aboutsummaryrefslogtreecommitdiffstats
path: root/arch/x86/xen/mmu.c
blob: ae173f6edd8ba044f5cc5372193ea7ce0af2e299 (plain) (blame)
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
648
649
650
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
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
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
844
845
846
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
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
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
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
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
1038
1039
1040
1041
1042
1043
1044
1045
1046
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
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
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
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
/*
 * 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/debugfs.h>
#include <linux/bug.h>

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/fixmap.h>
#include <asm/mmu_context.h>
#include <asm/paravirt.h>
#include <asm/linkage.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"
#include "debugfs.h"

#define MMU_UPDATE_HISTO	30

#ifdef CONFIG_XEN_DEBUG_FS

static struct {
	u32 pgd_update;
	u32 pgd_update_pinned;
	u32 pgd_update_batched;

	u32 pud_update;
	u32 pud_update_pinned;
	u32 pud_update_batched;

	u32 pmd_update;
	u32 pmd_update_pinned;
	u32 pmd_update_batched;

	u32 pte_update;
	u32 pte_update_pinned;
	u32 pte_update_batched;

	u32 mmu_update;
	u32 mmu_update_extended;
	u32 mmu_update_histo[MMU_UPDATE_HISTO];

	u32 prot_commit;
	u32 prot_commit_batched;

	u32 set_pte_at;
	u32 set_pte_at_batched;
	u32 set_pte_at_pinned;
	u32 set_pte_at_current;
	u32 set_pte_at_kernel;
} mmu_stats;

static u8 zero_stats;

static inline void check_zero(void)
{
	if (unlikely(zero_stats)) {
		memset(&mmu_stats, 0, sizeof(mmu_stats));
		zero_stats = 0;
	}
}

#define ADD_STATS(elem, val)			\
	do { check_zero(); mmu_stats.elem += (val); } while(0)

#else  /* !CONFIG_XEN_DEBUG_FS */

#define ADD_STATS(elem, val)	do { (void)(val); } while(0)

#endif /* CONFIG_XEN_DEBUG_FS */

/*
 * Just beyond the highest usermode address.  STACK_TOP_MAX has a
 * redzone above it, so round it up to a PGD boundary.
 */
#define USER_LIMIT	((STACK_TOP_MAX + PGDIR_SIZE - 1) & PGDIR_MASK)


#define P2M_ENTRIES_PER_PAGE	(PAGE_SIZE / sizeof(unsigned long))
#define TOP_ENTRIES		(MAX_DOMAIN_PAGES / P2M_ENTRIES_PER_PAGE)

/* Placeholder for holes in the address space */
static unsigned long p2m_missing[P2M_ENTRIES_PER_PAGE] __page_aligned_data =
		{ [ 0 ... P2M_ENTRIES_PER_PAGE-1 ] = ~0UL };

 /* Array of pointers to pages containing p2m entries */
static unsigned long *p2m_top[TOP_ENTRIES] __page_aligned_data =
		{ [ 0 ... TOP_ENTRIES - 1] = &p2m_missing[0] };

/* Arrays of p2m arrays expressed in mfns used for save/restore */
static unsigned long p2m_top_mfn[TOP_ENTRIES] __page_aligned_bss;

static unsigned long p2m_top_mfn_list[TOP_ENTRIES / P2M_ENTRIES_PER_PAGE]
	__page_aligned_bss;

static inline unsigned p2m_top_index(unsigned long pfn)
{
	BUG_ON(pfn >= MAX_DOMAIN_PAGES);
	return pfn / P2M_ENTRIES_PER_PAGE;
}

static inline unsigned p2m_index(unsigned long pfn)
{
	return pfn % P2M_ENTRIES_PER_PAGE;
}

/* Build the parallel p2m_top_mfn structures */
void xen_setup_mfn_list_list(void)
{
	unsigned pfn, idx;

	for(pfn = 0; pfn < MAX_DOMAIN_PAGES; pfn += P2M_ENTRIES_PER_PAGE) {
		unsigned topidx = p2m_top_index(pfn);

		p2m_top_mfn[topidx] = virt_to_mfn(p2m_top[topidx]);
	}

	for(idx = 0; idx < ARRAY_SIZE(p2m_top_mfn_list); idx++) {
		unsigned topidx = idx * P2M_ENTRIES_PER_PAGE;
		p2m_top_mfn_list[idx] = virt_to_mfn(&p2m_top_mfn[topidx]);
	}

	BUG_ON(HYPERVISOR_shared_info == &xen_dummy_shared_info);

	HYPERVISOR_shared_info->arch.pfn_to_mfn_frame_list_list =
		virt_to_mfn(p2m_top_mfn_list);
	HYPERVISOR_shared_info->arch.max_pfn = xen_start_info->nr_pages;
}

/* Set up p2m_top to point to the domain-builder provided p2m pages */
void __init xen_build_dynamic_phys_to_machine(void)
{
	unsigned long *mfn_list = (unsigned long *)xen_start_info->mfn_list;
	unsigned long max_pfn = min(MAX_DOMAIN_PAGES, xen_start_info->nr_pages);
	unsigned pfn;

	for(pfn = 0; pfn < max_pfn; pfn += P2M_ENTRIES_PER_PAGE) {
		unsigned topidx = p2m_top_index(pfn);

		p2m_top[topidx] = &mfn_list[pfn];
	}
}

unsigned long get_phys_to_machine(unsigned long pfn)
{
	unsigned topidx, idx;

	if (unlikely(pfn >= MAX_DOMAIN_PAGES))
		return INVALID_P2M_ENTRY;

	topidx = p2m_top_index(pfn);
	idx = p2m_index(pfn);
	return p2m_top[topidx][idx];
}
EXPORT_SYMBOL_GPL(get_phys_to_machine);

static void alloc_p2m(unsigned long **pp, unsigned long *mfnp)
{
	unsigned long *p;
	unsigned i;

	p = (void *)__get_free_page(GFP_KERNEL | __GFP_NOFAIL);
	BUG_ON(p == NULL);

	for(i = 0; i < P2M_ENTRIES_PER_PAGE; i++)
		p[i] = INVALID_P2M_ENTRY;

	if (cmpxchg(pp, p2m_missing, p) != p2m_missing)
		free_page((unsigned long)p);
	else
		*mfnp = virt_to_mfn(p);
}

void set_phys_to_machine(unsigned long pfn, unsigned long mfn)
{
	unsigned topidx, idx;

	if (unlikely(xen_feature(XENFEAT_auto_translated_physmap))) {
		BUG_ON(pfn != mfn && mfn != INVALID_P2M_ENTRY);
		return;
	}

	if (unlikely(pfn >= MAX_DOMAIN_PAGES)) {
		BUG_ON(mfn != INVALID_P2M_ENTRY);
		return;
	}

	topidx = p2m_top_index(pfn);
	if (p2m_top[topidx] == p2m_missing) {
		/* no need to allocate a page to store an invalid entry */
		if (mfn == INVALID_P2M_ENTRY)
			return;
		alloc_p2m(&p2m_top[topidx], &p2m_top_mfn[topidx]);
	}

	idx = p2m_index(pfn);
	p2m_top[topidx][idx] = mfn;
}

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

	BUG_ON(pte == NULL);

	return XMADDR(((phys_addr_t)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();
}


static bool xen_page_pinned(void *ptr)
{
	struct page *page = virt_to_page(ptr);

	return PagePinned(page);
}

static void xen_extend_mmu_update(const struct mmu_update *update)
{
	struct multicall_space mcs;
	struct mmu_update *u;

	mcs = xen_mc_extend_args(__HYPERVISOR_mmu_update, sizeof(*u));

	if (mcs.mc != NULL) {
		ADD_STATS(mmu_update_extended, 1);
		ADD_STATS(mmu_update_histo[mcs.mc->args[1]], -1);

		mcs.mc->args[1]++;

		if (mcs.mc->args[1] < MMU_UPDATE_HISTO)
			ADD_STATS(mmu_update_histo[mcs.mc->args[1]], 1);
		else
			ADD_STATS(mmu_update_histo[0], 1);
	} else {
		ADD_STATS(mmu_update, 1);
		mcs = __xen_mc_entry(sizeof(*u));
		MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
		ADD_STATS(mmu_update_histo[1], 1);
	}

	u = mcs.args;
	*u = *update;
}

void xen_set_pmd_hyper(pmd_t *ptr, pmd_t val)
{
	struct mmu_update u;

	preempt_disable();

	xen_mc_batch();

	/* ptr may be ioremapped for 64-bit pagetable setup */
	u.ptr = arbitrary_virt_to_machine(ptr).maddr;
	u.val = pmd_val_ma(val);
	xen_extend_mmu_update(&u);

	ADD_STATS(pmd_update_batched, paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

void xen_set_pmd(pmd_t *ptr, pmd_t val)
{
	ADD_STATS(pmd_update, 1);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		return;
	}

	ADD_STATS(pmd_update_pinned, 1);

	xen_set_pmd_hyper(ptr, val);
}

/*
 * 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)
{
	set_pte_vaddr(vaddr, mfn_pte(mfn, flags));
}

void xen_set_pte_at(struct mm_struct *mm, unsigned long addr,
		    pte_t *ptep, pte_t pteval)
{
	/* updates to init_mm may be done without lock */
	if (mm == &init_mm)
		preempt_disable();

	ADD_STATS(set_pte_at, 1);
//	ADD_STATS(set_pte_at_pinned, xen_page_pinned(ptep));
	ADD_STATS(set_pte_at_current, mm == current->mm);
	ADD_STATS(set_pte_at_kernel, mm == &init_mm);

	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);
			ADD_STATS(set_pte_at_batched, 1);
			xen_mc_issue(PARAVIRT_LAZY_MMU);
			goto out;
		} else
			if (HYPERVISOR_update_va_mapping(addr, pteval, 0) == 0)
				goto out;
	}
	xen_set_pte(ptep, pteval);

out:
	if (mm == &init_mm)
		preempt_enable();
}

pte_t xen_ptep_modify_prot_start(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
{
	/* Just return the pte as-is.  We preserve the bits on commit */
	return *ptep;
}

void xen_ptep_modify_prot_commit(struct mm_struct *mm, unsigned long addr,
				 pte_t *ptep, pte_t pte)
{
	struct mmu_update u;

	xen_mc_batch();

	u.ptr = virt_to_machine(ptep).maddr | MMU_PT_UPDATE_PRESERVE_AD;
	u.val = pte_val_ma(pte);
	xen_extend_mmu_update(&u);

	ADD_STATS(prot_commit, 1);
	ADD_STATS(prot_commit_batched, paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}

/* Assume pteval_t is equivalent to all the other *val_t types. */
static pteval_t pte_mfn_to_pfn(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long mfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		pteval_t flags = val & PTE_FLAGS_MASK;
		val = ((pteval_t)mfn_to_pfn(mfn) << PAGE_SHIFT) | flags;
	}

	return val;
}

static pteval_t pte_pfn_to_mfn(pteval_t val)
{
	if (val & _PAGE_PRESENT) {
		unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
		pteval_t flags = val & PTE_FLAGS_MASK;
		val = ((pteval_t)pfn_to_mfn(pfn) << PAGE_SHIFT) | flags;
	}

	return val;
}

pteval_t xen_pte_val(pte_t pte)
{
	return pte_mfn_to_pfn(pte.pte);
}

pgdval_t xen_pgd_val(pgd_t pgd)
{
	return pte_mfn_to_pfn(pgd.pgd);
}

pte_t xen_make_pte(pteval_t pte)
{
	pte = pte_pfn_to_mfn(pte);
	return native_make_pte(pte);
}

pgd_t xen_make_pgd(pgdval_t pgd)
{
	pgd = pte_pfn_to_mfn(pgd);
	return native_make_pgd(pgd);
}

pmdval_t xen_pmd_val(pmd_t pmd)
{
	return pte_mfn_to_pfn(pmd.pmd);
}

void xen_set_pud_hyper(pud_t *ptr, pud_t val)
{
	struct mmu_update u;

	preempt_disable();

	xen_mc_batch();

	/* ptr may be ioremapped for 64-bit pagetable setup */
	u.ptr = arbitrary_virt_to_machine(ptr).maddr;
	u.val = pud_val_ma(val);
	xen_extend_mmu_update(&u);

	ADD_STATS(pud_update_batched, paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

void xen_set_pud(pud_t *ptr, pud_t val)
{
	ADD_STATS(pud_update, 1);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		return;
	}

	ADD_STATS(pud_update_pinned, 1);

	xen_set_pud_hyper(ptr, val);
}

void xen_set_pte(pte_t *ptep, pte_t pte)
{
	ADD_STATS(pte_update, 1);
//	ADD_STATS(pte_update_pinned, xen_page_pinned(ptep));
	ADD_STATS(pte_update_batched, paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU);

#ifdef CONFIG_X86_PAE
	ptep->pte_high = pte.pte_high;
	smp_wmb();
	ptep->pte_low = pte.pte_low;
#else
	*ptep = pte;
#endif
}

#ifdef CONFIG_X86_PAE
void xen_set_pte_atomic(pte_t *ptep, pte_t pte)
{
	set_64bit((u64 *)ptep, native_pte_val(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)
{
	set_pmd(pmdp, __pmd(0));
}
#endif	/* CONFIG_X86_PAE */

pmd_t xen_make_pmd(pmdval_t pmd)
{
	pmd = pte_pfn_to_mfn(pmd);
	return native_make_pmd(pmd);
}

#if PAGETABLE_LEVELS == 4
pudval_t xen_pud_val(pud_t pud)
{
	return pte_mfn_to_pfn(pud.pud);
}

pud_t xen_make_pud(pudval_t pud)
{
	pud = pte_pfn_to_mfn(pud);

	return native_make_pud(pud);
}

pgd_t *xen_get_user_pgd(pgd_t *pgd)
{
	pgd_t *pgd_page = (pgd_t *)(((unsigned long)pgd) & PAGE_MASK);
	unsigned offset = pgd - pgd_page;
	pgd_t *user_ptr = NULL;

	if (offset < pgd_index(USER_LIMIT)) {
		struct page *page = virt_to_page(pgd_page);
		user_ptr = (pgd_t *)page->private;
		if (user_ptr)
			user_ptr += offset;
	}

	return user_ptr;
}

static void __xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
	struct mmu_update u;

	u.ptr = virt_to_machine(ptr).maddr;
	u.val = pgd_val_ma(val);
	xen_extend_mmu_update(&u);
}

/*
 * Raw hypercall-based set_pgd, intended for in early boot before
 * there's a page structure.  This implies:
 *  1. The only existing pagetable is the kernel's
 *  2. It is always pinned
 *  3. It has no user pagetable attached to it
 */
void __init xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
	preempt_disable();

	xen_mc_batch();

	__xen_set_pgd_hyper(ptr, val);

	xen_mc_issue(PARAVIRT_LAZY_MMU);

	preempt_enable();
}

void xen_set_pgd(pgd_t *ptr, pgd_t val)
{
	pgd_t *user_ptr = xen_get_user_pgd(ptr);

	ADD_STATS(pgd_update, 1);

	/* If page is not pinned, we can just update the entry
	   directly */
	if (!xen_page_pinned(ptr)) {
		*ptr = val;
		if (user_ptr) {
			WARN_ON(xen_page_pinned(user_ptr));
			*user_ptr = val;
		}
		return;
	}

	ADD_STATS(pgd_update_pinned, 1);
	ADD_STATS(pgd_update_batched, paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU);

	/* If it's pinned, then we can at least batch the kernel and
	   user updates together. */
	xen_mc_batch();

	__xen_set_pgd_hyper(ptr, val);
	if (user_ptr)
		__xen_set_pgd_hyper(user_ptr, val);

	xen_mc_issue(PARAVIRT_LAZY_MMU);
}
#endif	/* PAGETABLE_LEVELS == 4 */

/*
 * (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 limit.  In the normal case this
 * will be STACK_TOP_MAX, but at boot we need to pin up to
 * FIXADDR_TOP.
 *
 * For 32-bit the important bit is that we don't pin beyond there,
 * because then we start getting into Xen's ptes.
 *
 * For 64-bit, we must skip the Xen hole in the middle of the address
 * space, just after the big x86-64 virtual hole.
 */
static int xen_pgd_walk(struct mm_struct *mm,
			int (*func)(struct mm_struct *mm, struct page *,
				    enum pt_level),
			unsigned long limit)
{
	pgd_t *pgd = mm->pgd;
	int flush = 0;
	unsigned hole_low, hole_high;
	unsigned pgdidx_limit, pudidx_limit, pmdidx_limit;
	unsigned pgdidx, pudidx, pmdidx;

	/* The limit is the last byte to be touched */
	limit--;
	BUG_ON(limit >= FIXADDR_TOP);

	if (xen_feature(XENFEAT_auto_translated_physmap))
		return 0;

	/*
	 * 64-bit has a great big hole in the middle of the address
	 * space, which contains the Xen mappings.  On 32-bit these
	 * will end up making a zero-sized hole and so is a no-op.
	 */
	hole_low = pgd_index(USER_LIMIT);
	hole_high = pgd_index(PAGE_OFFSET);

	pgdidx_limit = pgd_index(limit);
#if PTRS_PER_PUD > 1
	pudidx_limit = pud_index(limit);
#else
	pudidx_limit = 0;
#endif
#if PTRS_PER_PMD > 1
	pmdidx_limit = pmd_index(limit);
#else
	pmdidx_limit = 0;
#endif

	for (pgdidx = 0; pgdidx <= pgdidx_limit; pgdidx++) {
		pud_t *pud;

		if (pgdidx >= hole_low && pgdidx < hole_high)
			continue;

		if (!pgd_val(pgd[pgdidx]))
			continue;

		pud = pud_offset(&pgd[pgdidx], 0);

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

		for (pudidx = 0; pudidx < PTRS_PER_PUD; pudidx++) {
			pmd_t *pmd;

			if (pgdidx == pgdidx_limit &&
			    pudidx > pudidx_limit)
				goto out;

			if (pud_none(pud[pudidx]))
				continue;

			pmd = pmd_offset(&pud[pudidx], 0);

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

			for (pmdidx = 0; pmdidx < PTRS_PER_PMD; pmdidx++) {
				struct page *pte;

				if (pgdidx == pgdidx_limit &&
				    pudidx == pudidx_limit &&
				    pmdidx > pmdidx_limit)
					goto out;

				if (pmd_none(pmd[pmdidx]))
					continue;

				pte = pmd_page(pmd[pmdidx]);
				flush |= (*func)(mm, pte, PT_PTE);
			}
		}
	}

out:
	/* Do the top level last, so that the callbacks can use it as
	   a cue to do final things like tlb flushes. */
	flush |= (*func)(mm, virt_to_page(pgd), PT_PGD);

	return flush;
}

/* If we're using split pte locks, then take the page's lock and
   return a pointer to it.  Otherwise return NULL. */
static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm)
{
	spinlock_t *ptl = NULL;

#if USE_SPLIT_PTLOCKS
	ptl = __pte_lockptr(page);
	spin_lock_nest_lock(ptl, &mm->page_table_lock);
#endif

	return ptl;
}

static void xen_pte_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 xen_pin_page(struct mm_struct *mm, struct page *page,
			enum pt_level level)
{
	unsigned pgfl = TestSetPagePinned(page);
	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;

		/*
		 * We need to hold the pagetable lock between the time
		 * we make the pagetable RO and when we actually pin
		 * it.  If we don't, then other users may come in and
		 * attempt to update the pagetable by writing it,
		 * which will fail because the memory is RO but not
		 * pinned, so Xen won't do the trap'n'emulate.
		 *
		 * If we're using split pte locks, we can't hold the
		 * entire pagetable's worth of locks during the
		 * traverse, because we may wrap the preempt count (8
		 * bits).  The solution is to mark RO and pin each PTE
		 * page while holding the lock.  This means the number
		 * of locks we end up holding is never more than a
		 * batch size (~32 entries, at present).
		 *
		 * If we're not using split pte locks, we needn't pin
		 * the PTE pages independently, because we're
		 * protected by the overall pagetable lock.
		 */
		ptl = NULL;
		if (level == PT_PTE)
			ptl = xen_pte_lock(page, mm);

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

		if (ptl) {
			xen_do_pin(MMUEXT_PIN_L1_TABLE, pfn);

			/* Queue a deferred unlock for when this batch
			   is completed. */
			xen_mc_callback(xen_pte_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. */
static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd)
{
	xen_mc_batch();

	if (xen_pgd_walk(mm, xen_pin_page, USER_LIMIT)) {
		/* re-enable interrupts for kmap_flush_unused */
		xen_mc_issue(0);
		kmap_flush_unused();
		xen_mc_batch();
	}

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(pgd);

		xen_do_pin(MMUEXT_PIN_L4_TABLE, PFN_DOWN(__pa(pgd)));

		if (user_pgd) {
			xen_pin_page(mm, virt_to_page(user_pgd), PT_PGD);
			xen_do_pin(MMUEXT_PIN_L4_TABLE, PFN_DOWN(__pa(user_pgd)));
		}
	}
#else /* CONFIG_X86_32 */
#ifdef CONFIG_X86_PAE
	/* Need to make sure unshared kernel PMD is pinnable */
	xen_pin_page(mm, virt_to_page(pgd_page(pgd[pgd_index(TASK_SIZE)])),
		     PT_PMD);
#endif
	xen_do_pin(MMUEXT_PIN_L3_TABLE, PFN_DOWN(__pa(pgd)));
#endif /* CONFIG_X86_64 */
	xen_mc_issue(0);
}

static void xen_pgd_pin(struct mm_struct *mm)
{
	__xen_pgd_pin(mm, mm->pgd);
}

/*
 * On save, we need to pin all pagetables to make sure they get their
 * mfns turned into pfns.  Search the list for any unpinned pgds and pin
 * them (unpinned pgds are not currently in use, probably because the
 * process is under construction or destruction).
 *
 * Expected to be called in stop_machine() ("equivalent to taking
 * every spinlock in the system"), so the locking doesn't really
 * matter all that much.
 */
void xen_mm_pin_all(void)
{
	unsigned long flags;
	struct page *page;

	spin_lock_irqsave(&pgd_lock, flags);

	list_for_each_entry(page, &pgd_list, lru) {
		if (!PagePinned(page)) {
			__xen_pgd_pin(&init_mm, (pgd_t *)page_address(page));
			SetPageSavePinned(page);
		}
	}

	spin_unlock_irqrestore(&pgd_lock, flags);
}

/*
 * 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 xen_mark_pinned(struct mm_struct *mm, struct page *page,
				  enum pt_level level)
{
	SetPagePinned(page);
	return 0;
}

void __init xen_mark_init_mm_pinned(void)
{
	xen_pgd_walk(&init_mm, xen_mark_pinned, FIXADDR_TOP);
}

static int xen_unpin_page(struct mm_struct *mm, struct page *page,
			  enum pt_level level)
{
	unsigned pgfl = TestClearPagePinned(page);

	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;

		/*
		 * Do the converse to pin_page.  If we're using split
		 * pte locks, we must be holding the lock for while
		 * the pte page is unpinned but still RO to prevent
		 * concurrent updates from seeing it in this
		 * partially-pinned state.
		 */
		if (level == PT_PTE) {
			ptl = xen_pte_lock(page, mm);

			if (ptl)
				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(xen_pte_unlock, ptl);
		}
	}

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

/* Release a pagetables pages back as normal RW */
static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd)
{
	xen_mc_batch();

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

#ifdef CONFIG_X86_64
	{
		pgd_t *user_pgd = xen_get_user_pgd(pgd);

		if (user_pgd) {
			xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(user_pgd)));
			xen_unpin_page(mm, virt_to_page(user_pgd), PT_PGD);
		}
	}
#endif

#ifdef CONFIG_X86_PAE
	/* Need to make sure unshared kernel PMD is unpinned */
	xen_unpin_page(mm, virt_to_page(pgd_page(pgd[pgd_index(TASK_SIZE)])),
		       PT_PMD);
#endif

	xen_pgd_walk(mm, xen_unpin_page, USER_LIMIT);

	xen_mc_issue(0);
}

static void xen_pgd_unpin(struct mm_struct *mm)
{
	__xen_pgd_unpin(mm, mm->pgd);
}

/*
 * On resume, undo any pinning done at save, so that the rest of the
 * kernel doesn't see any unexpected pinned pagetables.
 */
void xen_mm_unpin_all(void)
{
	unsigned long flags;
	struct page *page;

	spin_lock_irqsave(&pgd_lock, flags);

	list_for_each_entry(page, &pgd_list, lru) {
		if (PageSavePinned(page)) {
			BUG_ON(!PagePinned(page));
			__xen_pgd_unpin(&init_mm, (pgd_t *)page_address(page));
			ClearPageSavePinned(page);
		}
	}

	spin_unlock_irqrestore(&pgd_lock, flags);
}

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

#ifdef CONFIG_X86_64
	active_mm = read_pda(active_mm);
#else
	active_mm = __get_cpu_var(cpu_tlbstate).active_mm;
#endif

	if (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 xen_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))
		smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
}
#else
static void xen_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 */
	xen_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 (xen_page_pinned(mm->pgd))
		xen_pgd_unpin(mm);

	spin_unlock(&mm->page_table_lock);
}

#ifdef CONFIG_XEN_DEBUG_FS

static struct dentry *d_mmu_debug;

static int __init xen_mmu_debugfs(void)
{
	struct dentry *d_xen = xen_init_debugfs();

	if (d_xen == NULL)
		return -ENOMEM;

	d_mmu_debug = debugfs_create_dir("mmu", d_xen);

	debugfs_create_u8("zero_stats", 0644, d_mmu_debug, &zero_stats);

	debugfs_create_u32("pgd_update", 0444, d_mmu_debug, &mmu_stats.pgd_update);
	debugfs_create_u32("pgd_update_pinned", 0444, d_mmu_debug,
			   &mmu_stats.pgd_update_pinned);
	debugfs_create_u32("pgd_update_batched", 0444, d_mmu_debug,
			   &mmu_stats.pgd_update_pinned);

	debugfs_create_u32("pud_update", 0444, d_mmu_debug, &mmu_stats.pud_update);
	debugfs_create_u32("pud_update_pinned", 0444, d_mmu_debug,
			   &mmu_stats.pud_update_pinned);
	debugfs_create_u32("pud_update_batched", 0444, d_mmu_debug,
			   &mmu_stats.pud_update_pinned);

	debugfs_create_u32("pmd_update", 0444, d_mmu_debug, &mmu_stats.pmd_update);
	debugfs_create_u32("pmd_update_pinned", 0444, d_mmu_debug,
			   &mmu_stats.pmd_update_pinned);
	debugfs_create_u32("pmd_update_batched", 0444, d_mmu_debug,
			   &mmu_stats.pmd_update_pinned);

	debugfs_create_u32("pte_update", 0444, d_mmu_debug, &mmu_stats.pte_update);
//	debugfs_create_u32("pte_update_pinned", 0444, d_mmu_debug,
//			   &mmu_stats.pte_update_pinned);
	debugfs_create_u32("pte_update_batched", 0444, d_mmu_debug,
			   &mmu_stats.pte_update_pinned);

	debugfs_create_u32("mmu_update", 0444, d_mmu_debug, &mmu_stats.mmu_update);
	debugfs_create_u32("mmu_update_extended", 0444, d_mmu_debug,
			   &mmu_stats.mmu_update_extended);
	xen_debugfs_create_u32_array("mmu_update_histo", 0444, d_mmu_debug,
				     mmu_stats.mmu_update_histo, 20);

	debugfs_create_u32("set_pte_at", 0444, d_mmu_debug, &mmu_stats.set_pte_at);
	debugfs_create_u32("set_pte_at_batched", 0444, d_mmu_debug,
			   &mmu_stats.set_pte_at_batched);
	debugfs_create_u32("set_pte_at_current", 0444, d_mmu_debug,
			   &mmu_stats.set_pte_at_current);
	debugfs_create_u32("set_pte_at_kernel", 0444, d_mmu_debug,
			   &mmu_stats.set_pte_at_kernel);

	debugfs_create_u32("prot_commit", 0444, d_mmu_debug, &mmu_stats.prot_commit);
	debugfs_create_u32("prot_commit_batched", 0444, d_mmu_debug,
			   &mmu_stats.prot_commit_batched);

	return 0;
}
fs_initcall(xen_mmu_debugfs);

#endif	/* CONFIG_XEN_DEBUG_FS */