe,
S
Scott Wood 已提交
409
	struct tlbe_ref *ref)
410
{
411 412 413 414
	struct kvm_memory_slot *slot;
	unsigned long pfn, hva;
	int pfnmap = 0;
	int tsize = BOOK3E_PAGESZ_4K;
415

416 417 418
	/*
	 * Translate guest physical to true physical, acquiring
	 * a page reference if it is normal, non-reserved memory.
419 420 421 422
	 *
	 * gfn_to_memslot() must succeed because otherwise we wouldn't
	 * have gotten this far.  Eventually we should just pass the slot
	 * pointer through from the first lookup.
423
	 */
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
	slot = gfn_to_memslot(vcpu_e500->vcpu.kvm, gfn);
	hva = gfn_to_hva_memslot(slot, gfn);

	if (tlbsel == 1) {
		struct vm_area_struct *vma;
		down_read(&current->mm->mmap_sem);

		vma = find_vma(current->mm, hva);
		if (vma && hva >= vma->vm_start &&
		    (vma->vm_flags & VM_PFNMAP)) {
			/*
			 * This VMA is a physically contiguous region (e.g.
			 * /dev/mem) that bypasses normal Linux page
			 * management.  Find the overlap between the
			 * vma and the memslot.
			 */

			unsigned long start, end;
			unsigned long slot_start, slot_end;

			pfnmap = 1;

			start = vma->vm_pgoff;
			end = start +
			      ((vma->vm_end - vma->vm_start) >> PAGE_SHIFT);

			pfn = start + ((hva - vma->vm_start) >> PAGE_SHIFT);

			slot_start = pfn - (gfn - slot->base_gfn);
			slot_end = slot_start + slot->npages;

			if (start < slot_start)
				start = slot_start;
			if (end > slot_end)
				end = slot_end;

			tsize = (gtlbe->mas1 & MAS1_TSIZE_MASK) >>
				MAS1_TSIZE_SHIFT;

			/*
			 * e500 doesn't implement the lowest tsize bit,
			 * or 1K pages.
			 */
			tsize = max(BOOK3E_PAGESZ_4K, tsize & ~1);

			/*
			 * Now find the largest tsize (up to what the guest
			 * requested) that will cover gfn, stay within the
			 * range, and for which gfn and pfn are mutually
			 * aligned.
			 */

			for (; tsize > BOOK3E_PAGESZ_4K; tsize -= 2) {
				unsigned long gfn_start, gfn_end, tsize_pages;
				tsize_pages = 1 << (tsize - 2);

				gfn_start = gfn & ~(tsize_pages - 1);
				gfn_end = gfn_start + tsize_pages;

				if (gfn_start + pfn - gfn < start)
					continue;
				if (gfn_end + pfn - gfn > end)
					continue;
				if ((gfn & (tsize_pages - 1)) !=
				    (pfn & (tsize_pages - 1)))
					continue;

				gvaddr &= ~((tsize_pages << PAGE_SHIFT) - 1);
				pfn &= ~(tsize_pages - 1);
				break;
			}
495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
		} else if (vma && hva >= vma->vm_start &&
			   (vma->vm_flags & VM_HUGETLB)) {
			unsigned long psize = vma_kernel_pagesize(vma);

			tsize = (gtlbe->mas1 & MAS1_TSIZE_MASK) >>
				MAS1_TSIZE_SHIFT;

			/*
			 * Take the largest page size that satisfies both host
			 * and guest mapping
			 */
			tsize = min(__ilog2(psize) - 10, tsize);

			/*
			 * e500 doesn't implement the lowest tsize bit,
			 * or 1K pages.
			 */
			tsize = max(BOOK3E_PAGESZ_4K, tsize & ~1);
513 514 515 516 517 518
		}

		up_read(&current->mm->mmap_sem);
	}

	if (likely(!pfnmap)) {
519
		unsigned long tsize_pages = 1 << (tsize + 10 - PAGE_SHIFT);
520 521 522 523 524 525 526
		pfn = gfn_to_pfn_memslot(vcpu_e500->vcpu.kvm, slot, gfn);
		if (is_error_pfn(pfn)) {
			printk(KERN_ERR "Couldn't get real page for gfn %lx!\n",
					(long)gfn);
			kvm_release_pfn_clean(pfn);
			return;
		}
527 528 529 530

		/* Align guest and physical address to page map boundaries */
		pfn &= ~(tsize_pages - 1);
		gvaddr &= ~((tsize_pages << PAGE_SHIFT) - 1);
531 532
	}

533 534 535
	/* Drop old ref and setup new one. */
	kvmppc_e500_ref_release(ref);
	kvmppc_e500_ref_setup(ref, gtlbe, pfn);
536

537 538
	kvmppc_e500_setup_stlbe(&vcpu_e500->vcpu, gtlbe, tsize,
				ref, gvaddr, stlbe);
539 540 541
}

/* XXX only map the one-one case, for now use TLB0 */
542 543 544
static void kvmppc_e500_tlb0_map(struct kvmppc_vcpu_e500 *vcpu_e500,
				 int esel,
				 struct kvm_book3e_206_tlb_entry *stlbe)
545
{
S
Scott Wood 已提交
546
	struct kvm_book3e_206_tlb_entry *gtlbe;
547
	struct tlbe_ref *ref;
548

S
Scott Wood 已提交
549
	gtlbe = get_entry(vcpu_e500, 0, esel);
550 551
	ref = &vcpu_e500->gtlb_priv[0][esel].ref;

552 553
	kvmppc_e500_shadow_map(vcpu_e500, get_tlb_eaddr(gtlbe),
			get_tlb_raddr(gtlbe) >> PAGE_SHIFT,
554
			gtlbe, 0, stlbe, ref);
555 556 557 558 559 560
}

/* Caller must ensure that the specified guest TLB entry is safe to insert into
 * the shadow TLB. */
/* XXX for both one-one and one-to-many , for now use TLB1 */
static int kvmppc_e500_tlb1_map(struct kvmppc_vcpu_e500 *vcpu_e500,
S
Scott Wood 已提交
561
		u64 gvaddr, gfn_t gfn, struct kvm_book3e_206_tlb_entry *gtlbe,
562
		struct kvm_book3e_206_tlb_entry *stlbe, int esel)
563
{
564
	struct tlbe_ref *ref;
565 566
	unsigned int victim;

567
	victim = vcpu_e500->host_tlb1_nv++;
568

569 570
	if (unlikely(vcpu_e500->host_tlb1_nv >= tlb1_max_shadow_size()))
		vcpu_e500->host_tlb1_nv = 0;
571

572
	ref = &vcpu_e500->tlb_refs[1][victim];
573
	kvmppc_e500_shadow_map(vcpu_e500, gvaddr, gfn, gtlbe, 1, stlbe, ref);
574

575 576 577 578 579 580 581 582
	vcpu_e500->g2h_tlb1_map[esel] |= (u64)1 << victim;
	vcpu_e500->gtlb_priv[1][esel].ref.flags |= E500_TLB_BITMAP;
	if (vcpu_e500->h2g_tlb1_rmap[victim]) {
		unsigned int idx = vcpu_e500->h2g_tlb1_rmap[victim];
		vcpu_e500->g2h_tlb1_map[idx] &= ~(1ULL << victim);
	}
	vcpu_e500->h2g_tlb1_rmap[victim] = esel;

583 584 585
	return victim;
}

L
Liu Yu 已提交
586 587 588
static inline int kvmppc_e500_gtlbe_invalidate(
				struct kvmppc_vcpu_e500 *vcpu_e500,
				int tlbsel, int esel)
589
{
S
Scott Wood 已提交
590 591
	struct kvm_book3e_206_tlb_entry *gtlbe =
		get_entry(vcpu_e500, tlbsel, esel);
592 593 594 595 596 597 598 599 600

	if (unlikely(get_tlb_iprot(gtlbe)))
		return -1;

	gtlbe->mas1 = 0;

	return 0;
}

601 602 603 604 605
int kvmppc_e500_emul_mt_mmucsr0(struct kvmppc_vcpu_e500 *vcpu_e500, ulong value)
{
	int esel;

	if (value & MMUCSR0_TLB0FI)
S
Scott Wood 已提交
606
		for (esel = 0; esel < vcpu_e500->gtlb_params[0].entries; esel++)
607 608
			kvmppc_e500_gtlbe_invalidate(vcpu_e500, 0, esel);
	if (value & MMUCSR0_TLB1FI)
S
Scott Wood 已提交
609
		for (esel = 0; esel < vcpu_e500->gtlb_params[1].entries; esel++)
610 611
			kvmppc_e500_gtlbe_invalidate(vcpu_e500, 1, esel);

L
Liu Yu 已提交
612
	/* Invalidate all vcpu id mappings */
613
	kvmppc_e500_tlbil_all(vcpu_e500);
614 615 616 617

	return EMULATE_DONE;
}

618 619 620 621 622 623 624
int kvmppc_e500_emul_tlbivax(struct kvm_vcpu *vcpu, int ra, int rb)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
	unsigned int ia;
	int esel, tlbsel;
	gva_t ea;

625
	ea = ((ra) ? kvmppc_get_gpr(vcpu, ra) : 0) + kvmppc_get_gpr(vcpu, rb);
626 627 628

	ia = (ea >> 2) & 0x1;

629
	/* since we only have two TLBs, only lower bit is used. */
630 631 632 633
	tlbsel = (ea >> 3) & 0x1;

	if (ia) {
		/* invalidate all entries */
S
Scott Wood 已提交
634 635
		for (esel = 0; esel < vcpu_e500->gtlb_params[tlbsel].entries;
		     esel++)
636 637 638 639 640 641 642 643 644
			kvmppc_e500_gtlbe_invalidate(vcpu_e500, tlbsel, esel);
	} else {
		ea &= 0xfffff000;
		esel = kvmppc_e500_tlb_index(vcpu_e500, ea, tlbsel,
				get_cur_pid(vcpu), -1);
		if (esel >= 0)
			kvmppc_e500_gtlbe_invalidate(vcpu_e500, tlbsel, esel);
	}

L
Liu Yu 已提交
645
	/* Invalidate all vcpu id mappings */
646
	kvmppc_e500_tlbil_all(vcpu_e500);
647 648 649 650

	return EMULATE_DONE;
}

S
Scott Wood 已提交
651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702
static void tlbilx_all(struct kvmppc_vcpu_e500 *vcpu_e500, int tlbsel,
		       int pid, int rt)
{
	struct kvm_book3e_206_tlb_entry *tlbe;
	int tid, esel;

	/* invalidate all entries */
	for (esel = 0; esel < vcpu_e500->gtlb_params[tlbsel].entries; esel++) {
		tlbe = get_entry(vcpu_e500, tlbsel, esel);
		tid = get_tlb_tid(tlbe);
		if (rt == 0 || tid == pid) {
			inval_gtlbe_on_host(vcpu_e500, tlbsel, esel);
			kvmppc_e500_gtlbe_invalidate(vcpu_e500, tlbsel, esel);
		}
	}
}

static void tlbilx_one(struct kvmppc_vcpu_e500 *vcpu_e500, int pid,
		       int ra, int rb)
{
	int tlbsel, esel;
	gva_t ea;

	ea = kvmppc_get_gpr(&vcpu_e500->vcpu, rb);
	if (ra)
		ea += kvmppc_get_gpr(&vcpu_e500->vcpu, ra);

	for (tlbsel = 0; tlbsel < 2; tlbsel++) {
		esel = kvmppc_e500_tlb_index(vcpu_e500, ea, tlbsel, pid, -1);
		if (esel >= 0) {
			inval_gtlbe_on_host(vcpu_e500, tlbsel, esel);
			kvmppc_e500_gtlbe_invalidate(vcpu_e500, tlbsel, esel);
			break;
		}
	}
}

int kvmppc_e500_emul_tlbilx(struct kvm_vcpu *vcpu, int rt, int ra, int rb)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
	int pid = get_cur_spid(vcpu);

	if (rt == 0 || rt == 1) {
		tlbilx_all(vcpu_e500, 0, pid, rt);
		tlbilx_all(vcpu_e500, 1, pid, rt);
	} else if (rt == 3) {
		tlbilx_one(vcpu_e500, pid, ra, rb);
	}

	return EMULATE_DONE;
}

703 704 705 706
int kvmppc_e500_emul_tlbre(struct kvm_vcpu *vcpu)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
	int tlbsel, esel;
S
Scott Wood 已提交
707
	struct kvm_book3e_206_tlb_entry *gtlbe;
708

709 710
	tlbsel = get_tlb_tlbsel(vcpu);
	esel = get_tlb_esel(vcpu, tlbsel);
711

S
Scott Wood 已提交
712
	gtlbe = get_entry(vcpu_e500, tlbsel, esel);
713 714 715 716 717
	vcpu->arch.shared->mas0 &= ~MAS0_NV(~0);
	vcpu->arch.shared->mas0 |= MAS0_NV(vcpu_e500->gtlb_nv[tlbsel]);
	vcpu->arch.shared->mas1 = gtlbe->mas1;
	vcpu->arch.shared->mas2 = gtlbe->mas2;
	vcpu->arch.shared->mas7_3 = gtlbe->mas7_3;
718 719 720 721 722 723 724

	return EMULATE_DONE;
}

int kvmppc_e500_emul_tlbsx(struct kvm_vcpu *vcpu, int rb)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
725 726
	int as = !!get_cur_sas(vcpu);
	unsigned int pid = get_cur_spid(vcpu);
727
	int esel, tlbsel;
S
Scott Wood 已提交
728
	struct kvm_book3e_206_tlb_entry *gtlbe = NULL;
729 730
	gva_t ea;

731
	ea = kvmppc_get_gpr(vcpu, rb);
732 733 734 735

	for (tlbsel = 0; tlbsel < 2; tlbsel++) {
		esel = kvmppc_e500_tlb_index(vcpu_e500, ea, tlbsel, pid, as);
		if (esel >= 0) {
S
Scott Wood 已提交
736
			gtlbe = get_entry(vcpu_e500, tlbsel, esel);
737 738 739 740 741
			break;
		}
	}

	if (gtlbe) {
742 743
		esel &= vcpu_e500->gtlb_params[tlbsel].ways - 1;

744
		vcpu->arch.shared->mas0 = MAS0_TLBSEL(tlbsel) | MAS0_ESEL(esel)
L
Liu Yu 已提交
745
			| MAS0_NV(vcpu_e500->gtlb_nv[tlbsel]);
746 747 748
		vcpu->arch.shared->mas1 = gtlbe->mas1;
		vcpu->arch.shared->mas2 = gtlbe->mas2;
		vcpu->arch.shared->mas7_3 = gtlbe->mas7_3;
749 750 751
	} else {
		int victim;

752
		/* since we only have two TLBs, only lower bit is used. */
753
		tlbsel = vcpu->arch.shared->mas4 >> 28 & 0x1;
754
		victim = (tlbsel == 0) ? gtlb0_get_next_victim(vcpu_e500) : 0;
755

756 757
		vcpu->arch.shared->mas0 = MAS0_TLBSEL(tlbsel)
			| MAS0_ESEL(victim)
L
Liu Yu 已提交
758
			| MAS0_NV(vcpu_e500->gtlb_nv[tlbsel]);
759 760 761 762 763 764 765 766 767
		vcpu->arch.shared->mas1 =
			  (vcpu->arch.shared->mas6 & MAS6_SPID0)
			| (vcpu->arch.shared->mas6 & (MAS6_SAS ? MAS1_TS : 0))
			| (vcpu->arch.shared->mas4 & MAS4_TSIZED(~0));
		vcpu->arch.shared->mas2 &= MAS2_EPN;
		vcpu->arch.shared->mas2 |= vcpu->arch.shared->mas4 &
					   MAS2_ATTRIB_MASK;
		vcpu->arch.shared->mas7_3 &= MAS3_U0 | MAS3_U1 |
					     MAS3_U2 | MAS3_U3;
768 769
	}

770
	kvmppc_set_exit_type(vcpu, EMULATED_TLBSX_EXITS);
771 772 773
	return EMULATE_DONE;
}

774
/* sesel is for tlb1 only */
775
static void write_stlbe(struct kvmppc_vcpu_e500 *vcpu_e500,
S
Scott Wood 已提交
776 777
			struct kvm_book3e_206_tlb_entry *gtlbe,
			struct kvm_book3e_206_tlb_entry *stlbe,
778 779 780 781 782
			int stlbsel, int sesel)
{
	int stid;

	preempt_disable();
783
	stid = kvmppc_e500_get_tlb_stid(&vcpu_e500->vcpu, gtlbe);
784 785 786 787 788 789

	stlbe->mas1 |= MAS1_TID(stid);
	write_host_tlbe(vcpu_e500, stlbsel, sesel, stlbe);
	preempt_enable();
}

790 791 792
int kvmppc_e500_emul_tlbwe(struct kvm_vcpu *vcpu)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
793 794
	struct kvm_book3e_206_tlb_entry *gtlbe, stlbe;
	int tlbsel, esel, stlbsel, sesel;
795

796 797
	tlbsel = get_tlb_tlbsel(vcpu);
	esel = get_tlb_esel(vcpu, tlbsel);
798

S
Scott Wood 已提交
799
	gtlbe = get_entry(vcpu_e500, tlbsel, esel);
800

L
Liu Yu 已提交
801
	if (get_tlb_v(gtlbe))
802
		inval_gtlbe_on_host(vcpu_e500, tlbsel, esel);
803

804 805 806
	gtlbe->mas1 = vcpu->arch.shared->mas1;
	gtlbe->mas2 = vcpu->arch.shared->mas2;
	gtlbe->mas7_3 = vcpu->arch.shared->mas7_3;
807

L
Liu Yu 已提交
808 809
	trace_kvm_booke206_gtlb_write(vcpu->arch.shared->mas0, gtlbe->mas1,
	                              gtlbe->mas2, gtlbe->mas7_3);
810 811 812

	/* Invalidate shadow mappings for the about-to-be-clobbered TLBE. */
	if (tlbe_is_host_safe(vcpu, gtlbe)) {
L
Liu Yu 已提交
813 814 815
		u64 eaddr;
		u64 raddr;

816 817 818 819
		switch (tlbsel) {
		case 0:
			/* TLB0 */
			gtlbe->mas1 &= ~MAS1_TSIZE(~0);
820
			gtlbe->mas1 |= MAS1_TSIZE(BOOK3E_PAGESZ_4K);
821 822

			stlbsel = 0;
823 824
			kvmppc_e500_tlb0_map(vcpu_e500, esel, &stlbe);
			sesel = 0; /* unused */
825 826 827 828 829 830 831 832 833 834 835 836 837 838

			break;

		case 1:
			/* TLB1 */
			eaddr = get_tlb_eaddr(gtlbe);
			raddr = get_tlb_raddr(gtlbe);

			/* Create a 4KB mapping on the host.
			 * If the guest wanted a large page,
			 * only the first 4KB is mapped here and the rest
			 * are mapped on the fly. */
			stlbsel = 1;
			sesel = kvmppc_e500_tlb1_map(vcpu_e500, eaddr,
839
				    raddr >> PAGE_SHIFT, gtlbe, &stlbe, esel);
840 841 842 843 844
			break;

		default:
			BUG();
		}
845 846

		write_stlbe(vcpu_e500, gtlbe, &stlbe, stlbsel, sesel);
847 848
	}

849
	kvmppc_set_exit_type(vcpu, EMULATED_TLBWE_EXITS);
850 851 852
	return EMULATE_DONE;
}

853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894
static int kvmppc_e500_tlb_search(struct kvm_vcpu *vcpu,
				  gva_t eaddr, unsigned int pid, int as)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
	int esel, tlbsel;

	for (tlbsel = 0; tlbsel < 2; tlbsel++) {
		esel = kvmppc_e500_tlb_index(vcpu_e500, eaddr, tlbsel, pid, as);
		if (esel >= 0)
			return index_of(tlbsel, esel);
	}

	return -1;
}

/* 'linear_address' is actually an encoding of AS|PID|EADDR . */
int kvmppc_core_vcpu_translate(struct kvm_vcpu *vcpu,
                               struct kvm_translation *tr)
{
	int index;
	gva_t eaddr;
	u8 pid;
	u8 as;

	eaddr = tr->linear_address;
	pid = (tr->linear_address >> 32) & 0xff;
	as = (tr->linear_address >> 40) & 0x1;

	index = kvmppc_e500_tlb_search(vcpu, eaddr, pid, as);
	if (index < 0) {
		tr->valid = 0;
		return 0;
	}

	tr->physical_address = kvmppc_mmu_xlate(vcpu, index, eaddr);
	/* XXX what does "writeable" and "usermode" even mean? */
	tr->valid = 1;

	return 0;
}


895 896
int kvmppc_mmu_itlb_index(struct kvm_vcpu *vcpu, gva_t eaddr)
{
897
	unsigned int as = !!(vcpu->arch.shared->msr & MSR_IS);
898 899 900 901 902 903

	return kvmppc_e500_tlb_search(vcpu, eaddr, get_cur_pid(vcpu), as);
}

int kvmppc_mmu_dtlb_index(struct kvm_vcpu *vcpu, gva_t eaddr)
{
904
	unsigned int as = !!(vcpu->arch.shared->msr & MSR_DS);
905 906 907 908 909 910

	return kvmppc_e500_tlb_search(vcpu, eaddr, get_cur_pid(vcpu), as);
}

void kvmppc_mmu_itlb_miss(struct kvm_vcpu *vcpu)
{
911
	unsigned int as = !!(vcpu->arch.shared->msr & MSR_IS);
912 913 914 915 916 917

	kvmppc_e500_deliver_tlb_miss(vcpu, vcpu->arch.pc, as);
}

void kvmppc_mmu_dtlb_miss(struct kvm_vcpu *vcpu)
{
918
	unsigned int as = !!(vcpu->arch.shared->msr & MSR_DS);
919 920 921 922 923 924 925 926

	kvmppc_e500_deliver_tlb_miss(vcpu, vcpu->arch.fault_dear, as);
}

gpa_t kvmppc_mmu_xlate(struct kvm_vcpu *vcpu, unsigned int index,
			gva_t eaddr)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
S
Scott Wood 已提交
927 928 929 930 931
	struct kvm_book3e_206_tlb_entry *gtlbe;
	u64 pgmask;

	gtlbe = get_entry(vcpu_e500, tlbsel_of(index), esel_of(index));
	pgmask = get_tlb_bytes(gtlbe) - 1;
932 933 934 935 936 937 938 939 940 941 942 943

	return get_tlb_raddr(gtlbe) | (eaddr & pgmask);
}

void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu)
{
}

void kvmppc_mmu_map(struct kvm_vcpu *vcpu, u64 eaddr, gpa_t gpaddr,
			unsigned int index)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
L
Liu Yu 已提交
944
	struct tlbe_priv *priv;
S
Scott Wood 已提交
945
	struct kvm_book3e_206_tlb_entry *gtlbe, stlbe;
946 947 948 949
	int tlbsel = tlbsel_of(index);
	int esel = esel_of(index);
	int stlbsel, sesel;

S
Scott Wood 已提交
950
	gtlbe = get_entry(vcpu_e500, tlbsel, esel);
L
Liu Yu 已提交
951

952 953 954
	switch (tlbsel) {
	case 0:
		stlbsel = 0;
955
		sesel = 0; /* unused */
956
		priv = &vcpu_e500->gtlb_priv[tlbsel][esel];
L
Liu Yu 已提交
957

958
		kvmppc_e500_setup_stlbe(vcpu, gtlbe, BOOK3E_PAGESZ_4K,
959
					&priv->ref, eaddr, &stlbe);
960 961 962 963 964 965
		break;

	case 1: {
		gfn_t gfn = gpaddr >> PAGE_SHIFT;

		stlbsel = 1;
L
Liu Yu 已提交
966
		sesel = kvmppc_e500_tlb1_map(vcpu_e500, eaddr, gfn,
967
					     gtlbe, &stlbe, esel);
968 969 970 971 972 973 974
		break;
	}

	default:
		BUG();
		break;
	}
L
Liu Yu 已提交
975

976
	write_stlbe(vcpu_e500, gtlbe, &stlbe, stlbsel, sesel);
977 978
}

S
Scott Wood 已提交
979 980 981 982
static void free_gtlb(struct kvmppc_vcpu_e500 *vcpu_e500)
{
	int i;

983 984 985
	clear_tlb1_bitmap(vcpu_e500);
	kfree(vcpu_e500->g2h_tlb1_map);

S
Scott Wood 已提交
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
	clear_tlb_refs(vcpu_e500);
	kfree(vcpu_e500->gtlb_priv[0]);
	kfree(vcpu_e500->gtlb_priv[1]);

	if (vcpu_e500->shared_tlb_pages) {
		vfree((void *)(round_down((uintptr_t)vcpu_e500->gtlb_arch,
					  PAGE_SIZE)));

		for (i = 0; i < vcpu_e500->num_shared_tlb_pages; i++) {
			set_page_dirty_lock(vcpu_e500->shared_tlb_pages[i]);
			put_page(vcpu_e500->shared_tlb_pages[i]);
		}

		vcpu_e500->num_shared_tlb_pages = 0;
		vcpu_e500->shared_tlb_pages = NULL;
	} else {
		kfree(vcpu_e500->gtlb_arch);
	}

	vcpu_e500->gtlb_arch = NULL;
}

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037
void kvmppc_get_sregs_e500_tlb(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
{
	sregs->u.e.mas0 = vcpu->arch.shared->mas0;
	sregs->u.e.mas1 = vcpu->arch.shared->mas1;
	sregs->u.e.mas2 = vcpu->arch.shared->mas2;
	sregs->u.e.mas7_3 = vcpu->arch.shared->mas7_3;
	sregs->u.e.mas4 = vcpu->arch.shared->mas4;
	sregs->u.e.mas6 = vcpu->arch.shared->mas6;

	sregs->u.e.mmucfg = vcpu->arch.mmucfg;
	sregs->u.e.tlbcfg[0] = vcpu->arch.tlbcfg[0];
	sregs->u.e.tlbcfg[1] = vcpu->arch.tlbcfg[1];
	sregs->u.e.tlbcfg[2] = 0;
	sregs->u.e.tlbcfg[3] = 0;
}

int kvmppc_set_sregs_e500_tlb(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs)
{
	if (sregs->u.e.features & KVM_SREGS_E_ARCH206_MMU) {
		vcpu->arch.shared->mas0 = sregs->u.e.mas0;
		vcpu->arch.shared->mas1 = sregs->u.e.mas1;
		vcpu->arch.shared->mas2 = sregs->u.e.mas2;
		vcpu->arch.shared->mas7_3 = sregs->u.e.mas7_3;
		vcpu->arch.shared->mas4 = sregs->u.e.mas4;
		vcpu->arch.shared->mas6 = sregs->u.e.mas6;
	}

	return 0;
}

S
Scott Wood 已提交
1038 1039 1040 1041 1042 1043 1044 1045
int kvm_vcpu_ioctl_config_tlb(struct kvm_vcpu *vcpu,
			      struct kvm_config_tlb *cfg)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);
	struct kvm_book3e_206_tlb_params params;
	char *virt;
	struct page **pages;
	struct tlbe_priv *privs[2] = {};
1046
	u64 *g2h_bitmap = NULL;
S
Scott Wood 已提交
1047 1048 1049 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 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
	size_t array_len;
	u32 sets;
	int num_pages, ret, i;

	if (cfg->mmu_type != KVM_MMU_FSL_BOOKE_NOHV)
		return -EINVAL;

	if (copy_from_user(&params, (void __user *)(uintptr_t)cfg->params,
			   sizeof(params)))
		return -EFAULT;

	if (params.tlb_sizes[1] > 64)
		return -EINVAL;
	if (params.tlb_ways[1] != params.tlb_sizes[1])
		return -EINVAL;
	if (params.tlb_sizes[2] != 0 || params.tlb_sizes[3] != 0)
		return -EINVAL;
	if (params.tlb_ways[2] != 0 || params.tlb_ways[3] != 0)
		return -EINVAL;

	if (!is_power_of_2(params.tlb_ways[0]))
		return -EINVAL;

	sets = params.tlb_sizes[0] >> ilog2(params.tlb_ways[0]);
	if (!is_power_of_2(sets))
		return -EINVAL;

	array_len = params.tlb_sizes[0] + params.tlb_sizes[1];
	array_len *= sizeof(struct kvm_book3e_206_tlb_entry);

	if (cfg->array_len < array_len)
		return -EINVAL;

	num_pages = DIV_ROUND_UP(cfg->array + array_len - 1, PAGE_SIZE) -
		    cfg->array / PAGE_SIZE;
	pages = kmalloc(sizeof(struct page *) * num_pages, GFP_KERNEL);
	if (!pages)
		return -ENOMEM;

	ret = get_user_pages_fast(cfg->array, num_pages, 1, pages);
	if (ret < 0)
		goto err_pages;

	if (ret != num_pages) {
		num_pages = ret;
		ret = -EFAULT;
		goto err_put_page;
	}

	virt = vmap(pages, num_pages, VM_MAP, PAGE_KERNEL);
	if (!virt)
		goto err_put_page;

	privs[0] = kzalloc(sizeof(struct tlbe_priv) * params.tlb_sizes[0],
			   GFP_KERNEL);
	privs[1] = kzalloc(sizeof(struct tlbe_priv) * params.tlb_sizes[1],
			   GFP_KERNEL);

	if (!privs[0] || !privs[1])
		goto err_put_page;

1108 1109 1110 1111 1112
	g2h_bitmap = kzalloc(sizeof(u64) * params.tlb_sizes[1],
	                     GFP_KERNEL);
	if (!g2h_bitmap)
		goto err_put_page;

S
Scott Wood 已提交
1113 1114 1115 1116
	free_gtlb(vcpu_e500);

	vcpu_e500->gtlb_priv[0] = privs[0];
	vcpu_e500->gtlb_priv[1] = privs[1];
1117
	vcpu_e500->g2h_tlb1_map = g2h_bitmap;
S
Scott Wood 已提交
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127

	vcpu_e500->gtlb_arch = (struct kvm_book3e_206_tlb_entry *)
		(virt + (cfg->array & (PAGE_SIZE - 1)));

	vcpu_e500->gtlb_params[0].entries = params.tlb_sizes[0];
	vcpu_e500->gtlb_params[1].entries = params.tlb_sizes[1];

	vcpu_e500->gtlb_offset[0] = 0;
	vcpu_e500->gtlb_offset[1] = params.tlb_sizes[0];

1128 1129 1130
	vcpu->arch.mmucfg = mfspr(SPRN_MMUCFG) & ~MMUCFG_LPIDSIZE;

	vcpu->arch.tlbcfg[0] &= ~(TLBnCFG_N_ENTRY | TLBnCFG_ASSOC);
S
Scott Wood 已提交
1131
	if (params.tlb_sizes[0] <= 2048)
1132 1133
		vcpu->arch.tlbcfg[0] |= params.tlb_sizes[0];
	vcpu->arch.tlbcfg[0] |= params.tlb_ways[0] << TLBnCFG_ASSOC_SHIFT;
S
Scott Wood 已提交
1134

1135 1136 1137
	vcpu->arch.tlbcfg[1] &= ~(TLBnCFG_N_ENTRY | TLBnCFG_ASSOC);
	vcpu->arch.tlbcfg[1] |= params.tlb_sizes[1];
	vcpu->arch.tlbcfg[1] |= params.tlb_ways[1] << TLBnCFG_ASSOC_SHIFT;
S
Scott Wood 已提交
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168

	vcpu_e500->shared_tlb_pages = pages;
	vcpu_e500->num_shared_tlb_pages = num_pages;

	vcpu_e500->gtlb_params[0].ways = params.tlb_ways[0];
	vcpu_e500->gtlb_params[0].sets = sets;

	vcpu_e500->gtlb_params[1].ways = params.tlb_sizes[1];
	vcpu_e500->gtlb_params[1].sets = 1;

	return 0;

err_put_page:
	kfree(privs[0]);
	kfree(privs[1]);

	for (i = 0; i < num_pages; i++)
		put_page(pages[i]);

err_pages:
	kfree(pages);
	return ret;
}

int kvm_vcpu_ioctl_dirty_tlb(struct kvm_vcpu *vcpu,
			     struct kvm_dirty_tlb *dirty)
{
	struct kvmppc_vcpu_e500 *vcpu_e500 = to_e500(vcpu);

	clear_tlb_refs(vcpu_e500);
	return 0;
1169 1170 1171 1172
}

int kvmppc_e500_tlb_init(struct kvmppc_vcpu_e500 *vcpu_e500)
{
1173
	struct kvm_vcpu *vcpu = &vcpu_e500->vcpu;
S
Scott Wood 已提交
1174 1175 1176
	int entry_size = sizeof(struct kvm_book3e_206_tlb_entry);
	int entries = KVM_E500_TLB0_SIZE + KVM_E500_TLB1_SIZE;

1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
	host_tlb_params[0].entries = mfspr(SPRN_TLB0CFG) & TLBnCFG_N_ENTRY;
	host_tlb_params[1].entries = mfspr(SPRN_TLB1CFG) & TLBnCFG_N_ENTRY;

	/*
	 * This should never happen on real e500 hardware, but is
	 * architecturally possible -- e.g. in some weird nested
	 * virtualization case.
	 */
	if (host_tlb_params[0].entries == 0 ||
	    host_tlb_params[1].entries == 0) {
		pr_err("%s: need to know host tlb size\n", __func__);
		return -ENODEV;
	}

	host_tlb_params[0].ways = (mfspr(SPRN_TLB0CFG) & TLBnCFG_ASSOC) >>
				  TLBnCFG_ASSOC_SHIFT;
	host_tlb_params[1].ways = host_tlb_params[1].entries;

	if (!is_power_of_2(host_tlb_params[0].entries) ||
	    !is_power_of_2(host_tlb_params[0].ways) ||
	    host_tlb_params[0].entries < host_tlb_params[0].ways ||
	    host_tlb_params[0].ways == 0) {
		pr_err("%s: bad tlb0 host config: %u entries %u ways\n",
		       __func__, host_tlb_params[0].entries,
		       host_tlb_params[0].ways);
		return -ENODEV;
	}

	host_tlb_params[0].sets =
		host_tlb_params[0].entries / host_tlb_params[0].ways;
	host_tlb_params[1].sets = 1;
1208

S
Scott Wood 已提交
1209 1210
	vcpu_e500->gtlb_params[0].entries = KVM_E500_TLB0_SIZE;
	vcpu_e500->gtlb_params[1].entries = KVM_E500_TLB1_SIZE;
1211

S
Scott Wood 已提交
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
	vcpu_e500->gtlb_params[0].ways = KVM_E500_TLB0_WAY_NUM;
	vcpu_e500->gtlb_params[0].sets =
		KVM_E500_TLB0_SIZE / KVM_E500_TLB0_WAY_NUM;

	vcpu_e500->gtlb_params[1].ways = KVM_E500_TLB1_SIZE;
	vcpu_e500->gtlb_params[1].sets = 1;

	vcpu_e500->gtlb_arch = kmalloc(entries * entry_size, GFP_KERNEL);
	if (!vcpu_e500->gtlb_arch)
		return -ENOMEM;

	vcpu_e500->gtlb_offset[0] = 0;
	vcpu_e500->gtlb_offset[1] = KVM_E500_TLB0_SIZE;
1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237

	vcpu_e500->tlb_refs[0] =
		kzalloc(sizeof(struct tlbe_ref) * host_tlb_params[0].entries,
			GFP_KERNEL);
	if (!vcpu_e500->tlb_refs[0])
		goto err;

	vcpu_e500->tlb_refs[1] =
		kzalloc(sizeof(struct tlbe_ref) * host_tlb_params[1].entries,
			GFP_KERNEL);
	if (!vcpu_e500->tlb_refs[1])
		goto err;

S
Scott Wood 已提交
1238 1239 1240
	vcpu_e500->gtlb_priv[0] = kzalloc(sizeof(struct tlbe_ref) *
					  vcpu_e500->gtlb_params[0].entries,
					  GFP_KERNEL);
1241 1242 1243
	if (!vcpu_e500->gtlb_priv[0])
		goto err;

S
Scott Wood 已提交
1244 1245 1246
	vcpu_e500->gtlb_priv[1] = kzalloc(sizeof(struct tlbe_ref) *
					  vcpu_e500->gtlb_params[1].entries,
					  GFP_KERNEL);
1247 1248
	if (!vcpu_e500->gtlb_priv[1])
		goto err;
1249

1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261
	vcpu_e500->g2h_tlb1_map = kzalloc(sizeof(unsigned int) *
					  vcpu_e500->gtlb_params[1].entries,
					  GFP_KERNEL);
	if (!vcpu_e500->g2h_tlb1_map)
		goto err;

	vcpu_e500->h2g_tlb1_rmap = kzalloc(sizeof(unsigned int) *
					   host_tlb_params[1].entries,
					   GFP_KERNEL);
	if (!vcpu_e500->h2g_tlb1_rmap)
		goto err;

L
Liu Yu 已提交
1262
	/* Init TLB configuration register */
1263
	vcpu->arch.tlbcfg[0] = mfspr(SPRN_TLB0CFG) &
1264
			     ~(TLBnCFG_N_ENTRY | TLBnCFG_ASSOC);
1265 1266
	vcpu->arch.tlbcfg[0] |= vcpu_e500->gtlb_params[0].entries;
	vcpu->arch.tlbcfg[0] |=
1267 1268
		vcpu_e500->gtlb_params[0].ways << TLBnCFG_ASSOC_SHIFT;

1269
	vcpu->arch.tlbcfg[1] = mfspr(SPRN_TLB1CFG) &
1270
			     ~(TLBnCFG_N_ENTRY | TLBnCFG_ASSOC);
1271 1272
	vcpu->arch.tlbcfg[0] |= vcpu_e500->gtlb_params[1].entries;
	vcpu->arch.tlbcfg[0] |=
1273
		vcpu_e500->gtlb_params[1].ways << TLBnCFG_ASSOC_SHIFT;
L
Liu Yu 已提交
1274

1275 1276
	return 0;

1277
err:
S
Scott Wood 已提交
1278
	free_gtlb(vcpu_e500);
1279 1280
	kfree(vcpu_e500->tlb_refs[0]);
	kfree(vcpu_e500->tlb_refs[1]);
1281 1282 1283 1284 1285
	return -1;
}

void kvmppc_e500_tlb_uninit(struct kvmppc_vcpu_e500 *vcpu_e500)
{
S
Scott Wood 已提交
1286
	free_gtlb(vcpu_e500);
1287
	kfree(vcpu_e500->h2g_tlb1_rmap);
1288 1289
	kfree(vcpu_e500->tlb_refs[0]);
	kfree(vcpu_e500->tlb_refs[1]);
1290
}