Commit | Line | Data |
---|---|---|
c906108c SS |
1 | |
2 | PSIM - model the PowerPC environment | |
3 | ||
4 | Copyright (C) 1994-1996, Andrew Cagney <cagney@highland.com.au>. | |
5 | ||
6 | ---------------------------------------------------------------------- | |
7 | ||
8 | ||
9 | Building PSIM | |
10 | ||
11 | This file describes how to build the program PSIM | |
12 | ||
13 | o Walk through a basic build | |
14 | ||
15 | o Discussion of PSIM's components and | |
16 | how they relate to the build process | |
17 | ||
18 | o Detailed description of each of PSIM's | |
19 | compile time configuration options | |
20 | ||
21 | ||
22 | ---------------------------------------------------------------------- | |
23 | ||
24 | ||
25 | BUILDING PSIM: | |
26 | ||
27 | PSIM 1.0.2 is included in GDB-4.16. To build PSIM you will need the | |
28 | following: | |
29 | ||
30 | gdb-4.16.tar.gz Available from your favorite GNU | |
31 | ftp site | |
32 | ||
33 | gcc GCC version two includes suport | |
34 | for long long (64bit integer) | |
35 | arrithemetic which PSIM uses. Hence | |
36 | it is recommended that you build PSIM | |
37 | using GCC. | |
38 | ||
39 | Method: | |
40 | ||
41 | 1. Unpack gdb | |
42 | ||
43 | $ cd .../scratch | |
44 | $ gunzip < gdb-4.16.tar.gz | tar xf - | |
45 | ||
46 | ||
47 | 2. Configure gdb | |
48 | ||
49 | First consult the gdb documentation | |
50 | ||
51 | $ cd .../scratch | |
52 | $ cd gdb-4.16 | |
53 | $ more README | |
54 | $ more gdb/README | |
55 | ||
56 | then something like (I assume SH): | |
57 | ||
58 | $ CC=gcc ./configure \ | |
59 | --enable-sim-powerpc \ | |
60 | --target=powerpc-unknown-eabi \ | |
61 | --prefix=/applications/psim | |
62 | ||
63 | ||
64 | 4. Build (again specifying GCC) | |
65 | ||
66 | $ make CC=gcc | |
67 | ||
68 | alternatively, if you are short on disk space or only | |
69 | want to build the simulator: | |
70 | ||
71 | $ ( cd libiberty && make CC=gcc ) | |
72 | $ ( cd bfd && make CC=gcc ) | |
73 | $ ( cd sim/ppc && make CC=gcc ) | |
74 | ||
75 | ||
76 | 5. Install | |
77 | ||
78 | $ make CC=gcc install | |
79 | ||
80 | or just | |
81 | ||
82 | $ cp gdb/gdb ~/bin/powerpc-unknown-eabisim-gdb | |
83 | $ cp sim/ppc/run ~/bin/powerpc-unknown-eabisim-run | |
84 | ||
85 | ||
86 | ---------------------------------------------------------------------- | |
87 | ||
88 | ||
89 | UPDATING PSIM: | |
90 | ||
91 | ||
92 | A PSIM is an ongoing development. Occasional snapshots which both contain new | |
93 | features and fix old bugs are made available. See the ftp directory: | |
94 | ||
95 | ftp://ftp.ci.com.au/pub/psim/beta | |
96 | or ftp://cambridge.cygnus.com/pub/psim/beta | |
97 | ||
98 | for the latest version. To build/install one of these snapshots, you | |
99 | replace the sim/ppc found in the gdb archive with with one from the | |
100 | snapshot. Then just re-configure and rebuild/install. | |
101 | ||
102 | Procedure: | |
103 | ||
104 | 0. A starting point | |
105 | ||
106 | $ cd gdb-4.16 | |
107 | ||
108 | ||
109 | 1. Remove the old psim directory | |
110 | ||
111 | $ mv sim/ppc sim/old.ppc | |
112 | ||
113 | ||
114 | 2. Unpack the new one | |
115 | ||
116 | $ gunzip < ../psim-NNNNNN.tar.gz | tar tf - | |
117 | $ gunzip < ../psim-NNNNNN.tar.gz | tar tf - | |
118 | ||
119 | ||
120 | 3. Reconfigure/rebuild (as seen above): | |
121 | ||
122 | $ CC=gcc ./configure \ | |
123 | --enable-sim-powerpc \ | |
124 | --target=powerpc-unknown-eabi \ | |
125 | --prefix=/applications/psim | |
126 | $ make CC=gcc | |
127 | ||
128 | ||
129 | ---------------------------------------------------------------------- | |
130 | ||
131 | ||
132 | UPDATES TO GDB: | |
133 | ||
134 | From time to time, problems involving the integration of PSIM into gdb | |
135 | are found. While eventually each of these problems is resolved there | |
136 | can be periouds during which a local hack may be needed. | |
137 | ||
138 | At the time of writing the following were outstanding: | |
139 | ||
140 | ATTACH command: | |
141 | ||
142 | ftp://ftp.ci.com.au/pub/psim/gdb-4.15+attach.diff.gz | |
143 | or ftp://cambridge.cygnus.com/pub/psim/gdb-4.15+attach.diff.gz | |
144 | ||
145 | PSIM, unlike the other simulators found in GDB, is able to load | |
146 | the description of a target machine (including the initial | |
147 | state of all processor registers) from a file. | |
148 | ||
149 | Unfortunatly GDB does not yet have a standard command that | |
150 | facilitates the use of this feature. Until such a command is | |
151 | added, the patch (hack?) gdb-4.15+attach.diff.gz can be used to | |
152 | extend GDB's attach command so that it can be used to initialize | |
153 | the simulators configuration from a file. | |
154 | ||
155 | ||
156 | ||
157 | ---------------------------------------------------------------------- | |
158 | ||
159 | ||
160 | RUNNING PROGRAMS: | |
161 | ||
162 | ||
163 | See the file: | |
164 | ||
165 | ftp://ftp.ci.com.au/pub/psim/RUN | |
166 | or ftp://cambridge.cygnus.com/pub/psim/RUN | |
167 | ||
168 | ||
169 | ---------------------------------------------------------------------- | |
170 | ||
171 | ||
172 | COMPILE TIME CONFIGURATION OPTIONS: | |
173 | ||
174 | ||
175 | PSIM's compile time configuration is controlled by autoconf. PSIM's | |
176 | configure script recognises options of the form: | |
177 | ||
178 | --enable-sim-<option>[=<val>] | |
179 | ||
180 | And can be specified on the configure command line (at the top level | |
181 | of the gdb directory tree) vis: | |
182 | ||
183 | $ cd gdb-4.15 | |
184 | $ CC=gcc ./configure \ | |
185 | --target=powerpc-unknown-eabisim \ | |
186 | --prefix=/applications/psim \ | |
187 | --enable-sim-inline | |
188 | $ make CC=gcc | |
189 | ||
190 | For a brief list of PSIM's configuration options, configure --help | |
191 | will list them vis: | |
192 | ||
193 | $ cd sim/ppc | |
194 | $ ./configure --help | |
195 | ||
196 | Each PSIM specific option is discussed in detail below. | |
197 | ||
198 | ||
199 | ||
200 | --enable-sim-cflags=<opts> | |
201 | ||
202 | ||
203 | Specify additional C compiler flags that are to be used when compiling | |
204 | just PSIM. | |
205 | ||
206 | PSIM places heavy demands on both the host machine and its C compiler. So that | |
207 | the builder has better control over the compiler the above option can be used | |
208 | to pass additional options to the compiler while PSIM is being built. | |
209 | ||
210 | Ex: No debug information | |
211 | ||
212 | PSIM can be built with everything inline. Unfortunately, because of | |
213 | all the debugging information generated the C compiler can grow very | |
214 | very large as a result. For GCC, the debug information can be | |
215 | restricted with the `-g0' option. To specify that this option should | |
216 | be include in the CFLAGS when compiling the psim source code use: | |
217 | ||
218 | --enable-sim-cflags=-g0 | |
219 | ||
220 | Ex: Additional optimization flags | |
221 | ||
222 | A significant gain in performance can be achieved by tuning the | |
223 | optimization flags passed to the C compiler. For instance on an x86 | |
224 | you may consider: | |
225 | ||
226 | --enable-sim-cflags='-g0 -O2 -fno-strength-reduce -f...' | |
227 | ||
228 | ||
229 | ||
230 | --enable-sim-warnings=<flags> | |
231 | ||
232 | ||
233 | Turn on additional GCC specific checks. | |
234 | ||
235 | Some hosts (NetBSD, Linux, Solaris-2.5) have complete header files | |
236 | that include correct prototypes for all library functions. On such | |
237 | hosts, PSIM can be built with many more than the standard C checks | |
238 | enabled. The option --enable-sim-warnings controls this. | |
239 | ||
240 | Ex: Default warnings | |
241 | ||
242 | With just --enable-sim-warnings, the following -W options are enabled: | |
243 | -Werror -Wall -Wpointer-arith -Wmissing-prototypes. | |
244 | ||
245 | ||
246 | ||
247 | --enable-sim-opcode=which | |
248 | ||
249 | ||
250 | Specify the file containing the rules for generating the instruction | |
251 | decode and execute functions from the file ppc-instructions. | |
252 | ||
253 | The form of the instruction decode and execute functions is controlled | |
254 | by an opcode table. It specifies: the combination of switch | |
255 | statements and jump tables to use when decoding an instruction and how | |
256 | much of each instruction should be decoded before calling the | |
257 | instruction execute function. | |
258 | ||
259 | PSIM includes a number of opcode tables: | |
260 | ||
261 | psim-opcode-simple | |
262 | Generates a small compact two level switch statement | |
263 | that will compile quickly and run reasonably fast. | |
264 | ||
265 | This may be useful on a small machine. | |
266 | ||
267 | psim-opcode-complex | |
268 | (the default) A fairly aggressive instruction decode | |
269 | table that includes the breaking out of a number | |
270 | of special instruction cases (eg RA==0 vs RA!=0). | |
271 | ||
272 | psim-opcode-flat | |
273 | Identical to complex except a switch statement | |
274 | is used. Ideal for when the icache is being | |
275 | disabled. | |
276 | ||
277 | psim-opcode-stupid | |
278 | In addition to the instruction decodes performed | |
279 | by psim-opcode-complex, this also full decodes mtspr, | |
280 | mfspr, and branch instructions. The table generated | |
281 | is very large and, as a consequence, only performs | |
282 | well on machines with large caches. | |
283 | ||
284 | ppc-opcode-test-1 | |
285 | ppc-opcode-test-2 | |
286 | Generate test (but workable) tables. These exercise | |
287 | PSIM's ability to generate instruction decode functions | |
288 | that are a combination of jump-tables and switch statements. | |
289 | ||
290 | The program igen generates the instruction tables from the opcode | |
291 | table and the ppc-instruction table. | |
292 | ||
293 | ||
294 | ||
295 | --enable-sim-switch | |
296 | ||
297 | ||
298 | Enable/disable the use of a switch statement when looking up the | |
299 | attributes of a SPR register. | |
300 | ||
301 | The PowerPC architecture defines a number of Special Purpose Registers | |
302 | (SPR's). Associated with each of these registers are a number of | |
303 | attributes (such as validity or size) which the instructions | |
304 | mtspr/mfspr query as part of their execution. | |
305 | ||
306 | For PSIM, this information is kept in a table (ppc-spr-table). The | |
307 | program dgen converts this table into lookup routines (contained in | |
308 | the generated files spreg.h spreg.c) that can be used to query an | |
309 | SPR's attributes. Those lookup routines are either implemented as | |
310 | a table or alternatively as a number of switch statements: | |
311 | ||
312 | spr_table spr_info[] = { .... }; | |
313 | int spr_length(sprs spr) { return spr_info[spr].length; } | |
314 | ||
315 | vs | |
316 | ||
317 | int spr_length(sprs spr) { switch (spr) { case ..: return ..; } } | |
318 | ||
319 | In general the first implementation (a table) is the most efficient. | |
320 | It may, however, prove that when performing an aggressive optimization | |
321 | where both the SPR is known and the above function is being inlined | |
322 | (with the consequence that GCC can eliminate the switch statement) | |
323 | that the second choice is improves performance. | |
324 | ||
325 | In practice, only a marginal (if any benefit) has ever been seen. | |
326 | ||
327 | ||
328 | ||
329 | --enable-sim-duplicate | |
330 | ||
331 | ||
332 | Create a duplicate copy of each instruction function hardwiring | |
333 | instruction fields that would have otherwise have been variable. | |
334 | ||
335 | As discussed above, igen outputs a C function generated from the file | |
336 | ppc-instructions (using the opcode rules) for each of the | |
337 | instructions. Thus multiple entries in the instruction decode tables | |
338 | may be pointing back at the same function. Enabling duplicate, will | |
339 | result in psim creating a duplicate of the instruction's function for | |
340 | each different entry in the instruction decode tables. | |
341 | ||
342 | For instance, given the branch instruction: | |
343 | ||
344 | 0.19,6.BO,11.BI,16./,21.528,31.LK | |
345 | ... | |
346 | if (LK) LR = (spreg)IEA(CIA + 4); | |
347 | ... | |
348 | ||
349 | igen as part of its instruction lookup table may have generated two | |
350 | different entries - one for LK=0 and one for LK=1. With duplicate | |
351 | enabled, igen outputs (almost) duplicate copies of branch function, | |
352 | one with LK hardwired to 0 and one with LK hardwired to 1. | |
353 | ||
354 | By doing this the compiler is provided with additional information that | |
355 | will allow it possibly eliminate dead code. (such as the assignment | |
356 | to LK if LR==0). | |
357 | ||
358 | Ex: default | |
359 | ||
360 | Because this feature is such a big win, --enable-sim-duplicate is | |
361 | turned on by default. | |
362 | ||
363 | Ex: A small machine | |
364 | ||
365 | Only rarely (eg on a very small host) would this feature need to be | |
366 | disabled (using: --disable-sim-duplicate). | |
367 | ||
368 | ||
369 | ||
370 | --enable-sim-filter=rule | |
371 | ||
372 | ||
373 | Include/exclude PowerPC instructions that are specific to a particular | |
374 | implementation. | |
375 | ||
376 | Some of the PowerPC instructions included in the file ppc-instructions | |
377 | are limited to certain specific PPC implementations. For instance, | |
378 | the instruction: | |
379 | ||
380 | 0.58,6.RT,11.RA,16.DS,30.2:DS:64::Load Word Algebraic | |
381 | ||
382 | Is only valid for the 64bit architecture. The enable-sim-filter flag | |
383 | is passed to igen so that it can `filter out' any invalid | |
384 | instructions. The filter rule has the form: | |
385 | ||
386 | -f <name> | |
387 | ||
388 | thus: | |
389 | ||
390 | --enable-sim-filter='-f 64' | |
391 | ||
392 | (the default) would filter out all 64bit instructions. | |
393 | ||
394 | Ex: Remove floating point instructions | |
395 | ||
396 | A given 32bit PowerPC implementation may not include floating point | |
397 | hardware. Consequently there is little point in including floating | |
398 | point instructions in the instruction table. The option: | |
399 | ||
400 | --enable-sim-filter='-f 64 -f f' | |
401 | ||
402 | will eliminate all floating point instructions from the instruction | |
403 | table. | |
404 | ||
405 | ||
406 | ||
407 | --enable-sim-icache=size | |
408 | ||
409 | ||
410 | Set the size of the cache used to hold decoded instructions. | |
411 | ||
412 | Psim executes instructions in two separate steps: | |
413 | ||
414 | o instruction fetch/decode | |
415 | ||
416 | o instruction execution | |
417 | ||
418 | For a given instruction, the first stage need only be executed once | |
419 | (the first time the instruction is encountered) while the second stage | |
420 | must be executed every time the program `executes' that instruction. | |
421 | ||
422 | Exploiting this, PSIM can maintain a cache of decoded instructions. | |
423 | It will then use the decoded instruction from the cache in preference | |
424 | to fetching/decoding the real instruction from memory. | |
425 | ||
426 | Ex: default | |
427 | ||
428 | Because this feature is normally such a big win, it is enabled by | |
429 | default (with the cache size set to 1024 entries). | |
430 | ||
431 | The 1024 entries equals 4096 bytes (or one page) of instructions. | |
432 | Larger caches can be used but with caution - PSIM does not check for | |
433 | address aliasing within its instruction cache. | |
434 | ||
435 | Ex: disable the cache | |
436 | ||
437 | There may be cases (for instance where the cache has a low hit rate) | |
438 | where the psim performs better with no instruction cache. For such | |
439 | situations, the cache can be disabled vis: --disable-sim-icache. | |
440 | ||
441 | ||
442 | ||
443 | --enable-sim-inline[=module] | |
444 | ||
445 | ||
446 | Specify the inlining of one or more modules. | |
447 | ||
448 | Many architectures (in particular the x86) suffer from a large | |
449 | function call overhead. By eliminating function calls (through | |
450 | inlining of functions) a large performance gain can be achieved. | |
451 | ||
452 | In PSIM, modules are inlined in one of two possible ways. Some | |
453 | modules (such as the byte swapping code) can be inlined into any | |
454 | module that calls them. Other modules, due to complex | |
455 | interdependencies, are only inlined as a group when compiling the | |
456 | external interface module psim.c. | |
457 | ||
458 | Ex: default | |
459 | ||
460 | By default the modules endian (handle be/le), bits (manipulate | |
461 | bit-fields within words), cpu (the processor object) and events | |
462 | (timers) are inlined in any module that calls them. This gives a | |
463 | reasonable performance gain with little additional compilation | |
464 | overhead. | |
465 | ||
466 | Ex: recommended --enable-sim-inline | |
467 | ||
468 | Assuming you machine is reasonably well configured, this option is | |
469 | highly recommended. On the x86 several orders of magnitude | |
470 | improvement in performance is possible. | |
471 | ||
472 | Ex: fine tuning | |
473 | ||
474 | The file std-config.h contains a detailed description of how the | |
475 | inlining works. Individual modules can be inlined by specifying them. | |
476 | For if you have a very large cache the model module could be inlined | |
477 | with: | |
478 | ||
479 | --enable-sim-inline=MODEL | |
480 | ||
481 | ||
482 | ||
483 | --enable-sim-bswap | |
484 | ||
485 | ||
486 | (x86 specific) Use the i486/P5/P6 byte swap instruction. | |
487 | ||
488 | PSIM contains generic byte swapping code. For the x86 (P[4-6]) PSIM | |
489 | can be built so that it uses the bswap instruction instead of relying | |
490 | on the compiler to generate byte swap code. | |
491 | ||
492 | Ex: default | |
493 | ||
494 | By default, when compiling with GCC-2 on an i486/P5/P6 the bswap | |
495 | instruction is used. | |
496 | ||
497 | ||
498 | ||
499 | --enable-sim-endian=endian | |
500 | ||
501 | ||
502 | Specify the byte order of the target. | |
503 | ||
504 | By default, PSIM is able to execute both big and little endian | |
505 | executables. As a consequence, every byte swap routine includes a | |
506 | test to see if the byte swap is really needed. By specifying the byte | |
507 | order of the target (and the host below) the need for this test can be | |
508 | eliminated. | |
509 | ||
510 | Clearly setting the byte order of the target is only useful when known | |
511 | before hand. | |
512 | ||
513 | ||
514 | ||
515 | --enable-sim-hostendain=end | |
516 | ||
517 | ||
518 | As above but for the host. | |
519 | ||
520 | Normally this option should not be needed. configure (autoconf) should | |
521 | determine the byte order of the host automatically. However if for | |
522 | some reason there is a problem, this option can be used to override | |
523 | autoconf. | |
524 | ||
525 | ||
526 | ||
527 | --enable-sim-smp=n | |
528 | ||
529 | ||
530 | Set the maximum number of processors that PSIM can model. | |
531 | ||
532 | Psim can model (with small limitation discussed else where) a | |
533 | multi-processor PowerPC environment. While the overhead of | |
534 | co-ordinating the execution of a number of processors is relatively | |
535 | small it is still significant when compared to handling only one | |
536 | processor. | |
537 | ||
538 | This option only sets the maximum number of processors that can be | |
539 | simulated. The number active during a given simulation run us | |
540 | determined at run time. | |
541 | ||
542 | Ex: default | |
543 | ||
544 | By default 5 processors are configured but only one is enabled. | |
545 | Additional processors can be enabled with the runtime option: | |
546 | ||
547 | -o '/openprom/options/smp 5' | |
548 | ||
549 | Ex: recommended | |
550 | ||
551 | Unless you intend studying multi-processor systems there is little reason for | |
552 | having PSIM configured with SMP support. Specifying: | |
553 | ||
554 | --disable-sim-smp | |
555 | or --enable-sim-smp=0 | |
556 | ||
557 | will eliminate any SMP such as: | |
558 | ||
559 | for (cpu = 0; cpu < nr_cpus; cpu++) | |
560 | ... | |
561 | ||
562 | ||
563 | ||
564 | --enable-sim-xor-endian=n | |
565 | ||
566 | ||
567 | Set the byte-size of the bus involved in the PowerPC's xor endian byte | |
568 | swapping. | |
569 | ||
570 | The PowerPC's implementation of BE/LE mode is different to what a | |
571 | programmer may first expect. The details of this implementation are | |
572 | discussed at length in PowerPC documentation. | |
573 | ||
574 | Ex: default | |
575 | ||
576 | By default this is configured with a value of 8 (the bus size of most | |
577 | 60x processors). | |
578 | ||
579 | Ex: recommended | |
580 | ||
581 | Unless you are expecting to test/debug PowerPC be/le switching code | |
582 | this option is of little use and should be disabled: | |
583 | ||
584 | --disable-sim-xor-endian | |
585 | ||
586 | ||
587 | ||
588 | --enable-sim-bitsize=n | |
589 | ||
590 | ||
591 | Specify the bit size (32/64) of the PowerPC to be modelled. | |
592 | ||
593 | Note: By default 32 is specified. The implementation of the 64bit | |
594 | architecture is still under development. | |
595 | ||
596 | ||
597 | --enable-sim-hostbitsize=32|64 | |
598 | ||
599 | As above but for the host. | |
600 | ||
601 | NOTE: Psim has yet to be built on a 64bit host. | |
602 | ||
603 | ||
604 | ||
605 | --enable-sim-env=env | |
606 | ||
607 | ||
608 | Hardwire the PowerPC environment being modelled (user, virtual or | |
609 | operating). | |
610 | ||
611 | The PowerPC architecture defines three different levels of compliance to its | |
612 | architectural specification. These environments are discussed in detail in | |
613 | PowerPC publications. | |
614 | ||
615 | user - normal user programs | |
616 | virtual - an extension of the user environment (includes timers) | |
617 | operating - kernel code | |
618 | ||
619 | Ex: default | |
620 | ||
621 | By default all three environments are supported. | |
622 | ||
623 | Ex: recommended | |
624 | ||
625 | If you only intend running psim with user (or operating) code then | |
626 | PSIM should be configured accordingly. For user code, it eliminates: | |
627 | support for timers and events and redundant VM calls. | |
628 | ||
629 | ||
630 | ||
631 | --enable-sim-timebase | |
632 | ||
633 | ||
634 | Enable/disable the time base register. | |
635 | ||
636 | The PowerPC architecture (virtual environment) includes a time base | |
637 | register. Maintaining that register incurs an overhead in | |
638 | performance that can be eliminated by eliminating time-base register | |
639 | support. | |
640 | ||
641 | Ex: default | |
642 | ||
643 | Normally this option is not used. Instead --enable-sim-env (above) us | |
644 | used to disable/enable features such as the timebase register. | |
645 | ||
646 | ||
647 | ||
648 | --enable-sim-alignment=align | |
649 | ||
650 | ||
651 | Control the PowerPC's memory access alignment restrictions. | |
652 | ||
653 | The PowerPC in LE mode only allows memory transfers of a correctly | |
654 | aligned size/address. The above option controls how misaligned | |
655 | accesses are handled. | |
656 | ||
657 | strict All accesses must be correctly aligned | |
658 | ||
659 | nonstrict Unaligned access allowed (the are split | |
660 | into a number of aligned accesses). | |
661 | ||
662 | Ex: default | |
663 | ||
664 | Unless otherwise specified PSIM will auto configure a BE program to | |
665 | allow miss-aligned accesses while a LE program will not. | |
666 | ||
667 | Ex: 604e | |
668 | ||
669 | The recently announced 604e processor allows miss-aligned accesses in both | |
670 | BE and LE modes. If modeling the 604e then you should specify: | |
671 | ||
672 | --enable-sim-alignment=nonstrict | |
673 | ||
674 | ||
675 | ||
676 | --enable-sim-trace | |
677 | ||
678 | ||
679 | Include code to trace PSIM's internal progress (also controlled by the | |
680 | -t option). | |
681 | ||
682 | Checking to see if a trace message should be output slows down a | |
683 | simulation. Disabling this option (--disable-sim-trace) eliminates | |
684 | completely that code. | |
685 | ||
686 | ||
687 | ||
688 | --enable-sim-assert | |
689 | ||
690 | ||
691 | Include the code that checks the correctness of parts of PSIM. | |
692 | ||
693 | Eliminating such code (--disable-sim-assert) eliminates internal | |
694 | consistency tests and their overhead. | |
695 | ||
696 | ||
697 | ||
698 | --enable-sim-reserved-bits | |
699 | ||
700 | ||
701 | Include code to check that the reserved fields of the instruction are | |
702 | zero. | |
703 | ||
704 | The PowerPC architecture defines certain fields of some instructions | |
705 | as reserved (`/'). By default, for each instruction, PSIM will check | |
706 | the reserved fields causing an invalid instruction exception if a | |
707 | field is invalid. Disabling this option eliminates this test. This | |
708 | is at the slight risk of PSIM treating an invalid instruction as | |
709 | valid. | |
710 | ||
711 | ||
712 | ||
713 | --enable-sim-float | |
714 | ||
715 | ||
716 | Include support for hardware floating point. | |
717 | ||
718 | ||
719 | ||
720 | --enable-sim-monitor=mon | |
721 | ||
722 | ||
723 | Include support for basic instruction counting. | |
724 | ||
725 | If you are not interested in the performance of either you program or | |
726 | the simulator then you can disable this option. | |
727 | ||
728 | ||
729 | ||
730 | --enable-sim-model=which | |
731 | ||
732 | Hardwire the processor that will be used as a reference when modeling | |
733 | execution units. | |
734 | ||
735 | ||
736 | ||
737 | --enable-sim-default-model=which | |
738 | ||
739 | ||
740 | Specify the processor of choice for the execution unit model. | |
741 | ||
742 | ||
743 | ||
744 | --enable-sim-model-issue | |
745 | ||
746 | ||
747 | Include support for the modeling of processor execution units. | |
748 | ||
749 | ---------------------------------------------------------------------- | |
750 | ||
751 | TYPICAL CONFIGURATION OPTIONS: | |
752 | ||
753 | ||
754 | VEA CODE ONLY: | |
755 | ||
756 | Here of note are: | |
757 | ||
758 | o ramp up the compiler options (some | |
759 | of the below are P5 specific). | |
760 | ||
761 | o disable anything not used | |
762 | ||
763 | CC=gcc ./configure \ | |
764 | --prefix=/applications/psim \ | |
765 | --target=powerpc-unknown-eabi \ | |
766 | --enable-sim-powerpc \ | |
767 | --enable-sim-warnings \ | |
768 | --enable-sim-inline \ | |
769 | --disable-sim-smp \ | |
770 | --enable-sim-duplicate \ | |
771 | --enable-sim-endian=big \ | |
772 | --disable-sim-xor-endian \ | |
773 | --enable-sim-env=user \ | |
774 | --disable-sim-reserved-bits \ | |
775 | --disable-sim-assert \ | |
776 | --disable-sim-trace \ | |
777 | --enable-sim-cflags='-g0,-O2,-fno-strength-reduce,-fomit-frame-pointer' | |
778 | ||
779 | ||
780 | OEA CODE ONLY: | |
781 | ||
782 | The key configuration changes are: | |
783 | ||
784 | o turn off the instruction cache. The overhead | |
785 | of flushing and reloading it is greater than | |
786 | not having a cache. | |
787 | ||
788 | o use a switch statement (ppc-opcode-flat) for | |
789 | the instruction decode and then (-O3) fully | |
790 | inline all functions. | |
791 | ||
792 | o --enable-sim-warnings is not present. GCC (2.7.2) | |
793 | gets confused by the instruction decode table | |
794 | generated by igen (contains a perfect switch) | |
795 | and, as a consequence, generates a bogus warning. | |
796 | ||
797 | CC=gcc ./configure \ | |
798 | --prefix=/applications/psim \ | |
799 | --target=powerpc-unknown-eabi \ | |
800 | --enable-sim-powerpc \ | |
801 | --enable-sim-inline \ | |
802 | --disable-sim-smp \ | |
803 | --enable-sim-duplicate \ | |
804 | --enable-sim-endian=big \ | |
805 | --disable-sim-xor-endian \ | |
806 | --enable-sim-env=operating \ | |
807 | --disable-sim-reserved-bits \ | |
808 | --disable-sim-assert \ | |
809 | --disable-sim-trace \ | |
810 | --enable-sim-opcode=ppc-opcode-flat \ | |
811 | --disable-sim-icache \ | |
812 | --enable-sim-cflags='-g0,-O3,-fno-strength-reduce,-fomit-frame-pointer' |