aboutsummaryrefslogtreecommitdiff
path: root/arch/x86/xen/mmu.c
blob: 0144395448ae8775c2609ecfb0c0072197e9b35b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
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
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
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
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
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
/*
 * Xen mmu operations
 *
 * This file contains the various mmu fetch and update operations.
 * The most important job they must perform is the mapping between the
 * domain's pfn and the overall machine mfns.
 *
 * Xen allows guests to directly update the pagetable, in a controlled
 * fashion.  In other words, the guest modifies the same pagetable
 * that the CPU actually uses, which eliminates the overhead of having
 * a separate shadow pagetable.
 *
 * In order to allow this, it falls on the guest domain to map its
 * notion of a "physical" pfn - which is just a domain-local linear
 * address - into a real "machine address" which the CPU's MMU can
 * use.
 *
 * A pgd_t/pmd_t/pte_t will typically contain an mfn, and so can be
 * inserted directly into the pagetable.  When creating a new
 * pte/pmd/pgd, it converts the passed pfn into an mfn.  Conversely,
 * when reading the content back with __(pgd|pmd|pte)_val, it converts
 * the mfn back into a pfn.
 *
 * The other constraint is that all pages which make up a pagetable
 * must be mapped read-only in the guest.  This prevents uncontrolled
 * guest updates to the pagetable.  Xen strictly enforces this, and
 * will disallow any pagetable update which will end up mapping a
 * pagetable page RW, and will disallow using any writable page as a
 * pagetable.
 *
 * Naively, when loading %cr3 with the base of a new pagetable, Xen
 * would need to validate the whole pagetable before going on.
 * Naturally, this is quite slow.  The solution is to "pin" a
 * pagetable, which enforces all the constraints on the pagetable even
 * when it is not actively in use.  This menas that Xen can be assured
 * that it is still valid when you do load it into %cr3, and doesn't
 * need to revalidate it.
 *
 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
 */
#include <linux/sched.h>
#include <linux/highmem.h>
#include <linux/bug.h>

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/mmu_context.h>
#include <asm/paravirt.h>

#include <asm/xen/hypercall.h>
#include <asm/xen/hypervisor.h>

#include <xen/page.h>
#include <xen/interface/xen.h>

#include "multicalls.h"
#include "mmu.h"

xmaddr_t arbitrary_virt_to_machine(unsigned long address)
{
	unsigned int level;
	pte_t *pte = lookup_address(address, &level);
	unsigned offset = address & PAGE_MASK;

	BUG_ON(pte == NULL);

	return XMADDR((pte_mfn(*pte) << PAGE_SHIFT) + offset);
}

void make_lowmem_page_readonly(void *vaddr)
{
	pte_t *pte, ptev;
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;

	pte = lookup_address(address, &level);
	BUG_ON(pte == NULL);

	ptev = pte_wrprotect(*pte);

	if (HYPERVISOR_update_va_mapping(address, ptev, 0))
		BUG();
}

void make_lowmem_page_readwrite(void *vaddr)
{
	pte_t *pte, ptev;
	unsigned long address = (unsigned long)vaddr;
	unsigned int level;

	pte = lookup_address(address, &level);
	BUG_ON(pte == NULL);

	ptev = pte_mkwrite(*pte);

	if (HYPERVISOR_update_va_mapping(address, ptev, 0))
		BUG();
}


void xen_set_pmd(pmd_t *ptr, pmd_t val)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*u));
	u = mcs.args;
	u->ptr = virt_to_machine(ptr).maddr;
	u->val = pmd_val_ma(val);
	MULTI_mmu_update(mcs.mc, u, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

/*
 * Associate a virtual page frame with a given physical page frame
 * and protection flags for that frame.
 */
void set_pte_mfn(unsigned long vaddr, unsigned long mfn, pgprot_t flags)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;
	pte_t *pte;

	pgd = swapper_pg_dir + pgd_index(vaddr);
	if (pgd_none(*pgd)) {
		BUG();
		return;
	}
	pud = pud_offset(pgd, vaddr);
	if (pud_none(*pud)) {
		BUG();
		return;
	}
	pmd = pmd_offset(pud, vaddr);
	if (pmd_none(*pmd)) {
		BUG();
		return;
	}
	pte = pte_offset_kernel(pmd, vaddr);
	/* <mfn,flags> stored as-is, to permit clearing entries */
	xen_set_pte(pte, mfn_pte(mfn, flags));

	/*
	 * It's enough to flush this one mapping.
	 * (PGE mappings get flushed as well)
	 */
	__flush_tlb_one(vaddr);
}

void xen_set_pte_at(struct mm_struct *mm, unsigned long addr,
		    pte_t *ptep, pte_t pteval)
{
	if (mm == current->mm || mm == &init_mm) {
		if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
			struct multicall_space mcs;
			mcs = xen_mc_entry(0);

			MULTI_update_va_mapping(mcs.mc, addr, pteval, 0);
			xen_mc_issue(PARAVIRT_LAZY_MMU);
			return;
		} else
			if (HYPERVISOR_update_va_mapping(addr, pteval, 0) == 0)
				return;
	}
	xen_set_pte(ptep, pteval);
}

#ifdef CONFIG_X86_PAE
void xen_set_pud(pud_t *ptr, pud_t val)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	preempt_disable();

	mcs = xen_mc_entry(sizeof(*u));
	u = mcs.args;
	u->ptr = virt_to_machine(ptr).maddr;
	u->val = pud_val_ma(val);
	MULTI_mmu_update(mcs.mc, u, 1, NULL, DOMID_SELF);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

void xen_set_pte(pte_t *ptep, pte_t pte)
{
	ptep->pte_high = pte.pte_high;
	smp_wmb();
	ptep->pte_low = pte.pte_low;
}

void xen_set_pte_atomic(pte_t *ptep, pte_t pte)
{
	set_64bit((u64 *)ptep, pte_val_ma(pte));
}

void xen_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
{
	ptep->pte_low = 0;
	smp_wmb();		/* make sure low gets written first */
	ptep->pte_high = 0;
}

void xen_pmd_clear(pmd_t *pmdp)
{
	xen_set_pmd(pmdp, __pmd(0));
}

unsigned long long xen_pte_val(pte_t pte)
{
	unsigned long long ret = 0;

	if (pte.pte_low) {
		ret = ((unsigned long long)pte.pte_high << 32) | pte.pte_low;
		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
	}

	return ret;
}

unsigned long long xen_pmd_val(pmd_t pmd)
{
	unsigned long long ret = pmd.pmd;
	if (ret)
		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
	return ret;
}

unsigned long long xen_pgd_val(pgd_t pgd)
{
	unsigned long long ret = pgd.pgd;
	if (ret)
		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
	return ret;
}

pte_t xen_make_pte(unsigned long long pte)
{
	if (pte & _PAGE_PRESENT) {
		pte = phys_to_machine(XPADDR(pte)).maddr;
		pte &= ~(_PAGE_PCD | _PAGE_PWT);
	}

	return (pte_t){ .pte = pte };
}

pmd_t xen_make_pmd(unsigned long long pmd)
{
	if (pmd & 1)
		pmd = phys_to_machine(XPADDR(pmd)).maddr;

	return (pmd_t){ pmd };
}

pgd_t xen_make_pgd(unsigned long long pgd)
{
	if (pgd & _PAGE_PRESENT)
		pgd = phys_to_machine(XPADDR(pgd)).maddr;

	return (pgd_t){ pgd };
}
#else  /* !PAE */
void xen_set_pte(pte_t *ptep, pte_t pte)
{
	*ptep = pte;
}

unsigned long xen_pte_val(pte_t pte)
{
	unsigned long ret = pte.pte_low;

	if (ret & _PAGE_PRESENT)
		ret = machine_to_phys(XMADDR(ret)).paddr;

	return ret;
}

unsigned long xen_pgd_val(pgd_t pgd)
{
	unsigned long ret = pgd.pgd;
	if (ret)
		ret = machine_to_phys(XMADDR(ret)).paddr | 1;
	return ret;
}

pte_t xen_make_pte(unsigned long pte)
{
	if (pte & _PAGE_PRESENT) {
		pte = phys_to_machine(XPADDR(pte)).maddr;
		pte &= ~(_PAGE_PCD | _PAGE_PWT);
	}

	return (pte_t){ pte };
}

pgd_t xen_make_pgd(unsigned long pgd)
{
	if (pgd & _PAGE_PRESENT)
		pgd = phys_to_machine(XPADDR(pgd)).maddr;

	return (pgd_t){ pgd };
}
#endif	/* CONFIG_X86_PAE */

enum pt_level {
	PT_PGD,
	PT_PUD,
	PT_PMD,
	PT_PTE
};

/*
  (Yet another) pagetable walker.  This one is intended for pinning a
  pagetable.  This means that it walks a pagetable and calls the
  callback function on each page it finds making up the page table,
  at every level.  It walks the entire pagetable, but it only bothers
  pinning pte pages which are below pte_limit.  In the normal case
  this will be TASK_SIZE, but at boot we need to pin up to
  FIXADDR_TOP.  But the important bit is that we don't pin beyond
  there, because then we start getting into Xen's ptes.
*/
static int pgd_walk(pgd_t *pgd_base, int (*func)(struct page *, enum pt_level),
		    unsigned long limit)
{
	pgd_t *pgd = pgd_base;
	int flush = 0;
	unsigned long addr = 0;
	unsigned long pgd_next;

	BUG_ON(limit > FIXADDR_TOP);

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	for (; addr != FIXADDR_TOP; pgd++, addr = pgd_next) {
		pud_t *pud;
		unsigned long pud_limit, pud_next;

		pgd_next = pud_limit = pgd_addr_end(addr, FIXADDR_TOP);

		if (!pgd_val(*pgd))
			continue;

		pud = pud_offset(pgd, 0);

		if (PTRS_PER_PUD > 1) /* not folded */
			flush |= (*func)(virt_to_page(pud), PT_PUD);

		for (; addr != pud_limit; pud++, addr = pud_next) {
			pmd_t *pmd;
			unsigned long pmd_limit;

			pud_next = pud_addr_end(addr, pud_limit);

			if (pud_next < limit)
				pmd_limit = pud_next;
			else
				pmd_limit = limit;

			if (pud_none(*pud))
				continue;

			pmd = pmd_offset(pud, 0);

			if (PTRS_PER_PMD > 1) /* not folded */
				flush |= (*func)(virt_to_page(pmd), PT_PMD);

			for (; addr != pmd_limit; pmd++) {
				addr += (PAGE_SIZE * PTRS_PER_PTE);
				if ((pmd_limit-1) < (addr-1)) {
					addr = pmd_limit;
					break;
				}

				if (pmd_none(*pmd))
					continue;

				flush |= (*func)(pmd_page(*pmd), PT_PTE);
			}
		}
	}

	flush |= (*func)(virt_to_page(pgd_base), PT_PGD);

	return flush;
}

static spinlock_t *lock_pte(struct page *page)
{
	spinlock_t *ptl = NULL;

#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
	ptl = __pte_lockptr(page);
	spin_lock(ptl);
#endif

	return ptl;
}

static void do_unlock(void *v)
{
	spinlock_t *ptl = v;
	spin_unlock(ptl);
}

static void xen_do_pin(unsigned level, unsigned long pfn)
{
	struct mmuext_op *op;
	struct multicall_space mcs;

	mcs = __xen_mc_entry(sizeof(*op));
	op = mcs.args;
	op->cmd = level;
	op->arg1.mfn = pfn_to_mfn(pfn);
	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
}

static int pin_page(struct page *page, enum pt_level level)
{
	unsigned pgfl = test_and_set_bit(PG_pinned, &page->flags);
	int flush;

	if (pgfl)
		flush = 0;		/* already pinned */
	else if (PageHighMem(page))
		/* kmaps need flushing if we found an unpinned
		   highpage */
		flush = 1;
	else {
		void *pt = lowmem_page_address(page);
		unsigned long pfn = page_to_pfn(page);
		struct multicall_space mcs = __xen_mc_entry(0);
		spinlock_t *ptl;

		flush = 0;

		ptl = NULL;
		if (level == PT_PTE)
			ptl = lock_pte(page);

		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
					pfn_pte(pfn, PAGE_KERNEL_RO),
					level == PT_PGD ? UVMF_TLB_FLUSH : 0);

		if (level == PT_PTE)
			xen_do_pin(MMUEXT_PIN_L1_TABLE, pfn);

		if (ptl) {
			/* Queue a deferred unlock for when this batch
			   is completed. */
			xen_mc_callback(do_unlock, ptl);
		}
	}

	return flush;
}

/* This is called just after a mm has been created, but it has not
   been used yet.  We need to make sure that its pagetable is all
   read-only, and can be pinned. */
void xen_pgd_pin(pgd_t *pgd)
{
	unsigned level;

	xen_mc_batch();

	if (pgd_walk(pgd, pin_page, TASK_SIZE)) {
		/* re-enable interrupts for kmap_flush_unused */
		xen_mc_issue(0);
		kmap_flush_unused();
		xen_mc_batch();
	}

#ifdef CONFIG_X86_PAE
	level = MMUEXT_PIN_L3_TABLE;
#else
	level = MMUEXT_PIN_L2_TABLE;
#endif

	xen_do_pin(level, PFN_DOWN(__pa(pgd)));

	xen_mc_issue(0);
}

/* The init_mm pagetable is really pinned as soon as its created, but
   that's before we have page structures to store the bits.  So do all
   the book-keeping now. */
static __init int mark_pinned(struct page *page, enum pt_level level)
{
	SetPagePinned(page);
	return 0;
}

void __init xen_mark_init_mm_pinned(void)
{
	pgd_walk(init_mm.pgd, mark_pinned, FIXADDR_TOP);
}

static int unpin_page(struct page *page, enum pt_level level)
{
	unsigned pgfl = test_and_clear_bit(PG_pinned, &page->flags);

	if (pgfl && !PageHighMem(page)) {
		void *pt = lowmem_page_address(page);
		unsigned long pfn = page_to_pfn(page);
		spinlock_t *ptl = NULL;
		struct multicall_space mcs;

		if (level == PT_PTE) {
			ptl = lock_pte(page);

			xen_do_pin(MMUEXT_UNPIN_TABLE, pfn);
		}

		mcs = __xen_mc_entry(0);

		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
					pfn_pte(pfn, PAGE_KERNEL),
					level == PT_PGD ? UVMF_TLB_FLUSH : 0);

		if (ptl) {
			/* unlock when batch completed */
			xen_mc_callback(do_unlock, ptl);
		}
	}

	return 0;		/* never need to flush on unpin */
}

/* Release a pagetables pages back as normal RW */
static void xen_pgd_unpin(pgd_t *pgd)
{
	xen_mc_batch();

	xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));

	pgd_walk(pgd, unpin_page, TASK_SIZE);

	xen_mc_issue(0);
}

void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)
{
	spin_lock(&next->page_table_lock);
	xen_pgd_pin(next->pgd);
	spin_unlock(&next->page_table_lock);
}

void xen_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
{
	spin_lock(&mm->page_table_lock);
	xen_pgd_pin(mm->pgd);
	spin_unlock(&mm->page_table_lock);
}


#ifdef CONFIG_SMP
/* Another cpu may still have their %cr3 pointing at the pagetable, so
   we need to repoint it somewhere else before we can unpin it. */
static void drop_other_mm_ref(void *info)
{
	struct mm_struct *mm = info;

	if (__get_cpu_var(cpu_tlbstate).active_mm == mm)
		leave_mm(smp_processor_id());

	/* If this cpu still has a stale cr3 reference, then make sure
	   it has been flushed. */
	if (x86_read_percpu(xen_current_cr3) == __pa(mm->pgd)) {
		load_cr3(swapper_pg_dir);
		arch_flush_lazy_cpu_mode();
	}
}

static void drop_mm_ref(struct mm_struct *mm)
{
	cpumask_t mask;
	unsigned cpu;

	if (current->active_mm == mm) {
		if (current->mm == mm)
			load_cr3(swapper_pg_dir);
		else
			leave_mm(smp_processor_id());
		arch_flush_lazy_cpu_mode();
	}

	/* Get the "official" set of cpus referring to our pagetable. */
	mask = mm->cpu_vm_mask;

	/* It's possible that a vcpu may have a stale reference to our
	   cr3, because its in lazy mode, and it hasn't yet flushed
	   its set of pending hypercalls yet.  In this case, we can
	   look at its actual current cr3 value, and force it to flush
	   if needed. */
	for_each_online_cpu(cpu) {
		if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd))
			cpu_set(cpu, mask);
	}

	if (!cpus_empty(mask))
		xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
}
#else
static void drop_mm_ref(struct mm_struct *mm)
{
	if (current->active_mm == mm)
		load_cr3(swapper_pg_dir);
}
#endif

/*
 * While a process runs, Xen pins its pagetables, which means that the
 * hypervisor forces it to be read-only, and it controls all updates
 * to it.  This means that all pagetable updates have to go via the
 * hypervisor, which is moderately expensive.
 *
 * Since we're pulling the pagetable down, we switch to use init_mm,
 * unpin old process pagetable and mark it all read-write, which
 * allows further operations on it to be simple memory accesses.
 *
 * The only subtle point is that another CPU may be still using the
 * pagetable because of lazy tlb flushing.  This means we need need to
 * switch all CPUs off this pagetable before we can unpin it.
 */
void xen_exit_mmap(struct mm_struct *mm)
{
	get_cpu();		/* make sure we don't move around */
	drop_mm_ref(mm);
	put_cpu();

	spin_lock(&mm->page_table_lock);

	/* pgd may not be pinned in the error exit path of execve */
	if (PagePinned(virt_to_page(mm->pgd)))
		xen_pgd_unpin(mm->pgd);

	spin_unlock(&mm->page_table_lock);
}