Cortex-A8 Technical Reference Manual - ARM Information Center

More documents

Recommendations

Info

13.1 About the NEON and VFP programmers model 13.1.1 NEON media coprocessor 13.1.2 VFP coprocessor NEON and VFP Programmers Model The processor implements both the ARM Advanced SIMD and VFPv3 architectures. See the ARM Architecture Reference Manual for information on the Advanced SIMD and VFPv3 instruction sets. Note This chapter uses the older assembler language instruction mnemonics. See Appendix B Instruction Mnemonics for information about the Unified Assembler Language (UAL) equivalents of the Advanced SIMD and VFP data-processing instruction mnemonics. See the ARM Architecture Reference Manual for more information on the UAL syntax. The NEON coprocessor implements the Advanced SIMD media processing architecture. Advanced SIMD is an optional part of the ARMv7-A architecture. The components of the NEON coprocessor are: • NEON register file with 32x64-bit general-purpose registers • NEON integer execute pipeline (ALU, Shift, MAC) • NEON dual, single-precision floating-point execute pipeline (FADD, FMUL) • NEON load/store and permute pipeline • nonpipelined VFP coprocessor that implements VFPv3 data-processing floating-point operations. The NEON coprocessor can receive up to two valid Advanced SIMD instructions per cycle from the ARM integer instruction execute unit. In addition, it can receive 32-bit MCR data from or send 32-bit MRC data to the ARM integer instruction execute unit. The NEON coprocessor can load data from either the L1 data cache or the L2 memory system. Enable L1 data caching for best performance of the NEON coprocessor when the L2 memory system is off or not present. See c1, Auxiliary Control Register on page 3-47. The VFP coprocessor implements the VFPv3 architecture. The VFP coprocessor provides a floating-point computation coprocessor that is fully compliant with the ANSI/IEEE Std 754-1985, IEEE Standard for Binary Floating-Point Arithmetic, referred to in this document as the IEEE 754 standard. The VFP coprocessor supports all data-processing instructions and data types in the VFPv3 architecture and is described in the ARM Architecture Reference Manual. Designed for the processor, the VFP coprocessor fully supports single-precision and double-precision add, subtract, multiply, divide, multiply and accumulate, and square root operations. Conversions between fixed-point and floating-point data formats, and floating-point constant instruction are provided. ARM DDI 0344K Copyright © 2006-2010 ARM Limited. All rights reserved. 13-2 ID060510 Non-Confidential
13.2 General-purpose registers 13.2.1 NEON views of the register bank 13.2.2 VFP views of the register bank NEON and VFP Programmers Model The NEON and VFP coprocessor shares the same register bank. This is distinct from the ARM register bank. You can reference the NEON and VFP register bank using three explicitly aliased views, as described in the following sections. Figure 13-1 shows the three views of the register bank and the way the word, doubleword, and quadword registers overlap. NEON views the register bank as: • Sixteen 128-bit quadword registers, Q0-Q15. • Thirty-two 64-bit doubleword registers, D0-D31. This view is also available in VFP. • A combination of these 128-bit and 64-bit registers, Q0-Q15 and D0-D31. In VFP, you can view the register bank as: • Thirty-two 64-bit doubleword registers, D0-D31. This view is also available in NEON. • Thirty-two 32-bit single word registers, S0-S31. Only half of the register bank is accessible in this view. • A combination of these 128-bit and 64-bit registers, D0-D31 and S0-S31. The mapping between the registers is as follows: • S maps to the least significant half of D Figure 13-1 NEON and VFP register bank ARM DDI 0344K Copyright © 2006-2010 ARM Limited. All rights reserved. 13-3 ID060510 Non-Confidential S0 S1 S2 S3 S4 S5 S6 S7 S28 S29 S30 S31 D0 D1 D2 D3 D14 D15 D16 D17 D30 D31 Q0 Q1 Q7 Q8 Q15
Page 1 and 2:
Copyright © 2006-2010 ARM Limited.
Page 3 and 4:
Web Address http://www.arm.com ARM
Page 5 and 6:
Contents 2.14 The program status re
Page 7 and 8:
Contents 16.6 Instruction-specific
Page 9 and 10:
List of Tables Table 3-16 Debug Fea
Page 11 and 12:
List of Tables Table 3-136 Secure o
Page 13 and 14:
List of Tables Table 12-59 Values t
Page 15 and 16:
List of Figures Cortex-A8 Technical
Page 17 and 18:
List of Figures Figure 3-77 BTB arr
Page 19 and 20:
List of Figures Figure 15-16 CTI Ch
Page 21 and 22:
About this manual Product revision
Page 23 and 24:
old Highlights interface elements,
Page 25 and 26:
Feedback Feedback on the product Fe
Page 27 and 28:
1.1 About the processor Introductio
Page 29 and 30:
1.3 Components of the processor L1
Page 31 and 32:
1.3.6 NEON 1.3.7 ETM The NEON unit
Page 33 and 34:
1.5 Debug Introduction The processo
Page 35 and 36:
1.7 Configurable options Table 1-1
Page 37 and 38:
Introduction • The various data a
Page 39 and 40:
1.9.6 r2p0-r2p1 1.9.7 r2p1-r2p2 1.9
Page 41 and 42:
Chapter 2 Programmers Model This ch
Page 43 and 44:
2.2 Thumb-2 instruction set Program
Page 45 and 46:
Bits Field Function Programmers Mod
Page 47 and 48:
2.4 Jazelle Extension 2.4.1 Jazelle
Page 49 and 50:
2.5 Security Extensions architectur
Page 51 and 52:
2.6 Advanced SIMD architecture Prog
Page 53 and 54:
2.8 Processor operating states 2.8.
Page 55 and 56:
2.10 Memory formats 2.10.1 Byte-inv
Page 57 and 58:
2.12 Operating modes Programmers Mo
Page 59 and 60:
System and User r0 r1 r2 r3 r4 r5 r
Page 61 and 62:
2.14 The program status registers 2
Page 63 and 64:
2.14.5 The GE[3:0] bits Programmers
Page 65 and 66:
Mode bits Programmers Model M[4:0]
Page 67 and 68:
2.15 Exceptions 2.15.1 Exception en
Page 69 and 70:
2.15.5 Interrupt request 2.15.6 Abo
Page 71 and 72:
This restores both the PC and the C
Page 73 and 74:
2.15.13 Exception priorities Progra
Page 75 and 76:
2.17 Hardware consideration for Sec
Page 77 and 78:
SECMONOUT[78] instruction caches at
Page 79 and 80:
Chapter 3 System Control Coprocesso
Page 81 and 82:
System Control Coprocessor Register
Page 83 and 84:
3.1.3 MMU control and configuration
Page 85 and 86:
3.2 System control coprocessor regi
Page 87 and 88:
CRn Op1 CRm Op2 Register or operati
Page 89 and 90:
Page 91 and 92:
Page 93 and 94:
Page 95 and 96:
Page 97 and 98:
3.2.2 c0, Main ID Register System C
Page 99 and 100:
3.2.4 c0, TCM Type Register 3.2.5 c
Page 101 and 102:
3.2.7 c0, Processor Feature Registe
Page 103 and 104:
• accessible in privileged modes
Page 105 and 106:
System Control Coprocessor Table 3-
Page 107 and 108:
Bits Field Function [11:8] L1 Harva
Page 109 and 110:
Bits Field Function [11:8] Harvard
Page 111 and 112:
31 28 27 24 23 20 19 16 15 12 11 8
Page 113 and 114:
Table 3-30 shows the results of att
Page 115 and 116:
System Control Coprocessor 31 28 27
Page 117 and 118:
Table 3-36 shows the results of att
Page 119 and 120:
• accessible in privileged modes
Page 121 and 122:
CSSR Size 3.2.24 c0, Cache Size Sel
Page 123 and 124:
System Control Coprocessor 31 30 29
Page 125 and 126:
MCR p15, 0, , c1, c0, 0 ; Write Con
Page 127 and 128:
Bits Field [19] Clock stop request
Page 129 and 130:
Bits Field Security State NS S Tabl
Page 131 and 132:
3.2.28 c1, Secure Configuration Reg
Page 133 and 134:
AW EA Function Table 3-55 shows the
Page 135 and 136:
Page 137 and 138:
3.2.32 c2, Translation Table Base R
Page 139 and 140:
System Control Coprocessor Figure 3
Page 141 and 142:
Bits Field Function 3.2.35 c5, Data
Page 143 and 144:
3.2.36 c5, Instruction Fault Status
Page 145 and 146:
3.2.38 c6, Data Fault Address Regis
Page 147 and 148:
System Control Coprocessor Note •
Page 149 and 150:
System Control Coprocessor Figure 3
Page 151 and 152:
Page 153 and 154:
3.2.41 c8, TLB operations The data
Page 155 and 156:
Figure 3-38 shows the bit arrangeme
Page 157 and 158:
EN b 3.2.44 c9, Count Enable Clear
Page 159 and 160:
EN b 3.2.46 c9, Software Increment
Page 161 and 162:
EN b 3.2.48 c9, Cycle Count Registe
Page 163 and 164:
EN b Table 3-96 shows the results o
Page 165 and 166:
Value Description 0x44 Any cacheabl
Page 167 and 168:
DBGEN || NIDEN 3.2.51 c9, User Enab
Page 169 and 170:
EN 3.2.53 c9, Interrupt Enable Clea
Page 171 and 172:
Figure 3-48 shows the bit arrangeme
Page 173 and 174:
3.2.55 c9, L2 Cache Auxiliary Contr
Page 175 and 176:
Bits Field Function [8:6] Tag RAM l
Page 177 and 178:
TL bit value 3.2.57 c10, TLB preloa
Page 179 and 180:
Page 181 and 182:
Page 183 and 184:
The PLE Identification and Status R
Page 185 and 186:
Page 187 and 188:
3.2.62 c11, PLE enable commands U b
Page 189 and 190:
Bits Field Function System Control
Page 191 and 192:
U bit PLE bit Table 3-129 shows the
Page 193 and 194:
U bit PLE bit System Control Coproc
Page 195 and 196:
3.2.68 c12, Secure or Nonsecure Vec
Page 197 and 198:
3.2.70 c12, Interrupt Status Regist
Page 199 and 200:
The FCSE PID Register is: • a rea
Page 201 and 202:
Page 203 and 204:
TLB CAM read/write TLB ATTR read/wr
Page 205 and 206:
MCR p15 0, , c15, c3, 4 ; I-L1 TLB
Page 207 and 208:
31 27 26 22 21 0 Reserved L1 Data 0
Page 209 and 210:
System Control Coprocessor The L1 H
Page 211 and 212:
3.2.78 c15, L1 data array operation
Page 213 and 214:
Instruction L1 Data 0 register Inst
Page 215 and 216:
Parity/ECC RAM read/write Data RAM
Page 217 and 218:
3.2.82 c15, L2 parity/ECC array ope
Page 219 and 220:
3.2.84 c15, L2 data array operation
Page 221 and 222:
4.1 About unaligned and mixed-endia
Page 223 and 224:
Unaligned Data and Mixed-endian Dat
Page 225 and 226:
Chapter 5 Program Flow Prediction T
Page 227 and 228:
5.2 Predicted instructions Program
Page 229 and 230:
Program Flow Prediction Because ret
Page 231 and 232:
5.4 Guidelines for optimal performa
Page 233 and 234:
5.6 Operating system and predictor
Page 235 and 236:
6.1 About the MMU Memory Management
Page 237 and 238:
6.3 16MB supersection support Memor
Page 239 and 240:
6.5 External aborts 6.5.1 External
Page 241 and 242:
6.7 MMU software-accessible registe
Page 243 and 244:
7.1 About the L1 memory system Leve
Page 245 and 246:
7.2.6 Instruction cache maintenance
Page 247 and 248:
L1 inner policy a L2 outer policy N
Page 249 and 250:
7.5 Data cache features 7.5.1 Data
Page 251 and 252:
7.7 Hardware support for virtual al
Page 253 and 254:
Chapter 8 Level 2 Memory System Thi
Page 255 and 256:
8.2 Cache organization 8.2.1 L2 cac
Page 257 and 258:
8.3 Enabling and disabling the L2 c
Page 259 and 260:
Level 2 Memory System Note It is en
Page 261 and 262:
Level 2 Memory System Note You must
Page 263 and 264:
Lock address : LockAddr Lock free :
Page 265 and 266:
8.7 Parity and error correction cod
Page 267 and 268:
9.1 About the external memory inter
Page 269 and 270:
Integer data and CP14 writes Noncac
Page 271 and 272:
9.4 AXI data read/write transaction
Page 273 and 274:
Noncacheable, or strongly ordered,
Page 275 and 276:
Noncacheable store word Noncacheabl
Page 277 and 278:
LA Last Access External Memory Inte
Page 279 and 280:
10.1 Clock domains 10.1.1 AXI clock
Page 281 and 282:
10.2 Reset domains 10.2.1 Power-on
Page 283 and 284:
REFCLK (PLL input) nPORESET ARESETn
Page 285 and 286:
10.3 Power control 10.3.1 Dynamic p
Page 287 and 288:
Clock, Reset, and Power Control the
Page 289 and 290:
ATB APB I/O clamp I/O clamp L/S = L
Page 291 and 292:
ATB APB I/O clamp I/O clamp LS = Le
Page 293 and 294:
Clock, Reset, and Power Control 3.
Page 295 and 296:
Powering up the debug and ETM domai
Page 297 and 298:
7. Perform a normal software reset
Page 299 and 300:
Clock, Reset, and Power Control 5.
Page 301 and 302:
Chapter 11 Design for Test This cha
Page 303 and 304:
In L1 MBIST Instruction Register Fi
Page 305 and 306:
Note Do not test the CAMBIST arrays
Page 307 and 308:
• number of rows of the L2 data,
Page 309 and 310:
Design for Test Not all row setting
Page 311 and 312:
In Design for Test Figure 11-4 show
Page 313 and 314:
Design for Test When testing the ta
Page 315 and 316:
CLK ARESETn MBISTMODE MBISTSHIFT MB
Page 317 and 318:
End-of-test datalog retrieval Desig
Page 319 and 320:
Pattern N RWRXMARCH 8N Row-fast Sta
Page 321 and 322:
4. rscan array, data_seed = invert.
Page 323 and 324:
3. R_, W, R, decr. 4. rscan data fr
Page 325 and 326:
• the data after pass 1 of XADDRB
Page 327 and 328:
(1 + (2 17)) 2 17 = 4,587,520 cycle
Page 329 and 330:
processor input ports WEXTEST WINTE
Page 331 and 332:
Design for Test toggling during shi
Page 333 and 334:
12.1 Debug systems 12.1.1 Debug hos
Page 335 and 336:
12.2.3 Security extensions and debu
Page 337 and 338:
Instruction Mnemonic Description 12
Page 339 and 340:
12.3.6 Power domains and debug 12.3
Page 341 and 342:
Table 12-5 shows the APB interface
Page 343 and 344:
12.4 Debug register descriptions Te
Page 345 and 346:
MRC p14, 0, , c0, c0, 0 ; Read Debu
Page 347 and 348:
31 30 29 28 27 26 25 24 23 22 21 20
Page 349 and 350:
Bits Field Function [21:20] DTR acc
Page 351 and 352:
Bits Field Function To access the D
Page 353 and 354:
12.4.7 Watchpoint Fault Address Reg
Page 355 and 356:
Bits Access Normal address [2] RW V
Page 357 and 358:
12.4.12 Debug Run Control Register
Page 359 and 360:
Table 12-23 shows how the bit value
Page 361 and 362:
BVR[22:20] Meaning b011 The corresp
Page 363 and 364:
Bits Field Function [15:14] Secure
Page 365 and 366:
12.4.18 Operating System Lock Statu
Page 367 and 368:
Debug • The sequence can be aband
Page 369 and 370:
12.5 Management registers Offset Th
Page 371 and 372: Offset Register number Mnemonic Fun
Page 373 and 374: Bits Field Function [5] nDMAIRQ nDM
Page 375 and 376: 12.5.7 Claim Tag Clear Register 12.
Page 377 and 378: 12.5.10 Authentication Status Regis
Page 379 and 380: Table 12-45 shows fields that are i
Page 381 and 382: 12.6 Debug events 12.6.1 Software d
Page 383 and 384: 12.6.5 Watchpoint debug events If a
Page 385 and 386: Table 12-53 shows the values in the
Page 387 and 388: 12.8 Debug state 12.8.1 Entering de
Page 389 and 390: 12.8.4 Writing to the CPSR in debug
Page 391 and 392: Mode SCR[0] For CP15 instructions,
Page 393 and 394: 12.8.9 Leaving debug state Imprecis
Page 395 and 396: 12.9.3 Cache usage profiling Debug
Page 397 and 398: Note DBGPWRDWNREQ must be tied LOW
Page 399 and 400: If software running on the processo
Page 401 and 402: Rules for accessing the DCC At the
Page 403 and 404: } While the processor is running, i
Page 405 and 406: Debug Table 12-59 Values to write t
Page 407 and 408: 12.11.4 Debug state entry Example 1
Page 409 and 410: } // Step 1. Update the CPSR value
Page 411 and 412: Writing the CPSR in debug state Exa
Page 413 and 414: Debug Example 12-21 Reading a word
Page 415 and 416: } // Step 9. Check for aborts. abor
Page 417 and 418: 12.12 Debugging systems with energy
Page 419 and 420: MOV r0, r4 POP {r4, pc} Example 12-
Page 421: Chapter 13 NEON and VFP Programmers
Page 425 and 426: 13.3 Short vectors 13.3.1 About reg
Page 427 and 428: 13.3.2 Operations using register ba
Page 429 and 430: NEON and VFP Programmers Model The
Page 431 and 432: Note All hardware ID information is
Page 433 and 434: Bits Field Function [12:8] - Reserv
Page 435 and 436: NEON and VFP Programmers Model Tabl
Page 437 and 438: 13.6 Compliance with the IEEE 754 s
Page 439 and 440: Underflow NEON and VFP Programmers
Page 441 and 442: 14.1 About the ETM 14.1.1 ETM featu
Page 443 and 444: 14.1.3 NEON Bridge and bus matrix A
Page 445 and 446: 14.3 ETM register summary The ETM r
Page 447 and 448: Bits Field Function [11:8] Major ET
Page 449 and 450: 14.4.4 Peripheral Identification Re
Page 451 and 452: See the ETM Architecture Specificat
Page 453 and 454: Embedded Trace Macrocell Table 14-1
Page 457 and 458: 14.5.3 Enabling events 14.5.4 Addre
Page 459 and 460: 14.7 Context ID tracing Embedded Tr
Page 461 and 462: 14.9 Idle state control Embedded Tr
Page 465 and 466: Chapter 15 Cross Trigger Interface
Page 467 and 468: CTICHIN[0] CTICHIN[1] CTICHIN[2] CT
Page 469 and 470: 15.2 Trigger inputs and outputs Tri
Page 471 and 472: 15.3 Connecting asynchronous channe
Page 473 and 474:
15.5 CTI register summary Address o
Page 475 and 476:
15.6 CTI register descriptions This
Page 477 and 478:
15.6.4 CTI Application Trigger Clea
Page 479 and 480:
Table 15-10 shows how the bit value
Page 481 and 482:
15.6.12 ASIC Control Register, ASIC
Page 483 and 484:
15.7 CTI Integration Test Registers
Page 485 and 486:
15.7.4 ITTRIGOUTACK, 0xEF0 15.7.5 I
Page 487 and 488:
Cross Trigger Interface Table 15-26
Page 489 and 490:
15.8.3 Device Type Identifier, 0xFC
Page 491 and 492:
Cross Trigger Interface Table 15-31
Page 493 and 494:
16.1 About instruction cycle timing
Page 495 and 496:
Instruction Cycle Timing Example 16
Page 497 and 498:
Table 16-4 shows the operation of m
Page 499 and 500:
Table 16-9 shows the operation of l
Page 501 and 502:
d. See Load/store instructions on p
Page 503 and 504:
Instruction Cycle Timing Table 16-1
Page 505 and 506:
16.4 Other pipeline-dependent laten
Page 507 and 508:
16.4.5 Conditional instructions Ins
Page 509 and 510:
16.6 Instruction-specific schedulin
Page 511 and 512:
Instruction Cycle Timing VNEG Dd,Dm
Page 513 and 514:
VMLA a VMLS a VMLAa VMLSa VQDMLAa V
Page 515 and 516:
VSLI VSRI 16.6.5 Advanced SIMD floa
Page 517 and 518:
16.6.6 Advanced SIMD byte permute i
Page 519 and 520:
Instruction Table 16-23 shows the o
Page 521 and 522:
Instruction VST3 3-reg (unaligned)
Page 523 and 524:
Instruction VLD3 3-reg (unaligned)
Page 525 and 526:
Instruction Single precision cycles
Page 527 and 528:
• FMACS, FNMACS • FMSCS, FNMSCS
Page 529 and 530:
is dual issued with previous instru
Page 531 and 532:
17.1 About setup and hold times AC
Page 533 and 534:
17.2 AXI interface Table 17-2 shows
Page 535 and 536:
17.3 ATB and CTI interfaces Table 1
Page 537 and 538:
AC Characteristics Table 17-4 Timin
Page 539 and 540:
17.6 L2 preload interface AC Charac
Page 541 and 542:
17.8 Miscellaneous signals AC Chara
Page 543 and 544:
A.1 AXI interface Signal Descriptio
Page 545 and 546:
A.3 MBIST and DFT interface A.3.1 M
Page 547 and 548:
A.4 Preload engine interface Table
Page 549 and 550:
A.6 Miscellaneous signals Table A-7
Page 551 and 552:
Signal I/O Reset Description CFGNMF
Page 553 and 554:
Signal I/O Reset Description DBGNOP
Page 555 and 556:
Appendix B Instruction Mnemonics Th
Page 557 and 558:
Instruction Mnemonics Table B-1 Adv
Page 559 and 560:
Appendix C Revisions This appendix
Page 561 and 562:
Revisions Added text to clarify des
Page 563 and 564:
Glossary This glossary describes so
Page 565 and 566:
Glossary Automatic Test Pattern Gen
Page 567 and 568:
Big-endian memory Memory in which:
Page 569 and 570:
Conditional execution Glossary incl
Page 571 and 572:
Glossary Embedded Trace Macrocell (
Page 573 and 574:
Illegal instruction An instruction
Page 575 and 576:
Glossary NaN Not a number. A symbol
Page 577 and 578:
Glossary Significand The component
Page 579 and 580:
Glossary Watchpoint A watchpoint is
show all

Cortex-A8 Technical Reference Manual - ARM Information Center

Create successful ePaper yourself

Delete template?

Save as template?