OpenCores
URL https://opencores.org/ocsvn/lxp32/lxp32/trunk

Subversion Repositories lxp32

[/] [lxp32/] [trunk/] [tools/] [src/] [lxp32asm/] [assembler.cpp] - Blame information for rev 9

Details | Compare with Previous | View Log

Line No. Rev Author Line
1 9 ring0_mipt 
/*
2 
 * Copyright (c) 2016 by Alex I. Kuznetsov.
3 
 *
4 
 * Part of the LXP32 CPU IP core.
5 
 *
6 
 * This module implements members of the Assembler class.
7 
 */
8 
 
9 
#include "assembler.h"
10 
#include "utils.h"
11 
 
12 
#include <iostream>
13 
#include <fstream>
14 
#include <sstream>
15 
#include <stdexcept>
16 
#include <utility>
17 
#include <limits>
18 
#include <type_traits>
19 
#include <cctype>
20 
#include <cassert>
21 
#include <cstdlib>
22 
 
23 
void Assembler::processFile(const std::string &filename) {
24 
        auto nativePath=Utils::normalizeSeparators(filename);
25 
        auto pos=nativePath.find_last_of('/');
26 
        if(pos!=std::string::npos) nativePath=filename.substr(pos+1);
27 
        _obj.setName(nativePath);
28 
 
29 
        _line=0;
30 
        _state=Initial;
31 
        _currentFileName=filename;
32 
        processFileRecursive(filename);
33 
 
34 
        if(!_currentLabels.empty())
35 
                throw std::runtime_error("Symbol definition must be followed by an instruction or data definition statement");
36 
 
37 
        if(!_sectionEnabled.empty())
38 
                throw std::runtime_error("#endif expected");
39 
 
40 
// Examine symbol table
41 
        for(auto const &sym: _obj.symbols()) {
42 
                if(sym.second.type==LinkableObject::Unknown&&!sym.second.refs.empty()) {
43 
                        std::ostringstream msg;
44 
                        msg<<"Undefined symbol \""+sym.first+"\"";
45 
                        msg<<" (referenced from "<<sym.second.refs[0].source;
46 
                        msg<<":"<<sym.second.refs[0].line<<")";
47 
                        throw std::runtime_error(msg.str());
48 
                }
49 
        }
50 
 
51 
        for(auto const &sym: _exportedSymbols) _obj.exportSymbol(sym);
52 
}
53 
 
54 
void Assembler::processFileRecursive(const std::string &filename) {
55 
        std::ifstream in(filename,std::ios_base::in);
56 
        if(!in) throw std::runtime_error("Cannot open file \""+filename+"\"");
57 
 
58 
// Process input file line-by-line
59 
        auto savedLine=_line;
60 
        auto savedState=_state;
61 
        auto savedFileName=_currentFileName;
62 
 
63 
        _line=1;
64 
        _state=Initial;
65 
        _currentFileName=filename;
66 
 
67 
        std::string line;
68 
        while(std::getline(in,line)) {
69 
                auto tokens=tokenize(line);
70 
                expand(tokens);
71 
                elaborate(tokens);
72 
                _line++;
73 
        }
74 
 
75 
        if(_state!=Initial) throw std::runtime_error("Unexpected end of file");
76 
 
77 
        _line=savedLine;
78 
        _state=savedState;
79 
        _currentFileName=savedFileName;
80 
}
81 
 
82 
void Assembler::addIncludeSearchDir(const std::string &dir) {
83 
        auto ndir=Utils::normalizeSeparators(dir);
84 
        if(!ndir.empty()&&ndir.back()!='/') ndir.push_back('/');
85 
        _includeSearchDirs.push_back(std::move(ndir));
86 
}
87 
 
88 
int Assembler::line() const {
89 
        return _line;
90 
}
91 
 
92 
std::string Assembler::currentFileName() const {
93 
        return _currentFileName;
94 
}
95 
 
96 
LinkableObject &Assembler::object() {
97 
        return _obj;
98 
}
99 
 
100 
const LinkableObject &Assembler::object() const {
101 
        return _obj;
102 
}
103 
 
104 
Assembler::TokenList Assembler::tokenize(const std::string &str) {
105 
        TokenList tokenList;
106 
        std::string word;
107 
        std::size_t i;
108 
        for(i=0;i<str.size();i++) {
109 
                char ch=str[i];
110 
                switch(_state) {
111 
                case Initial:
112 
                        if(ch==' '||ch=='\t'||ch=='\n'||ch=='\r') continue; // skip whitespace
113 
                        else if(ch==','||ch==':') { // separator
114 
                                tokenList.push_back(std::string(1,ch));
115 
                        }
116 
                        else if(std::isalnum(ch)||ch=='.'||ch=='#'||ch=='_'||ch=='-'||ch=='+') {
117 
                                word=std::string(1,ch);
118 
                                _state=Word;
119 
                        }
120 
                        else if(ch=='\"') {
121 
                                word="\"";
122 
                                _state=StringLiteral;
123 
                        }
124 
                        else if(ch=='/') {
125 
                                if(++i>=str.size()) throw std::runtime_error("Unexpected end of line");
126 
                                ch=str[i];
127 
                                if(ch=='/') i=str.size(); // skip the rest of the line
128 
                                else if(ch=='*') _state=BlockComment;
129 
                                else throw std::runtime_error(std::string("Unexpected character: \"")+ch+"\"");
130 
                        }
131 
                        else throw std::runtime_error(std::string("Unexpected character: \"")+ch+"\"");
132 
                        break;
133 
                case Word:
134 
                        if(std::isalnum(ch)||ch=='_'||ch=='@'||ch=='+'||ch=='-') word+=ch;
135 
                        else {
136 
                                i--;
137 
                                _state=Initial;
138 
                                tokenList.push_back(std::move(word));
139 
                        }
140 
                        break;
141 
                case StringLiteral:
142 
                        if(ch=='\\') {
143 
                                if(++i>=str.size()) throw std::runtime_error("Unexpected end of line");
144 
                                ch=str[i];
145 
                                if(ch=='\\') word.push_back('\\');
146 
                                else if(ch=='\"') word.push_back('\"');
147 
                                else if(ch=='\'') word.push_back('\'');
148 
                                else if(ch=='t') word.push_back('\t');
149 
                                else if(ch=='n') word.push_back('\n');
150 
                                else if(ch=='r') word.push_back('\r');
151 
                                else if(ch=='x') { // hexadecimal sequence can be 1-2 digit long
152 
                                        std::string seq;
153 
                                        if(i+1<str.size()&&Utils::ishexdigit(str[i+1])) seq+=str[i+1];
154 
                                        if(i+2<str.size()&&Utils::ishexdigit(str[i+2])) seq+=str[i+2];
155 
                                        if(seq.empty()) throw std::runtime_error("Ill-formed escape sequence");
156 
                                        try {
157 
                                                word.push_back(static_cast<char>(std::stoul(seq,nullptr,16)));
158 
                                        }
159 
                                        catch(std::exception &) {
160 
                                                throw std::runtime_error("Ill-formed escape sequence");
161 
                                        }
162 
                                        i+=seq.size();
163 
                                }
164 
                                else if(Utils::isoctdigit(ch)) { // octal sequence can be 1-3 digit long
165 
                                        std::string seq(1,ch);
166 
                                        if(i+1<str.size()&&Utils::isoctdigit(str[i+1])) seq+=str[i+1];
167 
                                        if(i+2<str.size()&&Utils::isoctdigit(str[i+2])) seq+=str[i+2];
168 
                                        unsigned long value;
169 
                                        try {
170 
                                                value=std::stoul(seq,nullptr,8);
171 
                                        }
172 
                                        catch(std::exception &) {
173 
                                                throw std::runtime_error("Ill-formed escape sequence");
174 
                                        }
175 
 
176 
                                        if(value>255) throw std::runtime_error("Octal value is out of range");
177 
                                        word.push_back(static_cast<char>(value));
178 
 
179 
                                        i+=seq.size()-1;
180 
                                }
181 
                                else throw std::runtime_error(std::string("Unknown escape sequence: \"\\")+ch+"\"");
182 
                        }
183 
                        else if(ch=='\"') {
184 
                                word.push_back('\"');
185 
                                tokenList.push_back(std::move(word));
186 
                                _state=Initial;
187 
                        }
188 
                        else word.push_back(ch);
189 
                        break;
190 
                case BlockComment:
191 
                        if(ch=='*') {
192 
                                if(++i>=str.size()) break;
193 
                                ch=str[i];
194 
                                if(ch=='/') _state=Initial;
195 
                                else i--;
196 
                        }
197 
                        break;
198 
                }
199 
        }
200 
 
201 
        if(_state==StringLiteral) throw std::runtime_error("Unexpected end of line");
202 
        if(_state==Word) tokenList.push_back(std::move(word)); // store last word
203 
        if(_state!=BlockComment) _state=Initial; // reset state if not in block comment
204 
 
205 
        return tokenList;
206 
}
207 
 
208 
void Assembler::expand(TokenList &list) {
209 
        TokenList newlist;
210 
// Perform macro substitution
211 
        for(auto &token: list) {
212 
                auto it=_macros.find(token);
213 
                bool substitute=false;
214 
                if(it!=_macros.end()) {
215 
                        substitute=true;
216 
// Don't substitute macros for a second token in certain directives
217 
                        if(newlist.size()==1) {
218 
                                if(newlist[0]=="#define") substitute=false;
219 
                                else if(newlist[0]=="#ifdef") substitute=false;
220 
                                else if(newlist[0]=="#ifndef") substitute=false;
221 
                        }
222 
                        else if(newlist.size()==3&&newlist[1]==":") {
223 
                                if(newlist[2]=="#define") substitute=false;
224 
                                else if(newlist[2]=="#ifdef") substitute=false;
225 
                                else if(newlist[2]=="#ifndef") substitute=false;
226 
                        }
227 
                }
228 
 
229 
                if(substitute) {
230 
                        for(auto const &replace: it->second) newlist.push_back(replace);
231 
                }
232 
                else newlist.push_back(std::move(token));
233 
        }
234 
        list=std::move(newlist);
235 
}
236 
 
237 
void Assembler::elaborate(TokenList &list) {
238 
        if(list.empty()) return;
239 
 
240 
// Process label (if present)
241 
        if(list.size()>=2&&list[1]==":") {
242 
                if(!validateIdentifier(list[0]))
243 
                        throw std::runtime_error("Ill-formed identifier: \""+list[0]+"\"");
244 
                if(isSectionEnabled()) _currentLabels.push_back(std::move(list[0]));
245 
                list.erase(list.begin(),list.begin()+2);
246 
        }
247 
 
248 
        if(list.empty()) return;
249 
 
250 
// If the section is disabled, we look only for #ifdef, #ifndef, #else or #endif
251 
        if(!isSectionEnabled()&&list[0]!="#ifdef"&&list[0]!="#ifndef"&&
252 
                list[0]!="#else"&&list[0]!="#endif") return;
253 
 
254 
// Process statement itself
255 
        if(list[0][0]=='#') elaborateDirective(list);
256 
        else {
257 
                LinkableObject::Word rva;
258 
                if(list[0][0]=='.') rva=elaborateDataDefinition(list);
259 
                else rva=elaborateInstruction(list);
260 
 
261 
                for(auto const &label: _currentLabels) {
262 
                        _obj.addSymbol(label,rva);
263 
                }
264 
                _currentLabels.clear();
265 
        }
266 
}
267 
 
268 
void Assembler::elaborateDirective(TokenList &list) {
269 
        assert(!list.empty());
270 
 
271 
        if(list[0]=="#define") {
272 
                if(list.size()<2)
273 
                        throw std::runtime_error("Wrong number of tokens in the directive");
274 
                if(_macros.find(list[1])!=_macros.end())
275 
                        throw std::runtime_error("Macro \""+list[1]+"\" has been already defined");
276 
                if(!validateIdentifier(list[1]))
277 
                        throw std::runtime_error("Ill-formed identifier: \""+list[1]+"\"");
278 
                _macros.emplace(list[1],TokenList(list.begin()+2,list.end()));
279 
        }
280 
        else if(list[0]=="#export") {
281 
                if(list.size()!=2) throw std::runtime_error("Wrong number of tokens in the directive");
282 
                if(!validateIdentifier(list[1])) throw std::runtime_error("Ill-formed identifier: \""+list[1]+"\"");
283 
                _exportedSymbols.push_back(list[1]);
284 
        }
285 
        else if(list[0]=="#import") {
286 
                if(list.size()!=2) throw std::runtime_error("Wrong number of tokens in the directive");
287 
                if(!validateIdentifier(list[1])) throw std::runtime_error("Ill-formed identifier: \""+list[1]+"\"");
288 
                _obj.addImportedSymbol(list[1]);
289 
        }
290 
        else if(list[0]=="#include") {
291 
                if(list.size()!=2) throw std::runtime_error("Wrong number of tokens in the directive");
292 
                auto filename=Utils::dequoteString(list[1]);
293 
                if(Utils::isAbsolutePath(filename)) return processFileRecursive(filename);
294 
                else {
295 
                        auto path=Utils::relativePath(currentFileName(),filename);
296 
                        if(Utils::fileExists(path)) return processFileRecursive(path);
297 
                        else {
298 
                                for(auto const &dir: _includeSearchDirs) {
299 
                                        path=Utils::nativeSeparators(dir+filename);
300 
                                        if(Utils::fileExists(path)) return processFileRecursive(path);
301 
                                }
302 
                        }
303 
                }
304 
                throw std::runtime_error("Cannot locate include file \""+filename+"\"");
305 
        }
306 
        else if(list[0]=="#message") {
307 
                if(list.size()!=2) throw std::runtime_error("Wrong number of tokens in the directive");
308 
                auto msg=Utils::dequoteString(list[1]);
309 
                std::cout<<currentFileName()<<":"<<line()<<": "<<msg<<std::endl;
310 
        }
311 
        else if(list[0]=="#error") {
312 
                if(list.size()<2) throw std::runtime_error("#error directive encountered");
313 
                auto msg=Utils::dequoteString(list[1]);
314 
                throw std::runtime_error(msg);
315 
        }
316 
        else if(list[0]=="#ifdef") {
317 
                if(list.size()!=2) throw std::runtime_error("Wrong number of tokens in the directive");
318 
                if(_macros.find(list[1])!=_macros.end()) _sectionEnabled.push_back(true);
319 
                else _sectionEnabled.push_back(false);
320 
        }
321 
        else if(list[0]=="#ifndef") {
322 
                if(list.size()!=2) throw std::runtime_error("Wrong number of tokens in the directive");
323 
                if(_macros.find(list[1])!=_macros.end()) _sectionEnabled.push_back(false);
324 
                else _sectionEnabled.push_back(true);
325 
        }
326 
        else if(list[0]=="#else") {
327 
                if(list.size()!=1) throw std::runtime_error("Wrong number of tokens in the directive");
328 
                if(_sectionEnabled.empty()) throw std::runtime_error("Unexpected #else");
329 
                _sectionEnabled.back()=!_sectionEnabled.back();
330 
        }
331 
        else if(list[0]=="#endif") {
332 
                if(list.size()!=1) throw std::runtime_error("Wrong number of tokens in the directive");
333 
                if(_sectionEnabled.empty()) throw std::runtime_error("Unexpected #endif");
334 
                _sectionEnabled.pop_back();
335 
        }
336 
        else throw std::runtime_error("Unrecognized directive: \""+list[0]+"\"");
337 
}
338 
 
339 
LinkableObject::Word Assembler::elaborateDataDefinition(TokenList &list) {
340 
        assert(!list.empty());
341 
 
342 
        LinkableObject::Word rva=0;
343 
 
344 
        if(list[0]==".align") {
345 
                if(list.size()>2) throw std::runtime_error("Unexpected token: \""+list[2]+"\"");
346 
                std::size_t align=4;
347 
                if(list.size()>1) align=static_cast<std::size_t>(numericLiteral(list[1]));
348 
                if(!Utils::isPowerOf2(align)) throw std::runtime_error("Alignment must be a power of 2");
349 
                if(align<4) throw std::runtime_error("Alignment must be at least 4");
350 
                rva=_obj.addPadding(align);
351 
        }
352 
        else if(list[0]==".reserve") {
353 
                if(list.size()<2) throw std::runtime_error("Unexpected end of statement");
354 
                else if(list.size()>2) throw std::runtime_error("Unexpected token: \""+list[2]+"\"");
355 
                auto n=static_cast<std::size_t>(numericLiteral(list[1]));
356 
                rva=_obj.addZeros(n);
357 
        }
358 
        else if(list[0]==".word") {
359 
                if(list.size()<2) throw std::runtime_error("Unexpected end of statement");
360 
                for(std::size_t i=1;i<list.size();i++) {
361 
                        if(i%2!=0) {
362 
                                auto w=static_cast<LinkableObject::Word>(numericLiteral(list[i]));
363 
                                auto r=_obj.addWord(w);
364 
                                if(i==1) rva=r;
365 
                        }
366 
                        else {
367 
                                if(list[i]!=",") throw std::runtime_error("Comma expected");
368 
                                if(i+1==list.size()) throw std::runtime_error("Unexpected end of statement");
369 
                        }
370 
                }
371 
        }
372 
        else if(list[0]==".byte") {
373 
                if(list.size()<2) throw std::runtime_error("Unexpected end of statement");
374 
                for(std::size_t i=1;i<list.size();i++) {
375 
                        if(i%2!=0) {
376 
                                if(list[i].at(0)=='\"') { // string literal
377 
                                        auto bytes=Utils::dequoteString(list[i]);
378 
                                        auto r=_obj.addBytes(reinterpret_cast<const LinkableObject::Byte*>
379 
                                                (bytes.c_str()),bytes.size());
380 
                                        if(i==1) rva=r;
381 
                                }
382 
                                else {
383 
                                        auto n=numericLiteral(list[i]);
384 
 
385 
                                        if(n>255||n<-128) throw std::runtime_error("\""+list[i]+"\": out of range");
386 
 
387 
                                        auto b=static_cast<LinkableObject::Byte>(n);
388 
                                        auto r=_obj.addByte(b);
389 
                                        if(i==1) rva=r;
390 
                                }
391 
                        }
392 
                        else {
393 
                                if(list[i]!=",") throw std::runtime_error("Comma expected");
394 
                                if(i+1==list.size()) throw std::runtime_error("Unexpected end of statement");
395 
                        }
396 
                }
397 
        }
398 
        else throw std::runtime_error("Unrecognized statement: \""+list[0]+"\"");
399 
 
400 
        return rva;
401 
}
402 
 
403 
LinkableObject::Word Assembler::elaborateInstruction(TokenList &list) {
404 
        assert(!list.empty());
405 
        auto rva=_obj.addPadding();
406 
        if(list[0]=="add") encodeAdd(list);
407 
        else if(list[0]=="and") encodeAnd(list);
408 
        else if(list[0]=="call") encodeCall(list);
409 
        else if(list[0].substr(0,4)=="cjmp") encodeCjmpxx(list);
410 
        else if(list[0]=="divs") encodeDivs(list);
411 
        else if(list[0]=="divu") encodeDivu(list);
412 
        else if(list[0]=="hlt") encodeHlt(list);
413 
        else if(list[0]=="jmp") encodeJmp(list);
414 
        else if(list[0]=="iret") encodeIret(list);
415 
        else if(list[0]=="lc") encodeLc(list);
416 
        else if(list[0]=="lcs") encodeLcs(list);
417 
        else if(list[0]=="lsb") encodeLsb(list);
418 
        else if(list[0]=="lub") encodeLub(list);
419 
        else if(list[0]=="lw") encodeLw(list);
420 
        else if(list[0]=="mods") encodeMods(list);
421 
        else if(list[0]=="modu") encodeModu(list);
422 
        else if(list[0]=="mov") encodeMov(list);
423 
        else if(list[0]=="mul") encodeMul(list);
424 
        else if(list[0]=="neg") encodeNeg(list);
425 
        else if(list[0]=="nop") encodeNop(list);
426 
        else if(list[0]=="not") encodeNot(list);
427 
        else if(list[0]=="or") encodeOr(list);
428 
        else if(list[0]=="ret") encodeRet(list);
429 
        else if(list[0]=="sb") encodeSb(list);
430 
        else if(list[0]=="sl") encodeSl(list);
431 
        else if(list[0]=="srs") encodeSrs(list);
432 
        else if(list[0]=="sru") encodeSru(list);
433 
        else if(list[0]=="sub") encodeSub(list);
434 
        else if(list[0]=="sw") encodeSw(list);
435 
        else if(list[0]=="xor") encodeXor(list);
436 
        else throw std::runtime_error("Unrecognized instruction: \""+list[0]+"\"");
437 
        return rva;
438 
}
439 
 
440 
bool Assembler::isSectionEnabled() const {
441 
        if(_sectionEnabled.empty()) return true;
442 
        bool enabled=true;
443 
        for(auto b: _sectionEnabled) enabled=enabled&&b;
444 
        return enabled;
445 
}
446 
 
447 
bool Assembler::validateIdentifier(const std::string &str) {
448 
/*
449 
 * Valid identifier must satisfy the following requirements:
450 
 *  1. Must not be empty
451 
 *  2. The first character must be either alphabetic or an underscore
452 
 *  3. Subsequent characters must be either alphanumeric or underscores
453 
 */
454 
        if(str.empty()) return false;
455 
        for(std::size_t i=0;i<str.size();i++) {
456 
                char ch=str[i];
457 
                if(i==0) {
458 
                        if(!std::isalpha(ch)&&ch!='_') return false;
459 
                }
460 
                else {
461 
                        if(!std::isalnum(ch)&&ch!='_') return false;
462 
                }
463 
        }
464 
        return true;
465 
}
466 
 
467 
Assembler::Integer Assembler::numericLiteral(const std::string &str) {
468 
        std::size_t pos;
469 
        Integer i;
470 
        try {
471 
                i=std::stoll(str,&pos,0);
472 
        }
473 
        catch(std::exception &) {
474 
                throw std::runtime_error("Ill-formed numeric literal: \""+str+"\"");
475 
        }
476 
        if(pos<str.size()) throw std::runtime_error("Ill-formed numeric literal: \""+str+"\"");
477 
 
478 
        typedef std::make_signed<LinkableObject::Word>::type SignedWord;
479 
 
480 
        if(i>static_cast<Integer>(std::numeric_limits<LinkableObject::Word>::max())||
481 
                i<static_cast<Integer>(std::numeric_limits<SignedWord>::min()))
482 
                        throw std::runtime_error("\""+str+"\": out of range");
483 
 
484 
        return i;
485 
}
486 
 
487 
std::vector<Assembler::Operand> Assembler::getOperands(const TokenList &list) {
488 
        std::vector<Operand> arglist;
489 
        for(std::size_t i=1;i<list.size();i++) {
490 
                if(i%2!=0) {
491 
                        Operand a;
492 
                        a.str=list[i];
493 
 
494 
                        if(!list[i].empty()&&list[i][0]=='r') {
495 
// Is argument a register?
496 
                                char *endptr;
497 
                                auto regstr=list[i].substr(1);
498 
                                auto reg=std::strtol(regstr.c_str(),&endptr,10);
499 
 
500 
                                if(!*endptr&&reg>=0&&reg<=255) {
501 
                                        a.type=Operand::Register;
502 
                                        a.reg=static_cast<std::uint8_t>(reg);
503 
                                        arglist.push_back(std::move(a));
504 
                                        continue;
505 
                                }
506 
                        }
507 
 
508 
// Try alternative register names
509 
                        if(list[i]=="sp") { // stack pointer
510 
                                a.type=Operand::Register;
511 
                                a.reg=255;
512 
                                arglist.push_back(std::move(a));
513 
                        }
514 
                        else if(list[i]=="rp") { // return pointer
515 
                                a.type=Operand::Register;
516 
                                a.reg=254;
517 
                                arglist.push_back(std::move(a));
518 
                        }
519 
                        else if(list[i]=="irp") { // interrupt return pointer
520 
                                a.type=Operand::Register;
521 
                                a.reg=253;
522 
                                arglist.push_back(std::move(a));
523 
                        }
524 
                        else if(list[i]=="cr") { // control register
525 
                                a.type=Operand::Register;
526 
                                a.reg=252;
527 
                                arglist.push_back(std::move(a));
528 
                        }
529 
                        else if(list[i].size()==3&&list[i].substr(0,2)=="iv"&&
530 
                                list[i][2]>='0'&&list[i][2]<='7') // interrupt vector
531 
                        {
532 
                                a.type=Operand::Register;
533 
                                a.reg=240+(list[i][2]-'0');
534 
                                arglist.push_back(std::move(a));
535 
                        }
536 
                        else if(validateIdentifier(list[i])) {
537 
// Is argument an identifier?
538 
                                a.type=Operand::Identifier;
539 
                                arglist.push_back(std::move(a));
540 
                        }
541 
                        else {
542 
                                auto atpos=list[i].find_first_of('@');
543 
                                if(atpos!=std::string::npos) {
544 
// Identifier with an offset?
545 
                                        a.type=Operand::Identifier;
546 
                                        a.str=list[i].substr(0,atpos);
547 
                                        if(!validateIdentifier(a.str)) throw std::runtime_error("Ill-formed identifier");
548 
                                        a.i=numericLiteral(list[i].substr(atpos+1));
549 
                                        arglist.push_back(std::move(a));
550 
                                }
551 
                                else {
552 
// Numeric literal?
553 
                                        a.type=Operand::NumericLiteral;
554 
                                        a.i=numericLiteral(list[i]);
555 
                                        arglist.push_back(std::move(a));
556 
                                }
557 
                        }
558 
                }
559 
                else {
560 
                        if(list[i]!=",") throw std::runtime_error("Comma expected");
561 
                        if(i+1==list.size()) throw std::runtime_error("Unexpected end of line");
562 
                }
563 
        }
564 
        return arglist;
565 
}
566 
 
567 
/*
568 
 * Member functions to encode LXP32 instructions
569 
 */
570 
 
571 
void Assembler::encodeDstOperand(LinkableObject::Word &word,const Operand &arg) {
572 
        if(arg.type!=Operand::Register)
573 
                throw std::runtime_error("\""+arg.str+"\": must be a register");
574 
        word|=arg.reg<<16;
575 
}
576 
 
577 
void Assembler::encodeRd1Operand(LinkableObject::Word &word,const Operand &arg) {
578 
        if(arg.type==Operand::Register) {
579 
                word|=0x02000000;
580 
                word|=arg.reg<<8;
581 
        }
582 
        else if(arg.type==Operand::NumericLiteral) {
583 
                if((arg.i<-128||arg.i>127)&&(arg.i<0xFFFFFF80||arg.i>0xFFFFFFFF))
584 
                        throw std::runtime_error("\""+arg.str+"\": out of range");
585 
                auto b=static_cast<LinkableObject::Byte>(arg.i);
586 
                word|=b<<8;
587 
        }
588 
        else throw std::runtime_error("\""+arg.str+"\": bad argument");
589 
}
590 
 
591 
void Assembler::encodeRd2Operand(LinkableObject::Word &word,const Operand &arg) {
592 
        if(arg.type==Operand::Register) {
593 
                word|=0x01000000;
594 
                word|=arg.reg;
595 
        }
596 
        else if(arg.type==Operand::NumericLiteral) {
597 
                if((arg.i<-128||arg.i>127)&&(arg.i<0xFFFFFF80||arg.i>0xFFFFFFFF))
598 
                        throw std::runtime_error("\""+arg.str+"\": out of range");
599 
                auto b=static_cast<LinkableObject::Byte>(arg.i);
600 
                word|=b;
601 
        }
602 
        else throw std::runtime_error("\""+arg.str+"\": bad argument");
603 
}
604 
 
605 
void Assembler::encodeAdd(const TokenList &list) {
606 
        auto args=getOperands(list);
607 
        if(args.size()!=3) throw std::runtime_error("add instruction requires 3 operands");
608 
        LinkableObject::Word w=0x40000000;
609 
        encodeDstOperand(w,args[0]);
610 
        encodeRd1Operand(w,args[1]);
611 
        encodeRd2Operand(w,args[2]);
612 
        _obj.addWord(w);
613 
}
614 
 
615 
void Assembler::encodeAnd(const TokenList &list) {
616 
        auto args=getOperands(list);
617 
        if(args.size()!=3) throw std::runtime_error("and instruction requires 3 operands");
618 
        LinkableObject::Word w=0x60000000;
619 
        encodeDstOperand(w,args[0]);
620 
        encodeRd1Operand(w,args[1]);
621 
        encodeRd2Operand(w,args[2]);
622 
        _obj.addWord(w);
623 
}
624 
 
625 
void Assembler::encodeCall(const TokenList &list) {
626 
        auto args=getOperands(list);
627 
        if(args.size()!=1) throw std::runtime_error("call instruction requires 1 operand");
628 
        if(args[0].type!=Operand::Register) throw std::runtime_error("\""+args[0].str+"\": must be a register");
629 
        LinkableObject::Word w=0x86FE0000;
630 
        encodeRd1Operand(w,args[0]);
631 
        _obj.addWord(w);
632 
}
633 
 
634 
void Assembler::encodeCjmpxx(const TokenList &list) {
635 
        auto args=getOperands(list);
636 
        if(args.size()!=3) throw std::runtime_error("cjmpxx instruction requires 3 operands");
637 
 
638 
        LinkableObject::Word w;
639 
        bool reverse=false;
640 
/*
641 
 * Note: cjmpul, cjmpule, cjmpsl and cjmpsle don't have distinct opcodes;
642 
 * instead, they are aliases for respective "g" or "ge" instructions
643 
 * with reversed operand order.
644 
 */
645 
        if(list[0]=="cjmpe") w=0xE0000000;
646 
        else if(list[0]=="cjmpne") w=0xD0000000;
647 
        else if(list[0]=="cjmpug"||list[0]=="cjmpul") w=0xC8000000;
648 
        else if(list[0]=="cjmpuge"||list[0]=="cjmpule") w=0xE8000000;
649 
        else if(list[0]=="cjmpsg"||list[0]=="cjmpsl") w=0xC4000000;
650 
        else if(list[0]=="cjmpsge"||list[0]=="cjmpsle") w=0xE4000000;
651 
        else throw std::runtime_error("Unrecognized instruction: \""+list[0]+"\"");
652 
 
653 
        if(list[0]=="cjmpul"||list[0]=="cjmpule"||
654 
                list[0]=="cjmpsl"||list[0]=="cjmpsle") reverse=true;
655 
 
656 
        encodeDstOperand(w,args[0]);
657 
 
658 
        if(!reverse) {
659 
                encodeRd1Operand(w,args[1]);
660 
                encodeRd2Operand(w,args[2]);
661 
        }
662 
        else {
663 
                encodeRd1Operand(w,args[2]);
664 
                encodeRd2Operand(w,args[1]);
665 
        }
666 
        _obj.addWord(w);
667 
}
668 
 
669 
void Assembler::encodeDivs(const TokenList &list) {
670 
        auto args=getOperands(list);
671 
        if(args.size()!=3) throw std::runtime_error("divs instruction requires 3 operands");
672 
        LinkableObject::Word w=0x54000000;
673 
        encodeDstOperand(w,args[0]);
674 
        encodeRd1Operand(w,args[1]);
675 
        encodeRd2Operand(w,args[2]);
676 
        _obj.addWord(w);
677 
}
678 
 
679 
void Assembler::encodeDivu(const TokenList &list) {
680 
        auto args=getOperands(list);
681 
        if(args.size()!=3) throw std::runtime_error("divu instruction requires 3 operands");
682 
        LinkableObject::Word w=0x50000000;
683 
        encodeDstOperand(w,args[0]);
684 
        encodeRd1Operand(w,args[1]);
685 
        encodeRd2Operand(w,args[2]);
686 
        _obj.addWord(w);
687 
}
688 
 
689 
void Assembler::encodeHlt(const TokenList &list) {
690 
        auto args=getOperands(list);
691 
        if(!args.empty()) throw std::runtime_error("hlt instruction doesn't take operands");
692 
        _obj.addWord(0x08000000);
693 
}
694 
 
695 
void Assembler::encodeJmp(const TokenList &list) {
696 
        auto args=getOperands(list);
697 
        if(args.size()!=1) throw std::runtime_error("jmp instruction requires 1 operand");
698 
        if(args[0].type!=Operand::Register) throw std::runtime_error("\""+args[0].str+"\": must be a register");
699 
        LinkableObject::Word w=0x82000000;
700 
        encodeRd1Operand(w,args[0]);
701 
        _obj.addWord(w);
702 
}
703 
 
704 
void Assembler::encodeIret(const TokenList &list) {
705 
// Note: "iret" is not a real instruction, but an alias for "jmp irp"
706 
        auto args=getOperands(list);
707 
        if(!args.empty()) throw std::runtime_error("iret instruction doesn't take operands");
708 
        _obj.addWord(0x8200FD00);
709 
}
710 
 
711 
void Assembler::encodeLc(const TokenList &list) {
712 
        auto args=getOperands(list);
713 
        if(args.size()!=2) throw std::runtime_error("lc instruction requires 2 operands");
714 
 
715 
        LinkableObject::Word w=0x04000000;
716 
        encodeDstOperand(w,args[0]);
717 
        _obj.addWord(w);
718 
 
719 
        if(args[1].type==Operand::Identifier) {
720 
                LinkableObject::Reference ref;
721 
                ref.source=currentFileName();
722 
                ref.line=line();
723 
                ref.rva=_obj.addWord(0);
724 
                ref.offset=args[1].i;
725 
                ref.type=LinkableObject::Regular;
726 
                _obj.addReference(args[1].str,ref);
727 
        }
728 
        else if(args[1].type==Operand::NumericLiteral) {
729 
                _obj.addWord(static_cast<LinkableObject::Word>(args[1].i));
730 
        }
731 
        else throw std::runtime_error("\""+args[1].str+"\": bad argument");
732 
}
733 
 
734 
void Assembler::encodeLcs(const TokenList &list) {
735 
        auto args=getOperands(list);
736 
        if(args.size()!=2) throw std::runtime_error("lcs instruction requires 2 operands");
737 
 
738 
        LinkableObject::Word w=0xA0000000;
739 
        encodeDstOperand(w,args[0]);
740 
 
741 
        if(args[1].type==Operand::NumericLiteral) {
742 
                if((args[1].i<-1048576||args[1].i>1048575)&&(args[1].i<0xFFF00000||args[1].i>0xFFFFFFFF))
743 
                        throw std::runtime_error("\""+args[1].str+"\": out of range");
744 
                auto c=static_cast<LinkableObject::Word>(args[1].i)&0x1FFFFF;
745 
                w|=(c&0xFFFF);
746 
                w|=((c<<8)&0x1F000000);
747 
                _obj.addWord(w);
748 
        }
749 
        else if(args[1].type==Operand::Identifier) {
750 
                LinkableObject::Reference ref;
751 
                ref.source=currentFileName();
752 
                ref.line=line();
753 
                ref.rva=_obj.addWord(w);
754 
                ref.offset=args[1].i;
755 
                ref.type=LinkableObject::Short;
756 
                _obj.addReference(args[1].str,ref);
757 
        }
758 
        else throw std::runtime_error("\""+args[1].str+"\": bad argument");
759 
}
760 
 
761 
void Assembler::encodeLsb(const TokenList &list) {
762 
        auto args=getOperands(list);
763 
        if(args.size()!=2) throw std::runtime_error("lsb instruction requires 2 operands");
764 
        if(args[1].type!=Operand::Register) throw std::runtime_error("\""+args[1].str+"\": must be a register");
765 
        LinkableObject::Word w=0x2E000000;
766 
        encodeDstOperand(w,args[0]);
767 
        encodeRd1Operand(w,args[1]);
768 
        _obj.addWord(w);
769 
}
770 
 
771 
void Assembler::encodeLub(const TokenList &list) {
772 
        auto args=getOperands(list);
773 
        if(args.size()!=2) throw std::runtime_error("lub instruction requires 2 operands");
774 
        if(args[1].type!=Operand::Register) throw std::runtime_error("\""+args[1].str+"\": must be a register");
775 
        LinkableObject::Word w=0x2A000000;
776 
        encodeDstOperand(w,args[0]);
777 
        encodeRd1Operand(w,args[1]);
778 
        _obj.addWord(w);
779 
}
780 
 
781 
void Assembler::encodeLw(const TokenList &list) {
782 
        auto args=getOperands(list);
783 
        if(args.size()!=2) throw std::runtime_error("lw instruction requires 2 operands");
784 
        if(args[1].type!=Operand::Register) throw std::runtime_error("\""+args[1].str+"\": must be a register");
785 
        LinkableObject::Word w=0x22000000;
786 
        encodeDstOperand(w,args[0]);
787 
        encodeRd1Operand(w,args[1]);
788 
        _obj.addWord(w);
789 
}
790 
 
791 
void Assembler::encodeMods(const TokenList &list) {
792 
        auto args=getOperands(list);
793 
        if(args.size()!=3) throw std::runtime_error("mods instruction requires 3 operands");
794 
        LinkableObject::Word w=0x5C000000;
795 
        encodeDstOperand(w,args[0]);
796 
        encodeRd1Operand(w,args[1]);
797 
        encodeRd2Operand(w,args[2]);
798 
        _obj.addWord(w);
799 
}
800 
 
801 
void Assembler::encodeModu(const TokenList &list) {
802 
        auto args=getOperands(list);
803 
        if(args.size()!=3) throw std::runtime_error("modu instruction requires 3 operands");
804 
        LinkableObject::Word w=0x58000000;
805 
        encodeDstOperand(w,args[0]);
806 
        encodeRd1Operand(w,args[1]);
807 
        encodeRd2Operand(w,args[2]);
808 
        _obj.addWord(w);
809 
}
810 
 
811 
void Assembler::encodeMov(const TokenList &list) {
812 
// Note: "mov" is not a real instruction, but an alias for "add dst, src, 0"
813 
        auto args=getOperands(list);
814 
        if(args.size()!=2) throw std::runtime_error("mov instruction requires 2 operands");
815 
        LinkableObject::Word w=0x40000000;
816 
        encodeDstOperand(w,args[0]);
817 
        encodeRd1Operand(w,args[1]);
818 
        _obj.addWord(w);
819 
}
820 
 
821 
void Assembler::encodeMul(const TokenList &list) {
822 
        auto args=getOperands(list);
823 
        if(args.size()!=3) throw std::runtime_error("mul instruction requires 3 operands");
824 
        LinkableObject::Word w=0x48000000;
825 
        encodeDstOperand(w,args[0]);
826 
        encodeRd1Operand(w,args[1]);
827 
        encodeRd2Operand(w,args[2]);
828 
        _obj.addWord(w);
829 
}
830 
 
831 
void Assembler::encodeNeg(const TokenList &list) {
832 
// Note: "neg" is not a real instruction, but an alias for "sub dst, 0, src"
833 
        auto args=getOperands(list);
834 
        if(args.size()!=2) throw std::runtime_error("neg instruction requires 2 operands");
835 
        LinkableObject::Word w=0x44000000;
836 
        encodeDstOperand(w,args[0]);
837 
        encodeRd2Operand(w,args[1]);
838 
        _obj.addWord(w);
839 
}
840 
 
841 
void Assembler::encodeNop(const TokenList &list) {
842 
        auto args=getOperands(list);
843 
        if(!args.empty()) throw std::runtime_error("nop instruction doesn't take operands");
844 
        _obj.addWord(0);
845 
}
846 
 
847 
void Assembler::encodeNot(const TokenList &list) {
848 
// Note: "not" is not a real instruction, but an alias for "xor dst, src, -1"
849 
        auto args=getOperands(list);
850 
        if(args.size()!=2) throw std::runtime_error("not instruction requires 2 operands");
851 
        LinkableObject::Word w=0x680000FF;
852 
        encodeDstOperand(w,args[0]);
853 
        encodeRd1Operand(w,args[1]);
854 
        _obj.addWord(w);
855 
}
856 
 
857 
void Assembler::encodeOr(const TokenList &list) {
858 
        auto args=getOperands(list);
859 
        if(args.size()!=3) throw std::runtime_error("or instruction requires 3 operands");
860 
        LinkableObject::Word w=0x64000000;
861 
        encodeDstOperand(w,args[0]);
862 
        encodeRd1Operand(w,args[1]);
863 
        encodeRd2Operand(w,args[2]);
864 
        _obj.addWord(w);
865 
}
866 
 
867 
void Assembler::encodeRet(const TokenList &list) {
868 
// Note: "ret" is not a real instruction, but an alias for "jmp rp"
869 
        auto args=getOperands(list);
870 
        if(!args.empty()) throw std::runtime_error("ret instruction doesn't take operands");
871 
        _obj.addWord(0x8200FE00);
872 
}
873 
 
874 
void Assembler::encodeSb(const TokenList &list) {
875 
        auto args=getOperands(list);
876 
        if(args.size()!=2) throw std::runtime_error("sb instruction requires 2 operands");
877 
        if(args[0].type!=Operand::Register) throw std::runtime_error("\""+args[0].str+"\": must be a register");
878 
        if(args[1].type==Operand::NumericLiteral) {
879 
// If numeric literal value is between 128 and 255 (inclusive), convert
880 
// it to a signed byte to avoid exception in encodeRd2Operand()
881 
                if(args[1].i>=128&&args[1].i<=255) args[1].i-=256;
882 
        }
883 
        LinkableObject::Word w=0x3A000000;
884 
        encodeRd1Operand(w,args[0]);
885 
        encodeRd2Operand(w,args[1]);
886 
        _obj.addWord(w);
887 
}
888 
 
889 
void Assembler::encodeSl(const TokenList &list) {
890 
        auto args=getOperands(list);
891 
        if(args.size()!=3) throw std::runtime_error("sl instruction requires 3 operands");
892 
        if(args[2].type==Operand::NumericLiteral&&
893 
                (args[2].i<0||args[2].i>=static_cast<Integer>(8*sizeof(LinkableObject::Word))))
894 
        {
895 
                        std::cerr<<currentFileName()<<":"<<line()<<": ";
896 
                        std::cerr<<"Warning: Bitwise shift result is undefined when "
897 
                        "the second operand is negative or greater than 31"<<std::endl;
898 
        }
899 
 
900 
        LinkableObject::Word w=0x70000000;
901 
        encodeDstOperand(w,args[0]);
902 
        encodeRd1Operand(w,args[1]);
903 
        encodeRd2Operand(w,args[2]);
904 
        _obj.addWord(w);
905 
}
906 
 
907 
void Assembler::encodeSrs(const TokenList &list) {
908 
        auto args=getOperands(list);
909 
        if(args.size()!=3) throw std::runtime_error("srs instruction requires 3 operands");
910 
        if(args[2].type==Operand::NumericLiteral&&
911 
                (args[2].i<0||args[2].i>=static_cast<Integer>(8*sizeof(LinkableObject::Word))))
912 
        {
913 
                        std::cerr<<currentFileName()<<":"<<line()<<": ";
914 
                        std::cerr<<"Warning: Bitwise shift result is undefined when "
915 
                        "the second operand is negative or greater than 31"<<std::endl;
916 
        }
917 
 
918 
        LinkableObject::Word w=0x7C000000;
919 
        encodeDstOperand(w,args[0]);
920 
        encodeRd1Operand(w,args[1]);
921 
        encodeRd2Operand(w,args[2]);
922 
        _obj.addWord(w);
923 
}
924 
 
925 
void Assembler::encodeSru(const TokenList &list) {
926 
        auto args=getOperands(list);
927 
        if(args.size()!=3) throw std::runtime_error("sru instruction requires 3 operands");
928 
        if(args[2].type==Operand::NumericLiteral&&
929 
                (args[2].i<0||args[2].i>=static_cast<Integer>(8*sizeof(LinkableObject::Word))))
930 
        {
931 
                        std::cerr<<currentFileName()<<":"<<line()<<": ";
932 
                        std::cerr<<"Warning: Bitwise shift result is undefined when "
933 
                        "the second operand is negative or greater than 31"<<std::endl;
934 
        }
935 
 
936 
        LinkableObject::Word w=0x78000000;
937 
        encodeDstOperand(w,args[0]);
938 
        encodeRd1Operand(w,args[1]);
939 
        encodeRd2Operand(w,args[2]);
940 
        _obj.addWord(w);
941 
}
942 
 
943 
void Assembler::encodeSub(const TokenList &list) {
944 
        auto args=getOperands(list);
945 
        if(args.size()!=3) throw std::runtime_error("sub instruction requires 3 operands");
946 
        LinkableObject::Word w=0x44000000;
947 
        encodeDstOperand(w,args[0]);
948 
        encodeRd1Operand(w,args[1]);
949 
        encodeRd2Operand(w,args[2]);
950 
        _obj.addWord(w);
951 
}
952 
 
953 
void Assembler::encodeSw(const TokenList &list) {
954 
        auto args=getOperands(list);
955 
        if(args.size()!=2) throw std::runtime_error("sw instruction requires 2 operands");
956 
        if(args[0].type!=Operand::Register) throw std::runtime_error("\""+args[0].str+"\": must be a register");
957 
        LinkableObject::Word w=0x32000000;
958 
        encodeRd1Operand(w,args[0]);
959 
        encodeRd2Operand(w,args[1]);
960 
        _obj.addWord(w);
961 
}
962 
 
963 
void Assembler::encodeXor(const TokenList &list) {
964 
        auto args=getOperands(list);
965 
        if(args.size()!=3) throw std::runtime_error("xor instruction requires 3 operands");
966 
        LinkableObject::Word w=0x68000000;
967 
        encodeDstOperand(w,args[0]);
968 
        encodeRd1Operand(w,args[1]);
969 
        encodeRd2Operand(w,args[2]);
970 
        _obj.addWord(w);
971 
}

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.