blob: 816902e1f1bc0948ec409039af227647a7508469 [file] [log] [blame]
Jack Palevichae54f1f2009-05-08 14:54:15 -07001/*
Jack Paleviche7b59062009-05-19 17:12:17 -07002 * Android "Almost" C Compiler.
3 * This is a compiler for a small subset of the C language, intended for use
4 * in scripting environments where speed and memory footprint are important.
5 *
6 * This code is based upon the "unobfuscated" version of the
Jack Palevich1cdef202009-05-22 12:06:27 -07007 * Obfuscated Tiny C compiler, see the file LICENSE for details.
Jack Paleviche7b59062009-05-19 17:12:17 -07008 *
9 */
10
Jack Palevich77ae76e2009-05-10 19:59:24 -070011#include <ctype.h>
12#include <dlfcn.h>
Jack Palevich8dc662e2009-06-09 22:53:47 +000013#include <errno.h>
Jack Paleviche27bf3e2009-05-10 14:09:03 -070014#include <stdarg.h>
Jack Palevich8b0624c2009-05-20 12:12:06 -070015#include <stdint.h>
Jack Palevichae54f1f2009-05-08 14:54:15 -070016#include <stdio.h>
Jack Palevichf6b5a532009-05-10 19:16:42 -070017#include <stdlib.h>
18#include <string.h>
Jack Palevich2d11dfb2009-06-08 14:34:26 -070019#include <cutils/hashmap.h>
Jack Palevichae54f1f2009-05-08 14:54:15 -070020
Jack Palevich8dc662e2009-06-09 22:53:47 +000021#if defined(__i386__)
22#include <sys/mman.h>
23#endif
24
Jack Palevich546b2242009-05-13 15:10:04 -070025#if defined(__arm__)
26#include <unistd.h>
27#endif
28
Jack Paleviche7b59062009-05-19 17:12:17 -070029#if defined(__arm__)
30#define DEFAULT_ARM_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -070031#define PROVIDE_ARM_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -070032#elif defined(__i386__)
33#define DEFAULT_X86_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -070034#define PROVIDE_X86_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -070035#elif defined(__x86_64__)
36#define DEFAULT_X64_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -070037#define PROVIDE_X64_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -070038#endif
39
Jack Paleviche7b59062009-05-19 17:12:17 -070040
41#ifdef PROVIDE_ARM_CODEGEN
Jack Palevicha6535612009-05-13 16:24:17 -070042#include "disassem.h"
Jack Paleviche7b59062009-05-19 17:12:17 -070043#endif
Jack Palevicha6535612009-05-13 16:24:17 -070044
Jack Palevich1cdef202009-05-22 12:06:27 -070045#include <acc/acc.h>
46
Jack Palevich09555c72009-05-27 12:25:55 -070047#define LOG_API(...) do {} while(0)
48// #define LOG_API(...) fprintf (stderr, __VA_ARGS__)
Jack Palevich09555c72009-05-27 12:25:55 -070049// #define ENABLE_ARM_DISASSEMBLY
50
Jack Palevichb67b18f2009-06-11 21:12:23 -070051// #define PROVIDE_TRACE_CODEGEN
52
Jack Palevichbbf8ab52009-05-11 11:54:30 -070053namespace acc {
54
Jack Palevichac0e95e2009-05-29 13:53:44 -070055class ErrorSink {
56public:
57 void error(const char *fmt, ...) {
58 va_list ap;
59 va_start(ap, fmt);
60 verror(fmt, ap);
61 va_end(ap);
62 }
63
64 virtual void verror(const char* fmt, va_list ap) = 0;
65};
66
67class Compiler : public ErrorSink {
Jack Palevich21a15a22009-05-11 14:49:29 -070068 class CodeBuf {
Jack Palevich653f42d2009-05-28 17:15:32 -070069 char* ind; // Output code pointer
Jack Palevich21a15a22009-05-11 14:49:29 -070070 char* pProgramBase;
Jack Palevichac0e95e2009-05-29 13:53:44 -070071 ErrorSink* mErrorSink;
72 int mSize;
Jack Palevich0a280a02009-06-11 10:53:51 -070073 bool mOverflowed;
Jack Palevichf0cbc922009-05-08 16:35:13 -070074
Jack Palevich21a15a22009-05-11 14:49:29 -070075 void release() {
76 if (pProgramBase != 0) {
77 free(pProgramBase);
78 pProgramBase = 0;
Jack Palevichae54f1f2009-05-08 14:54:15 -070079 }
Jack Palevich21a15a22009-05-11 14:49:29 -070080 }
81
Jack Palevich0a280a02009-06-11 10:53:51 -070082 bool check(int n) {
Jack Palevichac0e95e2009-05-29 13:53:44 -070083 int newSize = ind - pProgramBase + n;
Jack Palevich0a280a02009-06-11 10:53:51 -070084 bool overflow = newSize > mSize;
85 if (overflow && !mOverflowed) {
86 mOverflowed = true;
Jack Palevichac0e95e2009-05-29 13:53:44 -070087 if (mErrorSink) {
88 mErrorSink->error("Code too large: %d bytes", newSize);
89 }
90 }
Jack Palevich0a280a02009-06-11 10:53:51 -070091 return overflow;
Jack Palevichac0e95e2009-05-29 13:53:44 -070092 }
93
Jack Palevich21a15a22009-05-11 14:49:29 -070094 public:
95 CodeBuf() {
96 pProgramBase = 0;
97 ind = 0;
Jack Palevichac0e95e2009-05-29 13:53:44 -070098 mErrorSink = 0;
99 mSize = 0;
Jack Palevich0a280a02009-06-11 10:53:51 -0700100 mOverflowed = false;
Jack Palevich21a15a22009-05-11 14:49:29 -0700101 }
102
103 ~CodeBuf() {
104 release();
105 }
106
107 void init(int size) {
108 release();
Jack Palevichac0e95e2009-05-29 13:53:44 -0700109 mSize = size;
Jack Palevich21a15a22009-05-11 14:49:29 -0700110 pProgramBase = (char*) calloc(1, size);
111 ind = pProgramBase;
112 }
113
Jack Palevichac0e95e2009-05-29 13:53:44 -0700114 void setErrorSink(ErrorSink* pErrorSink) {
115 mErrorSink = pErrorSink;
116 }
117
Jack Palevich546b2242009-05-13 15:10:04 -0700118 int o4(int n) {
Jack Palevich0a280a02009-06-11 10:53:51 -0700119 if(check(4)) {
120 return 0;
121 }
Jack Palevich8b0624c2009-05-20 12:12:06 -0700122 intptr_t result = (intptr_t) ind;
Jack Palevich546b2242009-05-13 15:10:04 -0700123 * (int*) ind = n;
124 ind += 4;
125 return result;
126 }
127
Jack Palevich21a15a22009-05-11 14:49:29 -0700128 /*
129 * Output a byte. Handles all values, 0..ff.
130 */
131 void ob(int n) {
Jack Palevich0a280a02009-06-11 10:53:51 -0700132 if(check(1)) {
133 return;
134 }
Jack Palevich21a15a22009-05-11 14:49:29 -0700135 *ind++ = n;
136 }
137
Jack Palevich21a15a22009-05-11 14:49:29 -0700138 inline void* getBase() {
139 return (void*) pProgramBase;
140 }
141
Jack Palevich8b0624c2009-05-20 12:12:06 -0700142 intptr_t getSize() {
Jack Palevich21a15a22009-05-11 14:49:29 -0700143 return ind - pProgramBase;
144 }
145
Jack Palevich8b0624c2009-05-20 12:12:06 -0700146 intptr_t getPC() {
147 return (intptr_t) ind;
Jack Palevich21a15a22009-05-11 14:49:29 -0700148 }
149 };
150
Jack Palevich1cdef202009-05-22 12:06:27 -0700151 /**
152 * A code generator creates an in-memory program, generating the code on
153 * the fly. There is one code generator implementation for each supported
154 * architecture.
155 *
156 * The code generator implements the following abstract machine:
157 * R0 - the main accumulator.
158 * R1 - the secondary accumulator.
159 * FP - a frame pointer for accessing function arguments and local
160 * variables.
161 * SP - a stack pointer for storing intermediate results while evaluating
162 * expressions. The stack pointer grows downwards.
163 *
164 * The function calling convention is that all arguments are placed on the
165 * stack such that the first argument has the lowest address.
166 * After the call, the result is in R0. The caller is responsible for
167 * removing the arguments from the stack.
168 * The R0 and R1 registers are not saved across function calls. The
169 * FP and SP registers are saved.
170 */
171
Jack Palevich21a15a22009-05-11 14:49:29 -0700172 class CodeGenerator {
173 public:
Jack Palevichac0e95e2009-05-29 13:53:44 -0700174 CodeGenerator() {
175 mErrorSink = 0;
176 pCodeBuf = 0;
177 }
Jack Palevich21a15a22009-05-11 14:49:29 -0700178 virtual ~CodeGenerator() {}
179
Jack Palevich22305132009-05-13 10:58:45 -0700180 virtual void init(CodeBuf* pCodeBuf) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700181 this->pCodeBuf = pCodeBuf;
Jack Palevichac0e95e2009-05-29 13:53:44 -0700182 pCodeBuf->setErrorSink(mErrorSink);
183 }
184
Jack Palevichb67b18f2009-06-11 21:12:23 -0700185 virtual void setErrorSink(ErrorSink* pErrorSink) {
Jack Palevichac0e95e2009-05-29 13:53:44 -0700186 mErrorSink = pErrorSink;
187 if (pCodeBuf) {
188 pCodeBuf->setErrorSink(mErrorSink);
189 }
Jack Palevich21a15a22009-05-11 14:49:29 -0700190 }
191
Jack Palevich1cdef202009-05-22 12:06:27 -0700192 /* Emit a function prolog.
193 * argCount is the number of arguments.
194 * Save the old value of the FP.
195 * Set the new value of the FP.
196 * Convert from the native platform calling convention to
197 * our stack-based calling convention. This may require
198 * pushing arguments from registers to the stack.
199 * Allocate "N" bytes of stack space. N isn't known yet, so
200 * just emit the instructions for adjusting the stack, and return
201 * the address to patch up. The patching will be done in
202 * functionExit().
203 * returns address to patch with local variable size.
Jack Palevich22305132009-05-13 10:58:45 -0700204 */
Jack Palevich546b2242009-05-13 15:10:04 -0700205 virtual int functionEntry(int argCount) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700206
Jack Palevich1cdef202009-05-22 12:06:27 -0700207 /* Emit a function epilog.
208 * Restore the old SP and FP register values.
209 * Return to the calling function.
210 * argCount - the number of arguments to the function.
211 * localVariableAddress - returned from functionEntry()
212 * localVariableSize - the size in bytes of the local variables.
213 */
214 virtual void functionExit(int argCount, int localVariableAddress,
215 int localVariableSize) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700216
Jack Palevich1cdef202009-05-22 12:06:27 -0700217 /* load immediate value to R0 */
Jack Palevich546b2242009-05-13 15:10:04 -0700218 virtual void li(int t) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700219
Jack Palevich1cdef202009-05-22 12:06:27 -0700220 /* Jump to a target, and return the address of the word that
221 * holds the target data, in case it needs to be fixed up later.
222 */
Jack Palevich22305132009-05-13 10:58:45 -0700223 virtual int gjmp(int t) = 0;
224
Jack Palevich1cdef202009-05-22 12:06:27 -0700225 /* Test R0 and jump to a target if the test succeeds.
226 * l = 0: je, l == 1: jne
227 * Return the address of the word that holds the targed data, in
228 * case it needs to be fixed up later.
229 */
Jack Palevich22305132009-05-13 10:58:45 -0700230 virtual int gtst(bool l, int t) = 0;
231
Jack Palevich1cdef202009-05-22 12:06:27 -0700232 /* Compare R1 against R0, and store the boolean result in R0.
233 * op specifies the comparison.
234 */
Jack Palevich22305132009-05-13 10:58:45 -0700235 virtual void gcmp(int op) = 0;
236
Jack Palevich1cdef202009-05-22 12:06:27 -0700237 /* Perform the arithmetic op specified by op. R1 is the
238 * left argument, R0 is the right argument.
239 */
Jack Palevich546b2242009-05-13 15:10:04 -0700240 virtual void genOp(int op) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700241
Jack Palevich1cdef202009-05-22 12:06:27 -0700242 /* Set R1 to 0.
243 */
244 virtual void clearR1() = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700245
Jack Palevich1cdef202009-05-22 12:06:27 -0700246 /* Push R0 onto the stack.
247 */
248 virtual void pushR0() = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700249
Jack Palevich1cdef202009-05-22 12:06:27 -0700250 /* Pop R1 off of the stack.
251 */
252 virtual void popR1() = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700253
Jack Palevich1cdef202009-05-22 12:06:27 -0700254 /* Store R0 to the address stored in R1.
255 * isInt is true if a whole 4-byte integer value
256 * should be stored, otherwise a 1-byte character
257 * value should be stored.
258 */
259 virtual void storeR0ToR1(bool isInt) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700260
Jack Palevich1cdef202009-05-22 12:06:27 -0700261 /* Load R0 from the address stored in R0.
262 * isInt is true if a whole 4-byte integer value
263 * should be loaded, otherwise a 1-byte character
264 * value should be loaded.
265 */
266 virtual void loadR0FromR0(bool isInt) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700267
Jack Palevich1cdef202009-05-22 12:06:27 -0700268 /* Load the absolute address of a variable to R0.
269 * If ea <= LOCAL, then this is a local variable, or an
270 * argument, addressed relative to FP.
271 * else it is an absolute global address.
272 */
273 virtual void leaR0(int ea) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700274
Jack Palevich1cdef202009-05-22 12:06:27 -0700275 /* Store R0 to a variable.
276 * If ea <= LOCAL, then this is a local variable, or an
277 * argument, addressed relative to FP.
278 * else it is an absolute global address.
279 */
280 virtual void storeR0(int ea) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700281
Jack Palevich1cdef202009-05-22 12:06:27 -0700282 /* load R0 from a variable.
283 * If ea <= LOCAL, then this is a local variable, or an
284 * argument, addressed relative to FP.
285 * else it is an absolute global address.
286 * If isIncDec is true, then the stored variable's value
287 * should be post-incremented or post-decremented, based
288 * on the value of op.
289 */
290 virtual void loadR0(int ea, bool isIncDec, int op) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700291
Jack Palevich1cdef202009-05-22 12:06:27 -0700292 /* Emit code to adjust the stack for a function call. Return the
293 * label for the address of the instruction that adjusts the
294 * stack size. This will be passed as argument "a" to
295 * endFunctionCallArguments.
296 */
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700297 virtual int beginFunctionCallArguments() = 0;
298
Jack Palevich1cdef202009-05-22 12:06:27 -0700299 /* Emit code to store R0 to the stack at byte offset l.
300 */
301 virtual void storeR0ToArg(int l) = 0;
Jack Palevich7810bc92009-05-15 14:31:47 -0700302
Jack Palevich1cdef202009-05-22 12:06:27 -0700303 /* Patch the function call preamble.
304 * a is the address returned from beginFunctionCallArguments
305 * l is the number of bytes the arguments took on the stack.
306 * Typically you would also emit code to convert the argument
307 * list into whatever the native function calling convention is.
308 * On ARM for example you would pop the first 5 arguments into
309 * R0..R4
310 */
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700311 virtual void endFunctionCallArguments(int a, int l) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700312
Jack Palevich1cdef202009-05-22 12:06:27 -0700313 /* Emit a call to an unknown function. The argument "symbol" needs to
314 * be stored in the location where the address should go. It forms
315 * a chain. The address will be patched later.
316 * Return the address of the word that has to be patched.
317 */
Jack Palevich22305132009-05-13 10:58:45 -0700318 virtual int callForward(int symbol) = 0;
319
Jack Palevich1cdef202009-05-22 12:06:27 -0700320 /* Call a function using PC-relative addressing. t is the PC-relative
321 * address of the function. It has already been adjusted for the
322 * architectural jump offset, so just store it as-is.
323 */
Jack Palevich22305132009-05-13 10:58:45 -0700324 virtual void callRelative(int t) = 0;
325
Jack Palevich1cdef202009-05-22 12:06:27 -0700326 /* Call a function pointer. L is the number of bytes the arguments
327 * take on the stack. The address of the function is stored at
328 * location SP + l.
329 */
Jack Palevich22305132009-05-13 10:58:45 -0700330 virtual void callIndirect(int l) = 0;
331
Jack Palevich1cdef202009-05-22 12:06:27 -0700332 /* Adjust SP after returning from a function call. l is the
333 * number of bytes of arguments stored on the stack. isIndirect
334 * is true if this was an indirect call. (In which case the
335 * address of the function is stored at location SP + l.)
336 */
Jack Palevich7810bc92009-05-15 14:31:47 -0700337 virtual void adjustStackAfterCall(int l, bool isIndirect) = 0;
Jack Palevich22305132009-05-13 10:58:45 -0700338
Jack Palevich1cdef202009-05-22 12:06:27 -0700339 /* Print a disassembly of the assembled code to out. Return
340 * non-zero if there is an error.
341 */
Jack Palevicha6535612009-05-13 16:24:17 -0700342 virtual int disassemble(FILE* out) = 0;
343
Jack Palevich1cdef202009-05-22 12:06:27 -0700344 /* Generate a symbol at the current PC. t is the head of a
345 * linked list of addresses to patch.
346 */
Jack Paleviche7b59062009-05-19 17:12:17 -0700347 virtual void gsym(int t) = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -0700348
Jack Palevich1cdef202009-05-22 12:06:27 -0700349 /*
350 * Do any cleanup work required at the end of a compile.
351 * For example, an instruction cache might need to be
352 * invalidated.
353 * Return non-zero if there is an error.
354 */
355 virtual int finishCompile() = 0;
Jack Palevich546b2242009-05-13 15:10:04 -0700356
Jack Palevicha6535612009-05-13 16:24:17 -0700357 /**
358 * Adjust relative branches by this amount.
359 */
360 virtual int jumpOffset() = 0;
361
Jack Palevich21a15a22009-05-11 14:49:29 -0700362 protected:
Jack Palevich21a15a22009-05-11 14:49:29 -0700363 /*
364 * Output a byte. Handles all values, 0..ff.
365 */
366 void ob(int n) {
367 pCodeBuf->ob(n);
368 }
369
Jack Palevich8b0624c2009-05-20 12:12:06 -0700370 intptr_t o4(int data) {
Jack Paleviche7b59062009-05-19 17:12:17 -0700371 return pCodeBuf->o4(data);
Jack Palevich21a15a22009-05-11 14:49:29 -0700372 }
373
Jack Palevich8b0624c2009-05-20 12:12:06 -0700374 intptr_t getBase() {
375 return (intptr_t) pCodeBuf->getBase();
Jack Palevicha6535612009-05-13 16:24:17 -0700376 }
377
Jack Palevich8b0624c2009-05-20 12:12:06 -0700378 intptr_t getPC() {
Jack Palevich21a15a22009-05-11 14:49:29 -0700379 return pCodeBuf->getPC();
380 }
Jack Palevich1cdef202009-05-22 12:06:27 -0700381
382 intptr_t getSize() {
383 return pCodeBuf->getSize();
384 }
Jack Palevichac0e95e2009-05-29 13:53:44 -0700385
386 void error(const char* fmt,...) {
387 va_list ap;
388 va_start(ap, fmt);
389 mErrorSink->verror(fmt, ap);
390 va_end(ap);
391 }
Jack Palevich21a15a22009-05-11 14:49:29 -0700392 private:
393 CodeBuf* pCodeBuf;
Jack Palevichac0e95e2009-05-29 13:53:44 -0700394 ErrorSink* mErrorSink;
Jack Palevich21a15a22009-05-11 14:49:29 -0700395 };
396
Jack Paleviche7b59062009-05-19 17:12:17 -0700397#ifdef PROVIDE_ARM_CODEGEN
398
Jack Palevich22305132009-05-13 10:58:45 -0700399 class ARMCodeGenerator : public CodeGenerator {
400 public:
401 ARMCodeGenerator() {}
402 virtual ~ARMCodeGenerator() {}
403
404 /* returns address to patch with local variable size
405 */
Jack Palevich546b2242009-05-13 15:10:04 -0700406 virtual int functionEntry(int argCount) {
Jack Palevichb7c81e92009-06-04 19:56:13 -0700407 LOG_API("functionEntry(%d);\n", argCount);
Jack Palevich69796b62009-05-14 15:42:26 -0700408 // sp -> arg4 arg5 ...
409 // Push our register-based arguments back on the stack
410 if (argCount > 0) {
411 int regArgCount = argCount <= 4 ? argCount : 4;
412 o4(0xE92D0000 | ((1 << argCount) - 1)); // stmfd sp!, {}
413 }
414 // sp -> arg0 arg1 ...
415 o4(0xE92D4800); // stmfd sp!, {fp, lr}
416 // sp, fp -> oldfp, retadr, arg0 arg1 ....
417 o4(0xE1A0B00D); // mov fp, sp
418 return o4(0xE24DD000); // sub sp, sp, # <local variables>
Jack Palevich22305132009-05-13 10:58:45 -0700419 }
420
Jack Palevich546b2242009-05-13 15:10:04 -0700421 virtual void functionExit(int argCount, int localVariableAddress, int localVariableSize) {
Jack Palevich09555c72009-05-27 12:25:55 -0700422 LOG_API("functionExit(%d, %d, %d);\n", argCount, localVariableAddress, localVariableSize);
Jack Palevich69796b62009-05-14 15:42:26 -0700423 // Patch local variable allocation code:
424 if (localVariableSize < 0 || localVariableSize > 255) {
Jack Palevich8de461d2009-05-14 17:21:45 -0700425 error("localVariables out of range: %d", localVariableSize);
Jack Palevich546b2242009-05-13 15:10:04 -0700426 }
Jack Palevich69796b62009-05-14 15:42:26 -0700427 *(char*) (localVariableAddress) = localVariableSize;
428
429 // sp -> locals .... fp -> oldfp, retadr, arg0, arg1, ...
430 o4(0xE1A0E00B); // mov lr, fp
431 o4(0xE59BB000); // ldr fp, [fp]
432 o4(0xE28ED004); // add sp, lr, #4
433 // sp -> retadr, arg0, ...
434 o4(0xE8BD4000); // ldmfd sp!, {lr}
435 // sp -> arg0 ....
436 if (argCount > 0) {
437 // We store the PC into the lr so we can adjust the sp before
Jack Palevich8de461d2009-05-14 17:21:45 -0700438 // returning. We need to pull off the registers we pushed
Jack Palevich69796b62009-05-14 15:42:26 -0700439 // earlier. We don't need to actually store them anywhere,
440 // just adjust the stack.
441 int regArgCount = argCount <= 4 ? argCount : 4;
442 o4(0xE28DD000 | (regArgCount << 2)); // add sp, sp, #argCount << 2
443 }
444 o4(0xE12FFF1E); // bx lr
Jack Palevich22305132009-05-13 10:58:45 -0700445 }
446
447 /* load immediate value */
Jack Palevich546b2242009-05-13 15:10:04 -0700448 virtual void li(int t) {
Jack Palevich09555c72009-05-27 12:25:55 -0700449 LOG_API("li(%d);\n", t);
Jack Palevicha6535612009-05-13 16:24:17 -0700450 if (t >= 0 && t < 255) {
Jack Palevich69796b62009-05-14 15:42:26 -0700451 o4(0xE3A00000 + t); // mov r0, #0
Jack Palevicha6535612009-05-13 16:24:17 -0700452 } else if (t >= -256 && t < 0) {
453 // mvn means move constant ^ ~0
Jack Palevich69796b62009-05-14 15:42:26 -0700454 o4(0xE3E00001 - t); // mvn r0, #0
Jack Palevicha6535612009-05-13 16:24:17 -0700455 } else {
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700456 o4(0xE51F0000); // ldr r0, .L3
457 o4(0xEA000000); // b .L99
458 o4(t); // .L3: .word 0
459 // .L99:
Jack Palevicha6535612009-05-13 16:24:17 -0700460 }
Jack Palevich22305132009-05-13 10:58:45 -0700461 }
462
463 virtual int gjmp(int t) {
Jack Palevich09555c72009-05-27 12:25:55 -0700464 LOG_API("gjmp(%d);\n", t);
Jack Palevich8de461d2009-05-14 17:21:45 -0700465 return o4(0xEA000000 | encodeAddress(t)); // b .L33
Jack Palevich22305132009-05-13 10:58:45 -0700466 }
467
468 /* l = 0: je, l == 1: jne */
469 virtual int gtst(bool l, int t) {
Jack Palevich09555c72009-05-27 12:25:55 -0700470 LOG_API("gtst(%d, %d);\n", l, t);
Jack Palevich8de461d2009-05-14 17:21:45 -0700471 o4(0xE3500000); // cmp r0,#0
472 int branch = l ? 0x1A000000 : 0x0A000000; // bne : beq
473 return o4(branch | encodeAddress(t));
Jack Palevich22305132009-05-13 10:58:45 -0700474 }
475
476 virtual void gcmp(int op) {
Jack Palevich09555c72009-05-27 12:25:55 -0700477 LOG_API("gcmp(%d);\n", op);
Jack Palevich8de461d2009-05-14 17:21:45 -0700478 o4(0xE1510000); // cmp r1, r1
479 switch(op) {
480 case OP_EQUALS:
481 o4(0x03A00001); // moveq r0,#1
482 o4(0x13A00000); // movne r0,#0
483 break;
484 case OP_NOT_EQUALS:
485 o4(0x03A00000); // moveq r0,#0
486 o4(0x13A00001); // movne r0,#1
487 break;
488 case OP_LESS_EQUAL:
489 o4(0xD3A00001); // movle r0,#1
490 o4(0xC3A00000); // movgt r0,#0
491 break;
492 case OP_GREATER:
493 o4(0xD3A00000); // movle r0,#0
494 o4(0xC3A00001); // movgt r0,#1
495 break;
496 case OP_GREATER_EQUAL:
497 o4(0xA3A00001); // movge r0,#1
498 o4(0xB3A00000); // movlt r0,#0
499 break;
500 case OP_LESS:
501 o4(0xA3A00000); // movge r0,#0
502 o4(0xB3A00001); // movlt r0,#1
503 break;
504 default:
505 error("Unknown comparison op %d", op);
506 break;
507 }
Jack Palevich22305132009-05-13 10:58:45 -0700508 }
509
Jack Palevich546b2242009-05-13 15:10:04 -0700510 virtual void genOp(int op) {
Jack Palevich09555c72009-05-27 12:25:55 -0700511 LOG_API("genOp(%d);\n", op);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700512 switch(op) {
513 case OP_MUL:
514 o4(0x0E0000091); // mul r0,r1,r0
515 break;
Jack Palevich3d474a72009-05-15 15:12:38 -0700516 case OP_DIV:
517 callRuntime(runtime_DIV);
518 break;
519 case OP_MOD:
520 callRuntime(runtime_MOD);
521 break;
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700522 case OP_PLUS:
523 o4(0xE0810000); // add r0,r1,r0
524 break;
525 case OP_MINUS:
526 o4(0xE0410000); // sub r0,r1,r0
527 break;
528 case OP_SHIFT_LEFT:
529 o4(0xE1A00011); // lsl r0,r1,r0
530 break;
531 case OP_SHIFT_RIGHT:
532 o4(0xE1A00051); // asr r0,r1,r0
533 break;
534 case OP_BIT_AND:
535 o4(0xE0010000); // and r0,r1,r0
536 break;
537 case OP_BIT_XOR:
538 o4(0xE0210000); // eor r0,r1,r0
539 break;
540 case OP_BIT_OR:
541 o4(0xE1810000); // orr r0,r1,r0
542 break;
543 case OP_BIT_NOT:
544 o4(0xE1E00000); // mvn r0, r0
545 break;
546 default:
Jack Palevich69796b62009-05-14 15:42:26 -0700547 error("Unimplemented op %d\n", op);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700548 break;
549 }
Jack Palevich22305132009-05-13 10:58:45 -0700550#if 0
551 o(decodeOp(op));
552 if (op == OP_MOD)
553 o(0x92); /* xchg %edx, %eax */
554#endif
555 }
556
Jack Palevich1cdef202009-05-22 12:06:27 -0700557 virtual void clearR1() {
Jack Palevich09555c72009-05-27 12:25:55 -0700558 LOG_API("clearR1();\n");
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700559 o4(0xE3A01000); // mov r1, #0
Jack Palevich22305132009-05-13 10:58:45 -0700560 }
561
Jack Palevich1cdef202009-05-22 12:06:27 -0700562 virtual void pushR0() {
Jack Palevich09555c72009-05-27 12:25:55 -0700563 LOG_API("pushR0();\n");
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700564 o4(0xE92D0001); // stmfd sp!,{r0}
Jack Palevich22305132009-05-13 10:58:45 -0700565 }
566
Jack Palevich1cdef202009-05-22 12:06:27 -0700567 virtual void popR1() {
Jack Palevich09555c72009-05-27 12:25:55 -0700568 LOG_API("popR1();\n");
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700569 o4(0xE8BD0002); // ldmfd sp!,{r1}
Jack Palevich22305132009-05-13 10:58:45 -0700570 }
571
Jack Palevich1cdef202009-05-22 12:06:27 -0700572 virtual void storeR0ToR1(bool isInt) {
Jack Palevich09555c72009-05-27 12:25:55 -0700573 LOG_API("storeR0ToR1(%d);\n", isInt);
Jack Palevichbd894902009-05-14 19:35:31 -0700574 if (isInt) {
575 o4(0xE5810000); // str r0, [r1]
576 } else {
577 o4(0xE5C10000); // strb r0, [r1]
578 }
Jack Palevich22305132009-05-13 10:58:45 -0700579 }
580
Jack Palevich1cdef202009-05-22 12:06:27 -0700581 virtual void loadR0FromR0(bool isInt) {
Jack Palevich09555c72009-05-27 12:25:55 -0700582 LOG_API("loadR0FromR0(%d);\n", isInt);
Jack Palevich22305132009-05-13 10:58:45 -0700583 if (isInt)
Jack Palevich69796b62009-05-14 15:42:26 -0700584 o4(0xE5900000); // ldr r0, [r0]
Jack Palevich22305132009-05-13 10:58:45 -0700585 else
Jack Palevich69796b62009-05-14 15:42:26 -0700586 o4(0xE5D00000); // ldrb r0, [r0]
Jack Palevich22305132009-05-13 10:58:45 -0700587 }
588
Jack Palevich1cdef202009-05-22 12:06:27 -0700589 virtual void leaR0(int ea) {
Jack Palevich09555c72009-05-27 12:25:55 -0700590 LOG_API("leaR0(%d);\n", ea);
Jack Palevich4d93f302009-05-15 13:30:00 -0700591 if (ea < LOCAL) {
592 // Local, fp relative
593 if (ea < -1023 || ea > 1023 || ((ea & 3) != 0)) {
594 error("Offset out of range: %08x", ea);
595 }
596 if (ea < 0) {
597 o4(0xE24B0F00 | (0xff & ((-ea) >> 2))); // sub r0, fp, #ea
598 } else {
599 o4(0xE28B0F00 | (0xff & (ea >> 2))); // add r0, fp, #ea
600 }
Jack Palevichbd894902009-05-14 19:35:31 -0700601 } else {
Jack Palevich4d93f302009-05-15 13:30:00 -0700602 // Global, absolute.
603 o4(0xE59F0000); // ldr r0, .L1
604 o4(0xEA000000); // b .L99
605 o4(ea); // .L1: .word 0
606 // .L99:
Jack Palevichbd894902009-05-14 19:35:31 -0700607 }
Jack Palevich22305132009-05-13 10:58:45 -0700608 }
609
Jack Palevich1cdef202009-05-22 12:06:27 -0700610 virtual void storeR0(int ea) {
Jack Palevich09555c72009-05-27 12:25:55 -0700611 LOG_API("storeR0(%d);\n", ea);
Jack Palevich4d93f302009-05-15 13:30:00 -0700612 if (ea < LOCAL) {
613 // Local, fp relative
614 if (ea < -4095 || ea > 4095) {
615 error("Offset out of range: %08x", ea);
616 }
617 if (ea < 0) {
618 o4(0xE50B0000 | (0xfff & (-ea))); // str r0, [fp,#-ea]
619 } else {
620 o4(0xE58B0000 | (0xfff & ea)); // str r0, [fp,#ea]
621 }
622 } else{
623 // Global, absolute
624 o4(0xE59F1000); // ldr r1, .L1
625 o4(0xEA000000); // b .L99
626 o4(ea); // .L1: .word 0
627 o4(0xE5810000); // .L99: str r0, [r1]
Jack Palevich69796b62009-05-14 15:42:26 -0700628 }
Jack Palevich22305132009-05-13 10:58:45 -0700629 }
630
Jack Palevich1cdef202009-05-22 12:06:27 -0700631 virtual void loadR0(int ea, bool isIncDec, int op) {
Jack Palevich09555c72009-05-27 12:25:55 -0700632 LOG_API("loadR0(%d, %d, %d);\n", ea, isIncDec, op);
Jack Palevich4d93f302009-05-15 13:30:00 -0700633 if (ea < LOCAL) {
634 // Local, fp relative
635 if (ea < -4095 || ea > 4095) {
636 error("Offset out of range: %08x", ea);
637 }
638 if (ea < 0) {
639 o4(0xE51B0000 | (0xfff & (-ea))); // ldr r0, [fp,#-ea]
640 } else {
641 o4(0xE59B0000 | (0xfff & ea)); // ldr r0, [fp,#ea]
642 }
Jack Palevich69796b62009-05-14 15:42:26 -0700643 } else {
Jack Palevich4d93f302009-05-15 13:30:00 -0700644 // Global, absolute
645 o4(0xE59F2000); // ldr r2, .L1
646 o4(0xEA000000); // b .L99
647 o4(ea); // .L1: .word ea
648 o4(0xE5920000); // .L99: ldr r0, [r2]
Jack Palevich69796b62009-05-14 15:42:26 -0700649 }
Jack Palevich22305132009-05-13 10:58:45 -0700650
Jack Palevich4d93f302009-05-15 13:30:00 -0700651 if (isIncDec) {
652 switch (op) {
653 case OP_INCREMENT:
654 o4(0xE2801001); // add r1, r0, #1
655 break;
656 case OP_DECREMENT:
657 o4(0xE2401001); // sub r1, r0, #1
658 break;
659 default:
660 error("unknown opcode: %d", op);
661 }
662 if (ea < LOCAL) {
663 // Local, fp relative
664 // Don't need range check, was already checked above
665 if (ea < 0) {
666 o4(0xE50B1000 | (0xfff & (-ea))); // str r1, [fp,#-ea]
667 } else {
668 o4(0xE58B1000 | (0xfff & ea)); // str r1, [fp,#ea]
669 }
670 } else{
671 // Global, absolute
672 // r2 is already set up from before.
673 o4(0xE5821000); // str r1, [r2]
674 }
Jack Palevichbd894902009-05-14 19:35:31 -0700675 }
Jack Palevich22305132009-05-13 10:58:45 -0700676 }
677
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700678 virtual int beginFunctionCallArguments() {
Jack Palevich09555c72009-05-27 12:25:55 -0700679 LOG_API("beginFunctionCallArguments();\n");
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700680 return o4(0xE24DDF00); // Placeholder
681 }
682
Jack Palevich1cdef202009-05-22 12:06:27 -0700683 virtual void storeR0ToArg(int l) {
Jack Palevich09555c72009-05-27 12:25:55 -0700684 LOG_API("storeR0ToArg(%d);\n", l);
Jack Palevich7810bc92009-05-15 14:31:47 -0700685 if (l < 0 || l > 4096-4) {
686 error("l out of range for stack offset: 0x%08x", l);
687 }
688 o4(0xE58D0000 + l); // str r0, [sp, #4]
689 }
690
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700691 virtual void endFunctionCallArguments(int a, int l) {
Jack Palevich09555c72009-05-27 12:25:55 -0700692 LOG_API("endFunctionCallArguments(0x%08x, %d);\n", a, l);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700693 if (l < 0 || l > 0x3FC) {
694 error("L out of range for stack adjustment: 0x%08x", l);
695 }
696 * (int*) a = 0xE24DDF00 | (l >> 2); // sub sp, sp, #0 << 2
697 int argCount = l >> 2;
698 if (argCount > 0) {
699 int regArgCount = argCount > 4 ? 4 : argCount;
700 o4(0xE8BD0000 | ((1 << regArgCount) - 1)); // ldmfd sp!,{}
701 }
Jack Palevich22305132009-05-13 10:58:45 -0700702 }
703
Jack Palevich22305132009-05-13 10:58:45 -0700704 virtual int callForward(int symbol) {
Jack Palevich09555c72009-05-27 12:25:55 -0700705 LOG_API("callForward(%d);\n", symbol);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700706 // Forward calls are always short (local)
707 return o4(0xEB000000 | encodeAddress(symbol));
Jack Palevich22305132009-05-13 10:58:45 -0700708 }
709
710 virtual void callRelative(int t) {
Jack Palevich09555c72009-05-27 12:25:55 -0700711 LOG_API("callRelative(%d);\n", t);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700712 int abs = t + getPC() + jumpOffset();
Jack Palevichac0e95e2009-05-29 13:53:44 -0700713 LOG_API("abs=%d (0x%08x)\n", abs, abs);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700714 if (t >= - (1 << 25) && t < (1 << 25)) {
715 o4(0xEB000000 | encodeAddress(t));
716 } else {
717 // Long call.
718 o4(0xE59FC000); // ldr r12, .L1
719 o4(0xEA000000); // b .L99
Jack Palevichbd894902009-05-14 19:35:31 -0700720 o4(t - 12); // .L1: .word 0
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700721 o4(0xE08CC00F); // .L99: add r12,pc
722 o4(0xE12FFF3C); // blx r12
723 }
Jack Palevich22305132009-05-13 10:58:45 -0700724 }
725
726 virtual void callIndirect(int l) {
Jack Palevich09555c72009-05-27 12:25:55 -0700727 LOG_API("callIndirect(%d);\n", l);
Jack Palevich7810bc92009-05-15 14:31:47 -0700728 int argCount = l >> 2;
729 int poppedArgs = argCount > 4 ? 4 : argCount;
730 int adjustedL = l - (poppedArgs << 2);
731 if (adjustedL < 0 || adjustedL > 4096-4) {
732 error("l out of range for stack offset: 0x%08x", l);
733 }
734 o4(0xE59DC000 | (0xfff & adjustedL)); // ldr r12, [sp,#adjustedL]
735 o4(0xE12FFF3C); // blx r12
Jack Palevich22305132009-05-13 10:58:45 -0700736 }
737
Jack Palevich7810bc92009-05-15 14:31:47 -0700738 virtual void adjustStackAfterCall(int l, bool isIndirect) {
Jack Palevich09555c72009-05-27 12:25:55 -0700739 LOG_API("adjustStackAfterCall(%d, %d);\n", l, isIndirect);
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700740 int argCount = l >> 2;
Jack Palevich7810bc92009-05-15 14:31:47 -0700741 int stackArgs = argCount > 4 ? argCount - 4 : 0;
742 int stackUse = stackArgs + (isIndirect ? 1 : 0);
743 if (stackUse) {
744 if (stackUse < 0 || stackUse > 255) {
745 error("L out of range for stack adjustment: 0x%08x", l);
746 }
747 o4(0xE28DDF00 | stackUse); // add sp, sp, #stackUse << 2
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700748 }
Jack Palevich22305132009-05-13 10:58:45 -0700749 }
750
Jack Palevicha6535612009-05-13 16:24:17 -0700751 virtual int jumpOffset() {
Jack Palevichbd894902009-05-14 19:35:31 -0700752 return 8;
Jack Palevicha6535612009-05-13 16:24:17 -0700753 }
754
755 /* output a symbol and patch all calls to it */
756 virtual void gsym(int t) {
Jack Palevich09555c72009-05-27 12:25:55 -0700757 LOG_API("gsym(0x%x)\n", t);
Jack Palevicha6535612009-05-13 16:24:17 -0700758 int n;
759 int base = getBase();
760 int pc = getPC();
Jack Palevich09555c72009-05-27 12:25:55 -0700761 LOG_API("pc = 0x%x\n", pc);
Jack Palevicha6535612009-05-13 16:24:17 -0700762 while (t) {
763 int data = * (int*) t;
764 int decodedOffset = ((BRANCH_REL_ADDRESS_MASK & data) << 2);
765 if (decodedOffset == 0) {
766 n = 0;
767 } else {
768 n = base + decodedOffset; /* next value */
769 }
770 *(int *) t = (data & ~BRANCH_REL_ADDRESS_MASK)
771 | encodeRelAddress(pc - t - 8);
772 t = n;
773 }
774 }
775
Jack Palevich1cdef202009-05-22 12:06:27 -0700776 virtual int finishCompile() {
777#if defined(__arm__)
778 const long base = long(getBase());
779 const long curr = long(getPC());
780 int err = cacheflush(base, curr, 0);
781 return err;
782#else
783 return 0;
784#endif
785 }
786
Jack Palevicha6535612009-05-13 16:24:17 -0700787 virtual int disassemble(FILE* out) {
Jack Palevich09555c72009-05-27 12:25:55 -0700788#ifdef ENABLE_ARM_DISASSEMBLY
789 disasmOut = out;
Jack Palevicha6535612009-05-13 16:24:17 -0700790 disasm_interface_t di;
791 di.di_readword = disassemble_readword;
792 di.di_printaddr = disassemble_printaddr;
793 di.di_printf = disassemble_printf;
794
795 int base = getBase();
796 int pc = getPC();
797 for(int i = base; i < pc; i += 4) {
798 fprintf(out, "%08x: %08x ", i, *(int*) i);
799 ::disasm(&di, i, 0);
800 }
Jack Palevich09555c72009-05-27 12:25:55 -0700801#endif
Jack Palevicha6535612009-05-13 16:24:17 -0700802 return 0;
803 }
Jack Palevich7810bc92009-05-15 14:31:47 -0700804
Jack Palevich22305132009-05-13 10:58:45 -0700805 private:
Jack Palevicha6535612009-05-13 16:24:17 -0700806 static FILE* disasmOut;
807
808 static u_int
809 disassemble_readword(u_int address)
810 {
811 return(*((u_int *)address));
812 }
813
814 static void
815 disassemble_printaddr(u_int address)
816 {
817 fprintf(disasmOut, "0x%08x", address);
818 }
819
820 static void
821 disassemble_printf(const char *fmt, ...) {
822 va_list ap;
823 va_start(ap, fmt);
824 vfprintf(disasmOut, fmt, ap);
825 va_end(ap);
826 }
827
828 static const int BRANCH_REL_ADDRESS_MASK = 0x00ffffff;
829
830 /** Encode a relative address that might also be
831 * a label.
832 */
833 int encodeAddress(int value) {
834 int base = getBase();
835 if (value >= base && value <= getPC() ) {
836 // This is a label, encode it relative to the base.
837 value = value - base;
838 }
839 return encodeRelAddress(value);
840 }
841
842 int encodeRelAddress(int value) {
843 return BRANCH_REL_ADDRESS_MASK & (value >> 2);
844 }
Jack Palevich22305132009-05-13 10:58:45 -0700845
Jack Palevich3d474a72009-05-15 15:12:38 -0700846 typedef int (*int2FnPtr)(int a, int b);
847 void callRuntime(int2FnPtr fn) {
848 o4(0xE59F2000); // ldr r2, .L1
849 o4(0xEA000000); // b .L99
850 o4((int) fn); //.L1: .word fn
851 o4(0xE12FFF32); //.L99: blx r2
852 }
853
854 static int runtime_DIV(int a, int b) {
855 return b / a;
856 }
857
858 static int runtime_MOD(int a, int b) {
859 return b % a;
860 }
Jack Palevich22305132009-05-13 10:58:45 -0700861 };
862
Jack Palevich09555c72009-05-27 12:25:55 -0700863#endif // PROVIDE_ARM_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -0700864
865#ifdef PROVIDE_X86_CODEGEN
866
Jack Palevich21a15a22009-05-11 14:49:29 -0700867 class X86CodeGenerator : public CodeGenerator {
868 public:
869 X86CodeGenerator() {}
870 virtual ~X86CodeGenerator() {}
871
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700872 /* returns address to patch with local variable size
873 */
Jack Palevich546b2242009-05-13 15:10:04 -0700874 virtual int functionEntry(int argCount) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700875 o(0xe58955); /* push %ebp, mov %esp, %ebp */
876 return oad(0xec81, 0); /* sub $xxx, %esp */
877 }
878
Jack Palevich546b2242009-05-13 15:10:04 -0700879 virtual void functionExit(int argCount, int localVariableAddress, int localVariableSize) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700880 o(0xc3c9); /* leave, ret */
Jack Palevich546b2242009-05-13 15:10:04 -0700881 *(int *) localVariableAddress = localVariableSize; /* save local variables */
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700882 }
883
Jack Palevich21a15a22009-05-11 14:49:29 -0700884 /* load immediate value */
Jack Palevich546b2242009-05-13 15:10:04 -0700885 virtual void li(int t) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700886 oad(0xb8, t); /* mov $xx, %eax */
887 }
888
Jack Palevich22305132009-05-13 10:58:45 -0700889 virtual int gjmp(int t) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700890 return psym(0xe9, t);
891 }
892
893 /* l = 0: je, l == 1: jne */
Jack Palevich22305132009-05-13 10:58:45 -0700894 virtual int gtst(bool l, int t) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700895 o(0x0fc085); /* test %eax, %eax, je/jne xxx */
896 return psym(0x84 + l, t);
897 }
898
Jack Palevich22305132009-05-13 10:58:45 -0700899 virtual void gcmp(int op) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700900 int t = decodeOp(op);
Jack Palevich21a15a22009-05-11 14:49:29 -0700901 o(0xc139); /* cmp %eax,%ecx */
902 li(0);
903 o(0x0f); /* setxx %al */
904 o(t + 0x90);
905 o(0xc0);
906 }
907
Jack Palevich546b2242009-05-13 15:10:04 -0700908 virtual void genOp(int op) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700909 o(decodeOp(op));
910 if (op == OP_MOD)
911 o(0x92); /* xchg %edx, %eax */
912 }
913
Jack Palevich1cdef202009-05-22 12:06:27 -0700914 virtual void clearR1() {
Jack Palevich21a15a22009-05-11 14:49:29 -0700915 oad(0xb9, 0); /* movl $0, %ecx */
916 }
917
Jack Palevich1cdef202009-05-22 12:06:27 -0700918 virtual void pushR0() {
Jack Palevich21a15a22009-05-11 14:49:29 -0700919 o(0x50); /* push %eax */
920 }
921
Jack Palevich1cdef202009-05-22 12:06:27 -0700922 virtual void popR1() {
Jack Palevich21a15a22009-05-11 14:49:29 -0700923 o(0x59); /* pop %ecx */
Jack Palevichbf42c9c2009-05-12 12:48:35 -0700924 }
925
Jack Palevich1cdef202009-05-22 12:06:27 -0700926 virtual void storeR0ToR1(bool isInt) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700927 o(0x0188 + isInt); /* movl %eax/%al, (%ecx) */
928 }
929
Jack Palevich1cdef202009-05-22 12:06:27 -0700930 virtual void loadR0FromR0(bool isInt) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700931 if (isInt)
932 o(0x8b); /* mov (%eax), %eax */
933 else
934 o(0xbe0f); /* movsbl (%eax), %eax */
935 ob(0); /* add zero in code */
936 }
937
Jack Palevich1cdef202009-05-22 12:06:27 -0700938 virtual void leaR0(int ea) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700939 gmov(10, ea); /* leal EA, %eax */
940 }
941
Jack Palevich1cdef202009-05-22 12:06:27 -0700942 virtual void storeR0(int ea) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700943 gmov(6, ea); /* mov %eax, EA */
944 }
945
Jack Palevich1cdef202009-05-22 12:06:27 -0700946 virtual void loadR0(int ea, bool isIncDec, int op) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700947 gmov(8, ea); /* mov EA, %eax */
Jack Palevich4d93f302009-05-15 13:30:00 -0700948 if (isIncDec) {
949 /* Implement post-increment or post decrement.
950 */
951 gmov(0, ea); /* 83 ADD */
952 o(decodeOp(op));
953 }
Jack Palevich21a15a22009-05-11 14:49:29 -0700954 }
955
Jack Palevichcb1c9ef2009-05-14 11:38:49 -0700956 virtual int beginFunctionCallArguments() {
Jack Palevich21a15a22009-05-11 14:49:29 -0700957 return oad(0xec81, 0); /* sub $xxx, %esp */
958 }
959
Jack Palevich1cdef202009-05-22 12:06:27 -0700960 virtual void storeR0ToArg(int l) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700961 oad(0x248489, l); /* movl %eax, xxx(%esp) */
962 }
963
Jack Palevich7810bc92009-05-15 14:31:47 -0700964 virtual void endFunctionCallArguments(int a, int l) {
965 * (int*) a = l;
966 }
967
Jack Palevich22305132009-05-13 10:58:45 -0700968 virtual int callForward(int symbol) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700969 return psym(0xe8, symbol); /* call xxx */
970 }
971
Jack Palevich22305132009-05-13 10:58:45 -0700972 virtual void callRelative(int t) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700973 psym(0xe8, t); /* call xxx */
974 }
975
Jack Palevich22305132009-05-13 10:58:45 -0700976 virtual void callIndirect(int l) {
Jack Palevich21a15a22009-05-11 14:49:29 -0700977 oad(0x2494ff, l); /* call *xxx(%esp) */
978 }
979
Jack Palevich7810bc92009-05-15 14:31:47 -0700980 virtual void adjustStackAfterCall(int l, bool isIndirect) {
981 if (isIndirect) {
982 l += 4;
983 }
Jack Palevich21a15a22009-05-11 14:49:29 -0700984 oad(0xc481, l); /* add $xxx, %esp */
985 }
986
Jack Palevicha6535612009-05-13 16:24:17 -0700987 virtual int jumpOffset() {
988 return 5;
989 }
990
991 virtual int disassemble(FILE* out) {
Jack Palevich1cdef202009-05-22 12:06:27 -0700992 return 0;
Jack Palevicha6535612009-05-13 16:24:17 -0700993 }
994
Jack Paleviche7b59062009-05-19 17:12:17 -0700995 /* output a symbol and patch all calls to it */
996 virtual void gsym(int t) {
997 int n;
998 int pc = getPC();
999 while (t) {
1000 n = *(int *) t; /* next value */
1001 *(int *) t = pc - t - 4;
1002 t = n;
1003 }
1004 }
1005
Jack Palevich1cdef202009-05-22 12:06:27 -07001006 virtual int finishCompile() {
Jack Palevich8dc662e2009-06-09 22:53:47 +00001007 size_t pagesize = 4096;
1008 size_t base = (size_t) getBase() & ~ (pagesize - 1);
1009 size_t top = ((size_t) getPC() + pagesize - 1) & ~ (pagesize - 1);
1010 int err = mprotect((void*) base, top - base, PROT_READ | PROT_WRITE | PROT_EXEC);
1011 if (err) {
1012 error("mprotect() failed: %d", errno);
1013 }
1014 return err;
Jack Palevich1cdef202009-05-22 12:06:27 -07001015 }
1016
Jack Palevich21a15a22009-05-11 14:49:29 -07001017 private:
Jack Paleviche7b59062009-05-19 17:12:17 -07001018
1019 /** Output 1 to 4 bytes.
1020 *
1021 */
1022 void o(int n) {
1023 /* cannot use unsigned, so we must do a hack */
1024 while (n && n != -1) {
1025 ob(n & 0xff);
1026 n = n >> 8;
1027 }
1028 }
1029
1030 /* psym is used to put an instruction with a data field which is a
1031 reference to a symbol. It is in fact the same as oad ! */
1032 int psym(int n, int t) {
1033 return oad(n, t);
1034 }
1035
1036 /* instruction + address */
1037 int oad(int n, int t) {
1038 o(n);
1039 int result = getPC();
1040 o4(t);
1041 return result;
1042 }
1043
1044
Jack Palevichbf42c9c2009-05-12 12:48:35 -07001045 static const int operatorHelper[];
1046
1047 int decodeOp(int op) {
1048 if (op < 0 || op > OP_COUNT) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07001049 error("Out-of-range operator: %d\n", op);
Jack Palevich0a280a02009-06-11 10:53:51 -07001050 op = 0;
Jack Palevichbf42c9c2009-05-12 12:48:35 -07001051 }
1052 return operatorHelper[op];
1053 }
Jack Palevich21a15a22009-05-11 14:49:29 -07001054
Jack Palevich546b2242009-05-13 15:10:04 -07001055 void gmov(int l, int t) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001056 o(l + 0x83);
Jack Palevich8dc662e2009-06-09 22:53:47 +00001057 oad((t > -LOCAL && t < LOCAL) << 7 | 5, t);
Jack Palevich21a15a22009-05-11 14:49:29 -07001058 }
1059 };
1060
Jack Paleviche7b59062009-05-19 17:12:17 -07001061#endif // PROVIDE_X86_CODEGEN
1062
Jack Palevichb67b18f2009-06-11 21:12:23 -07001063#ifdef PROVIDE_TRACE_CODEGEN
1064 class TraceCodeGenerator : public CodeGenerator {
1065 private:
1066 CodeGenerator* mpBase;
1067
1068 public:
1069 TraceCodeGenerator(CodeGenerator* pBase) {
1070 mpBase = pBase;
1071 }
1072
1073 virtual ~TraceCodeGenerator() {
1074 delete mpBase;
1075 }
1076
1077 virtual void init(CodeBuf* pCodeBuf) {
1078 mpBase->init(pCodeBuf);
1079 }
1080
1081 void setErrorSink(ErrorSink* pErrorSink) {
1082 mpBase->setErrorSink(pErrorSink);
1083 }
1084
1085 /* returns address to patch with local variable size
1086 */
1087 virtual int functionEntry(int argCount) {
1088 int result = mpBase->functionEntry(argCount);
1089 fprintf(stderr, "functionEntry(%d) -> %d\n", argCount, result);
1090 return result;
1091 }
1092
1093 virtual void functionExit(int argCount, int localVariableAddress, int localVariableSize) {
1094 fprintf(stderr, "functionExit(%d, %d, %d)\n",
1095 argCount, localVariableAddress, localVariableSize);
1096 mpBase->functionExit(argCount, localVariableAddress, localVariableSize);
1097 }
1098
1099 /* load immediate value */
1100 virtual void li(int t) {
1101 fprintf(stderr, "li(%d)\n", t);
1102 mpBase->li(t);
1103 }
1104
1105 virtual int gjmp(int t) {
1106 int result = mpBase->gjmp(t);
1107 fprintf(stderr, "gjmp(%d) = %d\n", t, result);
1108 return result;
1109 }
1110
1111 /* l = 0: je, l == 1: jne */
1112 virtual int gtst(bool l, int t) {
1113 int result = mpBase->gtst(l, t);
1114 fprintf(stderr, "gtst(%d,%d) = %d\n", l, t, result);
1115 return result;
1116 }
1117
1118 virtual void gcmp(int op) {
1119 fprintf(stderr, "gcmp(%d)\n", op);
1120 mpBase->gcmp(op);
1121 }
1122
1123 virtual void genOp(int op) {
1124 fprintf(stderr, "genOp(%d)\n", op);
1125 mpBase->genOp(op);
1126 }
1127
1128 virtual void clearR1() {
1129 fprintf(stderr, "clearR1()\n");
1130 mpBase->clearR1();
1131 }
1132
1133 virtual void pushR0() {
1134 fprintf(stderr, "pushR0()\n");
1135 mpBase->pushR0();
1136 }
1137
1138 virtual void popR1() {
1139 fprintf(stderr, "popR1()\n");
1140 mpBase->popR1();
1141 }
1142
1143 virtual void storeR0ToR1(bool isInt) {
1144 fprintf(stderr, "storeR0ToR1(%d)\n", isInt);
1145 mpBase->storeR0ToR1(isInt);
1146 }
1147
1148 virtual void loadR0FromR0(bool isInt) {
1149 fprintf(stderr, "loadR0FromR0(%d)\n", isInt);
1150 mpBase->loadR0FromR0(isInt);
1151 }
1152
1153 virtual void leaR0(int ea) {
1154 fprintf(stderr, "leaR0(%d)\n", ea);
1155 mpBase->leaR0(ea);
1156 }
1157
1158 virtual void storeR0(int ea) {
1159 fprintf(stderr, "storeR0(%d)\n", ea);
1160 mpBase->storeR0(ea);
1161 }
1162
1163 virtual void loadR0(int ea, bool isIncDec, int op) {
1164 fprintf(stderr, "loadR0(%d, %d, %d)\n", ea, isIncDec, op);
1165 mpBase->loadR0(ea, isIncDec, op);
1166 }
1167
1168 virtual int beginFunctionCallArguments() {
1169 int result = mpBase->beginFunctionCallArguments();
1170 fprintf(stderr, "beginFunctionCallArguments() = %d\n", result);
1171 return result;
1172 }
1173
1174 virtual void storeR0ToArg(int l) {
1175 fprintf(stderr, "storeR0ToArg(%d)\n", l);
1176 mpBase->storeR0ToArg(l);
1177 }
1178
1179 virtual void endFunctionCallArguments(int a, int l) {
1180 fprintf(stderr, "endFunctionCallArguments(%d, %d)\n", a, l);
1181 mpBase->endFunctionCallArguments(a, l);
1182 }
1183
1184 virtual int callForward(int symbol) {
1185 int result = mpBase->callForward(symbol);
1186 fprintf(stderr, "callForward(%d) = %d\n", symbol, result);
1187 return result;
1188 }
1189
1190 virtual void callRelative(int t) {
1191 fprintf(stderr, "callRelative(%d)\n", t);
1192 mpBase->callRelative(t);
1193 }
1194
1195 virtual void callIndirect(int l) {
1196 fprintf(stderr, "callIndirect(%d)\n", l);
1197 mpBase->callIndirect(l);
1198 }
1199
1200 virtual void adjustStackAfterCall(int l, bool isIndirect) {
1201 fprintf(stderr, "adjustStackAfterCall(%d, %d)\n", l, isIndirect);
1202 mpBase->adjustStackAfterCall(l, isIndirect);
1203 }
1204
1205 virtual int jumpOffset() {
1206 return mpBase->jumpOffset();
1207 }
1208
1209 virtual int disassemble(FILE* out) {
1210 return mpBase->disassemble(out);
1211 }
1212
1213 /* output a symbol and patch all calls to it */
1214 virtual void gsym(int t) {
1215 fprintf(stderr, "gsym(%d)\n", t);
1216 mpBase->gsym(t);
1217 }
1218
1219 virtual int finishCompile() {
1220 int result = mpBase->finishCompile();
1221 fprintf(stderr, "finishCompile() = %d\n", result);
1222 return result;
1223 }
1224 };
1225
1226#endif // PROVIDE_TRACE_CODEGEN
1227
Jack Palevich1cdef202009-05-22 12:06:27 -07001228 class InputStream {
1229 public:
Jack Palevicheedf9d22009-06-04 16:23:40 -07001230 int getChar() {
1231 if (bumpLine) {
1232 line++;
1233 bumpLine = false;
1234 }
1235 int ch = get();
1236 if (ch == '\n') {
1237 bumpLine = true;
1238 }
1239 return ch;
1240 }
1241 int getLine() {
1242 return line;
1243 }
1244 protected:
1245 InputStream() :
1246 line(1), bumpLine(false) {
1247 }
1248 private:
Jack Palevich1cdef202009-05-22 12:06:27 -07001249 virtual int get() = 0;
Jack Palevicheedf9d22009-06-04 16:23:40 -07001250 int line;
1251 bool bumpLine;
Jack Palevich1cdef202009-05-22 12:06:27 -07001252 };
1253
1254 class FileInputStream : public InputStream {
1255 public:
1256 FileInputStream(FILE* in) : f(in) {}
Jack Palevich1cdef202009-05-22 12:06:27 -07001257 private:
Jack Palevicheedf9d22009-06-04 16:23:40 -07001258 virtual int get() { return fgetc(f); }
Jack Palevich1cdef202009-05-22 12:06:27 -07001259 FILE* f;
1260 };
1261
1262 class TextInputStream : public InputStream {
1263 public:
1264 TextInputStream(const char* text, size_t textLength)
1265 : pText(text), mTextLength(textLength), mPosition(0) {
1266 }
Jack Palevicheedf9d22009-06-04 16:23:40 -07001267
1268 private:
Jack Palevich1cdef202009-05-22 12:06:27 -07001269 virtual int get() {
1270 return mPosition < mTextLength ? pText[mPosition++] : EOF;
1271 }
Jack Palevich1cdef202009-05-22 12:06:27 -07001272
Jack Palevich1cdef202009-05-22 12:06:27 -07001273 const char* pText;
1274 size_t mTextLength;
1275 size_t mPosition;
1276 };
1277
Jack Palevicheedf9d22009-06-04 16:23:40 -07001278 class String {
1279 public:
1280 String() {
1281 mpBase = 0;
1282 mUsed = 0;
1283 mSize = 0;
1284 }
1285
Jack Palevich303d8ff2009-06-11 19:06:24 -07001286 String(const char* item, int len, bool adopt) {
1287 if (len < 0) {
1288 len = strlen(item);
1289 }
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001290 if (adopt) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001291 mpBase = (char*) item;
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001292 mUsed = len;
1293 mSize = len + 1;
1294 } else {
1295 mpBase = 0;
1296 mUsed = 0;
1297 mSize = 0;
1298 appendBytes(item, len);
1299 }
1300 }
1301
Jack Palevich303d8ff2009-06-11 19:06:24 -07001302 String(const String& other) {
1303 mpBase = 0;
1304 mUsed = 0;
1305 mSize = 0;
1306 appendBytes(other.getUnwrapped(), other.len());
1307 }
1308
Jack Palevicheedf9d22009-06-04 16:23:40 -07001309 ~String() {
1310 if (mpBase) {
1311 free(mpBase);
1312 }
1313 }
1314
Jack Palevicha6baa232009-06-12 11:25:59 -07001315 String& operator=(const String& other) {
1316 clear();
1317 appendBytes(other.getUnwrapped(), other.len());
1318 return *this;
1319 }
1320
Jack Palevich303d8ff2009-06-11 19:06:24 -07001321 inline char* getUnwrapped() const {
Jack Palevicheedf9d22009-06-04 16:23:40 -07001322 return mpBase;
1323 }
1324
Jack Palevich303d8ff2009-06-11 19:06:24 -07001325 void clear() {
1326 mUsed = 0;
1327 if (mSize > 0) {
1328 mpBase[0] = 0;
1329 }
1330 }
1331
Jack Palevicheedf9d22009-06-04 16:23:40 -07001332 void appendCStr(const char* s) {
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001333 appendBytes(s, strlen(s));
1334 }
1335
1336 void appendBytes(const char* s, int n) {
Jack Palevicheedf9d22009-06-04 16:23:40 -07001337 memcpy(ensure(n), s, n + 1);
1338 }
1339
1340 void append(char c) {
1341 * ensure(1) = c;
1342 }
1343
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001344 char* orphan() {
1345 char* result = mpBase;
1346 mpBase = 0;
1347 mUsed = 0;
1348 mSize = 0;
1349 return result;
1350 }
1351
Jack Palevicheedf9d22009-06-04 16:23:40 -07001352 void printf(const char* fmt,...) {
1353 va_list ap;
1354 va_start(ap, fmt);
1355 vprintf(fmt, ap);
1356 va_end(ap);
1357 }
1358
1359 void vprintf(const char* fmt, va_list ap) {
1360 char* temp;
1361 int numChars = vasprintf(&temp, fmt, ap);
1362 memcpy(ensure(numChars), temp, numChars+1);
1363 free(temp);
1364 }
1365
Jack Palevich303d8ff2009-06-11 19:06:24 -07001366 inline size_t len() const {
Jack Palevicheedf9d22009-06-04 16:23:40 -07001367 return mUsed;
1368 }
1369
1370 private:
1371 char* ensure(int n) {
1372 size_t newUsed = mUsed + n;
1373 if (newUsed > mSize) {
1374 size_t newSize = mSize * 2 + 10;
1375 if (newSize < newUsed) {
1376 newSize = newUsed;
1377 }
1378 mpBase = (char*) realloc(mpBase, newSize + 1);
1379 mSize = newSize;
1380 }
1381 mpBase[newUsed] = '\0';
1382 char* result = mpBase + mUsed;
1383 mUsed = newUsed;
1384 return result;
1385 }
1386
1387 char* mpBase;
1388 size_t mUsed;
1389 size_t mSize;
1390 };
1391
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001392 /**
1393 * Wrap an externally allocated string for use as a hash key.
1394 */
1395 class FakeString : public String {
1396 public:
Jack Palevich2db168f2009-06-11 14:29:47 -07001397 FakeString(const char* string, size_t length) :
1398 String((char*) string, length, true) {}
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001399
1400 ~FakeString() {
1401 orphan();
1402 }
1403 };
1404
1405 template<class V> class StringTable {
1406 public:
Jack Palevich303d8ff2009-06-11 19:06:24 -07001407 StringTable() {
1408 init(10);
1409 }
1410
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001411 StringTable(size_t initialCapacity) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001412 init(initialCapacity);
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001413 }
1414
1415 ~StringTable() {
1416 clear();
Jack Palevich2db168f2009-06-11 14:29:47 -07001417 hashmapFree(mpMap);
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001418 }
1419
1420 void clear() {
1421 hashmapForEach(mpMap, freeKeyValue, this);
1422 }
1423
1424 bool contains(String* pKey) {
1425 bool result = hashmapContainsKey(mpMap, pKey);
1426 return result;
1427 }
1428
1429 V* get(String* pKey) {
1430 V* result = (V*) hashmapGet(mpMap, pKey);
1431 return result;
1432 }
1433
1434 V* remove(String* pKey) {
1435 V* result = (V*) hashmapRemove(mpMap, pKey);
1436 return result;
1437 }
1438
1439 V* put(String* pKey, V* value) {
1440 V* result = (V*) hashmapPut(mpMap, pKey, value);
1441 if (result) {
1442 // The key was not adopted by the map, so delete it here.
1443 delete pKey;
1444 }
1445 return result;
1446 }
1447
Jack Palevicha6baa232009-06-12 11:25:59 -07001448 void forEach(bool (*callback)(String* key, V* value, void* context),
1449 void* context) {
1450 hashmapForEach(mpMap, (bool (*)(void*, void*, void*)) callback,
1451 context);
1452 }
1453
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001454 protected:
Jack Palevich303d8ff2009-06-11 19:06:24 -07001455
1456 void init(size_t initialCapacity) {
1457 mpMap = hashmapCreate(initialCapacity, hashFn, equalsFn);
1458 }
1459
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001460 static int hashFn(void* pKey) {
1461 String* pString = (String*) pKey;
1462 return hashmapHash(pString->getUnwrapped(), pString->len());
1463 }
1464
1465 static bool equalsFn(void* keyA, void* keyB) {
1466 String* pStringA = (String*) keyA;
1467 String* pStringB = (String*) keyB;
1468 return pStringA->len() == pStringB->len()
1469 && strcmp(pStringA->getUnwrapped(), pStringB->getUnwrapped())
1470 == 0;
1471 }
1472
1473 static bool freeKeyValue(void* key, void* value, void* context) {
1474 delete (String*) key;
1475 delete (V*) value;
1476 return true;
1477 }
1478
1479 Hashmap* mpMap;
1480 };
1481
1482 class MacroTable : public StringTable<String> {
1483 public:
1484 MacroTable() : StringTable<String>(10) {}
1485 };
1486
Jack Palevich2db168f2009-06-11 14:29:47 -07001487 class KeywordTable {
1488 public:
1489
1490 KeywordTable(){
1491 mpMap = hashmapCreate(40, hashFn, equalsFn);
1492 put("int", TOK_INT);
1493 put("char", TOK_CHAR);
1494 put("void", TOK_VOID);
1495 put("if", TOK_IF);
1496 put("else", TOK_ELSE);
1497 put("while", TOK_WHILE);
1498 put("break", TOK_BREAK);
1499 put("return", TOK_RETURN);
1500 put("for", TOK_FOR);
Jack Palevich2ccc40d2009-06-12 11:53:07 -07001501 // TODO: remove these preprocessor-specific keywords. You should
1502 // be able to have symbols named pragma or define.
Jack Palevich2db168f2009-06-11 14:29:47 -07001503 put("pragma", TOK_PRAGMA);
Jack Palevich2ccc40d2009-06-12 11:53:07 -07001504 put("define", TOK_DEFINE);
Jack Palevich2db168f2009-06-11 14:29:47 -07001505 }
1506
1507 ~KeywordTable() {
1508 hashmapFree(mpMap);
1509 }
1510
Jack Palevich303d8ff2009-06-11 19:06:24 -07001511 int get(String* key) {
1512 return (int) hashmapGet(mpMap, key->getUnwrapped());
Jack Palevich2db168f2009-06-11 14:29:47 -07001513 }
1514
1515 const char* lookupKeyFor(int value) {
1516 FindValContext context;
1517 context.key = 0;
1518 hashmapForEach(mpMap, findKeyFn, &context);
1519 return context.key;
1520 }
1521
1522 private:
1523 void put(const char* kw, int val) {
1524 hashmapPut(mpMap, (void*) kw, (void*) val);
1525 }
1526
1527 static int hashFn(void* pKey) {
1528 char* pString = (char*) pKey;
1529 return hashmapHash(pString, strlen(pString));
1530 }
1531
1532 static bool equalsFn(void* keyA, void* keyB) {
1533 const char* pStringA = (const char*) keyA;
1534 const char* pStringB = (const char*) keyB;
1535 return strcmp(pStringA, pStringB) == 0;
1536 }
1537
1538 struct FindValContext {
1539 char* key;
1540 int value;
1541 };
1542
1543 static bool findKeyFn(void* key, void* value, void* context) {
1544 FindValContext* pContext = (FindValContext*) context;
1545 if ((int) value == pContext->value) {
1546 pContext->key = (char*) key;
1547 return false;
1548 }
1549 return true;
1550 }
1551
1552 Hashmap* mpMap;
1553 };
1554
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001555 template<class E> class Array {
1556 public:
1557 Array() {
1558 mpBase = 0;
1559 mUsed = 0;
1560 mSize = 0;
1561 }
1562
1563 ~Array() {
1564 if (mpBase) {
1565 free(mpBase);
1566 }
1567 }
1568
1569 E get(int i) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001570 if (i < 0 || i > (int) mUsed) {
1571 // error("internal error: Index out of range");
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001572 return E();
1573 }
1574 return mpBase[i];
1575 }
1576
1577 void set(int i, E val) {
1578 mpBase[i] = val;
1579 }
1580
1581 void pop() {
1582 if (mUsed > 0) {
1583 mUsed -= 1;
Jack Palevich36d94142009-06-08 15:55:32 -07001584 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001585 // error("internal error: Popped empty stack.");
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001586 }
1587 }
1588
1589 void push(E item) {
1590 * ensure(1) = item;
1591 }
1592
1593 size_t len() {
1594 return mUsed;
1595 }
1596
1597 private:
1598 E* ensure(int n) {
1599 size_t newUsed = mUsed + n;
1600 if (newUsed > mSize) {
1601 size_t newSize = mSize * 2 + 10;
1602 if (newSize < newUsed) {
1603 newSize = newUsed;
1604 }
1605 mpBase = (E*) realloc(mpBase, sizeof(E) * newSize);
1606 mSize = newSize;
1607 }
1608 E* result = mpBase + mUsed;
1609 mUsed = newUsed;
1610 return result;
1611 }
1612
1613 E* mpBase;
1614 size_t mUsed;
1615 size_t mSize;
1616 };
1617
Jack Palevich36d94142009-06-08 15:55:32 -07001618 struct InputState {
1619 InputStream* pStream;
1620 int oldCh;
1621 };
1622
Jack Palevich2db168f2009-06-11 14:29:47 -07001623 struct VariableInfo {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001624 VariableInfo() {
1625 pAddress = 0;
1626 pForward = 0;
1627 }
1628 void* pAddress;
1629 void* pForward; // For a forward direction, linked list of data to fix up
Jack Palevich2db168f2009-06-11 14:29:47 -07001630 };
1631
Jack Palevich303d8ff2009-06-11 19:06:24 -07001632 typedef StringTable<VariableInfo> SymbolTable;
1633
1634 class SymbolStack {
1635 public:
1636 SymbolStack() {
1637 mLevel = 0;
1638 addEntry();
1639 }
1640
1641 void pushLevel() {
1642 mLevel++;
1643 }
1644
1645 void popLevel() {
1646 mLevel--;
1647 Entry e = mStack.get(mStack.len()-1);
1648 if (mLevel < e.level) {
1649 mStack.pop();
1650 delete e.pTable;
1651 }
1652 }
1653
1654 VariableInfo* get(String* pName) {
1655 int len = mStack.len();
1656 VariableInfo* v = NULL;
1657 int level = -1;
1658 for (int i = len - 1; i >= 0; i--) {
1659 Entry e = mStack.get(i);
1660 v = e.pTable->get(pName);
1661 if (v) {
1662 level = e.level;
1663 break;
1664 }
1665 }
1666#if 0
1667 fprintf(stderr, "Lookup %s %08x level %d\n", pName->getUnwrapped(), v, level);
1668 if (v) {
1669 fprintf(stderr, " %08x %08x\n", v->pAddress, v->pForward);
1670 }
1671#endif
1672 return v;
1673 }
1674
1675 VariableInfo* addLocal(String* pName) {
1676 int len = mStack.len();
1677 if (mStack.get(len-1).level != mLevel) {
1678 addEntry();
1679 len++;
1680 }
1681 return addImp(len-1, pName);
1682 }
1683
1684 VariableInfo* addGlobal(String* pName) {
1685 return addImp(0, pName);
1686 }
1687
Jack Palevicha6baa232009-06-12 11:25:59 -07001688 void forEachGlobal(
1689 bool (*callback)(String* key, VariableInfo* value, void* context),
1690 void* context) {
1691 mStack.get(0).pTable->forEach(callback, context);
1692 }
1693
Jack Palevich303d8ff2009-06-11 19:06:24 -07001694 private:
1695 VariableInfo* addImp(int entryIndex, String* pName) {
1696 Entry e = mStack.get(entryIndex);
1697 SymbolTable* pTable = e.pTable;
Jack Palevicha6baa232009-06-12 11:25:59 -07001698 if (pTable->contains(pName)) {
1699 return NULL;
1700 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07001701 VariableInfo* v = new VariableInfo();
Jack Palevicha6baa232009-06-12 11:25:59 -07001702
Jack Palevich303d8ff2009-06-11 19:06:24 -07001703 delete pTable->put(pName, v);
1704#if 0
1705 fprintf(stderr, "Add \"%s\" %08x level %d\n", pName->getUnwrapped(), v, e.level);
1706#endif
1707 return v;
1708 }
1709
1710 void addEntry() {
1711 Entry e;
1712 e.level = mLevel;
1713 e.pTable = new SymbolTable();
1714 mStack.push(e);
1715 }
1716
1717 struct Entry {
1718 Entry() {
1719 level = 0;
1720 pTable = NULL;
1721 }
1722 int level;
1723 SymbolTable* pTable;
1724 };
1725
1726 int mLevel;
1727 Array<Entry> mStack;
1728 };
Jack Palevich36d94142009-06-08 15:55:32 -07001729
1730 int ch; // Current input character, or EOF
1731 intptr_t tok; // token
1732 intptr_t tokc; // token extra info
1733 int tokl; // token operator level
1734 intptr_t rsym; // return symbol
1735 intptr_t loc; // local variable index
1736 char* glo; // global variable index
Jack Palevich303d8ff2009-06-11 19:06:24 -07001737 String mTokenString;
Jack Palevich36d94142009-06-08 15:55:32 -07001738 char* dptr; // Macro state: Points to macro text during macro playback.
1739 int dch; // Macro state: Saves old value of ch during a macro playback.
Jack Palevich36d94142009-06-08 15:55:32 -07001740 char* pGlobalBase;
Jack Palevich2db168f2009-06-11 14:29:47 -07001741 KeywordTable mKeywords;
Jack Palevich303d8ff2009-06-11 19:06:24 -07001742 SymbolStack mSymbolTable;
Jack Palevich36d94142009-06-08 15:55:32 -07001743 InputStream* file;
1744
1745 CodeBuf codeBuf;
1746 CodeGenerator* pGen;
1747
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001748 MacroTable mMacros;
Jack Palevich36d94142009-06-08 15:55:32 -07001749 Array<InputState> mInputStateStack;
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001750
Jack Palevicheedf9d22009-06-04 16:23:40 -07001751 String mErrorBuf;
1752
Jack Palevicheedf9d22009-06-04 16:23:40 -07001753 String mPragmas;
1754 int mPragmaStringCount;
1755
Jack Palevich21a15a22009-05-11 14:49:29 -07001756 static const int ALLOC_SIZE = 99999;
1757
Jack Palevich303d8ff2009-06-11 19:06:24 -07001758 static const int TOK_DUMMY = 1;
1759 static const int TOK_NUM = 2;
1760
1761 // 3..255 are character and/or operators
1762
Jack Palevich2db168f2009-06-11 14:29:47 -07001763 // Keywords start at 0x100 and increase by 1
1764 static const int TOK_KEYWORD = 0x100;
1765 static const int TOK_INT = TOK_KEYWORD + 0;
1766 static const int TOK_CHAR = TOK_KEYWORD + 1;
1767 static const int TOK_VOID = TOK_KEYWORD + 2;
1768 static const int TOK_IF = TOK_KEYWORD + 3;
1769 static const int TOK_ELSE = TOK_KEYWORD + 4;
1770 static const int TOK_WHILE = TOK_KEYWORD + 5;
1771 static const int TOK_BREAK = TOK_KEYWORD + 6;
1772 static const int TOK_RETURN = TOK_KEYWORD + 7;
1773 static const int TOK_FOR = TOK_KEYWORD + 8;
1774 static const int TOK_PRAGMA = TOK_KEYWORD + 9;
1775 static const int TOK_DEFINE = TOK_KEYWORD + 10;
1776
Jack Palevich303d8ff2009-06-11 19:06:24 -07001777 static const int TOK_UNDEFINED_SYMBOL = 0x200;
Jack Palevich21a15a22009-05-11 14:49:29 -07001778
Jack Palevich303d8ff2009-06-11 19:06:24 -07001779 // Symbols start at 0x300, but are really pointers to VariableInfo structs.
1780 static const int TOK_SYMBOL = 0x300;
1781
Jack Palevich21a15a22009-05-11 14:49:29 -07001782
1783 static const int LOCAL = 0x200;
1784
1785 static const int SYM_FORWARD = 0;
1786 static const int SYM_DEFINE = 1;
1787
1788 /* tokens in string heap */
1789 static const int TAG_TOK = ' ';
Jack Palevich21a15a22009-05-11 14:49:29 -07001790
Jack Palevichbf42c9c2009-05-12 12:48:35 -07001791 static const int OP_INCREMENT = 0;
1792 static const int OP_DECREMENT = 1;
1793 static const int OP_MUL = 2;
1794 static const int OP_DIV = 3;
1795 static const int OP_MOD = 4;
1796 static const int OP_PLUS = 5;
1797 static const int OP_MINUS = 6;
1798 static const int OP_SHIFT_LEFT = 7;
1799 static const int OP_SHIFT_RIGHT = 8;
1800 static const int OP_LESS_EQUAL = 9;
1801 static const int OP_GREATER_EQUAL = 10;
1802 static const int OP_LESS = 11;
1803 static const int OP_GREATER = 12;
1804 static const int OP_EQUALS = 13;
1805 static const int OP_NOT_EQUALS = 14;
1806 static const int OP_LOGICAL_AND = 15;
1807 static const int OP_LOGICAL_OR = 16;
1808 static const int OP_BIT_AND = 17;
1809 static const int OP_BIT_XOR = 18;
1810 static const int OP_BIT_OR = 19;
1811 static const int OP_BIT_NOT = 20;
1812 static const int OP_LOGICAL_NOT = 21;
1813 static const int OP_COUNT = 22;
1814
1815 /* Operators are searched from front, the two-character operators appear
1816 * before the single-character operators with the same first character.
1817 * @ is used to pad out single-character operators.
1818 */
1819 static const char* operatorChars;
1820 static const char operatorLevel[];
1821
Jack Palevich21a15a22009-05-11 14:49:29 -07001822 void pdef(int t) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001823 mTokenString.append(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07001824 }
1825
1826 void inp() {
1827 if (dptr) {
Jack Palevich653f42d2009-05-28 17:15:32 -07001828 ch = *dptr++;
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001829 if (ch == 0) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001830 dptr = 0;
1831 ch = dch;
1832 }
1833 } else
Jack Palevicheedf9d22009-06-04 16:23:40 -07001834 ch = file->getChar();
Jack Palevichb7c81e92009-06-04 19:56:13 -07001835#if 0
1836 printf("ch='%c' 0x%x\n", ch, ch);
1837#endif
Jack Palevich21a15a22009-05-11 14:49:29 -07001838 }
1839
1840 int isid() {
Jack Palevich546b2242009-05-13 15:10:04 -07001841 return isalnum(ch) | (ch == '_');
Jack Palevich21a15a22009-05-11 14:49:29 -07001842 }
1843
Jack Palevichb4758ff2009-06-12 12:49:14 -07001844 /* read a character constant, advances ch to after end of constant */
1845 int getq() {
1846 int val = ch;
Jack Palevich21a15a22009-05-11 14:49:29 -07001847 if (ch == '\\') {
1848 inp();
Jack Palevichb4758ff2009-06-12 12:49:14 -07001849 if (isoctal(ch)) {
1850 // 1 to 3 octal characters.
1851 val = 0;
1852 for(int i = 0; i < 3; i++) {
1853 if (isoctal(ch)) {
1854 val = (val << 3) + ch - '0';
1855 inp();
1856 }
1857 }
1858 return val;
1859 } else if (ch == 'x' || ch == 'X') {
1860 // N hex chars
1861 inp();
1862 if (! isxdigit(ch)) {
1863 error("'x' character escape requires at least one digit.");
1864 } else {
1865 val = 0;
1866 while (isxdigit(ch)) {
1867 int d = ch;
1868 if (isdigit(d)) {
1869 d -= '0';
1870 } else if (d <= 'F') {
1871 d = d - 'A' + 10;
1872 } else {
1873 d = d - 'a' + 10;
1874 }
1875 val = (val << 4) + d;
1876 inp();
1877 }
1878 }
1879 } else {
1880 int val = ch;
1881 switch (ch) {
1882 case 'a':
1883 val = '\a';
1884 break;
1885 case 'b':
1886 val = '\b';
1887 break;
1888 case 'f':
1889 val = '\f';
1890 break;
1891 case 'n':
1892 val = '\n';
1893 break;
1894 case 'r':
1895 val = '\r';
1896 break;
1897 case 't':
1898 val = '\t';
1899 break;
1900 case 'v':
1901 val = '\v';
1902 break;
1903 case '\\':
1904 val = '\\';
1905 break;
1906 case '\'':
1907 val = '\'';
1908 break;
1909 case '"':
1910 val = '"';
1911 break;
1912 case '?':
1913 val = '?';
1914 break;
1915 default:
1916 error("Undefined character escape %c", ch);
1917 break;
1918 }
1919 inp();
1920 return val;
1921 }
1922 } else {
1923 inp();
Jack Palevich21a15a22009-05-11 14:49:29 -07001924 }
Jack Palevichb4758ff2009-06-12 12:49:14 -07001925 return val;
1926 }
1927
1928 static bool isoctal(int ch) {
1929 return ch >= '0' && ch <= '7';
Jack Palevich21a15a22009-05-11 14:49:29 -07001930 }
1931
1932 void next() {
1933 int l, a;
1934
Jack Palevich546b2242009-05-13 15:10:04 -07001935 while (isspace(ch) | (ch == '#')) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001936 if (ch == '#') {
1937 inp();
1938 next();
1939 if (tok == TOK_DEFINE) {
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001940 doDefine();
Jack Palevicheedf9d22009-06-04 16:23:40 -07001941 } else if (tok == TOK_PRAGMA) {
1942 doPragma();
1943 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001944 error("Unsupported preprocessor directive \"%s\"",
1945 mTokenString.getUnwrapped());
Jack Palevich21a15a22009-05-11 14:49:29 -07001946 }
Jack Palevich21a15a22009-05-11 14:49:29 -07001947 }
1948 inp();
1949 }
1950 tokl = 0;
1951 tok = ch;
1952 /* encode identifiers & numbers */
1953 if (isid()) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001954 mTokenString.clear();
Jack Palevich21a15a22009-05-11 14:49:29 -07001955 while (isid()) {
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001956 pdef(ch);
1957 inp();
Jack Palevichae54f1f2009-05-08 14:54:15 -07001958 }
Jack Palevich21a15a22009-05-11 14:49:29 -07001959 if (isdigit(tok)) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001960 tokc = strtol(mTokenString.getUnwrapped(), 0, 0);
Jack Palevich21a15a22009-05-11 14:49:29 -07001961 tok = TOK_NUM;
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001962 } else {
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001963 // Is this a macro?
Jack Palevich303d8ff2009-06-11 19:06:24 -07001964 String* pValue = mMacros.get(&mTokenString);
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001965 if (pValue) {
1966 // Yes, it is a macro
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001967 dptr = pValue->getUnwrapped();
1968 dch = ch;
1969 inp();
1970 next();
1971 } else {
Jack Palevich2db168f2009-06-11 14:29:47 -07001972 // Is this a keyword?
Jack Palevich303d8ff2009-06-11 19:06:24 -07001973 int kwtok = mKeywords.get(&mTokenString);
Jack Palevich2db168f2009-06-11 14:29:47 -07001974 if (kwtok) {
1975 tok = kwtok;
1976 // fprintf(stderr, "tok= keyword %s %x\n", last_id, tok);
1977 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001978 tok = (intptr_t) mSymbolTable.get(&mTokenString);
1979 if (!tok) {
1980 tok = TOK_UNDEFINED_SYMBOL;
1981 }
Jack Palevich2db168f2009-06-11 14:29:47 -07001982 // fprintf(stderr, "tok= symbol %s %x\n", last_id, tok);
Jack Palevich21a15a22009-05-11 14:49:29 -07001983 }
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001984 }
1985 }
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001986 } else {
Jack Palevich21a15a22009-05-11 14:49:29 -07001987 inp();
1988 if (tok == '\'') {
1989 tok = TOK_NUM;
Jack Palevichb4758ff2009-06-12 12:49:14 -07001990 tokc = getq();
1991 if (ch != '\'') {
1992 error("Expected a ' character, got %c", ch);
1993 } else {
1994 inp();
1995 }
Jack Palevich546b2242009-05-13 15:10:04 -07001996 } else if ((tok == '/') & (ch == '*')) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001997 inp();
1998 while (ch) {
1999 while (ch != '*')
2000 inp();
2001 inp();
2002 if (ch == '/')
2003 ch = 0;
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002004 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002005 inp();
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002006 next();
Jack Palevichbd894902009-05-14 19:35:31 -07002007 } else if ((tok == '/') & (ch == '/')) {
2008 inp();
2009 while (ch && (ch != '\n')) {
2010 inp();
2011 }
2012 inp();
2013 next();
Jack Palevich21a15a22009-05-11 14:49:29 -07002014 } else {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002015 const char* t = operatorChars;
2016 int opIndex = 0;
Jack Palevich546b2242009-05-13 15:10:04 -07002017 while ((l = *t++) != 0) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002018 a = *t++;
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002019 tokl = operatorLevel[opIndex];
2020 tokc = opIndex;
Jack Palevich546b2242009-05-13 15:10:04 -07002021 if ((l == tok) & ((a == ch) | (a == '@'))) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002022#if 0
2023 printf("%c%c -> tokl=%d tokc=0x%x\n",
2024 l, a, tokl, tokc);
2025#endif
2026 if (a == ch) {
2027 inp();
2028 tok = TOK_DUMMY; /* dummy token for double tokens */
2029 }
2030 break;
2031 }
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002032 opIndex++;
2033 }
2034 if (l == 0) {
2035 tokl = 0;
2036 tokc = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002037 }
2038 }
2039 }
2040#if 0
2041 {
Jack Palevich2db168f2009-06-11 14:29:47 -07002042 const char* p;
Jack Palevich21a15a22009-05-11 14:49:29 -07002043
2044 printf("tok=0x%x ", tok);
Jack Palevich2db168f2009-06-11 14:29:47 -07002045 if (tok >= TOK_KEYWORD) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002046 printf("'");
Jack Palevich2db168f2009-06-11 14:29:47 -07002047 if (tok>= TOK_SYMBOL)
2048 p = sym_stk + 1 + ((char*) tok - (char*) pVarsBase) / 8;
2049 else {
2050 p = mKeywords.lookupKeyFor(tok);
2051 if (!p) {
2052 p = "unknown keyword";
2053 }
2054 }
Jack Palevich653f42d2009-05-28 17:15:32 -07002055 while (*p != TAG_TOK && *p)
2056 printf("%c", *p++);
Jack Palevich21a15a22009-05-11 14:49:29 -07002057 printf("'\n");
2058 } else if (tok == TOK_NUM) {
2059 printf("%d\n", tokc);
2060 } else {
2061 printf("'%c'\n", tok);
2062 }
2063 }
2064#endif
2065 }
2066
Jack Palevich2d11dfb2009-06-08 14:34:26 -07002067 void doDefine() {
2068 String* pName = new String();
2069 while (isspace(ch)) {
2070 inp();
2071 }
2072 while (isid()) {
2073 pName->append(ch);
2074 inp();
2075 }
2076 if (ch == '(') {
2077 delete pName;
2078 error("Defines with arguments not supported");
Jack Palevich0a280a02009-06-11 10:53:51 -07002079 return;
Jack Palevich2d11dfb2009-06-08 14:34:26 -07002080 }
2081 while (isspace(ch)) {
2082 inp();
2083 }
2084 String* pValue = new String();
2085 while (ch != '\n' && ch != EOF) {
2086 pValue->append(ch);
2087 inp();
2088 }
2089 delete mMacros.put(pName, pValue);
2090 }
2091
Jack Palevicheedf9d22009-06-04 16:23:40 -07002092 void doPragma() {
2093 // # pragma name(val)
2094 int state = 0;
2095 while(ch != EOF && ch != '\n' && state < 10) {
2096 switch(state) {
2097 case 0:
2098 if (isspace(ch)) {
2099 inp();
2100 } else {
2101 state++;
2102 }
2103 break;
2104 case 1:
2105 if (isalnum(ch)) {
2106 mPragmas.append(ch);
2107 inp();
2108 } else if (ch == '(') {
2109 mPragmas.append(0);
2110 inp();
2111 state++;
2112 } else {
2113 state = 11;
2114 }
2115 break;
2116 case 2:
2117 if (isalnum(ch)) {
2118 mPragmas.append(ch);
2119 inp();
2120 } else if (ch == ')') {
2121 mPragmas.append(0);
2122 inp();
2123 state = 10;
2124 } else {
2125 state = 11;
2126 }
2127 break;
2128 }
2129 }
2130 if(state != 10) {
2131 error("Unexpected pragma syntax");
2132 }
2133 mPragmaStringCount += 2;
2134 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002135
Jack Palevichac0e95e2009-05-29 13:53:44 -07002136 virtual void verror(const char* fmt, va_list ap) {
Jack Palevicheedf9d22009-06-04 16:23:40 -07002137 mErrorBuf.printf("%ld: ", file->getLine());
2138 mErrorBuf.vprintf(fmt, ap);
2139 mErrorBuf.printf("\n");
Jack Palevich21a15a22009-05-11 14:49:29 -07002140 }
2141
Jack Palevich8b0624c2009-05-20 12:12:06 -07002142 void skip(intptr_t c) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002143 if (tok != c) {
2144 error("'%c' expected", c);
2145 }
2146 next();
2147 }
2148
Jack Palevich21a15a22009-05-11 14:49:29 -07002149 /* l is one if '=' parsing wanted (quick hack) */
Jack Palevich8b0624c2009-05-20 12:12:06 -07002150 void unary(intptr_t l) {
Jack Palevich653f42d2009-05-28 17:15:32 -07002151 intptr_t n, t, a;
2152 int c;
Jack Palevicha6baa232009-06-12 11:25:59 -07002153 String tString;
Jack Palevich546b2242009-05-13 15:10:04 -07002154 t = 0;
Jack Palevichb4758ff2009-06-12 12:49:14 -07002155 n = 1; /* type of expression 0 = forward, 1 = value, other = lvalue */
Jack Palevich21a15a22009-05-11 14:49:29 -07002156 if (tok == '\"') {
Jack Palevich653f42d2009-05-28 17:15:32 -07002157 pGen->li((int) glo);
Jack Palevichb4758ff2009-06-12 12:49:14 -07002158 while (ch != '\"' && ch != EOF) {
2159 *allocGlobalSpace(1) = getq();
2160 }
2161 if (ch != '\"') {
2162 error("Unterminated string constant.");
Jack Palevich21a15a22009-05-11 14:49:29 -07002163 }
Jack Palevich653f42d2009-05-28 17:15:32 -07002164 *glo = 0;
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002165 /* align heap */
2166 allocGlobalSpace((char*) (((intptr_t) glo + 4) & -4) - glo);
Jack Palevich21a15a22009-05-11 14:49:29 -07002167 inp();
2168 next();
2169 } else {
2170 c = tokl;
2171 a = tokc;
2172 t = tok;
Jack Palevicha6baa232009-06-12 11:25:59 -07002173 tString = mTokenString;
Jack Palevich21a15a22009-05-11 14:49:29 -07002174 next();
2175 if (t == TOK_NUM) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002176 pGen->li(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002177 } else if (c == 2) {
2178 /* -, +, !, ~ */
2179 unary(0);
Jack Palevich1cdef202009-05-22 12:06:27 -07002180 pGen->clearR1();
Jack Palevich21a15a22009-05-11 14:49:29 -07002181 if (t == '!')
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002182 pGen->gcmp(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002183 else
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002184 pGen->genOp(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002185 } else if (t == '(') {
2186 expr();
2187 skip(')');
2188 } else if (t == '*') {
2189 /* parse cast */
2190 skip('(');
2191 t = tok; /* get type */
2192 next(); /* skip int/char/void */
2193 next(); /* skip '*' or '(' */
2194 if (tok == '*') {
2195 /* function type */
2196 skip('*');
2197 skip(')');
2198 skip('(');
2199 skip(')');
2200 t = 0;
2201 }
2202 skip(')');
2203 unary(0);
2204 if (tok == '=') {
2205 next();
Jack Palevich1cdef202009-05-22 12:06:27 -07002206 pGen->pushR0();
Jack Palevich21a15a22009-05-11 14:49:29 -07002207 expr();
Jack Palevich1cdef202009-05-22 12:06:27 -07002208 pGen->popR1();
2209 pGen->storeR0ToR1(t == TOK_INT);
Jack Palevich21a15a22009-05-11 14:49:29 -07002210 } else if (t) {
Jack Palevich1cdef202009-05-22 12:06:27 -07002211 pGen->loadR0FromR0(t == TOK_INT);
Jack Palevich21a15a22009-05-11 14:49:29 -07002212 }
2213 } else if (t == '&') {
Jack Palevich1cdef202009-05-22 12:06:27 -07002214 pGen->leaR0(*(int *) tok);
Jack Palevich21a15a22009-05-11 14:49:29 -07002215 next();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002216 } else if (t == EOF ) {
2217 error("Unexpected EOF.");
2218 } else if (t < TOK_UNDEFINED_SYMBOL) {
2219 error("Unexpected symbol or keyword");
Jack Palevich21a15a22009-05-11 14:49:29 -07002220 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002221 if (t == TOK_UNDEFINED_SYMBOL) {
2222 t = (intptr_t) mSymbolTable.addGlobal(
Jack Palevicha6baa232009-06-12 11:25:59 -07002223 new String(tString));
Jack Palevich303d8ff2009-06-11 19:06:24 -07002224 }
2225
Jack Palevicha6baa232009-06-12 11:25:59 -07002226 n = (intptr_t) ((VariableInfo*) t)->pAddress;
Jack Palevich21a15a22009-05-11 14:49:29 -07002227 /* forward reference: try dlsym */
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002228 if (!n) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002229 n = (intptr_t) dlsym(RTLD_DEFAULT,
Jack Palevicha6baa232009-06-12 11:25:59 -07002230 tString.getUnwrapped());
2231 ((VariableInfo*) t)->pAddress = (void*) n;
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002232 }
Jack Palevich546b2242009-05-13 15:10:04 -07002233 if ((tok == '=') & l) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002234 /* assignment */
2235 next();
2236 expr();
Jack Palevich1cdef202009-05-22 12:06:27 -07002237 pGen->storeR0(n);
Jack Palevich21a15a22009-05-11 14:49:29 -07002238 } else if (tok != '(') {
2239 /* variable */
Jack Palevicha6baa232009-06-12 11:25:59 -07002240 if (!n) {
2241 error("Undefined variable %s", tString.getUnwrapped());
2242 }
Jack Palevich1cdef202009-05-22 12:06:27 -07002243 pGen->loadR0(n, tokl == 11, tokc);
Jack Palevich21a15a22009-05-11 14:49:29 -07002244 if (tokl == 11) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002245 next();
2246 }
2247 }
2248 }
2249 }
2250
2251 /* function call */
2252 if (tok == '(') {
2253 if (n == 1)
Jack Palevich1cdef202009-05-22 12:06:27 -07002254 pGen->pushR0();
Jack Palevich21a15a22009-05-11 14:49:29 -07002255
2256 /* push args and invert order */
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002257 a = pGen->beginFunctionCallArguments();
Jack Palevich21a15a22009-05-11 14:49:29 -07002258 next();
2259 l = 0;
Jack Palevichb4758ff2009-06-12 12:49:14 -07002260 while (tok != ')' && tok != EOF) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002261 expr();
Jack Palevich1cdef202009-05-22 12:06:27 -07002262 pGen->storeR0ToArg(l);
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002263 if (tok == ',')
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002264 next();
Jack Palevich21a15a22009-05-11 14:49:29 -07002265 l = l + 4;
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002266 }
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002267 pGen->endFunctionCallArguments(a, l);
Jack Palevichb4758ff2009-06-12 12:49:14 -07002268 skip(')');
Jack Palevich21a15a22009-05-11 14:49:29 -07002269 if (!n) {
2270 /* forward reference */
2271 t = t + 4;
2272 *(int *) t = pGen->callForward(*(int *) t);
2273 } else if (n == 1) {
2274 pGen->callIndirect(l);
Jack Palevich21a15a22009-05-11 14:49:29 -07002275 } else {
Jack Palevich7810bc92009-05-15 14:31:47 -07002276 pGen->callRelative(n - codeBuf.getPC() - pGen->jumpOffset());
Jack Palevich21a15a22009-05-11 14:49:29 -07002277 }
Jack Palevich3d474a72009-05-15 15:12:38 -07002278 if (l | (n == 1))
Jack Palevich7810bc92009-05-15 14:31:47 -07002279 pGen->adjustStackAfterCall(l, n == 1);
Jack Palevich21a15a22009-05-11 14:49:29 -07002280 }
2281 }
2282
Jack Palevich653f42d2009-05-28 17:15:32 -07002283 void sum(int l) {
Jack Palevich8b0624c2009-05-20 12:12:06 -07002284 intptr_t t, n, a;
Jack Palevich546b2242009-05-13 15:10:04 -07002285 t = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002286 if (l-- == 1)
2287 unary(1);
2288 else {
2289 sum(l);
2290 a = 0;
2291 while (l == tokl) {
2292 n = tok;
2293 t = tokc;
2294 next();
2295
2296 if (l > 8) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002297 a = pGen->gtst(t == OP_LOGICAL_OR, a); /* && and || output code generation */
Jack Palevich21a15a22009-05-11 14:49:29 -07002298 sum(l);
2299 } else {
Jack Palevich1cdef202009-05-22 12:06:27 -07002300 pGen->pushR0();
Jack Palevich21a15a22009-05-11 14:49:29 -07002301 sum(l);
Jack Palevich1cdef202009-05-22 12:06:27 -07002302 pGen->popR1();
Jack Palevich21a15a22009-05-11 14:49:29 -07002303
Jack Palevich546b2242009-05-13 15:10:04 -07002304 if ((l == 4) | (l == 5)) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002305 pGen->gcmp(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07002306 } else {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002307 pGen->genOp(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07002308 }
2309 }
2310 }
2311 /* && and || output code generation */
2312 if (a && l > 8) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002313 a = pGen->gtst(t == OP_LOGICAL_OR, a);
2314 pGen->li(t != OP_LOGICAL_OR);
Jack Palevicha6535612009-05-13 16:24:17 -07002315 pGen->gjmp(5); /* jmp $ + 5 (sizeof li, FIXME for ARM) */
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002316 pGen->gsym(a);
2317 pGen->li(t == OP_LOGICAL_OR);
Jack Palevich21a15a22009-05-11 14:49:29 -07002318 }
2319 }
2320 }
2321
2322 void expr() {
2323 sum(11);
2324 }
2325
2326 int test_expr() {
2327 expr();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002328 return pGen->gtst(0, 0);
Jack Palevich21a15a22009-05-11 14:49:29 -07002329 }
2330
Jack Palevicha6baa232009-06-12 11:25:59 -07002331 void block(intptr_t l, bool outermostFunctionBlock) {
Jack Palevich8b0624c2009-05-20 12:12:06 -07002332 intptr_t a, n, t;
Jack Palevich21a15a22009-05-11 14:49:29 -07002333
2334 if (tok == TOK_IF) {
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002335 next();
2336 skip('(');
Jack Palevich21a15a22009-05-11 14:49:29 -07002337 a = test_expr();
2338 skip(')');
Jack Palevicha6baa232009-06-12 11:25:59 -07002339 block(l, false);
Jack Palevich21a15a22009-05-11 14:49:29 -07002340 if (tok == TOK_ELSE) {
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002341 next();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002342 n = pGen->gjmp(0); /* jmp */
2343 pGen->gsym(a);
Jack Palevicha6baa232009-06-12 11:25:59 -07002344 block(l, false);
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002345 pGen->gsym(n); /* patch else jmp */
Jack Palevich21a15a22009-05-11 14:49:29 -07002346 } else {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002347 pGen->gsym(a); /* patch if test */
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002348 }
Jack Palevich546b2242009-05-13 15:10:04 -07002349 } else if ((tok == TOK_WHILE) | (tok == TOK_FOR)) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002350 t = tok;
2351 next();
2352 skip('(');
2353 if (t == TOK_WHILE) {
Jack Palevicha6535612009-05-13 16:24:17 -07002354 n = codeBuf.getPC(); // top of loop, target of "next" iteration
Jack Palevich21a15a22009-05-11 14:49:29 -07002355 a = test_expr();
2356 } else {
2357 if (tok != ';')
2358 expr();
2359 skip(';');
2360 n = codeBuf.getPC();
2361 a = 0;
2362 if (tok != ';')
2363 a = test_expr();
2364 skip(';');
2365 if (tok != ')') {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002366 t = pGen->gjmp(0);
Jack Palevich21a15a22009-05-11 14:49:29 -07002367 expr();
Jack Palevicha6535612009-05-13 16:24:17 -07002368 pGen->gjmp(n - codeBuf.getPC() - pGen->jumpOffset());
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002369 pGen->gsym(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07002370 n = t + 4;
2371 }
2372 }
2373 skip(')');
Jack Palevicha6baa232009-06-12 11:25:59 -07002374 block((intptr_t) &a, false);
Jack Palevicha6535612009-05-13 16:24:17 -07002375 pGen->gjmp(n - codeBuf.getPC() - pGen->jumpOffset()); /* jmp */
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002376 pGen->gsym(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002377 } else if (tok == '{') {
Jack Palevicha6baa232009-06-12 11:25:59 -07002378 if (! outermostFunctionBlock) {
2379 mSymbolTable.pushLevel();
2380 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002381 next();
2382 /* declarations */
Jack Palevichb7c81e92009-06-04 19:56:13 -07002383 localDeclarations();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002384 while (tok != '}' && tok != EOF)
Jack Palevicha6baa232009-06-12 11:25:59 -07002385 block(l, false);
Jack Palevich303d8ff2009-06-11 19:06:24 -07002386 skip('}');
Jack Palevicha6baa232009-06-12 11:25:59 -07002387 if (! outermostFunctionBlock) {
2388 mSymbolTable.popLevel();
2389 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002390 } else {
2391 if (tok == TOK_RETURN) {
2392 next();
2393 if (tok != ';')
2394 expr();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002395 rsym = pGen->gjmp(rsym); /* jmp */
Jack Palevich21a15a22009-05-11 14:49:29 -07002396 } else if (tok == TOK_BREAK) {
2397 next();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002398 *(int *) l = pGen->gjmp(*(int *) l);
Jack Palevich21a15a22009-05-11 14:49:29 -07002399 } else if (tok != ';')
2400 expr();
2401 skip(';');
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002402 }
2403 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002404
Jack Palevichb7c81e92009-06-04 19:56:13 -07002405 typedef int Type;
2406 static const Type TY_UNKNOWN = 0;
2407 static const Type TY_INT = 1;
2408 static const Type TY_CHAR = 2;
2409 static const Type TY_VOID = 3;
2410 static const int TY_BASE_TYPE_MASK = 0xf;
2411 static const int TY_INDIRECTION_MASK = 0xf0;
2412 static const int TY_INDIRECTION_SHIFT = 4;
2413 static const int MAX_INDIRECTION_COUNT = 15;
Jack Palevich21a15a22009-05-11 14:49:29 -07002414
Jack Palevichb7c81e92009-06-04 19:56:13 -07002415 Type getBaseType(Type t) {
2416 return t & TY_BASE_TYPE_MASK;
2417 }
2418
2419 int getIndirectionCount(Type t) {
2420 return (TY_INDIRECTION_MASK & t) >> TY_INDIRECTION_SHIFT;
2421 }
2422
2423 void setIndirectionCount(Type& t, int count) {
2424 t = ((TY_INDIRECTION_MASK & (count << TY_INDIRECTION_SHIFT))
2425 | (t & ~TY_INDIRECTION_MASK));
2426 }
2427
2428 bool acceptType(Type& t) {
2429 t = TY_UNKNOWN;
2430 if (tok == TOK_INT) {
2431 t = TY_INT;
2432 } else if (tok == TOK_CHAR) {
2433 t = TY_CHAR;
2434 } else if (tok == TOK_VOID) {
2435 t = TY_VOID;
2436 } else {
2437 return false;
2438 }
2439 next();
2440 return true;
2441 }
2442
2443 Type acceptPointerDeclaration(Type& base) {
2444 Type t = base;
2445 int indirectionCount = 0;
2446 while (tok == '*' && indirectionCount <= MAX_INDIRECTION_COUNT) {
2447 next();
2448 indirectionCount++;
2449 }
2450 if (indirectionCount > MAX_INDIRECTION_COUNT) {
2451 error("Too many levels of pointer. Max %d", MAX_INDIRECTION_COUNT);
2452 }
2453 setIndirectionCount(t, indirectionCount);
2454 return t;
2455 }
2456
2457 void expectType(Type& t) {
2458 if (!acceptType(t)) {
2459 error("Expected a type.");
2460 }
2461 }
2462
2463 void checkSymbol() {
Jack Palevich2db168f2009-06-11 14:29:47 -07002464 if (tok < TOK_SYMBOL) {
Jack Palevichb7c81e92009-06-04 19:56:13 -07002465 error("Expected a symbol");
2466 }
2467 }
2468
Jack Palevicha6baa232009-06-12 11:25:59 -07002469 void addGlobalSymbol() {
2470 tok = (intptr_t) mSymbolTable.addGlobal(
2471 new String(mTokenString));
2472 reportIfDuplicate();
2473 }
2474
2475 void reportIfDuplicate() {
2476 if (!tok) {
2477 error("Duplicate definition of %s", mTokenString.getUnwrapped());
Jack Palevich303d8ff2009-06-11 19:06:24 -07002478 }
2479 }
2480
Jack Palevicha6baa232009-06-12 11:25:59 -07002481 void addLocalSymbol() {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002482 tok = (intptr_t) mSymbolTable.addLocal(
2483 new String(mTokenString));
Jack Palevicha6baa232009-06-12 11:25:59 -07002484 reportIfDuplicate();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002485 }
2486
Jack Palevichb7c81e92009-06-04 19:56:13 -07002487 void localDeclarations() {
2488 intptr_t a;
2489 Type base;
2490
2491 while (acceptType(base)) {
2492 while (tok != ';') {
2493 Type t = acceptPointerDeclaration(t);
Jack Palevicha6baa232009-06-12 11:25:59 -07002494 addLocalSymbol();
2495 if (tok) {
2496 loc = loc + 4;
2497 *(int *) tok = -loc;
2498 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002499 next();
Jack Palevichb7c81e92009-06-04 19:56:13 -07002500 if (tok == ',')
2501 next();
2502 }
2503 skip(';');
2504 }
2505 }
2506
2507 void globalDeclarations() {
2508 while (tok != EOF) {
2509 Type base;
2510 expectType(base);
2511 Type t = acceptPointerDeclaration(t);
Jack Palevicha6baa232009-06-12 11:25:59 -07002512 if (tok == TOK_UNDEFINED_SYMBOL) {
2513 addGlobalSymbol();
2514 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07002515 VariableInfo* name = (VariableInfo*) tok;
Jack Palevicha6baa232009-06-12 11:25:59 -07002516 if (name && name->pAddress) {
2517 error("Already defined global %s",
2518 mTokenString.getUnwrapped());
2519 }
Jack Palevichb7c81e92009-06-04 19:56:13 -07002520 next();
2521 if (tok == ',' || tok == ';') {
2522 // it's a variable declaration
2523 for(;;) {
Jack Palevicha6baa232009-06-12 11:25:59 -07002524 if (name) {
2525 name->pAddress = (int*) allocGlobalSpace(4);
2526 }
Jack Palevichb7c81e92009-06-04 19:56:13 -07002527 if (tok != ',') {
2528 break;
Jack Palevich21a15a22009-05-11 14:49:29 -07002529 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07002530 skip(',');
Jack Palevichb7c81e92009-06-04 19:56:13 -07002531 t = acceptPointerDeclaration(t);
Jack Palevicha6baa232009-06-12 11:25:59 -07002532 addGlobalSymbol();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002533 name = (VariableInfo*) tok;
Jack Palevichb7c81e92009-06-04 19:56:13 -07002534 next();
Jack Palevich21a15a22009-05-11 14:49:29 -07002535 }
2536 skip(';');
2537 } else {
Jack Palevicha6baa232009-06-12 11:25:59 -07002538 if (name) {
2539 /* patch forward references (XXX: does not work for function
2540 pointers) */
2541 pGen->gsym((int) name->pForward);
2542 /* put function address */
2543 name->pAddress = (void*) codeBuf.getPC();
2544 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002545 skip('(');
Jack Palevich303d8ff2009-06-11 19:06:24 -07002546 mSymbolTable.pushLevel();
Jack Palevichb7c81e92009-06-04 19:56:13 -07002547 intptr_t a = 8;
Jack Palevich546b2242009-05-13 15:10:04 -07002548 int argCount = 0;
Jack Palevich303d8ff2009-06-11 19:06:24 -07002549 while (tok != ')' && tok != EOF) {
Jack Palevichb7c81e92009-06-04 19:56:13 -07002550 Type aType;
2551 expectType(aType);
2552 aType = acceptPointerDeclaration(aType);
Jack Palevicha6baa232009-06-12 11:25:59 -07002553 addLocalSymbol();
2554 if (tok) {
2555 /* read param name and compute offset */
2556 *(int *) tok = a;
2557 a = a + 4;
2558 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002559 next();
2560 if (tok == ',')
2561 next();
Jack Palevich546b2242009-05-13 15:10:04 -07002562 argCount++;
Jack Palevich21a15a22009-05-11 14:49:29 -07002563 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07002564 skip(')');
Jack Palevich21a15a22009-05-11 14:49:29 -07002565 rsym = loc = 0;
Jack Palevich546b2242009-05-13 15:10:04 -07002566 a = pGen->functionEntry(argCount);
Jack Palevicha6baa232009-06-12 11:25:59 -07002567 block(0, true);
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002568 pGen->gsym(rsym);
Jack Palevich546b2242009-05-13 15:10:04 -07002569 pGen->functionExit(argCount, a, loc);
Jack Palevich303d8ff2009-06-11 19:06:24 -07002570 mSymbolTable.popLevel();
Jack Palevich21a15a22009-05-11 14:49:29 -07002571 }
2572 }
2573 }
2574
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002575 char* allocGlobalSpace(int bytes) {
2576 if (glo - pGlobalBase + bytes > ALLOC_SIZE) {
2577 error("Global space exhausted");
Jack Palevich0a280a02009-06-11 10:53:51 -07002578 return NULL;
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002579 }
2580 char* result = glo;
2581 glo += bytes;
2582 return result;
2583 }
2584
Jack Palevich21a15a22009-05-11 14:49:29 -07002585 void cleanup() {
Jack Palevich21a15a22009-05-11 14:49:29 -07002586 if (pGlobalBase != 0) {
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002587 free(pGlobalBase);
Jack Palevich21a15a22009-05-11 14:49:29 -07002588 pGlobalBase = 0;
2589 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002590 if (pGen) {
2591 delete pGen;
2592 pGen = 0;
2593 }
Jack Palevich1cdef202009-05-22 12:06:27 -07002594 if (file) {
2595 delete file;
2596 file = 0;
2597 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002598 }
2599
2600 void clear() {
2601 tok = 0;
2602 tokc = 0;
2603 tokl = 0;
2604 ch = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002605 rsym = 0;
2606 loc = 0;
2607 glo = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002608 dptr = 0;
2609 dch = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002610 file = 0;
2611 pGlobalBase = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002612 pGen = 0;
Jack Palevicheedf9d22009-06-04 16:23:40 -07002613 mPragmaStringCount = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002614 }
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002615
Jack Palevich22305132009-05-13 10:58:45 -07002616 void setArchitecture(const char* architecture) {
2617 delete pGen;
2618 pGen = 0;
2619
2620 if (architecture != NULL) {
Jack Paleviche7b59062009-05-19 17:12:17 -07002621#ifdef PROVIDE_ARM_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -07002622 if (! pGen && strcmp(architecture, "arm") == 0) {
Jack Palevich22305132009-05-13 10:58:45 -07002623 pGen = new ARMCodeGenerator();
Jack Palevich8b0624c2009-05-20 12:12:06 -07002624 }
Jack Paleviche7b59062009-05-19 17:12:17 -07002625#endif
Jack Paleviche7b59062009-05-19 17:12:17 -07002626#ifdef PROVIDE_X86_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -07002627 if (! pGen && strcmp(architecture, "x86") == 0) {
Jack Palevich22305132009-05-13 10:58:45 -07002628 pGen = new X86CodeGenerator();
Jack Palevich8b0624c2009-05-20 12:12:06 -07002629 }
Jack Paleviche7b59062009-05-19 17:12:17 -07002630#endif
Jack Palevich8b0624c2009-05-20 12:12:06 -07002631 if (!pGen ) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002632 error("Unknown architecture %s\n", architecture);
Jack Palevich22305132009-05-13 10:58:45 -07002633 }
2634 }
2635
2636 if (pGen == NULL) {
Jack Paleviche7b59062009-05-19 17:12:17 -07002637#if defined(DEFAULT_ARM_CODEGEN)
Jack Palevich22305132009-05-13 10:58:45 -07002638 pGen = new ARMCodeGenerator();
Jack Paleviche7b59062009-05-19 17:12:17 -07002639#elif defined(DEFAULT_X86_CODEGEN)
2640 pGen = new X86CodeGenerator();
2641#endif
2642 }
2643 if (pGen == NULL) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002644 error("No code generator defined.");
Jack Palevich0a280a02009-06-11 10:53:51 -07002645 } else {
2646 pGen->setErrorSink(this);
Jack Palevich22305132009-05-13 10:58:45 -07002647 }
2648 }
2649
Jack Palevich77ae76e2009-05-10 19:59:24 -07002650public:
Jack Palevich22305132009-05-13 10:58:45 -07002651 struct args {
2652 args() {
2653 architecture = 0;
2654 }
2655 const char* architecture;
2656 };
2657
Jack Paleviche7b59062009-05-19 17:12:17 -07002658 Compiler() {
Jack Palevich21a15a22009-05-11 14:49:29 -07002659 clear();
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002660 }
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002661
Jack Paleviche7b59062009-05-19 17:12:17 -07002662 ~Compiler() {
Jack Palevich21a15a22009-05-11 14:49:29 -07002663 cleanup();
2664 }
2665
Jack Palevich1cdef202009-05-22 12:06:27 -07002666 int compile(const char* text, size_t textLength) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002667 int result;
Jack Palevich0a280a02009-06-11 10:53:51 -07002668
2669 cleanup();
2670 clear();
2671 codeBuf.init(ALLOC_SIZE);
2672 setArchitecture(NULL);
2673 if (!pGen) {
2674 return -1;
2675 }
Jack Palevichb67b18f2009-06-11 21:12:23 -07002676#ifdef PROVIDE_TRACE_CODEGEN
2677 pGen = new TraceCodeGenerator(pGen);
2678#endif
2679 pGen->setErrorSink(this);
Jack Palevich0a280a02009-06-11 10:53:51 -07002680 pGen->init(&codeBuf);
2681 file = new TextInputStream(text, textLength);
Jack Palevich0a280a02009-06-11 10:53:51 -07002682 pGlobalBase = (char*) calloc(1, ALLOC_SIZE);
2683 glo = pGlobalBase;
Jack Palevich0a280a02009-06-11 10:53:51 -07002684 inp();
2685 next();
2686 globalDeclarations();
Jack Palevicha6baa232009-06-12 11:25:59 -07002687 checkForUndefinedForwardReferences();
Jack Palevich0a280a02009-06-11 10:53:51 -07002688 result = pGen->finishCompile();
2689 if (result == 0) {
2690 if (mErrorBuf.len()) {
2691 result = -2;
Jack Palevichac0e95e2009-05-29 13:53:44 -07002692 }
Jack Palevich8b0624c2009-05-20 12:12:06 -07002693 }
Jack Palevichac0e95e2009-05-29 13:53:44 -07002694 return result;
Jack Palevich21a15a22009-05-11 14:49:29 -07002695 }
2696
Jack Palevicha6baa232009-06-12 11:25:59 -07002697 void checkForUndefinedForwardReferences() {
2698 mSymbolTable.forEachGlobal(static_ufrcFn, this);
2699 }
2700
2701 static bool static_ufrcFn(String* key, VariableInfo* value,
2702 void* context) {
2703 Compiler* pCompiler = (Compiler*) context;
2704 return pCompiler->undefinedForwardReferenceCheck(key, value);
2705 }
2706
2707 bool undefinedForwardReferenceCheck(String* key, VariableInfo* value) {
2708#if 0
2709 fprintf(stderr, "%s 0x%8x 0x%08x\n", key->getUnwrapped(),
2710 value->pAddress, value->pForward);
2711#endif
2712 if (!value->pAddress && value->pForward) {
2713 error("Undefined forward reference: %s", key->getUnwrapped());
2714 }
2715 return true;
2716 }
2717
Jack Palevich21a15a22009-05-11 14:49:29 -07002718 int dump(FILE* out) {
2719 fwrite(codeBuf.getBase(), 1, codeBuf.getSize(), out);
2720 return 0;
2721 }
Jack Palevich77ae76e2009-05-10 19:59:24 -07002722
Jack Palevicha6535612009-05-13 16:24:17 -07002723 int disassemble(FILE* out) {
2724 return pGen->disassemble(out);
2725 }
2726
Jack Palevich1cdef202009-05-22 12:06:27 -07002727 /* Look through the symbol table to find a symbol.
2728 * If found, return its value.
2729 */
2730 void* lookup(const char* name) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002731 String string(name, -1, false);
2732 VariableInfo* pVariableInfo = mSymbolTable.get(&string);
2733 if (pVariableInfo) {
2734 return pVariableInfo->pAddress;
Jack Palevich1cdef202009-05-22 12:06:27 -07002735 }
2736 return NULL;
2737 }
2738
Jack Palevicheedf9d22009-06-04 16:23:40 -07002739 void getPragmas(ACCsizei* actualStringCount,
2740 ACCsizei maxStringCount, ACCchar** strings) {
2741 int stringCount = mPragmaStringCount;
2742 if (actualStringCount) {
2743 *actualStringCount = stringCount;
2744 }
2745 if (stringCount > maxStringCount) {
2746 stringCount = maxStringCount;
2747 }
2748 if (strings) {
2749 char* pPragmas = mPragmas.getUnwrapped();
2750 while (stringCount-- > 0) {
2751 *strings++ = pPragmas;
2752 pPragmas += strlen(pPragmas) + 1;
2753 }
2754 }
2755 }
2756
Jack Palevichac0e95e2009-05-29 13:53:44 -07002757 char* getErrorMessage() {
Jack Palevicheedf9d22009-06-04 16:23:40 -07002758 return mErrorBuf.getUnwrapped();
Jack Palevichac0e95e2009-05-29 13:53:44 -07002759 }
2760
Jack Palevich77ae76e2009-05-10 19:59:24 -07002761};
2762
Jack Paleviche7b59062009-05-19 17:12:17 -07002763const char* Compiler::operatorChars =
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002764 "++--*@/@%@+@-@<<>><=>=<@>@==!=&&||&@^@|@~@!@";
2765
Jack Paleviche7b59062009-05-19 17:12:17 -07002766const char Compiler::operatorLevel[] =
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002767 {11, 11, 1, 1, 1, 2, 2, 3, 3, 4, 4, 4, 4,
2768 5, 5, /* ==, != */
2769 9, 10, /* &&, || */
2770 6, 7, 8, /* & ^ | */
2771 2, 2 /* ~ ! */
2772 };
2773
Jack Palevich8b0624c2009-05-20 12:12:06 -07002774#ifdef PROVIDE_ARM_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -07002775FILE* Compiler::ARMCodeGenerator::disasmOut;
Jack Palevich8b0624c2009-05-20 12:12:06 -07002776#endif
Jack Palevicha6535612009-05-13 16:24:17 -07002777
Jack Palevich8b0624c2009-05-20 12:12:06 -07002778#ifdef PROVIDE_X86_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -07002779const int Compiler::X86CodeGenerator::operatorHelper[] = {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002780 0x1, // ++
2781 0xff, // --
2782 0xc1af0f, // *
2783 0xf9f79991, // /
2784 0xf9f79991, // % (With manual assist to swap results)
2785 0xc801, // +
2786 0xd8f7c829, // -
2787 0xe0d391, // <<
2788 0xf8d391, // >>
2789 0xe, // <=
2790 0xd, // >=
2791 0xc, // <
2792 0xf, // >
2793 0x4, // ==
2794 0x5, // !=
2795 0x0, // &&
2796 0x1, // ||
2797 0xc821, // &
2798 0xc831, // ^
2799 0xc809, // |
2800 0xd0f7, // ~
2801 0x4 // !
2802};
Jack Palevich8b0624c2009-05-20 12:12:06 -07002803#endif
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002804
Jack Palevich1cdef202009-05-22 12:06:27 -07002805struct ACCscript {
2806 ACCscript() {
2807 text = 0;
2808 textLength = 0;
2809 accError = ACC_NO_ERROR;
2810 }
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002811
Jack Palevich1cdef202009-05-22 12:06:27 -07002812 ~ACCscript() {
2813 delete text;
2814 }
Jack Palevich546b2242009-05-13 15:10:04 -07002815
Jack Palevich1cdef202009-05-22 12:06:27 -07002816 void setError(ACCenum error) {
2817 if (accError == ACC_NO_ERROR && error != ACC_NO_ERROR) {
2818 accError = error;
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002819 }
2820 }
2821
Jack Palevich1cdef202009-05-22 12:06:27 -07002822 ACCenum getError() {
2823 ACCenum result = accError;
2824 accError = ACC_NO_ERROR;
Jack Palevich22305132009-05-13 10:58:45 -07002825 return result;
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002826 }
2827
Jack Palevich1cdef202009-05-22 12:06:27 -07002828 Compiler compiler;
2829 char* text;
2830 int textLength;
2831 ACCenum accError;
2832};
2833
2834
2835extern "C"
2836ACCscript* accCreateScript() {
2837 return new ACCscript();
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002838}
Jack Palevich1cdef202009-05-22 12:06:27 -07002839
2840extern "C"
2841ACCenum accGetError( ACCscript* script ) {
2842 return script->getError();
2843}
2844
2845extern "C"
2846void accDeleteScript(ACCscript* script) {
2847 delete script;
2848}
2849
2850extern "C"
2851void accScriptSource(ACCscript* script,
2852 ACCsizei count,
2853 const ACCchar ** string,
2854 const ACCint * length) {
2855 int totalLength = 0;
2856 for(int i = 0; i < count; i++) {
2857 int len = -1;
2858 const ACCchar* s = string[i];
2859 if (length) {
2860 len = length[i];
2861 }
2862 if (len < 0) {
2863 len = strlen(s);
2864 }
2865 totalLength += len;
2866 }
2867 delete script->text;
2868 char* text = new char[totalLength + 1];
2869 script->text = text;
2870 script->textLength = totalLength;
Jack Palevich09555c72009-05-27 12:25:55 -07002871 char* dest = text;
Jack Palevich1cdef202009-05-22 12:06:27 -07002872 for(int i = 0; i < count; i++) {
2873 int len = -1;
2874 const ACCchar* s = string[i];
2875 if (length) {
2876 len = length[i];
2877 }
2878 if (len < 0) {
2879 len = strlen(s);
2880 }
Jack Palevich09555c72009-05-27 12:25:55 -07002881 memcpy(dest, s, len);
2882 dest += len;
Jack Palevich1cdef202009-05-22 12:06:27 -07002883 }
2884 text[totalLength] = '\0';
2885}
2886
2887extern "C"
2888void accCompileScript(ACCscript* script) {
2889 int result = script->compiler.compile(script->text, script->textLength);
2890 if (result) {
2891 script->setError(ACC_INVALID_OPERATION);
2892 }
2893}
2894
2895extern "C"
2896void accGetScriptiv(ACCscript* script,
2897 ACCenum pname,
2898 ACCint * params) {
2899 switch (pname) {
2900 case ACC_INFO_LOG_LENGTH:
2901 *params = 0;
2902 break;
2903 }
2904}
2905
2906extern "C"
2907void accGetScriptInfoLog(ACCscript* script,
2908 ACCsizei maxLength,
2909 ACCsizei * length,
2910 ACCchar * infoLog) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002911 char* message = script->compiler.getErrorMessage();
2912 int messageLength = strlen(message) + 1;
Jack Palevich1cdef202009-05-22 12:06:27 -07002913 if (length) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002914 *length = messageLength;
Jack Palevich1cdef202009-05-22 12:06:27 -07002915 }
Jack Palevichac0e95e2009-05-29 13:53:44 -07002916 if (infoLog && maxLength > 0) {
2917 int trimmedLength = maxLength < messageLength ?
2918 maxLength : messageLength;
2919 memcpy(infoLog, message, trimmedLength);
2920 infoLog[trimmedLength] = 0;
Jack Palevich1cdef202009-05-22 12:06:27 -07002921 }
2922}
2923
2924extern "C"
2925void accGetScriptLabel(ACCscript* script, const ACCchar * name,
2926 ACCvoid ** address) {
2927 void* value = script->compiler.lookup(name);
2928 if (value) {
2929 *address = value;
2930 } else {
2931 script->setError(ACC_INVALID_VALUE);
2932 }
2933}
2934
Jack Palevicheedf9d22009-06-04 16:23:40 -07002935extern "C"
2936void accGetPragmas(ACCscript* script, ACCsizei* actualStringCount,
2937 ACCsizei maxStringCount, ACCchar** strings){
2938 script->compiler.getPragmas(actualStringCount, maxStringCount, strings);
2939}
2940
2941
Jack Palevich1cdef202009-05-22 12:06:27 -07002942} // namespace acc
2943