Completely changed the assembler.

It now has a lexer/tokenizer, along with a parser.
I have also made the emulator even smaller.

1 files changed, 214 insertions, 0 deletions

diff --git a/asmmon.h b/asmmon.h
new file mode 100644
index 0000000..cc21c14
--- /dev/null
+++ b/asmmon.h
@@ -0,0 +1,214 @@
+#include "opcode.h"
+#include <ctype.h>
+#include <string.h>
+
+#define debug 1
+
+char lexeme[0x1000];
+uint8_t lex_type;
+
+static const uint8_t opcodes[OPNUM][9] = {
+	/*	 IMM	ZM   ZMX   ZMY	 IND  INDX  INDY   ABS	IMPL*/
+	[ 0] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00},
+	[ 1] = {0x01, 0x06, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x04, 0xFF},
+	[ 2] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x02},
+	[ 3] = {0x08, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[ 4] = {0x09, 0x2D, 0xFF, 0xFF, 0x55, 0xAD, 0xAC, 0x2C, 0xFF},
+	[ 5] = {0x0A, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[ 6] = {0xFF, 0x0D, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0C, 0xE5},
+	[ 7] = {0xFF, 0x0E, 0xFF, 0xFF, 0xCE, 0xFF, 0xFF, 0x10, 0xFF},
+	[ 8] = {0x11, 0x16, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x14, 0xFF},
+	[ 9] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x12},
+	[10] = {0x18, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[11] = {0x19, 0x3D, 0xFF, 0xFF, 0x85, 0xFF, 0xFF, 0x4C, 0xFF},
+	[12] = {0x1A, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[13] = {0xFF, 0x1D, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x1C, 0xF5},
+	[14] = {0xFF, 0x1E, 0xFF, 0xFF, 0xBE, 0xFF, 0xFF, 0xFF, 0xFF},
+	[15] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x20, 0xFF},
+	[16] = {0x21, 0x26, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x24, 0xFF},
+	[17] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x22},
+	[18] = {0x28, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[19] = {0x29, 0x4D, 0xFF, 0xFF, 0xB5, 0xFF, 0xFF, 0x3C, 0xFF},
+	[20] = {0x2A, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[21] = {0xFF, 0x2E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x30, 0xFF},
+	[22] = {0x31, 0x36, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x34, 0xFF},
+	[23] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x32},
+	[24] = {0x38, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[25] = {0x3A, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[26] = {0xFF, 0x3E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x40, 0xFF},
+	[27] = {0x41, 0x46, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x44, 0xFF},
+	[28] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x42},
+	[29] = {0x48, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[30] = {0x4A, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[31] = {0xFF, 0x4E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x50, 0xFF},
+	[32] = {0x51, 0x56, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x54, 0xFF},
+	[33] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x52},
+	[34] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x58},
+	[35] = {0x5A, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[36] = {0xFF, 0x5E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x60, 0xFF},
+	[37] = {0x61, 0x66, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x64, 0xFF},
+	[38] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x62},
+	[39] = {0x68, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[40] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x6A},
+	[41] = {0xFF, 0x6E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x70, 0xFF},
+	[42] = {0x71, 0x76, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x74, 0xFF},
+	[43] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x72},
+	[44] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x78},
+	[45] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x7A},
+	[46] = {0xFF, 0x7E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x80, 0xFF},
+	[47] = {0x81, 0x86, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x84, 0xFF},
+	[48] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x82},
+	[49] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x88},
+	[50] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x8A},
+	[51] = {0xFF, 0x8E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x90, 0xFF},
+	[52] = {0x91, 0x96, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x94, 0xFF},
+	[53] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x92},
+	[54] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x98},
+	[55] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x9A},
+	[56] = {0xFF, 0x9E, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xA0, 0xFF},
+	[57] = {0xA1, 0xA6, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xA4, 0xFF},
+	[58] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xA2},
+	[59] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xA8},
+	[60] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xAA},
+	[61] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xAE},
+	[62] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xB0},
+	[63] = {0xB1, 0xB6, 0xFF, 0xFF, 0x25, 0x7D, 0x7C, 0xB4, 0xFF},
+	[64] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xB2},
+	[65] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xB8},
+	[66] = {0xB9, 0xBD, 0xFF, 0xC9, 0x95, 0xFF, 0xFF, 0xBC, 0xFF},
+	[67] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xBA},
+	[68] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xC0},
+	[69] = {0xC1, 0xC6, 0x79, 0x39, 0x05, 0x5D, 0x5C, 0xC4, 0xFF},
+	[70] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xC5},
+	[71] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xC8},
+	[72] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xCA},
+	[73] = {0xFF, 0xCD, 0x89, 0x49, 0x15, 0x6D, 0x6C, 0xCC, 0xFF},
+	[74] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xD0},
+	[75] = {0xD1, 0xD6, 0x99, 0x59, 0x35, 0x8D, 0x8C, 0xD4, 0xFF},
+	[76] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xD5},
+	[77] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xD8},
+	[78] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xDA},
+	[79] = {0xFF, 0xDD, 0xA9, 0x69, 0x45, 0x9D, 0x9C, 0xDC, 0xFF},
+	[80] = {0xE0, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF},
+	[81] = {0xE1, 0xE6, 0xE9, 0xFF, 0x65, 0xFF, 0xFF, 0xE4, 0xFF},
+	[82] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xE8},
+	[83] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xEA},
+	[84] = {0xFF, 0xED, 0xF9, 0xFF, 0x75, 0xFF, 0xFF, 0xEC, 0xFF},
+	[85] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xEE},
+	[86] = {0xF1, 0xF6, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xF4, 0xFF},
+	[87] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xF2},
+	[88] = {0xFF, 0xFD, 0xFF, 0xD9, 0xA5, 0xFF, 0xFF, 0xFC, 0xFF},
+	[89] = {0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFE}
+};
+
+static const char *dir_t[5] = {
+	[0] = "org",
+	[1] = "byte",
+	[2] = "word",
+	[3] = "dword",
+	[4] = "qword"
+};
+
+static const char *rs_t[4] = {
+	[0] = "",
+	[1] = ".w",
+	[2] = ".d",
+	[3] = ".q"
+};
+
+static const char *lex_tok[14] = {
+	[0x0] = "TOK_DIR",
+	[0x1] = "TOK_LABEL",
+	[0x2] = "TOK_SYM",
+	[0x3] = "TOK_PLUS",
+	[0x4] = "TOK_MINUS",
+	[0x5] = "TOK_STRING",
+	[0x6] = "TOK_CHAR",
+	[0x7] = "TOK_IMM",
+	[0x8] = "TOK_OPCODE",
+	[0x9] = "TOK_RS",
+	[0xA] = "TOK_COMMENT",
+	[0xB] = "TOK_HEX",
+	[0xC] = "TOK_DEC",
+	[0xD] = "TOK_BIN"
+};
+
+static const uint8_t bitsize[4] = {
+	[0] = 0x07,
+	[1] = 0x0F,
+	[2] = 0x1F,
+	[3] = 0x3F
+};
+
+static const uint8_t amp[8] = {
+	[0] = 0x00,
+	[1] = 0x00,
+	[2] = 0x07,
+	[4] = 0x07,
+	[5] = 0x0B,
+	[6] = 0x0B,
+	[3] = 0x0F,
+	[7] = 0x0F
+};
+
+struct line {
+	uint8_t stab;
+	uint8_t sspace;
+	uint8_t dir;
+	uint8_t mne;
+	uint8_t rs;
+	uint8_t opbase;
+	uint8_t aopbase;
+	uint8_t am;
+	uint8_t cm;
+	uint8_t etab;
+	uint8_t espace;
+	uint8_t islabel;
+	uint8_t issym;
+	uint16_t sym;
+	uint16_t com;
+	uint16_t str;
+	uint16_t linenum;
+	uint64_t op;
+	uint64_t aop;
+	uint64_t addr;
+};
+
+extern uint16_t linenum;
+extern uint16_t lineidx;
+extern uint16_t stridx;
+extern uint16_t comidx;
+
+char *string[0x1000];
+char *comment[0x1000];
+
+struct line tokline[0x1000];
+
+struct fixup {
+	struct fixup *nxt;
+	struct symbol *s;
+	uint16_t ln;
+	uint64_t adr;
+};
+
+struct symbol {
+	struct symbol* nxt;
+	uint64_t val;
+	uint8_t def;
+	char name[128];
+	uint16_t id;
+};
+
+extern struct symbol *symbols;
+extern struct fixup *fixups;
+extern uint8_t defined;
+extern uint8_t isfixup;
+
+extern struct symbol *mksymbol(const char *name, uint64_t val, uint8_t def, uint8_t useid, uint16_t id, uint8_t dbg);
+extern uint64_t use_symbol(const char *name, uint16_t id, uint64_t val, uint8_t useid, uint8_t dbg);
+extern uint8_t set_symval(const char *name, uint16_t id, uint64_t val, uint8_t useid, uint8_t dbg);
+extern char *get_symname(uint16_t id, uint8_t dbg);
+extern uint16_t get_symid(const char *name, uint64_t val, uint16_t ln, uint8_t dbg);
+extern uint16_t get_comment(const char *cmnt, uint8_t dbg);
+extern uint16_t reslv_fixups(uint8_t dbg);
+extern uint64_t lex(char *str, uint64_t address, uint8_t dbg);