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
|
/* $OpenBSD: uvm_km.c,v 1.117 2014/11/16 12:31:00 deraadt Exp $ */
/* $NetBSD: uvm_km.c,v 1.42 2001/01/14 02:10:01 thorpej Exp $ */
/*
* Copyright (c) 1997 Charles D. Cranor and Washington University.
* Copyright (c) 1991, 1993, The Regents of the University of California.
*
* All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* The Mach Operating System project at Carnegie-Mellon University.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)vm_kern.c 8.3 (Berkeley) 1/12/94
* from: Id: uvm_km.c,v 1.1.2.14 1998/02/06 05:19:27 chs Exp
*
*
* Copyright (c) 1987, 1990 Carnegie-Mellon University.
* All rights reserved.
*
* Permission to use, copy, modify and distribute this software and
* its documentation is hereby granted, provided that both the copyright
* notice and this permission notice appear in all copies of the
* software, derivative works or modified versions, and any portions
* thereof, and that both notices appear in supporting documentation.
*
* CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
* CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
* FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
*
* Carnegie Mellon requests users of this software to return to
*
* Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
* School of Computer Science
* Carnegie Mellon University
* Pittsburgh PA 15213-3890
*
* any improvements or extensions that they make and grant Carnegie the
* rights to redistribute these changes.
*/
/*
* uvm_km.c: handle kernel memory allocation and management
*/
/*
* overview of kernel memory management:
*
* the kernel virtual address space is mapped by "kernel_map." kernel_map
* starts at VM_MIN_KERNEL_ADDRESS and goes to VM_MAX_KERNEL_ADDRESS.
* note that VM_MIN_KERNEL_ADDRESS is equal to vm_map_min(kernel_map).
*
* the kernel_map has several "submaps." submaps can only appear in
* the kernel_map (user processes can't use them). submaps "take over"
* the management of a sub-range of the kernel's address space. submaps
* are typically allocated at boot time and are never released. kernel
* virtual address space that is mapped by a submap is locked by the
* submap's lock -- not the kernel_map's lock.
*
* thus, the useful feature of submaps is that they allow us to break
* up the locking and protection of the kernel address space into smaller
* chunks.
*
* The VM system has several standard kernel submaps:
* kmem_map: Contains only wired kernel memory for malloc(9).
* Note: All access to this map must be protected by splvm as
* calls to malloc(9) are allowed in interrupt handlers.
* exec_map: Memory to hold arguments to system calls are allocated from
* this map.
* XXX: This is primeraly used to artificially limit the number
* of concurrent processes doing an exec.
* phys_map: Buffers for vmapbuf (physio) are allocated from this map.
*
* the kernel allocates its private memory out of special uvm_objects whose
* reference count is set to UVM_OBJ_KERN (thus indicating that the objects
* are "special" and never die). all kernel objects should be thought of
* as large, fixed-sized, sparsely populated uvm_objects. each kernel
* object is equal to the size of kernel virtual address space (i.e. the
* value "VM_MAX_KERNEL_ADDRESS - VM_MIN_KERNEL_ADDRESS").
*
* most kernel private memory lives in kernel_object. the only exception
* to this is for memory that belongs to submaps that must be protected
* by splvm(). each of these submaps manages their own pages.
*
* note that just because a kernel object spans the entire kernel virtual
* address space doesn't mean that it has to be mapped into the entire space.
* large chunks of a kernel object's space go unused either because
* that area of kernel VM is unmapped, or there is some other type of
* object mapped into that range (e.g. a vnode). for submap's kernel
* objects, the only part of the object that can ever be populated is the
* offsets that are managed by the submap.
*
* note that the "offset" in a kernel object is always the kernel virtual
* address minus the VM_MIN_KERNEL_ADDRESS (aka vm_map_min(kernel_map)).
* example:
* suppose VM_MIN_KERNEL_ADDRESS is 0xf8000000 and the kernel does a
* uvm_km_alloc(kernel_map, PAGE_SIZE) [allocate 1 wired down page in the
* kernel map]. if uvm_km_alloc returns virtual address 0xf8235000,
* then that means that the page at offset 0x235000 in kernel_object is
* mapped at 0xf8235000.
*
* kernel objects have one other special property: when the kernel virtual
* memory mapping them is unmapped, the backing memory in the object is
* freed right away. this is done with the uvm_km_pgremove() function.
* this has to be done because there is no backing store for kernel pages
* and no need to save them after they are no longer referenced.
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kthread.h>
#include <uvm/uvm.h>
/*
* global data structures
*/
struct vm_map *kernel_map = NULL;
/* Unconstraint range. */
struct uvm_constraint_range no_constraint = { 0x0, (paddr_t)-1 };
/*
* local data structues
*/
static struct vm_map kernel_map_store;
/*
* uvm_km_init: init kernel maps and objects to reflect reality (i.e.
* KVM already allocated for text, data, bss, and static data structures).
*
* => KVM is defined by VM_MIN_KERNEL_ADDRESS/VM_MAX_KERNEL_ADDRESS.
* we assume that [min -> start] has already been allocated and that
* "end" is the end.
*/
void
uvm_km_init(vaddr_t start, vaddr_t end)
{
vaddr_t base = VM_MIN_KERNEL_ADDRESS;
/* next, init kernel memory objects. */
/* kernel_object: for pageable anonymous kernel memory */
uao_init();
uvm.kernel_object = uao_create(VM_MAX_KERNEL_ADDRESS -
VM_MIN_KERNEL_ADDRESS, UAO_FLAG_KERNOBJ);
/*
* init the map and reserve already allocated kernel space
* before installing.
*/
uvm_map_setup(&kernel_map_store, base, end,
#ifdef KVA_GUARDPAGES
VM_MAP_PAGEABLE | VM_MAP_GUARDPAGES
#else
VM_MAP_PAGEABLE
#endif
);
kernel_map_store.pmap = pmap_kernel();
if (base != start && uvm_map(&kernel_map_store, &base, start - base,
NULL, UVM_UNKNOWN_OFFSET, 0, UVM_MAPFLAG(PROT_MASK, PROT_MASK,
UVM_INH_NONE, POSIX_MADV_RANDOM, UVM_FLAG_FIXED)) != 0)
panic("uvm_km_init: could not reserve space for kernel");
kernel_map = &kernel_map_store;
}
/*
* uvm_km_suballoc: allocate a submap in the kernel map. once a submap
* is allocated all references to that area of VM must go through it. this
* allows the locking of VAs in kernel_map to be broken up into regions.
*
* => if `fixed' is true, *min specifies where the region described
* by the submap must start
* => if submap is non NULL we use that as the submap, otherwise we
* alloc a new map
*/
struct vm_map *
uvm_km_suballoc(struct vm_map *map, vaddr_t *min, vaddr_t *max, vsize_t size,
int flags, boolean_t fixed, struct vm_map *submap)
{
int mapflags = UVM_FLAG_NOMERGE | (fixed ? UVM_FLAG_FIXED : 0);
size = round_page(size); /* round up to pagesize */
/* first allocate a blank spot in the parent map */
if (uvm_map(map, min, size, NULL, UVM_UNKNOWN_OFFSET, 0,
UVM_MAPFLAG(PROT_MASK, PROT_MASK, UVM_INH_NONE,
POSIX_MADV_RANDOM, mapflags)) != 0) {
panic("uvm_km_suballoc: unable to allocate space in parent map");
}
/* set VM bounds (min is filled in by uvm_map) */
*max = *min + size;
/* add references to pmap and create or init the submap */
pmap_reference(vm_map_pmap(map));
if (submap == NULL) {
submap = uvm_map_create(vm_map_pmap(map), *min, *max, flags);
if (submap == NULL)
panic("uvm_km_suballoc: unable to create submap");
} else {
uvm_map_setup(submap, *min, *max, flags);
submap->pmap = vm_map_pmap(map);
}
/* now let uvm_map_submap plug in it... */
if (uvm_map_submap(map, *min, *max, submap) != 0)
panic("uvm_km_suballoc: submap allocation failed");
return(submap);
}
/*
* uvm_km_pgremove: remove pages from a kernel uvm_object.
*
* => when you unmap a part of anonymous kernel memory you want to toss
* the pages right away. (this gets called from uvm_unmap_...).
*/
void
uvm_km_pgremove(struct uvm_object *uobj, vaddr_t start, vaddr_t end)
{
struct vm_page *pp;
voff_t curoff;
int slot;
KASSERT(uobj->pgops == &aobj_pager);
for (curoff = start ; curoff < end ; curoff += PAGE_SIZE) {
pp = uvm_pagelookup(uobj, curoff);
if (pp && pp->pg_flags & PG_BUSY) {
atomic_setbits_int(&pp->pg_flags, PG_WANTED);
UVM_WAIT(pp, 0, "km_pgrm", 0);
curoff -= PAGE_SIZE; /* loop back to us */
continue;
}
/* free the swap slot, then the page */
slot = uao_dropswap(uobj, curoff >> PAGE_SHIFT);
if (pp != NULL) {
uvm_lock_pageq();
uvm_pagefree(pp);
uvm_unlock_pageq();
} else if (slot != 0) {
uvmexp.swpgonly--;
}
}
}
/*
* uvm_km_pgremove_intrsafe: like uvm_km_pgremove(), but for "intrsafe"
* objects
*
* => when you unmap a part of anonymous kernel memory you want to toss
* the pages right away. (this gets called from uvm_unmap_...).
* => none of the pages will ever be busy, and none of them will ever
* be on the active or inactive queues (because these objects are
* never allowed to "page").
*/
void
uvm_km_pgremove_intrsafe(vaddr_t start, vaddr_t end)
{
struct vm_page *pg;
vaddr_t va;
paddr_t pa;
for (va = start; va < end; va += PAGE_SIZE) {
if (!pmap_extract(pmap_kernel(), va, &pa))
continue;
pg = PHYS_TO_VM_PAGE(pa);
if (pg == NULL)
panic("uvm_km_pgremove_intrsafe: no page");
uvm_pagefree(pg);
}
}
/*
* uvm_km_kmemalloc: lower level kernel memory allocator for malloc()
*
* => we map wired memory into the specified map using the obj passed in
* => NOTE: we can return NULL even if we can wait if there is not enough
* free VM space in the map... caller should be prepared to handle
* this case.
* => we return KVA of memory allocated
* => flags: NOWAIT, VALLOC - just allocate VA, TRYLOCK - fail if we can't
* lock the map
* => low, high, alignment, boundary, nsegs are the corresponding parameters
* to uvm_pglistalloc
* => flags: ZERO - correspond to uvm_pglistalloc flags
*/
vaddr_t
uvm_km_kmemalloc_pla(struct vm_map *map, struct uvm_object *obj, vsize_t size,
vsize_t valign, int flags, paddr_t low, paddr_t high, paddr_t alignment,
paddr_t boundary, int nsegs)
{
vaddr_t kva, loopva;
voff_t offset;
struct vm_page *pg;
struct pglist pgl;
int pla_flags;
KASSERT(vm_map_pmap(map) == pmap_kernel());
/* UVM_KMF_VALLOC => !UVM_KMF_ZERO */
KASSERT(!(flags & UVM_KMF_VALLOC) ||
!(flags & UVM_KMF_ZERO));
/* setup for call */
size = round_page(size);
kva = vm_map_min(map); /* hint */
if (nsegs == 0)
nsegs = atop(size);
/* allocate some virtual space */
if (__predict_false(uvm_map(map, &kva, size, obj, UVM_UNKNOWN_OFFSET,
valign, UVM_MAPFLAG(PROT_READ | PROT_WRITE,
PROT_READ | PROT_WRITE, UVM_INH_NONE,
POSIX_MADV_RANDOM, (flags & UVM_KMF_TRYLOCK))) != 0)) {
return(0);
}
/* if all we wanted was VA, return now */
if (flags & UVM_KMF_VALLOC) {
return(kva);
}
/* recover object offset from virtual address */
if (obj != NULL)
offset = kva - vm_map_min(kernel_map);
else
offset = 0;
/*
* now allocate and map in the memory... note that we are the only ones
* whom should ever get a handle on this area of VM.
*/
TAILQ_INIT(&pgl);
pla_flags = 0;
KASSERT(uvmexp.swpgonly <= uvmexp.swpages);
if ((flags & UVM_KMF_NOWAIT) ||
((flags & UVM_KMF_CANFAIL) &&
uvmexp.swpages - uvmexp.swpgonly <= atop(size)))
pla_flags |= UVM_PLA_NOWAIT;
else
pla_flags |= UVM_PLA_WAITOK;
if (flags & UVM_KMF_ZERO)
pla_flags |= UVM_PLA_ZERO;
if (uvm_pglistalloc(size, low, high, alignment, boundary, &pgl, nsegs,
pla_flags) != 0) {
/* Failed. */
uvm_unmap(map, kva, kva + size);
return (0);
}
loopva = kva;
while (loopva != kva + size) {
pg = TAILQ_FIRST(&pgl);
TAILQ_REMOVE(&pgl, pg, pageq);
uvm_pagealloc_pg(pg, obj, offset, NULL);
atomic_clearbits_int(&pg->pg_flags, PG_BUSY);
UVM_PAGE_OWN(pg, NULL);
/*
* map it in: note that we call pmap_enter with the map and
* object unlocked in case we are kmem_map.
*/
if (obj == NULL) {
pmap_kenter_pa(loopva, VM_PAGE_TO_PHYS(pg),
PROT_READ | PROT_WRITE);
} else {
pmap_enter(map->pmap, loopva, VM_PAGE_TO_PHYS(pg),
PROT_READ | PROT_WRITE,
PROT_READ | PROT_WRITE | PMAP_WIRED);
}
loopva += PAGE_SIZE;
offset += PAGE_SIZE;
}
KASSERT(TAILQ_EMPTY(&pgl));
pmap_update(pmap_kernel());
return(kva);
}
/*
* uvm_km_free: free an area of kernel memory
*/
void
uvm_km_free(struct vm_map *map, vaddr_t addr, vsize_t size)
{
uvm_unmap(map, trunc_page(addr), round_page(addr+size));
}
/*
* uvm_km_free_wakeup: free an area of kernel memory and wake up
* anyone waiting for vm space.
*
* => XXX: "wanted" bit + unlock&wait on other end?
*/
void
uvm_km_free_wakeup(struct vm_map *map, vaddr_t addr, vsize_t size)
{
struct uvm_map_deadq dead_entries;
vm_map_lock(map);
TAILQ_INIT(&dead_entries);
uvm_unmap_remove(map, trunc_page(addr), round_page(addr+size),
&dead_entries, FALSE, TRUE);
wakeup(map);
vm_map_unlock(map);
uvm_unmap_detach(&dead_entries, 0);
}
/*
* uvm_km_alloc1: allocate wired down memory in the kernel map.
*
* => we can sleep if needed
*/
vaddr_t
uvm_km_alloc1(struct vm_map *map, vsize_t size, vsize_t align, boolean_t zeroit)
{
vaddr_t kva, loopva;
voff_t offset;
struct vm_page *pg;
KASSERT(vm_map_pmap(map) == pmap_kernel());
size = round_page(size);
kva = vm_map_min(map); /* hint */
/* allocate some virtual space */
if (__predict_false(uvm_map(map, &kva, size, uvm.kernel_object,
UVM_UNKNOWN_OFFSET, align, UVM_MAPFLAG(PROT_MASK, PROT_MASK,
UVM_INH_NONE, POSIX_MADV_RANDOM, 0)) != 0)) {
return(0);
}
/* recover object offset from virtual address */
offset = kva - vm_map_min(kernel_map);
/* now allocate the memory. we must be careful about released pages. */
loopva = kva;
while (size) {
/* allocate ram */
pg = uvm_pagealloc(uvm.kernel_object, offset, NULL, 0);
if (pg) {
atomic_clearbits_int(&pg->pg_flags, PG_BUSY);
UVM_PAGE_OWN(pg, NULL);
}
if (__predict_false(pg == NULL)) {
if (curproc == uvm.pagedaemon_proc) {
/*
* It is unfeasible for the page daemon to
* sleep for memory, so free what we have
* allocated and fail.
*/
uvm_unmap(map, kva, loopva - kva);
return (0);
} else {
uvm_wait("km_alloc1w"); /* wait for memory */
continue;
}
}
/*
* map it in; note we're never called with an intrsafe
* object, so we always use regular old pmap_enter().
*/
pmap_enter(map->pmap, loopva, VM_PAGE_TO_PHYS(pg),
PROT_READ | PROT_WRITE | PROT_EXEC,
PROT_READ | PROT_WRITE | PMAP_WIRED);
/* XXX why is the above executable? */
loopva += PAGE_SIZE;
offset += PAGE_SIZE;
size -= PAGE_SIZE;
}
pmap_update(map->pmap);
/*
* zero on request (note that "size" is now zero due to the above loop
* so we need to subtract kva from loopva to reconstruct the size).
*/
if (zeroit)
memset((caddr_t)kva, 0, loopva - kva);
return(kva);
}
/*
* uvm_km_valloc: allocate zero-fill memory in the kernel's address space
*
* => memory is not allocated until fault time
*/
vaddr_t
uvm_km_valloc(struct vm_map *map, vsize_t size)
{
return(uvm_km_valloc_align(map, size, 0, 0));
}
vaddr_t
uvm_km_valloc_try(struct vm_map *map, vsize_t size)
{
return(uvm_km_valloc_align(map, size, 0, UVM_FLAG_TRYLOCK));
}
vaddr_t
uvm_km_valloc_align(struct vm_map *map, vsize_t size, vsize_t align, int flags)
{
vaddr_t kva;
KASSERT(vm_map_pmap(map) == pmap_kernel());
size = round_page(size);
kva = vm_map_min(map); /* hint */
/* allocate some virtual space, demand filled by kernel_object. */
if (__predict_false(uvm_map(map, &kva, size, uvm.kernel_object,
UVM_UNKNOWN_OFFSET, align, UVM_MAPFLAG(PROT_MASK, PROT_MASK,
UVM_INH_NONE, POSIX_MADV_RANDOM, flags)) != 0)) {
return(0);
}
return(kva);
}
/*
* uvm_km_valloc_wait: allocate zero-fill memory in the kernel's address space
*
* => memory is not allocated until fault time
* => if no room in map, wait for space to free, unless requested size
* is larger than map (in which case we return 0)
*/
vaddr_t
uvm_km_valloc_prefer_wait(struct vm_map *map, vsize_t size, voff_t prefer)
{
vaddr_t kva;
KASSERT(vm_map_pmap(map) == pmap_kernel());
size = round_page(size);
if (size > vm_map_max(map) - vm_map_min(map))
return(0);
while (1) {
kva = vm_map_min(map); /* hint */
/*
* allocate some virtual space. will be demand filled
* by kernel_object.
*/
if (__predict_true(uvm_map(map, &kva, size, uvm.kernel_object,
prefer, 0, UVM_MAPFLAG(PROT_MASK,
PROT_MASK, UVM_INH_NONE, POSIX_MADV_RANDOM, 0)) == 0)) {
return(kva);
}
/* failed. sleep for a while (on map) */
tsleep(map, PVM, "vallocwait", 0);
}
/*NOTREACHED*/
}
vaddr_t
uvm_km_valloc_wait(struct vm_map *map, vsize_t size)
{
return uvm_km_valloc_prefer_wait(map, size, UVM_UNKNOWN_OFFSET);
}
#if defined(__HAVE_PMAP_DIRECT)
/*
* uvm_km_page allocator, __HAVE_PMAP_DIRECT arch
* On architectures with machine memory direct mapped into a portion
* of KVM, we have very little work to do. Just get a physical page,
* and find and return its VA.
*/
void
uvm_km_page_init(void)
{
/* nothing */
}
#else
/*
* uvm_km_page allocator, non __HAVE_PMAP_DIRECT archs
* This is a special allocator that uses a reserve of free pages
* to fulfill requests. It is fast and interrupt safe, but can only
* return page sized regions. Its primary use is as a backend for pool.
*
* The memory returned is allocated from the larger kernel_map, sparing
* pressure on the small interrupt-safe kmem_map. It is wired, but
* not zero filled.
*/
struct uvm_km_pages uvm_km_pages;
void uvm_km_createthread(void *);
void uvm_km_thread(void *);
struct uvm_km_free_page *uvm_km_doputpage(struct uvm_km_free_page *);
/*
* Allocate the initial reserve, and create the thread which will
* keep the reserve full. For bootstrapping, we allocate more than
* the lowat amount, because it may be a while before the thread is
* running.
*/
void
uvm_km_page_init(void)
{
int lowat_min;
int i;
int len, bulk;
vaddr_t addr;
mtx_init(&uvm_km_pages.mtx, IPL_VM);
if (!uvm_km_pages.lowat) {
/* based on physmem, calculate a good value here */
uvm_km_pages.lowat = physmem / 256;
lowat_min = physmem < atop(16 * 1024 * 1024) ? 32 : 128;
if (uvm_km_pages.lowat < lowat_min)
uvm_km_pages.lowat = lowat_min;
}
if (uvm_km_pages.lowat > UVM_KM_PAGES_LOWAT_MAX)
uvm_km_pages.lowat = UVM_KM_PAGES_LOWAT_MAX;
uvm_km_pages.hiwat = 4 * uvm_km_pages.lowat;
if (uvm_km_pages.hiwat > UVM_KM_PAGES_HIWAT_MAX)
uvm_km_pages.hiwat = UVM_KM_PAGES_HIWAT_MAX;
/* Allocate all pages in as few allocations as possible. */
len = 0;
bulk = uvm_km_pages.hiwat;
while (len < uvm_km_pages.hiwat && bulk > 0) {
bulk = MIN(bulk, uvm_km_pages.hiwat - len);
addr = vm_map_min(kernel_map);
if (uvm_map(kernel_map, &addr, (vsize_t)bulk << PAGE_SHIFT,
NULL, UVM_UNKNOWN_OFFSET, 0,
UVM_MAPFLAG(PROT_READ | PROT_WRITE,
PROT_READ | PROT_WRITE, UVM_INH_NONE,
POSIX_MADV_RANDOM, UVM_KMF_TRYLOCK)) != 0) {
bulk /= 2;
continue;
}
for (i = len; i < len + bulk; i++, addr += PAGE_SIZE)
uvm_km_pages.page[i] = addr;
len += bulk;
}
uvm_km_pages.free = len;
for (i = len; i < UVM_KM_PAGES_HIWAT_MAX; i++)
uvm_km_pages.page[i] = 0;
/* tone down if really high */
if (uvm_km_pages.lowat > 512)
uvm_km_pages.lowat = 512;
kthread_create_deferred(uvm_km_createthread, NULL);
}
void
uvm_km_createthread(void *arg)
{
kthread_create(uvm_km_thread, NULL, &uvm_km_pages.km_proc, "kmthread");
}
/*
* Endless loop. We grab pages in increments of 16 pages, then
* quickly swap them into the list. At some point we can consider
* returning memory to the system if we have too many free pages,
* but that's not implemented yet.
*/
void
uvm_km_thread(void *arg)
{
vaddr_t pg[16];
int i;
int allocmore = 0;
int flags;
struct uvm_km_free_page *fp = NULL;
for (;;) {
mtx_enter(&uvm_km_pages.mtx);
if (uvm_km_pages.free >= uvm_km_pages.lowat &&
uvm_km_pages.freelist == NULL) {
msleep(&uvm_km_pages.km_proc, &uvm_km_pages.mtx,
PVM, "kmalloc", 0);
}
allocmore = uvm_km_pages.free < uvm_km_pages.lowat;
fp = uvm_km_pages.freelist;
uvm_km_pages.freelist = NULL;
uvm_km_pages.freelistlen = 0;
mtx_leave(&uvm_km_pages.mtx);
if (allocmore) {
/*
* If there was nothing on the freelist, then we
* must obtain at least one page to make progress.
* So, only use UVM_KMF_TRYLOCK for the first page
* if fp != NULL
*/
flags = UVM_MAPFLAG(PROT_READ | PROT_WRITE,
PROT_READ | PROT_WRITE,
UVM_INH_NONE, POSIX_MADV_RANDOM,
fp != NULL ? UVM_KMF_TRYLOCK : 0);
memset(pg, 0, sizeof(pg));
for (i = 0; i < nitems(pg); i++) {
pg[i] = vm_map_min(kernel_map);
if (uvm_map(kernel_map, &pg[i], PAGE_SIZE,
NULL, UVM_UNKNOWN_OFFSET, 0, flags) != 0) {
pg[i] = 0;
break;
}
/* made progress, so don't sleep for more */
flags = UVM_MAPFLAG(PROT_READ | PROT_WRITE,
PROT_READ | PROT_WRITE,
UVM_INH_NONE, POSIX_MADV_RANDOM,
UVM_KMF_TRYLOCK);
}
mtx_enter(&uvm_km_pages.mtx);
for (i = 0; i < nitems(pg); i++) {
if (uvm_km_pages.free ==
nitems(uvm_km_pages.page))
break;
else if (pg[i] != 0)
uvm_km_pages.page[uvm_km_pages.free++]
= pg[i];
}
wakeup(&uvm_km_pages.free);
mtx_leave(&uvm_km_pages.mtx);
/* Cleanup left-over pages (if any). */
for (; i < nitems(pg); i++) {
if (pg[i] != 0) {
uvm_unmap(kernel_map,
pg[i], pg[i] + PAGE_SIZE);
}
}
}
while (fp) {
fp = uvm_km_doputpage(fp);
}
}
}
struct uvm_km_free_page *
uvm_km_doputpage(struct uvm_km_free_page *fp)
{
vaddr_t va = (vaddr_t)fp;
struct vm_page *pg;
int freeva = 1;
struct uvm_km_free_page *nextfp = fp->next;
pg = uvm_atopg(va);
pmap_kremove(va, PAGE_SIZE);
pmap_update(kernel_map->pmap);
mtx_enter(&uvm_km_pages.mtx);
if (uvm_km_pages.free < uvm_km_pages.hiwat) {
uvm_km_pages.page[uvm_km_pages.free++] = va;
freeva = 0;
}
mtx_leave(&uvm_km_pages.mtx);
if (freeva)
uvm_unmap(kernel_map, va, va + PAGE_SIZE);
uvm_pagefree(pg);
return (nextfp);
}
#endif /* !__HAVE_PMAP_DIRECT */
void *
km_alloc(size_t sz, const struct kmem_va_mode *kv,
const struct kmem_pa_mode *kp, const struct kmem_dyn_mode *kd)
{
struct vm_map *map;
struct vm_page *pg;
struct pglist pgl;
int mapflags = 0;
vm_prot_t prot;
int pla_flags;
int pla_maxseg;
#ifdef __HAVE_PMAP_DIRECT
paddr_t pa;
#endif
vaddr_t va, sva;
KASSERT(sz == round_page(sz));
TAILQ_INIT(&pgl);
if (kp->kp_nomem || kp->kp_pageable)
goto alloc_va;
pla_flags = kd->kd_waitok ? UVM_PLA_WAITOK : UVM_PLA_NOWAIT;
pla_flags |= UVM_PLA_TRYCONTIG;
if (kp->kp_zero)
pla_flags |= UVM_PLA_ZERO;
pla_maxseg = kp->kp_maxseg;
if (pla_maxseg == 0)
pla_maxseg = sz / PAGE_SIZE;
if (uvm_pglistalloc(sz, kp->kp_constraint->ucr_low,
kp->kp_constraint->ucr_high, kp->kp_align, kp->kp_boundary,
&pgl, pla_maxseg, pla_flags)) {
return (NULL);
}
#ifdef __HAVE_PMAP_DIRECT
if (kv->kv_align || kv->kv_executable)
goto alloc_va;
#if 1
/*
* For now, only do DIRECT mappings for single page
* allocations, until we figure out a good way to deal
* with contig allocations in km_free.
*/
if (!kv->kv_singlepage)
goto alloc_va;
#endif
/*
* Dubious optimization. If we got a contig segment, just map it
* through the direct map.
*/
TAILQ_FOREACH(pg, &pgl, pageq) {
if (pg != TAILQ_FIRST(&pgl) &&
VM_PAGE_TO_PHYS(pg) != pa + PAGE_SIZE)
break;
pa = VM_PAGE_TO_PHYS(pg);
}
if (pg == NULL) {
TAILQ_FOREACH(pg, &pgl, pageq) {
vaddr_t v;
v = pmap_map_direct(pg);
if (pg == TAILQ_FIRST(&pgl))
va = v;
}
return ((void *)va);
}
#endif
alloc_va:
if (kv->kv_executable) {
prot = PROT_READ | PROT_WRITE | PROT_EXEC;
} else {
prot = PROT_READ | PROT_WRITE;
}
if (kp->kp_pageable) {
KASSERT(kp->kp_object);
KASSERT(!kv->kv_singlepage);
} else {
KASSERT(kp->kp_object == NULL);
}
if (kv->kv_singlepage) {
KASSERT(sz == PAGE_SIZE);
#ifdef __HAVE_PMAP_DIRECT
panic("km_alloc: DIRECT single page");
#else
mtx_enter(&uvm_km_pages.mtx);
while (uvm_km_pages.free == 0) {
if (kd->kd_waitok == 0) {
mtx_leave(&uvm_km_pages.mtx);
uvm_pglistfree(&pgl);
return NULL;
}
msleep(&uvm_km_pages.free, &uvm_km_pages.mtx, PVM,
"getpage", 0);
}
va = uvm_km_pages.page[--uvm_km_pages.free];
if (uvm_km_pages.free < uvm_km_pages.lowat &&
curproc != uvm_km_pages.km_proc) {
if (kd->kd_slowdown)
*kd->kd_slowdown = 1;
wakeup(&uvm_km_pages.km_proc);
}
mtx_leave(&uvm_km_pages.mtx);
#endif
} else {
struct uvm_object *uobj = NULL;
if (kd->kd_trylock)
mapflags |= UVM_KMF_TRYLOCK;
if (kp->kp_object)
uobj = *kp->kp_object;
try_map:
map = *kv->kv_map;
va = vm_map_min(map);
if (uvm_map(map, &va, sz, uobj, kd->kd_prefer,
kv->kv_align, UVM_MAPFLAG(prot, prot, UVM_INH_NONE,
POSIX_MADV_RANDOM, mapflags))) {
if (kv->kv_wait && kd->kd_waitok) {
tsleep(map, PVM, "km_allocva", 0);
goto try_map;
}
uvm_pglistfree(&pgl);
return (NULL);
}
}
sva = va;
TAILQ_FOREACH(pg, &pgl, pageq) {
if (kp->kp_pageable)
pmap_enter(pmap_kernel(), va, VM_PAGE_TO_PHYS(pg),
prot, prot | PMAP_WIRED);
else
pmap_kenter_pa(va, VM_PAGE_TO_PHYS(pg), prot);
va += PAGE_SIZE;
}
pmap_update(pmap_kernel());
return ((void *)sva);
}
void
km_free(void *v, size_t sz, const struct kmem_va_mode *kv,
const struct kmem_pa_mode *kp)
{
vaddr_t sva, eva, va;
struct vm_page *pg;
struct pglist pgl;
sva = va = (vaddr_t)v;
eva = va + sz;
if (kp->kp_nomem) {
goto free_va;
}
if (kv->kv_singlepage) {
#ifdef __HAVE_PMAP_DIRECT
pg = pmap_unmap_direct(va);
uvm_pagefree(pg);
#else
struct uvm_km_free_page *fp = v;
mtx_enter(&uvm_km_pages.mtx);
fp->next = uvm_km_pages.freelist;
uvm_km_pages.freelist = fp;
if (uvm_km_pages.freelistlen++ > 16)
wakeup(&uvm_km_pages.km_proc);
mtx_leave(&uvm_km_pages.mtx);
#endif
return;
}
if (kp->kp_pageable) {
pmap_remove(pmap_kernel(), sva, eva);
pmap_update(pmap_kernel());
} else {
TAILQ_INIT(&pgl);
for (va = sva; va < eva; va += PAGE_SIZE) {
paddr_t pa;
if (!pmap_extract(pmap_kernel(), va, &pa))
continue;
pg = PHYS_TO_VM_PAGE(pa);
if (pg == NULL) {
panic("km_free: unmanaged page 0x%lx\n", pa);
}
TAILQ_INSERT_TAIL(&pgl, pg, pageq);
}
pmap_kremove(sva, sz);
pmap_update(pmap_kernel());
uvm_pglistfree(&pgl);
}
free_va:
uvm_unmap(*kv->kv_map, sva, eva);
if (kv->kv_wait)
wakeup(*kv->kv_map);
}
const struct kmem_va_mode kv_any = {
.kv_map = &kernel_map,
};
const struct kmem_va_mode kv_intrsafe = {
.kv_map = &kmem_map,
};
const struct kmem_va_mode kv_page = {
.kv_singlepage = 1
};
const struct kmem_pa_mode kp_dirty = {
.kp_constraint = &no_constraint
};
const struct kmem_pa_mode kp_dma = {
.kp_constraint = &dma_constraint
};
const struct kmem_pa_mode kp_dma_contig = {
.kp_constraint = &dma_constraint,
.kp_maxseg = 1
};
const struct kmem_pa_mode kp_dma_zero = {
.kp_constraint = &dma_constraint,
.kp_zero = 1
};
const struct kmem_pa_mode kp_zero = {
.kp_constraint = &no_constraint,
.kp_zero = 1
};
const struct kmem_pa_mode kp_pageable = {
.kp_object = &uvm.kernel_object,
.kp_pageable = 1
/* XXX - kp_nomem, maybe, but we'll need to fix km_free. */
};
const struct kmem_pa_mode kp_none = {
.kp_nomem = 1
};
const struct kmem_dyn_mode kd_waitok = {
.kd_waitok = 1,
.kd_prefer = UVM_UNKNOWN_OFFSET
};
const struct kmem_dyn_mode kd_nowait = {
.kd_prefer = UVM_UNKNOWN_OFFSET
};
const struct kmem_dyn_mode kd_trylock = {
.kd_trylock = 1,
.kd_prefer = UVM_UNKNOWN_OFFSET
};
|