ncc.h

   1 /*
   2  * THE NEATCC C COMPILER
   3  *
   4  * This header file is organized as follows:
   5  *
   6  * 0. helper functions and data structures
   7  * 1. ncc.c -> tok.c: the interface for reading tokens
   8  * 2. ncc.c -> int.c: the interface for generating the intermediate code
   9  * 3. int.c -> gen.c: the intermediate code
  10  * 4. gen.c -> x64.c: the interface for generating the final code
  11  * 5. gen.c -> out.c: the interface for generating object files
  12  */
  13
  14 /* SECTION ZERO: Helper Functions */
  15 /* predefined array limits; (p.f. means per function) */
  16 #define NARGS           32              /* number of function/macro arguments */
  17 #define NTMPS           64              /* number of expression temporaries */
  18 #define NFIELDS         128             /* number of fields in structs */
  19 #define NAMELEN         128             /* size of identifiers */
  20 #define NDEFS           4096            /* number of macros */
  21 #define MARGLEN         1024            /* size of macro arguments */
  22 #define MDEFLEN         2048            /* size of macro definitions */
  23 #define NBUFS           32              /* macro expansion stack depth */
  24 #define NLOCS           1024            /* number of header search paths */
  25
  26 #define LEN(a)          (sizeof(a) / sizeof((a)[0]))
  27 #define ALIGN(x, a)     (((x) + (a) - 1) & ~((a) - 1))
  28 #define MIN(a, b)       ((a) < (b) ? (a) : (b))
  29 #define MAX(a, b)       ((a) < (b) ? (b) : (a))
  30
  31 void *mextend(void *old, long oldsz, long newsz, long memsz);
  32 void die(char *msg, ...);
  33 void err(char *fmt, ...);
  34 int opt(int level);
  35
  36 /* variable length buffer */
  37 struct mem {
  38         char *s;                /* allocated buffer */
  39         long sz;                /* buffer size */
  40         long n;                 /* length of data stored in s */
  41 };
  42
  43 void mem_init(struct mem *mem);
  44 void mem_done(struct mem *mem);
  45 void mem_cut(struct mem *mem, long pos);
  46 void *mem_buf(struct mem *mem);
  47 void mem_put(struct mem *mem, void *buf, long len);
  48 void mem_putc(struct mem *mem, int c);
  49 void mem_putz(struct mem *mem, long sz);
  50 void mem_cpy(struct mem *mem, long off, void *buf, long len);
  51 long mem_len(struct mem *mem);
  52 void *mem_get(struct mem *mem);
  53
  54 /* SECTION ONE: Tokenisation */
  55 void tok_init(char *path);
  56 void tok_done(void);
  57 char *tok_see(void);            /* return the current token; a static buffer */
  58 char *tok_get(void);            /* return and consume the current token */
  59 long tok_len(void);             /* the length of the last token */
  60 long tok_num(char *tok, long *n);
  61 long tok_addr(void);
  62 void tok_jump(long addr);
  63
  64 int cpp_init(char *path);
  65 void cpp_path(char *s);
  66 void cpp_define(char *name, char *def);
  67 char *cpp_loc(long addr);
  68 int cpp_read(char **buf, long *len);
  69
  70 /* SECTION TWO: Intermediate Code Generation */
  71 /* basic type meaning */
  72 #define T_MSIZE         0x000f
  73 #define T_MSIGN         0x0010
  74 #define T_SZ(bt)        ((bt) & T_MSIZE)
  75 #define T_SG(bt)        ((bt) & T_MSIGN)
  76 #define T_MK(sign, size)        (((sign) & T_MSIGN) | ((size) & T_MSIZE))
  77
  78 /* number of bytes in basic types */
  79 #define ULNG            (LONGSZ)
  80 #define UINT            (4)
  81 #define USHT            (2)
  82 #define UCHR            (1)
  83 /* basic types */
  84 #define SLNG            (ULNG | T_MSIGN)
  85 #define SINT            (UINT | T_MSIGN)
  86 #define SSHT            (USHT | T_MSIGN)
  87 #define SCHR            (UCHR | T_MSIGN)
  88
  89 /*
  90  * Intermediate instruction operands
  91  * R: register, N: immediate, S: symbol, L: local,
  92  * D: displacement, G: label, C: arguments
  93  */
  94 /* Instruction                          rd      r1      r2      r3 */
  95 #define O_ADD   0x000010        /*      R       R       RN      -  */
  96 #define O_SHL   0x000020        /*      R       R       RN      -  */
  97 #define O_MUL   0x000040        /*      R       R       RN      -  */
  98 #define O_CMP   0x000080        /*      R       R       RN      -  */
  99 #define O_UOP   0x000100        /*      R       R       -       -  */
 100 #define O_CALL  0x000200        /*      R       RS      D       C  */
 101 #define O_MOV   0x000400        /*      R       RNSL    D       -  */
 102 #define O_MEM   0x000800        /*      -       R       R       R  */
 103 #define O_JMP   0x001000        /*      -       -       -       G  */
 104 #define O_JZ    0x002000        /*      -       R       -       G  */
 105 #define O_JCC   0x004000        /*      -       R       RN      G  */
 106 #define O_RET   0x008000        /*      -       R       -       -  */
 107 #define O_LD    0x010000        /*      R       RSL     D       -  */
 108 #define O_ST    0x020000        /*      -       R       RSL     D  */
 109 /* opcode flags: num, loc, sym */
 110 #define O_NUM   0x100000        /* instruction immediate */
 111 #define O_LOC   0x200000        /* local (frame pointer displacement) */
 112 #define O_SYM   0x400000        /* symbols (relocations and offset) */
 113 /* other members of instruction groups */
 114 #define O_SUB           (1 | O_ADD)
 115 #define O_AND           (2 | O_ADD)
 116 #define O_OR            (3 | O_ADD)
 117 #define O_XOR           (4 | O_ADD)
 118 #define O_SHR           (1 | O_SHL)
 119 #define O_DIV           (1 | O_MUL)
 120 #define O_MOD           (2 | O_MUL)
 121 #define O_LT            (0 | O_CMP)
 122 #define O_GE            (1 | O_CMP)
 123 #define O_EQ            (2 | O_CMP)
 124 #define O_NE            (3 | O_CMP)
 125 #define O_LE            (4 | O_CMP)
 126 #define O_GT            (5 | O_CMP)
 127 #define O_NEG           (0 | O_UOP)
 128 #define O_NOT           (1 | O_UOP)
 129 #define O_LNOT          (2 | O_UOP)
 130 #define O_MSET          (0 | O_MEM)
 131 #define O_MCPY          (1 | O_MEM)
 132 #define O_JNZ           (1 | O_JZ)
 133 /* instruction masks */
 134 #define O_BOP           (O_ADD | O_MUL | O_CMP | O_SHL)
 135 #define O_OUT           (O_BOP | O_UOP | O_CALL | O_MOV | O_LD)
 136 #define O_JXX           (O_JMP | O_JZ | O_JCC)
 137 /* instruction operand type */
 138 #define O_C(op)         ((op) & 0xffffff)       /* operation code */
 139 #define O_T(op)         ((op) >> 24)    /* instruction operand type */
 140 #define O_MK(op, bt)    ((op) | ((bt) << 24))
 141
 142 /* operations on the stack */
 143 void o_bop(long op);            /* binary operation */
 144 void o_uop(long op);            /* unary operation */
 145 void o_cast(long bt);
 146 void o_memcpy(void);
 147 void o_memset(void);
 148 void o_call(int argc, int ret);
 149 void o_ret(int ret);
 150 void o_assign(long bt);
 151 void o_deref(long bt);
 152 void o_load(void);
 153 int o_popnum(long *num);
 154 int o_popsym(long *sym, long *off);
 155 /* pushing values to the stack */
 156 void o_num(long n);
 157 void o_local(long addr);
 158 void o_sym(char *sym);
 159 void o_tmpdrop(int n);
 160 void o_tmpswap(void);
 161 void o_tmpcopy(void);
 162 /* handling locals */
 163 long o_mklocal(long size);
 164 void o_rmlocal(long addr, long sz);
 165 long o_arg2loc(int i);
 166 /* branches */
 167 void o_label(long id);
 168 void o_jmp(long id);
 169 void o_jz(long id);
 170 long o_mark(void);
 171 void o_back(long mark);
 172 /* data/bss sections */
 173 long o_dsnew(char *name, long size, int global);
 174 void o_dscpy(long addr, void *buf, long len);
 175 void o_dsset(char *name, long off, long bt);
 176 void o_bsnew(char *name, long size, int global);
 177 /* functions */
 178 void o_func_beg(char *name, int argc, int global, int vararg);
 179 void o_func_end(void);
 180 void o_code(char *name, char *c, long c_len);
 181 /* output */
 182 void o_write(int fd);
 183
 184 /* SECTION THREE: The Intermediate Code */
 185 /* intermediate code instructions */
 186 struct ic {
 187         long op;                /* instruction opcode */
 188         long a1;                /* first argument */
 189         long a2;                /* second argument */
 190         long a3;                /* more information, like jump target */
 191         long *args;             /* call arguments */
 192 };
 193
 194 /* get the generated intermediate code */
 195 void ic_get(struct ic **c, long *n);
 196 int ic_num(struct ic *ic, long iv, long *num);
 197 int ic_sym(struct ic *ic, long iv, long *sym, long *off);
 198 long *ic_lastuse(struct ic *ic, long ic_n);
 199 void ic_free(struct ic *ic);
 200 int ic_regcnt(struct ic *ic);
 201
 202 /* global register allocation */
 203 void reg_init(struct ic *ic, long ic_n);
 204 long reg_mask(void);
 205 int reg_lmap(long ic, long loc);
 206 int reg_rmap(long ic, long reg);
 207 int reg_safe(long loc);
 208 void reg_done(void);
 209
 210 /* SECTION FOUR: Final Code Generation */
 211 /*
 212  * To make maintaining different architectures easier and to unify the
 213  * optimizations, I have merged the code generation for different
 214  * architectures.  The i_*() functions are now the low level
 215  * architecture-specific code generation entry points.  The
 216  * differences between RISC and CISC architectures, actually the
 217  * annoying asymmetry in CISC architecture, has made this interface
 218  * more complex than it could have ideally been.  Nevertheless,
 219  * the benefits of extracting gen.c and the cleaner design,
 220  * especially with the presence of the optimizations, outweighs the
 221  * added complexity.  Overall, there were many challenges for
 222  * extracting gen.c including:
 223  * + Different register sets; caller/callee saved and argument registers
 224  * + CISC-style instructions that work on limited registers and parameters
 225  * + Different instruction formats and immediate value limitations
 226  * + Generating epilog, prolog, and local variable addresses when optimizing
 227  *
 228  * I tried to make this interface as small as possible.  The key
 229  * functions and macros described next.
 230  *
 231  * i_reg() returns the mask of allowed registers for each
 232  * operand of an instruction.  The first argument op, specifies
 233  * the instruction (O_* macros); i_reg() sets the value r0, r1,
 234  * and r2 to indicate the mask of acceptable registers for the
 235  * first, second, and third operands of the instruction.
 236  * The value of these masks may be changed to zero to indicate
 237  * fewer than three operands.  If md is zero while m1 is not,
 238  * the destination register should be equal to the first register,
 239  * as in CISC architectures.  mt denotes the mask of registers
 240  * that may lose their contents after the instruction.
 241  *
 242  * i_ins() generates code for the given instruction.  The arguments
 243  * indicate the instruction and its operands.  The code is generated
 244  * by calling os() and oi() functions and the current position in
 245  * the code segment is obtained by calling opos().  For branch
 246  * instructions, i_ins() returns the position of branch offset in
 247  * code segment, to be filled later with i_fill().
 248  *
 249  * Some macros should be defined in architecture-dependent headers
 250  * and a few variables should be defined for each architecture,
 251  * such as tmpregs, which is an array of register numbers that
 252  * can be used for holding temporaries and argregs, which is an
 253  * array of register numbers for holding the first N_ARGS arguments.
 254  * Consult x64.h as an example, for the macros defined for each
 255  * architecture.
 256  *
 257  */
 258 #ifdef NEATCC_ARM
 259 #include "arm.h"
 260 #endif
 261 #ifdef NEATCC_X64
 262 #include "x64.h"
 263 #endif
 264 #ifdef NEATCC_X86
 265 #include "x86.h"
 266 #endif
 267
 268 /* architecture-specific operations */
 269 long i_reg(long op, long *rd, long *r1, long *r2, long *r3, long *mt);
 270 long i_ins(long op, long rd, long r1, long r2, long r3);
 271 int i_imm(long lim, long n);
 272 void i_label(long id);
 273 void i_wrap(int argc, long sargs, long spsub, int initfp, long sregs, long sregs_pos);
 274 void i_code(char **c, long *c_len, long **rsym, long **rflg, long **roff, long *rcnt);
 275 void i_done(void);
 276
 277 extern int tmpregs[];
 278 extern int argregs[];
 279
 280 /* SECTION FIVE: Object File Generation */
 281 #define OUT_CS          0x0001          /* code segment symbol */
 282 #define OUT_DS          0x0002          /* data segment symbol */
 283 #define OUT_BSS         0x0004          /* bss segment symbol */
 284
 285 #define OUT_GLOB        0x0010          /* global symbol */
 286
 287 #define OUT_RLREL       0x0020          /* relative relocation */
 288 #define OUT_RLSX        0x0040          /* sign extend relocation */
 289 #define OUT_RL24        0x0400          /* 3-byte relocation */
 290 #define OUT_RL32        0x0800          /* 4-byte relocation */
 291
 292 #define OUT_ALIGNMENT   16              /* section alignment */
 293
 294 void out_init(long flags);
 295
 296 long out_sym(char *name);
 297 void out_def(char *name, long flags, long off, long len);
 298 void out_rel(long id, long flags, long off);
 299
 300 void out_write(int fd, char *cs, long cslen, char *ds, long dslen);