blob: 7903c6ee17e66410120fadc268f182fe9f57df70 [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
1844 /* read a character constant */
1845 void getq() {
1846 if (ch == '\\') {
1847 inp();
1848 if (ch == 'n')
1849 ch = '\n';
1850 }
1851 }
1852
1853 void next() {
1854 int l, a;
1855
Jack Palevich546b2242009-05-13 15:10:04 -07001856 while (isspace(ch) | (ch == '#')) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001857 if (ch == '#') {
1858 inp();
1859 next();
1860 if (tok == TOK_DEFINE) {
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001861 doDefine();
Jack Palevicheedf9d22009-06-04 16:23:40 -07001862 } else if (tok == TOK_PRAGMA) {
1863 doPragma();
1864 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001865 error("Unsupported preprocessor directive \"%s\"",
1866 mTokenString.getUnwrapped());
Jack Palevich21a15a22009-05-11 14:49:29 -07001867 }
Jack Palevich21a15a22009-05-11 14:49:29 -07001868 }
1869 inp();
1870 }
1871 tokl = 0;
1872 tok = ch;
1873 /* encode identifiers & numbers */
1874 if (isid()) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001875 mTokenString.clear();
Jack Palevich21a15a22009-05-11 14:49:29 -07001876 while (isid()) {
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001877 pdef(ch);
1878 inp();
Jack Palevichae54f1f2009-05-08 14:54:15 -07001879 }
Jack Palevich21a15a22009-05-11 14:49:29 -07001880 if (isdigit(tok)) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001881 tokc = strtol(mTokenString.getUnwrapped(), 0, 0);
Jack Palevich21a15a22009-05-11 14:49:29 -07001882 tok = TOK_NUM;
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001883 } else {
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001884 // Is this a macro?
Jack Palevich303d8ff2009-06-11 19:06:24 -07001885 String* pValue = mMacros.get(&mTokenString);
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001886 if (pValue) {
1887 // Yes, it is a macro
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001888 dptr = pValue->getUnwrapped();
1889 dch = ch;
1890 inp();
1891 next();
1892 } else {
Jack Palevich2db168f2009-06-11 14:29:47 -07001893 // Is this a keyword?
Jack Palevich303d8ff2009-06-11 19:06:24 -07001894 int kwtok = mKeywords.get(&mTokenString);
Jack Palevich2db168f2009-06-11 14:29:47 -07001895 if (kwtok) {
1896 tok = kwtok;
1897 // fprintf(stderr, "tok= keyword %s %x\n", last_id, tok);
1898 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07001899 tok = (intptr_t) mSymbolTable.get(&mTokenString);
1900 if (!tok) {
1901 tok = TOK_UNDEFINED_SYMBOL;
1902 }
Jack Palevich2db168f2009-06-11 14:29:47 -07001903 // fprintf(stderr, "tok= symbol %s %x\n", last_id, tok);
Jack Palevich21a15a22009-05-11 14:49:29 -07001904 }
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001905 }
1906 }
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001907 } else {
Jack Palevich21a15a22009-05-11 14:49:29 -07001908 inp();
1909 if (tok == '\'') {
1910 tok = TOK_NUM;
1911 getq();
1912 tokc = ch;
1913 inp();
1914 inp();
Jack Palevich546b2242009-05-13 15:10:04 -07001915 } else if ((tok == '/') & (ch == '*')) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001916 inp();
1917 while (ch) {
1918 while (ch != '*')
1919 inp();
1920 inp();
1921 if (ch == '/')
1922 ch = 0;
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001923 }
Jack Palevich21a15a22009-05-11 14:49:29 -07001924 inp();
Jack Paleviche27bf3e2009-05-10 14:09:03 -07001925 next();
Jack Palevichbd894902009-05-14 19:35:31 -07001926 } else if ((tok == '/') & (ch == '/')) {
1927 inp();
1928 while (ch && (ch != '\n')) {
1929 inp();
1930 }
1931 inp();
1932 next();
Jack Palevich21a15a22009-05-11 14:49:29 -07001933 } else {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07001934 const char* t = operatorChars;
1935 int opIndex = 0;
Jack Palevich546b2242009-05-13 15:10:04 -07001936 while ((l = *t++) != 0) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001937 a = *t++;
Jack Palevichbf42c9c2009-05-12 12:48:35 -07001938 tokl = operatorLevel[opIndex];
1939 tokc = opIndex;
Jack Palevich546b2242009-05-13 15:10:04 -07001940 if ((l == tok) & ((a == ch) | (a == '@'))) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001941#if 0
1942 printf("%c%c -> tokl=%d tokc=0x%x\n",
1943 l, a, tokl, tokc);
1944#endif
1945 if (a == ch) {
1946 inp();
1947 tok = TOK_DUMMY; /* dummy token for double tokens */
1948 }
1949 break;
1950 }
Jack Palevichbf42c9c2009-05-12 12:48:35 -07001951 opIndex++;
1952 }
1953 if (l == 0) {
1954 tokl = 0;
1955 tokc = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07001956 }
1957 }
1958 }
1959#if 0
1960 {
Jack Palevich2db168f2009-06-11 14:29:47 -07001961 const char* p;
Jack Palevich21a15a22009-05-11 14:49:29 -07001962
1963 printf("tok=0x%x ", tok);
Jack Palevich2db168f2009-06-11 14:29:47 -07001964 if (tok >= TOK_KEYWORD) {
Jack Palevich21a15a22009-05-11 14:49:29 -07001965 printf("'");
Jack Palevich2db168f2009-06-11 14:29:47 -07001966 if (tok>= TOK_SYMBOL)
1967 p = sym_stk + 1 + ((char*) tok - (char*) pVarsBase) / 8;
1968 else {
1969 p = mKeywords.lookupKeyFor(tok);
1970 if (!p) {
1971 p = "unknown keyword";
1972 }
1973 }
Jack Palevich653f42d2009-05-28 17:15:32 -07001974 while (*p != TAG_TOK && *p)
1975 printf("%c", *p++);
Jack Palevich21a15a22009-05-11 14:49:29 -07001976 printf("'\n");
1977 } else if (tok == TOK_NUM) {
1978 printf("%d\n", tokc);
1979 } else {
1980 printf("'%c'\n", tok);
1981 }
1982 }
1983#endif
1984 }
1985
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001986 void doDefine() {
1987 String* pName = new String();
1988 while (isspace(ch)) {
1989 inp();
1990 }
1991 while (isid()) {
1992 pName->append(ch);
1993 inp();
1994 }
1995 if (ch == '(') {
1996 delete pName;
1997 error("Defines with arguments not supported");
Jack Palevich0a280a02009-06-11 10:53:51 -07001998 return;
Jack Palevich2d11dfb2009-06-08 14:34:26 -07001999 }
2000 while (isspace(ch)) {
2001 inp();
2002 }
2003 String* pValue = new String();
2004 while (ch != '\n' && ch != EOF) {
2005 pValue->append(ch);
2006 inp();
2007 }
2008 delete mMacros.put(pName, pValue);
2009 }
2010
Jack Palevicheedf9d22009-06-04 16:23:40 -07002011 void doPragma() {
2012 // # pragma name(val)
2013 int state = 0;
2014 while(ch != EOF && ch != '\n' && state < 10) {
2015 switch(state) {
2016 case 0:
2017 if (isspace(ch)) {
2018 inp();
2019 } else {
2020 state++;
2021 }
2022 break;
2023 case 1:
2024 if (isalnum(ch)) {
2025 mPragmas.append(ch);
2026 inp();
2027 } else if (ch == '(') {
2028 mPragmas.append(0);
2029 inp();
2030 state++;
2031 } else {
2032 state = 11;
2033 }
2034 break;
2035 case 2:
2036 if (isalnum(ch)) {
2037 mPragmas.append(ch);
2038 inp();
2039 } else if (ch == ')') {
2040 mPragmas.append(0);
2041 inp();
2042 state = 10;
2043 } else {
2044 state = 11;
2045 }
2046 break;
2047 }
2048 }
2049 if(state != 10) {
2050 error("Unexpected pragma syntax");
2051 }
2052 mPragmaStringCount += 2;
2053 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002054
Jack Palevichac0e95e2009-05-29 13:53:44 -07002055 virtual void verror(const char* fmt, va_list ap) {
Jack Palevicheedf9d22009-06-04 16:23:40 -07002056 mErrorBuf.printf("%ld: ", file->getLine());
2057 mErrorBuf.vprintf(fmt, ap);
2058 mErrorBuf.printf("\n");
Jack Palevich21a15a22009-05-11 14:49:29 -07002059 }
2060
Jack Palevich8b0624c2009-05-20 12:12:06 -07002061 void skip(intptr_t c) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002062 if (tok != c) {
2063 error("'%c' expected", c);
2064 }
2065 next();
2066 }
2067
Jack Palevich21a15a22009-05-11 14:49:29 -07002068 /* l is one if '=' parsing wanted (quick hack) */
Jack Palevich8b0624c2009-05-20 12:12:06 -07002069 void unary(intptr_t l) {
Jack Palevich653f42d2009-05-28 17:15:32 -07002070 intptr_t n, t, a;
2071 int c;
Jack Palevicha6baa232009-06-12 11:25:59 -07002072 String tString;
Jack Palevich546b2242009-05-13 15:10:04 -07002073 t = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002074 n = 1; /* type of expression 0 = forward, 1 = value, other =
2075 lvalue */
2076 if (tok == '\"') {
Jack Palevich653f42d2009-05-28 17:15:32 -07002077 pGen->li((int) glo);
Jack Palevich21a15a22009-05-11 14:49:29 -07002078 while (ch != '\"') {
2079 getq();
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002080 *allocGlobalSpace(1) = ch;
Jack Palevich21a15a22009-05-11 14:49:29 -07002081 inp();
2082 }
Jack Palevich653f42d2009-05-28 17:15:32 -07002083 *glo = 0;
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002084 /* align heap */
2085 allocGlobalSpace((char*) (((intptr_t) glo + 4) & -4) - glo);
Jack Palevich21a15a22009-05-11 14:49:29 -07002086 inp();
2087 next();
2088 } else {
2089 c = tokl;
2090 a = tokc;
2091 t = tok;
Jack Palevicha6baa232009-06-12 11:25:59 -07002092 tString = mTokenString;
Jack Palevich21a15a22009-05-11 14:49:29 -07002093 next();
2094 if (t == TOK_NUM) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002095 pGen->li(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002096 } else if (c == 2) {
2097 /* -, +, !, ~ */
2098 unary(0);
Jack Palevich1cdef202009-05-22 12:06:27 -07002099 pGen->clearR1();
Jack Palevich21a15a22009-05-11 14:49:29 -07002100 if (t == '!')
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002101 pGen->gcmp(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002102 else
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002103 pGen->genOp(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002104 } else if (t == '(') {
2105 expr();
2106 skip(')');
2107 } else if (t == '*') {
2108 /* parse cast */
2109 skip('(');
2110 t = tok; /* get type */
2111 next(); /* skip int/char/void */
2112 next(); /* skip '*' or '(' */
2113 if (tok == '*') {
2114 /* function type */
2115 skip('*');
2116 skip(')');
2117 skip('(');
2118 skip(')');
2119 t = 0;
2120 }
2121 skip(')');
2122 unary(0);
2123 if (tok == '=') {
2124 next();
Jack Palevich1cdef202009-05-22 12:06:27 -07002125 pGen->pushR0();
Jack Palevich21a15a22009-05-11 14:49:29 -07002126 expr();
Jack Palevich1cdef202009-05-22 12:06:27 -07002127 pGen->popR1();
2128 pGen->storeR0ToR1(t == TOK_INT);
Jack Palevich21a15a22009-05-11 14:49:29 -07002129 } else if (t) {
Jack Palevich1cdef202009-05-22 12:06:27 -07002130 pGen->loadR0FromR0(t == TOK_INT);
Jack Palevich21a15a22009-05-11 14:49:29 -07002131 }
2132 } else if (t == '&') {
Jack Palevich1cdef202009-05-22 12:06:27 -07002133 pGen->leaR0(*(int *) tok);
Jack Palevich21a15a22009-05-11 14:49:29 -07002134 next();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002135 } else if (t == EOF ) {
2136 error("Unexpected EOF.");
2137 } else if (t < TOK_UNDEFINED_SYMBOL) {
2138 error("Unexpected symbol or keyword");
Jack Palevich21a15a22009-05-11 14:49:29 -07002139 } else {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002140 if (t == TOK_UNDEFINED_SYMBOL) {
2141 t = (intptr_t) mSymbolTable.addGlobal(
Jack Palevicha6baa232009-06-12 11:25:59 -07002142 new String(tString));
Jack Palevich303d8ff2009-06-11 19:06:24 -07002143 }
2144
Jack Palevicha6baa232009-06-12 11:25:59 -07002145 n = (intptr_t) ((VariableInfo*) t)->pAddress;
Jack Palevich21a15a22009-05-11 14:49:29 -07002146 /* forward reference: try dlsym */
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002147 if (!n) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002148 n = (intptr_t) dlsym(RTLD_DEFAULT,
Jack Palevicha6baa232009-06-12 11:25:59 -07002149 tString.getUnwrapped());
2150 ((VariableInfo*) t)->pAddress = (void*) n;
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002151 }
Jack Palevich546b2242009-05-13 15:10:04 -07002152 if ((tok == '=') & l) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002153 /* assignment */
2154 next();
2155 expr();
Jack Palevich1cdef202009-05-22 12:06:27 -07002156 pGen->storeR0(n);
Jack Palevich21a15a22009-05-11 14:49:29 -07002157 } else if (tok != '(') {
2158 /* variable */
Jack Palevicha6baa232009-06-12 11:25:59 -07002159 if (!n) {
2160 error("Undefined variable %s", tString.getUnwrapped());
2161 }
Jack Palevich1cdef202009-05-22 12:06:27 -07002162 pGen->loadR0(n, tokl == 11, tokc);
Jack Palevich21a15a22009-05-11 14:49:29 -07002163 if (tokl == 11) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002164 next();
2165 }
2166 }
2167 }
2168 }
2169
2170 /* function call */
2171 if (tok == '(') {
2172 if (n == 1)
Jack Palevich1cdef202009-05-22 12:06:27 -07002173 pGen->pushR0();
Jack Palevich21a15a22009-05-11 14:49:29 -07002174
2175 /* push args and invert order */
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002176 a = pGen->beginFunctionCallArguments();
Jack Palevich21a15a22009-05-11 14:49:29 -07002177 next();
2178 l = 0;
2179 while (tok != ')') {
2180 expr();
Jack Palevich1cdef202009-05-22 12:06:27 -07002181 pGen->storeR0ToArg(l);
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002182 if (tok == ',')
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002183 next();
Jack Palevich21a15a22009-05-11 14:49:29 -07002184 l = l + 4;
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002185 }
Jack Palevichcb1c9ef2009-05-14 11:38:49 -07002186 pGen->endFunctionCallArguments(a, l);
Jack Palevich21a15a22009-05-11 14:49:29 -07002187 next();
2188 if (!n) {
2189 /* forward reference */
2190 t = t + 4;
2191 *(int *) t = pGen->callForward(*(int *) t);
2192 } else if (n == 1) {
2193 pGen->callIndirect(l);
Jack Palevich21a15a22009-05-11 14:49:29 -07002194 } else {
Jack Palevich7810bc92009-05-15 14:31:47 -07002195 pGen->callRelative(n - codeBuf.getPC() - pGen->jumpOffset());
Jack Palevich21a15a22009-05-11 14:49:29 -07002196 }
Jack Palevich3d474a72009-05-15 15:12:38 -07002197 if (l | (n == 1))
Jack Palevich7810bc92009-05-15 14:31:47 -07002198 pGen->adjustStackAfterCall(l, n == 1);
Jack Palevich21a15a22009-05-11 14:49:29 -07002199 }
2200 }
2201
Jack Palevich653f42d2009-05-28 17:15:32 -07002202 void sum(int l) {
Jack Palevich8b0624c2009-05-20 12:12:06 -07002203 intptr_t t, n, a;
Jack Palevich546b2242009-05-13 15:10:04 -07002204 t = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002205 if (l-- == 1)
2206 unary(1);
2207 else {
2208 sum(l);
2209 a = 0;
2210 while (l == tokl) {
2211 n = tok;
2212 t = tokc;
2213 next();
2214
2215 if (l > 8) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002216 a = pGen->gtst(t == OP_LOGICAL_OR, a); /* && and || output code generation */
Jack Palevich21a15a22009-05-11 14:49:29 -07002217 sum(l);
2218 } else {
Jack Palevich1cdef202009-05-22 12:06:27 -07002219 pGen->pushR0();
Jack Palevich21a15a22009-05-11 14:49:29 -07002220 sum(l);
Jack Palevich1cdef202009-05-22 12:06:27 -07002221 pGen->popR1();
Jack Palevich21a15a22009-05-11 14:49:29 -07002222
Jack Palevich546b2242009-05-13 15:10:04 -07002223 if ((l == 4) | (l == 5)) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002224 pGen->gcmp(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07002225 } else {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002226 pGen->genOp(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07002227 }
2228 }
2229 }
2230 /* && and || output code generation */
2231 if (a && l > 8) {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002232 a = pGen->gtst(t == OP_LOGICAL_OR, a);
2233 pGen->li(t != OP_LOGICAL_OR);
Jack Palevicha6535612009-05-13 16:24:17 -07002234 pGen->gjmp(5); /* jmp $ + 5 (sizeof li, FIXME for ARM) */
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002235 pGen->gsym(a);
2236 pGen->li(t == OP_LOGICAL_OR);
Jack Palevich21a15a22009-05-11 14:49:29 -07002237 }
2238 }
2239 }
2240
2241 void expr() {
2242 sum(11);
2243 }
2244
2245 int test_expr() {
2246 expr();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002247 return pGen->gtst(0, 0);
Jack Palevich21a15a22009-05-11 14:49:29 -07002248 }
2249
Jack Palevicha6baa232009-06-12 11:25:59 -07002250 void block(intptr_t l, bool outermostFunctionBlock) {
Jack Palevich8b0624c2009-05-20 12:12:06 -07002251 intptr_t a, n, t;
Jack Palevich21a15a22009-05-11 14:49:29 -07002252
2253 if (tok == TOK_IF) {
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002254 next();
2255 skip('(');
Jack Palevich21a15a22009-05-11 14:49:29 -07002256 a = test_expr();
2257 skip(')');
Jack Palevicha6baa232009-06-12 11:25:59 -07002258 block(l, false);
Jack Palevich21a15a22009-05-11 14:49:29 -07002259 if (tok == TOK_ELSE) {
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002260 next();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002261 n = pGen->gjmp(0); /* jmp */
2262 pGen->gsym(a);
Jack Palevicha6baa232009-06-12 11:25:59 -07002263 block(l, false);
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002264 pGen->gsym(n); /* patch else jmp */
Jack Palevich21a15a22009-05-11 14:49:29 -07002265 } else {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002266 pGen->gsym(a); /* patch if test */
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002267 }
Jack Palevich546b2242009-05-13 15:10:04 -07002268 } else if ((tok == TOK_WHILE) | (tok == TOK_FOR)) {
Jack Palevich21a15a22009-05-11 14:49:29 -07002269 t = tok;
2270 next();
2271 skip('(');
2272 if (t == TOK_WHILE) {
Jack Palevicha6535612009-05-13 16:24:17 -07002273 n = codeBuf.getPC(); // top of loop, target of "next" iteration
Jack Palevich21a15a22009-05-11 14:49:29 -07002274 a = test_expr();
2275 } else {
2276 if (tok != ';')
2277 expr();
2278 skip(';');
2279 n = codeBuf.getPC();
2280 a = 0;
2281 if (tok != ';')
2282 a = test_expr();
2283 skip(';');
2284 if (tok != ')') {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002285 t = pGen->gjmp(0);
Jack Palevich21a15a22009-05-11 14:49:29 -07002286 expr();
Jack Palevicha6535612009-05-13 16:24:17 -07002287 pGen->gjmp(n - codeBuf.getPC() - pGen->jumpOffset());
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002288 pGen->gsym(t);
Jack Palevich21a15a22009-05-11 14:49:29 -07002289 n = t + 4;
2290 }
2291 }
2292 skip(')');
Jack Palevicha6baa232009-06-12 11:25:59 -07002293 block((intptr_t) &a, false);
Jack Palevicha6535612009-05-13 16:24:17 -07002294 pGen->gjmp(n - codeBuf.getPC() - pGen->jumpOffset()); /* jmp */
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002295 pGen->gsym(a);
Jack Palevich21a15a22009-05-11 14:49:29 -07002296 } else if (tok == '{') {
Jack Palevicha6baa232009-06-12 11:25:59 -07002297 if (! outermostFunctionBlock) {
2298 mSymbolTable.pushLevel();
2299 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002300 next();
2301 /* declarations */
Jack Palevichb7c81e92009-06-04 19:56:13 -07002302 localDeclarations();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002303 while (tok != '}' && tok != EOF)
Jack Palevicha6baa232009-06-12 11:25:59 -07002304 block(l, false);
Jack Palevich303d8ff2009-06-11 19:06:24 -07002305 skip('}');
Jack Palevicha6baa232009-06-12 11:25:59 -07002306 if (! outermostFunctionBlock) {
2307 mSymbolTable.popLevel();
2308 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002309 } else {
2310 if (tok == TOK_RETURN) {
2311 next();
2312 if (tok != ';')
2313 expr();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002314 rsym = pGen->gjmp(rsym); /* jmp */
Jack Palevich21a15a22009-05-11 14:49:29 -07002315 } else if (tok == TOK_BREAK) {
2316 next();
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002317 *(int *) l = pGen->gjmp(*(int *) l);
Jack Palevich21a15a22009-05-11 14:49:29 -07002318 } else if (tok != ';')
2319 expr();
2320 skip(';');
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002321 }
2322 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002323
Jack Palevichb7c81e92009-06-04 19:56:13 -07002324 typedef int Type;
2325 static const Type TY_UNKNOWN = 0;
2326 static const Type TY_INT = 1;
2327 static const Type TY_CHAR = 2;
2328 static const Type TY_VOID = 3;
2329 static const int TY_BASE_TYPE_MASK = 0xf;
2330 static const int TY_INDIRECTION_MASK = 0xf0;
2331 static const int TY_INDIRECTION_SHIFT = 4;
2332 static const int MAX_INDIRECTION_COUNT = 15;
Jack Palevich21a15a22009-05-11 14:49:29 -07002333
Jack Palevichb7c81e92009-06-04 19:56:13 -07002334 Type getBaseType(Type t) {
2335 return t & TY_BASE_TYPE_MASK;
2336 }
2337
2338 int getIndirectionCount(Type t) {
2339 return (TY_INDIRECTION_MASK & t) >> TY_INDIRECTION_SHIFT;
2340 }
2341
2342 void setIndirectionCount(Type& t, int count) {
2343 t = ((TY_INDIRECTION_MASK & (count << TY_INDIRECTION_SHIFT))
2344 | (t & ~TY_INDIRECTION_MASK));
2345 }
2346
2347 bool acceptType(Type& t) {
2348 t = TY_UNKNOWN;
2349 if (tok == TOK_INT) {
2350 t = TY_INT;
2351 } else if (tok == TOK_CHAR) {
2352 t = TY_CHAR;
2353 } else if (tok == TOK_VOID) {
2354 t = TY_VOID;
2355 } else {
2356 return false;
2357 }
2358 next();
2359 return true;
2360 }
2361
2362 Type acceptPointerDeclaration(Type& base) {
2363 Type t = base;
2364 int indirectionCount = 0;
2365 while (tok == '*' && indirectionCount <= MAX_INDIRECTION_COUNT) {
2366 next();
2367 indirectionCount++;
2368 }
2369 if (indirectionCount > MAX_INDIRECTION_COUNT) {
2370 error("Too many levels of pointer. Max %d", MAX_INDIRECTION_COUNT);
2371 }
2372 setIndirectionCount(t, indirectionCount);
2373 return t;
2374 }
2375
2376 void expectType(Type& t) {
2377 if (!acceptType(t)) {
2378 error("Expected a type.");
2379 }
2380 }
2381
2382 void checkSymbol() {
Jack Palevich2db168f2009-06-11 14:29:47 -07002383 if (tok < TOK_SYMBOL) {
Jack Palevichb7c81e92009-06-04 19:56:13 -07002384 error("Expected a symbol");
2385 }
2386 }
2387
Jack Palevicha6baa232009-06-12 11:25:59 -07002388 void addGlobalSymbol() {
2389 tok = (intptr_t) mSymbolTable.addGlobal(
2390 new String(mTokenString));
2391 reportIfDuplicate();
2392 }
2393
2394 void reportIfDuplicate() {
2395 if (!tok) {
2396 error("Duplicate definition of %s", mTokenString.getUnwrapped());
Jack Palevich303d8ff2009-06-11 19:06:24 -07002397 }
2398 }
2399
Jack Palevicha6baa232009-06-12 11:25:59 -07002400 void addLocalSymbol() {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002401 tok = (intptr_t) mSymbolTable.addLocal(
2402 new String(mTokenString));
Jack Palevicha6baa232009-06-12 11:25:59 -07002403 reportIfDuplicate();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002404 }
2405
Jack Palevichb7c81e92009-06-04 19:56:13 -07002406 void localDeclarations() {
2407 intptr_t a;
2408 Type base;
2409
2410 while (acceptType(base)) {
2411 while (tok != ';') {
2412 Type t = acceptPointerDeclaration(t);
Jack Palevicha6baa232009-06-12 11:25:59 -07002413 addLocalSymbol();
2414 if (tok) {
2415 loc = loc + 4;
2416 *(int *) tok = -loc;
2417 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002418 next();
Jack Palevichb7c81e92009-06-04 19:56:13 -07002419 if (tok == ',')
2420 next();
2421 }
2422 skip(';');
2423 }
2424 }
2425
2426 void globalDeclarations() {
2427 while (tok != EOF) {
2428 Type base;
2429 expectType(base);
2430 Type t = acceptPointerDeclaration(t);
Jack Palevicha6baa232009-06-12 11:25:59 -07002431 if (tok == TOK_UNDEFINED_SYMBOL) {
2432 addGlobalSymbol();
2433 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07002434 VariableInfo* name = (VariableInfo*) tok;
Jack Palevicha6baa232009-06-12 11:25:59 -07002435 if (name && name->pAddress) {
2436 error("Already defined global %s",
2437 mTokenString.getUnwrapped());
2438 }
Jack Palevichb7c81e92009-06-04 19:56:13 -07002439 next();
2440 if (tok == ',' || tok == ';') {
2441 // it's a variable declaration
2442 for(;;) {
Jack Palevicha6baa232009-06-12 11:25:59 -07002443 if (name) {
2444 name->pAddress = (int*) allocGlobalSpace(4);
2445 }
Jack Palevichb7c81e92009-06-04 19:56:13 -07002446 if (tok != ',') {
2447 break;
Jack Palevich21a15a22009-05-11 14:49:29 -07002448 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07002449 skip(',');
Jack Palevichb7c81e92009-06-04 19:56:13 -07002450 t = acceptPointerDeclaration(t);
Jack Palevicha6baa232009-06-12 11:25:59 -07002451 addGlobalSymbol();
Jack Palevich303d8ff2009-06-11 19:06:24 -07002452 name = (VariableInfo*) tok;
Jack Palevichb7c81e92009-06-04 19:56:13 -07002453 next();
Jack Palevich21a15a22009-05-11 14:49:29 -07002454 }
2455 skip(';');
2456 } else {
Jack Palevicha6baa232009-06-12 11:25:59 -07002457 if (name) {
2458 /* patch forward references (XXX: does not work for function
2459 pointers) */
2460 pGen->gsym((int) name->pForward);
2461 /* put function address */
2462 name->pAddress = (void*) codeBuf.getPC();
2463 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002464 skip('(');
Jack Palevich303d8ff2009-06-11 19:06:24 -07002465 mSymbolTable.pushLevel();
Jack Palevichb7c81e92009-06-04 19:56:13 -07002466 intptr_t a = 8;
Jack Palevich546b2242009-05-13 15:10:04 -07002467 int argCount = 0;
Jack Palevich303d8ff2009-06-11 19:06:24 -07002468 while (tok != ')' && tok != EOF) {
Jack Palevichb7c81e92009-06-04 19:56:13 -07002469 Type aType;
2470 expectType(aType);
2471 aType = acceptPointerDeclaration(aType);
Jack Palevicha6baa232009-06-12 11:25:59 -07002472 addLocalSymbol();
2473 if (tok) {
2474 /* read param name and compute offset */
2475 *(int *) tok = a;
2476 a = a + 4;
2477 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002478 next();
2479 if (tok == ',')
2480 next();
Jack Palevich546b2242009-05-13 15:10:04 -07002481 argCount++;
Jack Palevich21a15a22009-05-11 14:49:29 -07002482 }
Jack Palevich303d8ff2009-06-11 19:06:24 -07002483 skip(')');
Jack Palevich21a15a22009-05-11 14:49:29 -07002484 rsym = loc = 0;
Jack Palevich546b2242009-05-13 15:10:04 -07002485 a = pGen->functionEntry(argCount);
Jack Palevicha6baa232009-06-12 11:25:59 -07002486 block(0, true);
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002487 pGen->gsym(rsym);
Jack Palevich546b2242009-05-13 15:10:04 -07002488 pGen->functionExit(argCount, a, loc);
Jack Palevich303d8ff2009-06-11 19:06:24 -07002489 mSymbolTable.popLevel();
Jack Palevich21a15a22009-05-11 14:49:29 -07002490 }
2491 }
2492 }
2493
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002494 char* allocGlobalSpace(int bytes) {
2495 if (glo - pGlobalBase + bytes > ALLOC_SIZE) {
2496 error("Global space exhausted");
Jack Palevich0a280a02009-06-11 10:53:51 -07002497 return NULL;
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002498 }
2499 char* result = glo;
2500 glo += bytes;
2501 return result;
2502 }
2503
Jack Palevich21a15a22009-05-11 14:49:29 -07002504 void cleanup() {
Jack Palevich21a15a22009-05-11 14:49:29 -07002505 if (pGlobalBase != 0) {
Jack Palevichf1f39cc2009-05-29 18:03:15 -07002506 free(pGlobalBase);
Jack Palevich21a15a22009-05-11 14:49:29 -07002507 pGlobalBase = 0;
2508 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002509 if (pGen) {
2510 delete pGen;
2511 pGen = 0;
2512 }
Jack Palevich1cdef202009-05-22 12:06:27 -07002513 if (file) {
2514 delete file;
2515 file = 0;
2516 }
Jack Palevich21a15a22009-05-11 14:49:29 -07002517 }
2518
2519 void clear() {
2520 tok = 0;
2521 tokc = 0;
2522 tokl = 0;
2523 ch = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002524 rsym = 0;
2525 loc = 0;
2526 glo = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002527 dptr = 0;
2528 dch = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002529 file = 0;
2530 pGlobalBase = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002531 pGen = 0;
Jack Palevicheedf9d22009-06-04 16:23:40 -07002532 mPragmaStringCount = 0;
Jack Palevich21a15a22009-05-11 14:49:29 -07002533 }
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002534
Jack Palevich22305132009-05-13 10:58:45 -07002535 void setArchitecture(const char* architecture) {
2536 delete pGen;
2537 pGen = 0;
2538
2539 if (architecture != NULL) {
Jack Paleviche7b59062009-05-19 17:12:17 -07002540#ifdef PROVIDE_ARM_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -07002541 if (! pGen && strcmp(architecture, "arm") == 0) {
Jack Palevich22305132009-05-13 10:58:45 -07002542 pGen = new ARMCodeGenerator();
Jack Palevich8b0624c2009-05-20 12:12:06 -07002543 }
Jack Paleviche7b59062009-05-19 17:12:17 -07002544#endif
Jack Paleviche7b59062009-05-19 17:12:17 -07002545#ifdef PROVIDE_X86_CODEGEN
Jack Palevich8b0624c2009-05-20 12:12:06 -07002546 if (! pGen && strcmp(architecture, "x86") == 0) {
Jack Palevich22305132009-05-13 10:58:45 -07002547 pGen = new X86CodeGenerator();
Jack Palevich8b0624c2009-05-20 12:12:06 -07002548 }
Jack Paleviche7b59062009-05-19 17:12:17 -07002549#endif
Jack Palevich8b0624c2009-05-20 12:12:06 -07002550 if (!pGen ) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002551 error("Unknown architecture %s\n", architecture);
Jack Palevich22305132009-05-13 10:58:45 -07002552 }
2553 }
2554
2555 if (pGen == NULL) {
Jack Paleviche7b59062009-05-19 17:12:17 -07002556#if defined(DEFAULT_ARM_CODEGEN)
Jack Palevich22305132009-05-13 10:58:45 -07002557 pGen = new ARMCodeGenerator();
Jack Paleviche7b59062009-05-19 17:12:17 -07002558#elif defined(DEFAULT_X86_CODEGEN)
2559 pGen = new X86CodeGenerator();
2560#endif
2561 }
2562 if (pGen == NULL) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002563 error("No code generator defined.");
Jack Palevich0a280a02009-06-11 10:53:51 -07002564 } else {
2565 pGen->setErrorSink(this);
Jack Palevich22305132009-05-13 10:58:45 -07002566 }
2567 }
2568
Jack Palevich77ae76e2009-05-10 19:59:24 -07002569public:
Jack Palevich22305132009-05-13 10:58:45 -07002570 struct args {
2571 args() {
2572 architecture = 0;
2573 }
2574 const char* architecture;
2575 };
2576
Jack Paleviche7b59062009-05-19 17:12:17 -07002577 Compiler() {
Jack Palevich21a15a22009-05-11 14:49:29 -07002578 clear();
Jack Paleviche27bf3e2009-05-10 14:09:03 -07002579 }
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002580
Jack Paleviche7b59062009-05-19 17:12:17 -07002581 ~Compiler() {
Jack Palevich21a15a22009-05-11 14:49:29 -07002582 cleanup();
2583 }
2584
Jack Palevich1cdef202009-05-22 12:06:27 -07002585 int compile(const char* text, size_t textLength) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002586 int result;
Jack Palevich0a280a02009-06-11 10:53:51 -07002587
2588 cleanup();
2589 clear();
2590 codeBuf.init(ALLOC_SIZE);
2591 setArchitecture(NULL);
2592 if (!pGen) {
2593 return -1;
2594 }
Jack Palevichb67b18f2009-06-11 21:12:23 -07002595#ifdef PROVIDE_TRACE_CODEGEN
2596 pGen = new TraceCodeGenerator(pGen);
2597#endif
2598 pGen->setErrorSink(this);
Jack Palevich0a280a02009-06-11 10:53:51 -07002599 pGen->init(&codeBuf);
2600 file = new TextInputStream(text, textLength);
Jack Palevich0a280a02009-06-11 10:53:51 -07002601 pGlobalBase = (char*) calloc(1, ALLOC_SIZE);
2602 glo = pGlobalBase;
Jack Palevich0a280a02009-06-11 10:53:51 -07002603 inp();
2604 next();
2605 globalDeclarations();
Jack Palevicha6baa232009-06-12 11:25:59 -07002606 checkForUndefinedForwardReferences();
Jack Palevich0a280a02009-06-11 10:53:51 -07002607 result = pGen->finishCompile();
2608 if (result == 0) {
2609 if (mErrorBuf.len()) {
2610 result = -2;
Jack Palevichac0e95e2009-05-29 13:53:44 -07002611 }
Jack Palevich8b0624c2009-05-20 12:12:06 -07002612 }
Jack Palevichac0e95e2009-05-29 13:53:44 -07002613 return result;
Jack Palevich21a15a22009-05-11 14:49:29 -07002614 }
2615
Jack Palevicha6baa232009-06-12 11:25:59 -07002616 void checkForUndefinedForwardReferences() {
2617 mSymbolTable.forEachGlobal(static_ufrcFn, this);
2618 }
2619
2620 static bool static_ufrcFn(String* key, VariableInfo* value,
2621 void* context) {
2622 Compiler* pCompiler = (Compiler*) context;
2623 return pCompiler->undefinedForwardReferenceCheck(key, value);
2624 }
2625
2626 bool undefinedForwardReferenceCheck(String* key, VariableInfo* value) {
2627#if 0
2628 fprintf(stderr, "%s 0x%8x 0x%08x\n", key->getUnwrapped(),
2629 value->pAddress, value->pForward);
2630#endif
2631 if (!value->pAddress && value->pForward) {
2632 error("Undefined forward reference: %s", key->getUnwrapped());
2633 }
2634 return true;
2635 }
2636
Jack Palevich21a15a22009-05-11 14:49:29 -07002637 int dump(FILE* out) {
2638 fwrite(codeBuf.getBase(), 1, codeBuf.getSize(), out);
2639 return 0;
2640 }
Jack Palevich77ae76e2009-05-10 19:59:24 -07002641
Jack Palevicha6535612009-05-13 16:24:17 -07002642 int disassemble(FILE* out) {
2643 return pGen->disassemble(out);
2644 }
2645
Jack Palevich1cdef202009-05-22 12:06:27 -07002646 /* Look through the symbol table to find a symbol.
2647 * If found, return its value.
2648 */
2649 void* lookup(const char* name) {
Jack Palevich303d8ff2009-06-11 19:06:24 -07002650 String string(name, -1, false);
2651 VariableInfo* pVariableInfo = mSymbolTable.get(&string);
2652 if (pVariableInfo) {
2653 return pVariableInfo->pAddress;
Jack Palevich1cdef202009-05-22 12:06:27 -07002654 }
2655 return NULL;
2656 }
2657
Jack Palevicheedf9d22009-06-04 16:23:40 -07002658 void getPragmas(ACCsizei* actualStringCount,
2659 ACCsizei maxStringCount, ACCchar** strings) {
2660 int stringCount = mPragmaStringCount;
2661 if (actualStringCount) {
2662 *actualStringCount = stringCount;
2663 }
2664 if (stringCount > maxStringCount) {
2665 stringCount = maxStringCount;
2666 }
2667 if (strings) {
2668 char* pPragmas = mPragmas.getUnwrapped();
2669 while (stringCount-- > 0) {
2670 *strings++ = pPragmas;
2671 pPragmas += strlen(pPragmas) + 1;
2672 }
2673 }
2674 }
2675
Jack Palevichac0e95e2009-05-29 13:53:44 -07002676 char* getErrorMessage() {
Jack Palevicheedf9d22009-06-04 16:23:40 -07002677 return mErrorBuf.getUnwrapped();
Jack Palevichac0e95e2009-05-29 13:53:44 -07002678 }
2679
Jack Palevich77ae76e2009-05-10 19:59:24 -07002680};
2681
Jack Paleviche7b59062009-05-19 17:12:17 -07002682const char* Compiler::operatorChars =
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002683 "++--*@/@%@+@-@<<>><=>=<@>@==!=&&||&@^@|@~@!@";
2684
Jack Paleviche7b59062009-05-19 17:12:17 -07002685const char Compiler::operatorLevel[] =
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002686 {11, 11, 1, 1, 1, 2, 2, 3, 3, 4, 4, 4, 4,
2687 5, 5, /* ==, != */
2688 9, 10, /* &&, || */
2689 6, 7, 8, /* & ^ | */
2690 2, 2 /* ~ ! */
2691 };
2692
Jack Palevich8b0624c2009-05-20 12:12:06 -07002693#ifdef PROVIDE_ARM_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -07002694FILE* Compiler::ARMCodeGenerator::disasmOut;
Jack Palevich8b0624c2009-05-20 12:12:06 -07002695#endif
Jack Palevicha6535612009-05-13 16:24:17 -07002696
Jack Palevich8b0624c2009-05-20 12:12:06 -07002697#ifdef PROVIDE_X86_CODEGEN
Jack Paleviche7b59062009-05-19 17:12:17 -07002698const int Compiler::X86CodeGenerator::operatorHelper[] = {
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002699 0x1, // ++
2700 0xff, // --
2701 0xc1af0f, // *
2702 0xf9f79991, // /
2703 0xf9f79991, // % (With manual assist to swap results)
2704 0xc801, // +
2705 0xd8f7c829, // -
2706 0xe0d391, // <<
2707 0xf8d391, // >>
2708 0xe, // <=
2709 0xd, // >=
2710 0xc, // <
2711 0xf, // >
2712 0x4, // ==
2713 0x5, // !=
2714 0x0, // &&
2715 0x1, // ||
2716 0xc821, // &
2717 0xc831, // ^
2718 0xc809, // |
2719 0xd0f7, // ~
2720 0x4 // !
2721};
Jack Palevich8b0624c2009-05-20 12:12:06 -07002722#endif
Jack Palevichbf42c9c2009-05-12 12:48:35 -07002723
Jack Palevich1cdef202009-05-22 12:06:27 -07002724struct ACCscript {
2725 ACCscript() {
2726 text = 0;
2727 textLength = 0;
2728 accError = ACC_NO_ERROR;
2729 }
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002730
Jack Palevich1cdef202009-05-22 12:06:27 -07002731 ~ACCscript() {
2732 delete text;
2733 }
Jack Palevich546b2242009-05-13 15:10:04 -07002734
Jack Palevich1cdef202009-05-22 12:06:27 -07002735 void setError(ACCenum error) {
2736 if (accError == ACC_NO_ERROR && error != ACC_NO_ERROR) {
2737 accError = error;
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002738 }
2739 }
2740
Jack Palevich1cdef202009-05-22 12:06:27 -07002741 ACCenum getError() {
2742 ACCenum result = accError;
2743 accError = ACC_NO_ERROR;
Jack Palevich22305132009-05-13 10:58:45 -07002744 return result;
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002745 }
2746
Jack Palevich1cdef202009-05-22 12:06:27 -07002747 Compiler compiler;
2748 char* text;
2749 int textLength;
2750 ACCenum accError;
2751};
2752
2753
2754extern "C"
2755ACCscript* accCreateScript() {
2756 return new ACCscript();
Jack Palevichbbf8ab52009-05-11 11:54:30 -07002757}
Jack Palevich1cdef202009-05-22 12:06:27 -07002758
2759extern "C"
2760ACCenum accGetError( ACCscript* script ) {
2761 return script->getError();
2762}
2763
2764extern "C"
2765void accDeleteScript(ACCscript* script) {
2766 delete script;
2767}
2768
2769extern "C"
2770void accScriptSource(ACCscript* script,
2771 ACCsizei count,
2772 const ACCchar ** string,
2773 const ACCint * length) {
2774 int totalLength = 0;
2775 for(int i = 0; i < count; i++) {
2776 int len = -1;
2777 const ACCchar* s = string[i];
2778 if (length) {
2779 len = length[i];
2780 }
2781 if (len < 0) {
2782 len = strlen(s);
2783 }
2784 totalLength += len;
2785 }
2786 delete script->text;
2787 char* text = new char[totalLength + 1];
2788 script->text = text;
2789 script->textLength = totalLength;
Jack Palevich09555c72009-05-27 12:25:55 -07002790 char* dest = text;
Jack Palevich1cdef202009-05-22 12:06:27 -07002791 for(int i = 0; i < count; i++) {
2792 int len = -1;
2793 const ACCchar* s = string[i];
2794 if (length) {
2795 len = length[i];
2796 }
2797 if (len < 0) {
2798 len = strlen(s);
2799 }
Jack Palevich09555c72009-05-27 12:25:55 -07002800 memcpy(dest, s, len);
2801 dest += len;
Jack Palevich1cdef202009-05-22 12:06:27 -07002802 }
2803 text[totalLength] = '\0';
2804}
2805
2806extern "C"
2807void accCompileScript(ACCscript* script) {
2808 int result = script->compiler.compile(script->text, script->textLength);
2809 if (result) {
2810 script->setError(ACC_INVALID_OPERATION);
2811 }
2812}
2813
2814extern "C"
2815void accGetScriptiv(ACCscript* script,
2816 ACCenum pname,
2817 ACCint * params) {
2818 switch (pname) {
2819 case ACC_INFO_LOG_LENGTH:
2820 *params = 0;
2821 break;
2822 }
2823}
2824
2825extern "C"
2826void accGetScriptInfoLog(ACCscript* script,
2827 ACCsizei maxLength,
2828 ACCsizei * length,
2829 ACCchar * infoLog) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002830 char* message = script->compiler.getErrorMessage();
2831 int messageLength = strlen(message) + 1;
Jack Palevich1cdef202009-05-22 12:06:27 -07002832 if (length) {
Jack Palevichac0e95e2009-05-29 13:53:44 -07002833 *length = messageLength;
Jack Palevich1cdef202009-05-22 12:06:27 -07002834 }
Jack Palevichac0e95e2009-05-29 13:53:44 -07002835 if (infoLog && maxLength > 0) {
2836 int trimmedLength = maxLength < messageLength ?
2837 maxLength : messageLength;
2838 memcpy(infoLog, message, trimmedLength);
2839 infoLog[trimmedLength] = 0;
Jack Palevich1cdef202009-05-22 12:06:27 -07002840 }
2841}
2842
2843extern "C"
2844void accGetScriptLabel(ACCscript* script, const ACCchar * name,
2845 ACCvoid ** address) {
2846 void* value = script->compiler.lookup(name);
2847 if (value) {
2848 *address = value;
2849 } else {
2850 script->setError(ACC_INVALID_VALUE);
2851 }
2852}
2853
Jack Palevicheedf9d22009-06-04 16:23:40 -07002854extern "C"
2855void accGetPragmas(ACCscript* script, ACCsizei* actualStringCount,
2856 ACCsizei maxStringCount, ACCchar** strings){
2857 script->compiler.getPragmas(actualStringCount, maxStringCount, strings);
2858}
2859
2860
Jack Palevich1cdef202009-05-22 12:06:27 -07002861} // namespace acc
2862