page.c
26 KB
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
/* ***************************************************************************/
/* * */
/* * page.c - does some memory allocation, plus paging for "virtual memory" */
/* * */
/* ***************************************************************************/
/* This code was liberated from flashlite, hence the copyright: */
/* ************************************************************************* */
/* * * */
/* * Copyright (C) 1993-1998 Stanford University * */
/* * * */
/* * These coded instructions, statements, and computer programs contain * */
/* * unpublished proprietary information of Stanford University, and * */
/* * are protected by Federal copyright law. They may not be disclosed * */
/* * to third parties or copied or duplicated in any form, in whole or * */
/* * in part, without the prior written consent of Stanford University. * */
/* * * */
/* ************************************************************************* */
/*#define COMA*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "solo.h"
#include "simtypes.h"
#include "sim_error.h"
#include "solo_page.h"
#include "cp0.h"
#include "tcl_init.h"
#include "params.h"
#ifdef USE_FLASHLITE
# include "machine.h"
# ifdef FLASHPOINT
# include "mipsy_interface.h"
# include "flash_interface.h"
# endif
#endif
#include "pcache.h" /* So we know the MHT size */
#include "scache.h" /* So we know the MHT size */
#include "bstring.h"
#define MINIMUM_VA_VALUE 100
#define ONE_MEGABYTE 0x100000
/* padding so that comaHACKADDR
* is on a FLASH cacheline of its
* own. We need to avoid a situation
* where it is on the same cache line
* as soloActiveProcs...bad things
* happen if it is!!!
*/
static struct garbage {
int PAD[128];
int comaHACKADDR;
int PAD1[128];
} junk;
#ifdef COMA
void SysPlaceRangeCOMAHack(unsigned procNum, FLASHAddress addrToDelete);
#else
void SysPlaceRangeCOMAHack(unsigned procNum, FLASHAddress addrToDelete) {}
#endif
int soloPACompressNodeShift;
uint soloPACompressOffsetMask;
uint soloLockBase;
uint soloBarrierBase;
uint soloTotalMemory;
static int nextclus = 0; /* next processor to assign pages to */
static unsigned *cachepage; /* array of next cacheable space page numbers */
static int totalProcs; /* Number of processors on system */
static int totalMemPerProc; /* Memory per node */
static int applicationMemoryStart; /* ??? */
static enum { ROUND_ROBIN, FIRST_TOUCHED} allocationPolicy;
/* Global FlashLite declarations */
/* local function prototypes */
static void InitPageTrans(void);
static void InitMemAllocation(void);
static int NumBits(unsigned number);
static bool RemoveMapping(VA v_addr);
#ifdef FLASHPOINT
// I need to call this from solo_anl.c when using flashpoint
bool GetPhysicalAddr(VA v_addr, SoloPA *p_addr, uint *flavor);
SoloPA AllocatePage(int module, VA va);
void NewMapping(VA v_addr, SoloPA p_addr, uint flavor);
#else
static bool GetPhysicalAddr(VA v_addr, SoloPA *p_addr, uint *flavor);
static SoloPA AllocatePage(int module, VA va);
static void NewMapping(VA v_addr, SoloPA p_addr, uint flavor);
#endif
static void TrySysPlaceRange(int cpuNum, EventCallbackHdr *hdr, void *v);
static EventCallbackHdr sys_place_range_callback;
static int sys_place_range_tries;
static VA sys_place_range_start;
static VA sys_place_range_stop;
static int sys_place_range_node;
int sys_place_range_StallCPUS = 0;
static unsigned *nextPageFrame;
static char *protocol;
#define SYS_PLACE_RANGE_QUIESCENT_TIME 100000
#define SYS_PLACE_RANGE_MAX_TRIES 15
void
SoloInitPageTranslation(void)
{
InitMemAllocation();
InitPageTrans();
ParamLookup(&protocol, "MEMSYS.FLASH.Protocol", PARAM_STRING);
}
void
Solosys_place_range(VA start, VA stop, int node)
{
SoloPA pa;
VA va;
int oldNode;
uint flavor;
int i;
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "sys_place_range %d: Allocating range %x to %x\n",
node, start, stop);
#endif
pa = 0LL;
va = start & SOLO_PAGE_NUMBER_MASK;
do {
if (!GetPhysicalAddr(va, &pa, &flavor)) {
ASSERT(node < totalProcs);
pa = AllocatePage(node, va);
NewMapping( va, pa, TLB_CACHED);
} else {
oldNode = SOLO_PA_NODE(pa);
if (oldNode != node) {
#ifdef DEBUG_PAGE_VERBOSE
CPUWarning("sys_place_range %u: Moving page %x (%llx) from %d to %d\n",
node, va, pa, oldNode, node);
#endif
sys_place_range_StallCPUS = 1;
/* Tell all the CPU's not to do anything */
sys_place_range_tries = 0;
sys_place_range_start = va;
sys_place_range_stop = stop;
sys_place_range_node = node;
sys_place_range_callback.active = 0;
EventDoCallback(soloCPUNum, TrySysPlaceRange,
&sys_place_range_callback, 0,
SYS_PLACE_RANGE_QUIESCENT_TIME);
break; /* TrySysPlaceRange will loop for me */
}
}
va += SOLO_PAGE_SIZE;
} while (va < stop);
}
static void
TrySysPlaceRange(int cpuNum, EventCallbackHdr *hdr, void *v)
{
SoloPA pa, currentPA;
PA realPA;
FLASHAddress transit;
VA va, currentVA;
int cpu, i, j, result, wasDirty;
int oldNode;
uint flavor;
int node = sys_place_range_node;
#define LINE_BUFFER_SIZE 1024
char data[LINE_BUFFER_SIZE];
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "TrySysPlaceRange %d: Moving VA range %8.8x to %8.8x\n",
node, sys_place_range_start, sys_place_range_stop);
#endif
ASSERT(LINE_BUFFER_SIZE >= SCACHE_LINE_SIZE);
/* Check MHT's to see if anything is outstanding */
for (i=0; i<soloActiveProcs; i++) {
for (j = 0; j < MHT_SIZE; j++) {
if (CACHE[i].MHT[j].inuse) {
CPUPrint("sys_place_range %u: wakeup %d found MHT %d, entry %d in use\n",
cpuNum, sys_place_range_tries, i, j);
sys_place_range_tries++;
if (sys_place_range_tries >= SYS_PLACE_RANGE_MAX_TRIES) {
CPUError("sys_place_range %u: too many tries with MHT full\n",cpuNum);
}
EventDoCallback(cpuNum, TrySysPlaceRange,
&sys_place_range_callback, 0,
SYS_PLACE_RANGE_QUIESCENT_TIME);
return;
}
}
}
/* Ok, MHT's seem clean, so let's clean up the caches. Now when I find mappings
already established, I go ahead and move them. Grab on to your socks.*/
ASSERT(node < totalProcs);
va = sys_place_range_start;
do {
if (!GetPhysicalAddr(va, &pa, &flavor)) {
pa = AllocatePage(node, va);
NewMapping( va, pa, TLB_CACHED);
} else {
oldNode = SOLO_PA_NODE(pa);
if (oldNode != node) {
for (cpu=0; cpu<soloActiveProcs; cpu++) {
for (currentPA = pa, currentVA = va;
currentPA < pa + SOLO_PAGE_SIZE;
currentPA += SCACHE_LINE_SIZE,
currentVA += SCACHE_LINE_SIZE) {
realPA = SoloCompressAddr(currentPA);
if (CacheExtract(cpu, realPA, SCACHE_LINE_SIZE,
&wasDirty, (char *)data)) {
if (wasDirty) {
bcopy(data, (char *)currentVA, SCACHE_LINE_SIZE);
if (!strcmp(protocol,"COMA")) {
transit.ll = currentPA;
SysPlaceRangeCOMAHack(cpu, transit);
}
}
}
}
}
/* Do the actual remapping operation */
CPUPrint("sys_place_range %u: Moving page %x (%llx) to %d\n",
cpuNum, va, pa, node);
if (!RemoveMapping(va)) {
CPUError("RemoveMapping failed though mapping exists!\n");
}
pa = AllocatePage(node, va);
NewMapping(va, pa, TLB_CACHED);
CPUPrint("sys_place_range %u: VA %x now at PA %llx\n",
cpuNum, va, pa);
}
}
va += SOLO_PAGE_SIZE;
} while (va < sys_place_range_stop);
sys_place_range_StallCPUS = 0;
}
/* ****************************************************************************/
/* * V_to_P maps virtual to physical addresses **/
/* * - global cacheble space assigned round-robin **/
/* * - private cacheable space assigned to home cluster **/
/* ****************************************************************************/
SoloPA
SoloV_to_P(int proc, VA va, int clus, bool isFrame, uint *tlbFlavor)
{
SoloPA pa;
bool retVal;
uint flavor;
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "V_to_P %u: Translating %x", proc, va);
#endif
#ifdef T5_MODEL
pa = (LL) va;
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "\nV_to_P %u: Returning PA==VA\n", proc);
#endif
return pa;
#else
pa = 0LL;
if (va < MINIMUM_VA_VALUE) {
CPUError("\nV_to_P %d: Got virtual address 0x%x\n", proc,va);
}
/* Check for references to other spaces */
if (va & VIRTUAL_MSG_MASK) {
/* This is MSG space */
#ifdef USE_FLASHLITE
retVal = GetPhysicalAddr((va & 0x7fffffff), &pa, &flavor);
ASSERT(retVal); ASSERT(flavor == TLB_CACHED);
pa |= (MSG << SOLO_PA_SPACE_SHIFT);
*tlbFlavor = TLB_UNCACHED;
flavor = TLB_UNCACHED; /* this is so the (*tlbFlavor) below */
/* doesn't squash the uncached setting. */
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("\nMSG_SPACE Returning translation %16.16llx\n",
pa);
#endif
#else
CPUError("\nBad access to VA 0x%x\n",va);
#endif
} else if (!GetPhysicalAddr(va, &pa, &flavor)) {
int node;
if (isFrame == FALSE) { /* global */
if (allocationPolicy == ROUND_ROBIN) {
nextclus = ((va & SOLO_PAGE_NUMBER_MASK) >> SOLO_PAGE_SHIFT) &
(totalProcs - 1);
node = nextclus;
} else { /* first touched */
node = clus;
}
} else { /* private */
node = proc;
}
ASSERT(node < totalProcs);
pa = AllocatePage(node, va);
NewMapping(va, pa, TLB_CACHED);
flavor = TLB_CACHED;
}
(*tlbFlavor) = flavor;
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "Resulting translation --> %16.16llx\n", pa);
#endif
return pa;
#endif /* T5_MODEL */
}
bool
SoloPageHasBeenMapped(VA vaddr)
{
SoloPA paddr;
uint flavor;
return (GetPhysicalAddr(vaddr, &paddr, &flavor));
}
void
SoloEstablishMapping(SoloPA pa, VA va, int space, uint tlbFlavor)
{
SoloPA addr;
uint flavor;
/* Make sure we don't have a normal mapping of this virtual address */
/* to some other physical address */
if (GetPhysicalAddr(va, &addr, &flavor)) {
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "EstablishMapping: flushing old mapping!\n");
#endif
if (!RemoveMapping(va)) {
CPUError("RemoveMapping failed though old mapping exists!\n");
}
}
addr = SOLO_FORM_PA(SOLO_PA_NODE(pa),space,SOLO_PA_OFFSET(pa));
/* Be sure space is set in mapping */
NewMapping(va, addr, tlbFlavor);
}
int
SoloGetHome(unsigned vaddr)
{
SoloPA paddr;
uint flavor;
if (GetPhysicalAddr(vaddr, &paddr, &flavor) == FALSE) {
return -1;
}
else {
return SOLO_PA_NODE(paddr);
}
}
/* *************************************************************************** */
/* * AllocatePage allocates the next page of cacheable space * */
/* *************************************************************************** */
SoloPA
AllocatePage(int module, VA va)
{
VA pageFrame;
unsigned numFrames;
PA offset;
/* Used for COMA only */
LL current, upperBound, lowerBound;
unsigned reservedMemoryPerNode = 0;
if (!strcmp(protocol,"COMA")) {
ParamLookup(&reservedMemoryPerNode, "MEMSYS.FLASH.ReservedMemoryPerNode",
PARAM_INT);
}
ASSERT(module < totalProcs);
numFrames = ONE_MEGABYTE >> SOLO_PAGE_SHIFT;
/* pageFrame = (va & (ONE_MEGABYTE - 1)) >> SOLO_PAGE_SHIFT;*/
if (nextPageFrame[module] >= soloLockBase) {
nextPageFrame[module] = 0;
}
pageFrame = nextPageFrame[module];
nextPageFrame[module] = (nextPageFrame[module] + 1) %
(ONE_MEGABYTE >> SOLO_PAGE_SHIFT);
ASSERT(cachepage[(module * numFrames) + pageFrame] !=
(totalMemPerProc/ONE_MEGABYTE));
cachepage[(module * numFrames) + pageFrame]++;
#define LOG_ONE_MB 20
if (!strcmp(protocol,"COMA")) {
if (reservedMemoryPerNode > 0) {
lowerBound = (LL)((module*reservedMemoryPerNode)%totalMemPerProc);
upperBound = (LL)(((module*reservedMemoryPerNode)%totalMemPerProc) +
reservedMemoryPerNode);
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("LOOK: lower reserved bound is %llx, "
"curr = %llx, upper bound is %llx\n", lowerBound,
((LL)(((cachepage[(module * numFrames) + pageFrame] - 1) <<
LOG_ONE_MB) + (pageFrame << SOLO_PAGE_SHIFT))),
upperBound);
#endif
while ((lowerBound <=
(((cachepage[(module * numFrames) + pageFrame] - 1) << LOG_ONE_MB) +
(pageFrame << SOLO_PAGE_SHIFT) + applicationMemoryStart)) &&
((((cachepage[(module * numFrames) + pageFrame] - 1) << LOG_ONE_MB) +
(pageFrame << SOLO_PAGE_SHIFT) + applicationMemoryStart) <
upperBound)) {
/* complicated, but the gist of it is that you fell into
the reserved memory portion of memory.
increment the stripe so that this is no longer true */
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("LOOK: lower reserved bound is %llx, "
"curr = %llx, upper bound is %llx\n",lowerBound,
((LL)(((cachepage[(module * numFrames) + pageFrame] - 1) <<
LOG_ONE_MB) + (pageFrame << SOLO_PAGE_SHIFT) +
applicationMemoryStart)),
upperBound);
CPUPrint("NextCacheablePage reserved memory conflict: "
"module = %d, pageFrame = %X, count = %x, MemPerProc = %X\n",
module, pageFrame, cachepage[(module * numFrames)+pageFrame],
totalMemPerProc);
#endif
ASSERT(cachepage[(module * numFrames) + pageFrame] !=
(totalMemPerProc/ONE_MEGABYTE));
cachepage[(module * numFrames) + pageFrame]++;
}
}
}
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint( "AllocatePage: module = %d, pageFrame = %X, count = %x, MemPerProc = %X\n", module, pageFrame, cachepage[(module * numFrames)+pageFrame],
totalMemPerProc);
#endif
offset = (((PA)(((cachepage[(module * numFrames) + pageFrame] - 1) << 20)
+ (pageFrame << SOLO_PAGE_SHIFT))) + applicationMemoryStart) +
(va & SOLO_PAGE_OFFSET_MASK);
if ((PA) offset > (PA)soloLockBase) {
CPUError("Mipsy: application brk() would have encroached on lock space:\nCurrent heap top: %8.8x soloLockBase: %8.8x\n",
offset, soloLockBase);
}
return SOLO_FORM_PA(module,0,offset);
}
/* *************************************************************************** */
/* * InitMemAllocation initializes the per-module page counters * */
/* *************************************************************************** */
static void
InitMemAllocation(void)
{
unsigned i;
unsigned j;
unsigned numFrames;
int membits;
char *buf;
applicationMemoryStart = 0;
totalProcs = TOTAL_CPUS;
ParamLookup(&totalMemPerProc, "MEMSYS.FLASH.TotalMemPerProc", PARAM_INT);
/*
* Compute values needed for compress/decompress of Solo 64 bit addresses
* into 32bit PAs. We do this by dividing the 32 bits into a node and
* memory offset field. To preserve the cache coloring we do we require
* that the memory offset value at least enough bits for the cache color.
*/
membits = NumBits(totalMemPerProc+applicationMemoryStart);
if (membits < NumBits(ONE_MEGABYTE-1)) {
membits = NumBits(ONE_MEGABYTE-1);
}
if ((NumBits(totalProcs)+membits) > sizeof(PA)*8) {
CPUError("%d processors with %d bytes of memory is too much\n",
totalProcs, totalMemPerProc);
}
soloPACompressNodeShift = membits;
soloPACompressOffsetMask = (1 << soloPACompressNodeShift)-1;
soloTotalMemory = totalMemPerProc;
soloLockBase = totalMemPerProc - 64*1024;
/* Reserve 32K for space 1 */
soloBarrierBase = totalMemPerProc - 32*1024;
/* Reserve 32K for space 2 */
ParamLookup(&buf, "MEMSYS.FLASH.PageAllocationPolicy", PARAM_STRING);
if (strcasecmp(buf, "ROUND_ROBIN") == 0) {
allocationPolicy = ROUND_ROBIN;
} else if (strcasecmp(buf, "FIRST_TOUCHED") == 0) {
allocationPolicy = FIRST_TOUCHED;
} else {
CPUError("Unknown page allocation policy %s\n", buf);
}
numFrames = ONE_MEGABYTE >> SOLO_PAGE_SHIFT;
cachepage = (unsigned *) malloc (numFrames*sizeof(int)*totalProcs);
if (cachepage == (unsigned *) NULL) {
CPUError("Out of memory in SoloInitMemAllocation");
}
nextPageFrame = (unsigned *) malloc(sizeof(unsigned)*totalProcs);
if (nextPageFrame == (unsigned *) NULL) {
CPUError("Out of memory in SoloInitMemAllocation");
}
/* starting page for each cluster's cacheable space */
for (i = 0; i < totalProcs; i++) {
/* NOTE: This isn't so good for P > 256 processors since numFrames
* is 256 */
nextPageFrame[i] = ((numFrames/totalProcs) % numFrames) * i;
for (j=0; j < numFrames; j++) {
cachepage[(i * numFrames) + j] = 0;
}
}
}
typedef struct mapping_struct mapping_pair;
struct mapping_struct
{
VA va; /* virtual address */
SoloPA pa; /* physical address */
int flavor; /* flavor of mapping */
mapping_pair *v_next; /* ptr to next virtual mapping_pair;
nil if last mapping_pair */
mapping_pair *p_next; /* ptr to next physical mapping_pair;
nil if last mapping_pair */
};
static mapping_pair *v_map[MAPPING_HASH_SIZE]; /* array of lists that make up tlb */
static mapping_pair *p_map[MAPPING_HASH_SIZE]; /* array of lists for reverse map */
/* WARNING -> this routine is terribly slow due to the compiler not
handling long long worth squat. I've hard coded the routine in
assembly */
/*
#define mapping_hash(_a) (((_a) >> SOLO_PAGE_SHIFT) % MAPPING_HASH_SIZE)
*/
extern unsigned mapping_hash(unsigned long long addr);
/* InitPageTrans initializes address translation. This is not necessary if */
/* a map is about to be read in, as ReadMap also performs the necessary */
/* initialization. */
static void
InitPageTrans(void)
{
int i;
for (i = 0; i < MAPPING_HASH_SIZE; i++) {
v_map[i] = 0;
p_map[i] = 0;
}
}
/* NewMapping takes care of the map hash table for a new virtual-to- */
/* physical mapping. It inserts the new mapping at the head of the */
/* appropriate virtual hash bin. The user is responsible for ensuring */
/* that the mapping does not already exist before calling this routine. */
void
NewMapping(VA v_addr, SoloPA p_addr, uint flavor)
/* cpu #(private); SHAREDBLK(shared) */
/* virtual address */
/* physical block number */
/* address space type */
{
VA my_virtual;
SoloPA physical;
mapping_pair *current;
unsigned v_hash_index, p_hash_index;
VA alignedVA;
FLASHAddress alignedPA;
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("New Mapping: va %#x pa %#16.16llx\n",
v_addr, p_addr);
#endif
my_virtual = v_addr & SOLO_PAGE_NUMBER_MASK;
physical = p_addr & (SoloPA)SOLO_PAGE_NUMBER_MASK_LONG;
/* create new mapping mapping_pair */
if ((current = (mapping_pair *) malloc(sizeof(mapping_pair))) == 0) {
CPUError("malloc of new mapping mapping_pair failed\n");
}
current->va = my_virtual;
current->pa = physical;
current->flavor = flavor;
/* insert new mapping into hash table */
v_hash_index = mapping_hash(my_virtual);
current->v_next = v_map[v_hash_index];
v_map[v_hash_index] = current;
p_hash_index = mapping_hash(physical);
current->p_next = p_map[p_hash_index];
p_map[p_hash_index] = current;
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("New Mapping: VA: %8.8x PA: %16.16llx\n",
v_addr, p_addr);
#endif
}
/* GetPhysicalAddr performs a virtual-to-physical address translation */
/* using the map hash table. TRUE is returned if the translation */
/* was successful. */
bool
GetPhysicalAddr(VA v_addr, SoloPA *p_addr, uint *flavor)
{
VA my_virtual;
mapping_pair *current;
unsigned hash_index;
my_virtual = v_addr & SOLO_PAGE_NUMBER_MASK;
hash_index = mapping_hash(my_virtual);
/* search for this virtual page's mapping in hash table */
current = v_map[hash_index];
while (current != 0) {
if (current->va == my_virtual) {
*p_addr = current->pa + (v_addr & SOLO_PAGE_OFFSET_MASK);
*flavor = current->flavor;
return TRUE;
}
current = current->v_next;
}
return FALSE;
}
/* GetMemoryAddr performs a physical-to-virtual address translation */
/* using the map hash table. TRUE is returned if the translation */
/* was successful. */
char *
SoloGetMemoryAddr(SoloPA p_addr)
{
SoloPA physical;
mapping_pair *current;
unsigned hash_index;
unsigned offset;
VA v_addr;
/* ADDED PURELY BECAUSE OF THE COMA PROBLEM
* I DESCRIBE BELOW
*/
unsigned fake_coma_i = 0;
unsigned fake_coma_offset;
unsigned fake_coma_space = 0;
SoloPA new_node_id;
/*
// because we initialize all directory entries to be EXCL at
// (fake) boot time, when you access data, it may return and force
// displacement of something that has not even been accessed
// yet (this displaced data would have its state at EXCL due to
// the initialization, so it would be thought that replacement is
// necessary). In this case, it will not have a valid VA->PA translation
// so the above test will fail and the simulation will die
// unnecessarily. The hack I am using to avoid this is to pick
// a different PA to access if the current one has no translation.
// We print a warning to make sure that if it is a real bug we
// catch it.
*/
physical = p_addr & (SoloPA)SOLO_PAGE_NUMBER_MASK_LONG;
hash_index = mapping_hash(physical);
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("GetVirtualAddr: physical: %16.16llx\n",
p_addr);
#endif
#ifdef T5_MODEL
v_addr = (VA) (p_addr & 0xffffffffLL);
/* Extract low bits... */
CPUPrint("GetVirtualAddr: Returning VA==PA: %8.8x\n",
v_addr);
return (char*) v_addr;
#else
/* search for this virtual page's mapping in hash table */
current = p_map[hash_index];
while (current != 0) {
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("GetVirtualAddr: VA: %8.8x PA: %16.16llx args: physical %llx\n",
current->va, current->pa, physical);
#endif
if (current->pa == physical) {
v_addr = current->va;
offset = (unsigned) (p_addr & (SoloPA)SOLO_PAGE_OFFSET_MASK_LONG);
v_addr |= offset;
return (char *) v_addr;
}
else current = current->p_next;
}
if (!strcmp(protocol,"COMA")) {
/* CPUWarning("Bad argument passed to GetMemoryAddress\n");*/
/*
// CPUWarning("Bad argument passed to GetMemoryAddress, addr = %16.16llx\n",
// physical);
// COMA HACK...SEE COMMENT AT TOP OF FUNCTION
// we actually choose a known valid VA so we do
// not accidentally overwrite a valid PA. We
// should do extra checks to make sure we are
// doing this because of our replacement problem
// rather than because there is an actual bug
*/
while (fake_coma_i < totalProcs){
fake_coma_space = SOLO_PA_SPACE(p_addr);
fake_coma_offset = SOLO_PA_OFFSET(p_addr);
/* we choose a different physical address: same node number, */
/* CPUPrint("Checking for COMA rplc case, node %u, curr_addr = %16.16llx\n",
fake_coma_i, physical); */
p_addr = SOLO_FORM_PA(fake_coma_i, fake_coma_space, fake_coma_offset);
physical = p_addr & (SoloPA)SOLO_PAGE_NUMBER_MASK;
hash_index = mapping_hash(physical);
/* search for this virtual page's mapping in hash table */
current = p_map[hash_index];
while (current != 0) {
if (current->pa == physical) {
v_addr = current->va;
offset = (unsigned) (p_addr & (SoloPA)SOLO_PAGE_OFFSET_MASK);
v_addr |= offset;
/* return (char *) v_addr; */
/* here, we realize this is probably due to the
* COMA replacement problem, so we return the
* fake VA that we set up above
*/
return (char *)&(junk.comaHACKADDR);
}
else current = current->p_next;
}
fake_coma_i++;
}
}
return NULL;
#endif /* T5_MODEL */
}
static int
NumBits(unsigned int number)
{
int i;
for (i = 0; i < 32; i++) {
if ((1 << i) > number)
return i;
}
return 32;
}
/* RemoveMapping removes an existing mapping for a VA. */
static bool
RemoveMapping(VA v_addr)
{
VA my_virtual;
mapping_pair *current, *prev;
int hash_index;
bool found;
my_virtual = v_addr & SOLO_PAGE_NUMBER_MASK;
hash_index = mapping_hash(my_virtual);
/* search for this virtual page's mapping in hash table */
current = v_map[hash_index];
prev = 0; /* Indicate first element is head */
found = FALSE;
while ((!found) && (current != 0)) {
if (current->va == my_virtual) {
found = TRUE;
break;
} else {
prev = current;
current = current->v_next;
}
}
if (found) {
#ifdef DEBUG_PAGE_VERBOSE
CPUPrint("RemoveMapping: VA: %8.8x PA: %16.16llx\n",
current->va, current->pa);
#endif
if (prev == 0) {
v_map[hash_index] = current->v_next;
/* Remove current from hash bucket head */
} else {
prev->v_next = current->v_next;
/* Remove current from prev's chain */
}
}
return found;
}