lex.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

#define	EXTERN
#include "a.h"
#include "y.tab.h"
#include <ctype.h>

void
main(int argc, char *argv[])
{
	char *p;
	int nout, nproc, status, i, c;

	thechar = 'q';
	thestring = "power";
	memset(debug, 0, sizeof(debug));
	cinit();
	outfile = 0;
	include[ninclude++] = ".";
	ARGBEGIN {
	default:
		c = ARGC();
		if(c >= 0 || c < sizeof(debug))
			debug[c] = 1;
		break;

	case 'o':
		outfile = ARGF();
		break;

	case 'D':
		p = ARGF();
		if(p)
			Dlist[nDlist++] = p;
		break;

	case 'I':
		p = ARGF();
		setinclude(p);
		break;
	} ARGEND
	if(*argv == 0) {
		print("usage: %ca [-options] file.s\n", thechar);
		errorexit();
	}
	if(argc > 1 && systemtype(Windows)){
		print("can't assemble multiple files on windows\n");
		errorexit();
	}
	if(argc > 1) {
		nproc = 1;
		if(p = getenv("NPROC"))
			nproc = atol(p);
		c = 0;
		nout = 0;
		for(;;) {
			while(nout < nproc && argc > 0) {
				i = myfork();
				if(i < 0) {
					i = mywait(&status);
					if(i < 0)
						errorexit();
					if(status)
						c++;
					nout--;
					continue;
				}
				if(i == 0) {
					print("%s:\n", *argv);
					if(assemble(*argv))
						errorexit();
					exits(0);
				}
				nout++;
				argc--;
				argv++;
			}
			i = mywait(&status);
			if(i < 0) {
				if(c)
					errorexit();
				exits(0);
			}
			if(status)
				c++;
			nout--;
		}
	}
	if(assemble(argv[0]))
		errorexit();
	exits(0);
}

int
assemble(char *file)
{
	char ofile[100], incfile[20], *p;
	int i, of;

	strcpy(ofile, file);
	if(p = strrchr(ofile, pathchar())) {
		include[0] = ofile;
		*p++ = 0;
	} else
		p = ofile;
	if(outfile == 0) {
		outfile = p;
		if(p = strrchr(outfile, '.'))
			if(p[1] == 's' && p[2] == 0)
				p[0] = 0;
		p = strrchr(outfile, 0);
		p[0] = '.';
		p[1] = thechar;
		p[2] = 0;
	}
	p = getenv("INCLUDE");
	if(p) {
		setinclude(p);
	} else {
		if(systemtype(Plan9)) {
			sprint(incfile,"/%s/include", thestring);
			setinclude(strdup(incfile));
		}
	}

	of = mycreat(outfile, 0664);
	if(of < 0) {
		yyerror("%ca: cannot create %s", thechar, outfile);
		errorexit();
	}
	Binit(&obuf, of, OWRITE);

	pass = 1;
	nosched = 0;
	pinit(file);
	for(i=0; i<nDlist; i++)
		dodefine(Dlist[i]);
	yyparse();
	if(nerrors) {
		cclean();
		return nerrors;
	}

	pass = 2;
	nosched = 0;
	outhist();
	pinit(file);
	for(i=0; i<nDlist; i++)
		dodefine(Dlist[i]);
	yyparse();
	cclean();
	return nerrors;
}

struct
{
	char	*name;
	ushort	type;
	ushort	value;
} itab[] =
{
	"SP",		LSP,	D_AUTO,
	"SB",		LSB,	D_EXTERN,
	"FP",		LFP,	D_PARAM,
	"PC",		LPC,	D_BRANCH,

	"LR",		LLR,	D_LR,
	"CTR",		LCTR,	D_CTR,

	"XER",		LSPREG,	D_XER,
	"MSR",		LMSR,	D_MSR,
	"FPSCR",	LFPSCR,	D_FPSCR,
	"SPR",		LSPR,	D_SPR,
	"DCR",		LSPR,	D_DCR,

	"SEG",		LSEG,	D_SREG,

	"CR",		LCR,	0,
	"CR0",		LCREG,	0,
	"CR1",		LCREG,	1,
	"CR2",		LCREG,	2,
	"CR3",		LCREG,	3,
	"CR4",		LCREG,	4,
	"CR5",		LCREG,	5,
	"CR6",		LCREG,	6,
	"CR7",		LCREG,	7,

	"R",		LR,	0,
	"R0",		LREG,	0,
	"R1",		LREG,	1,
	"R2",		LREG,	2,
	"R3",		LREG,	3,
	"R4",		LREG,	4,
	"R5",		LREG,	5,
	"R6",		LREG,	6,
	"R7",		LREG,	7,
	"R8",		LREG,	8,
	"R9",		LREG,	9,
	"R10",		LREG,	10,
	"R11",		LREG,	11,
	"R12",		LREG,	12,
	"R13",		LREG,	13,
	"R14",		LREG,	14,
	"R15",		LREG,	15,
	"R16",		LREG,	16,
	"R17",		LREG,	17,
	"R18",		LREG,	18,
	"R19",		LREG,	19,
	"R20",		LREG,	20,
	"R21",		LREG,	21,
	"R22",		LREG,	22,
	"R23",		LREG,	23,
	"R24",		LREG,	24,
	"R25",		LREG,	25,
	"R26",		LREG,	26,
	"R27",		LREG,	27,
	"R28",		LREG,	28,
	"R29",		LREG,	29,
	"R30",		LREG,	30,
	"R31",		LREG,	31,

	"F",		LF,	0,
	"F0",		LFREG,	0,
	"F1",		LFREG,	1,
	"F2",		LFREG,	2,
	"F3",		LFREG,	3,
	"F4",		LFREG,	4,
	"F5",		LFREG,	5,
	"F6",		LFREG,	6,
	"F7",		LFREG,	7,
	"F8",		LFREG,	8,
	"F9",		LFREG,	9,
	"F10",		LFREG,	10,
	"F11",		LFREG,	11,
	"F12",		LFREG,	12,
	"F13",		LFREG,	13,
	"F14",		LFREG,	14,
	"F15",		LFREG,	15,
	"F16",		LFREG,	16,
	"F17",		LFREG,	17,
	"F18",		LFREG,	18,
	"F19",		LFREG,	19,
	"F20",		LFREG,	20,
	"F21",		LFREG,	21,
	"F22",		LFREG,	22,
	"F23",		LFREG,	23,
	"F24",		LFREG,	24,
	"F25",		LFREG,	25,
	"F26",		LFREG,	26,
	"F27",		LFREG,	27,
	"F28",		LFREG,	28,
	"F29",		LFREG,	29,
	"F30",		LFREG,	30,
	"F31",		LFREG,	31,

	"CREQV",	LCROP, ACREQV,
	"CRXOR",	LCROP, ACRXOR,
	"CRAND",	LCROP, ACRAND,
	"CROR",		LCROP, ACROR,
	"CRANDN",	LCROP, ACRANDN,
	"CRORN",	LCROP, ACRORN,
	"CRNAND",	LCROP, ACRNAND,
	"CRNOR",	LCROP, ACRNOR,

	"ADD",		LADDW, AADD,
	"ADDV",		LADDW, AADDV,
	"ADDCC",	LADDW, AADDCC,
	"ADDVCC",	LADDW, AADDVCC,
	"ADDC",		LADDW, AADDC,
	"ADDCV",	LADDW, AADDCV,
	"ADDCCC",	LADDW, AADDCCC,
	"ADDCVCC",	LADDW, AADDCVCC,
	"ADDE",		LLOGW, AADDE,
	"ADDEV",	LLOGW, AADDEV,
	"ADDECC",	LLOGW, AADDECC,
	"ADDEVCC",	LLOGW, AADDEVCC,

	"ADDME",	LABS, AADDME,
	"ADDMEV",	LABS, AADDMEV,
	"ADDMECC",	LABS, AADDMECC,
	"ADDMEVCC",	LABS, AADDMEVCC,
	"ADDZE",	LABS, AADDZE,
	"ADDZEV",	LABS, AADDZEV,
	"ADDZECC",	LABS, AADDZECC,
	"ADDZEVCC",	LABS, AADDZEVCC,

	"SUB",		LADDW, ASUB,
	"SUBV",		LADDW, ASUBV,
	"SUBCC",	LADDW, ASUBCC,
	"SUBVCC",	LADDW, ASUBVCC,
	"SUBE",		LLOGW, ASUBE,
	"SUBECC",	LLOGW, ASUBECC,
	"SUBEV",	LLOGW, ASUBEV,
	"SUBEVCC",	LLOGW, ASUBEVCC,
	"SUBC",		LADDW, ASUBC,
	"SUBCCC",	LADDW, ASUBCCC,
	"SUBCV",	LADDW, ASUBCV,
	"SUBCVCC",	LADDW, ASUBCVCC,

	"SUBME",	LABS, ASUBME,
	"SUBMEV",	LABS, ASUBMEV,
	"SUBMECC",	LABS, ASUBMECC,
	"SUBMEVCC",	LABS, ASUBMEVCC,
	"SUBZE",	LABS, ASUBZE,
	"SUBZEV",	LABS, ASUBZEV,
	"SUBZECC",	LABS, ASUBZECC,
	"SUBZEVCC",	LABS, ASUBZEVCC,

	"AND",		LADDW, AAND,
	"ANDCC",	LADDW, AANDCC,	/* includes andil & andiu */
	"ANDN",		LLOGW, AANDN,
	"ANDNCC",	LLOGW, AANDNCC,
	"EQV",		LLOGW, AEQV,
	"EQVCC",	LLOGW, AEQVCC,
	"NAND",		LLOGW, ANAND,
	"NANDCC",	LLOGW, ANANDCC,
	"NOR",		LLOGW, ANOR,
	"NORCC",	LLOGW, ANORCC,
	"OR",		LADDW, AOR,	/* includes oril & oriu */
	"ORCC",		LADDW, AORCC,
	"ORN",		LLOGW, AORN,
	"ORNCC",	LLOGW, AORNCC,
	"XOR",		LADDW, AXOR,	/* includes xoril & xoriu */
	"XORCC",	LLOGW, AXORCC,

	"EXTSB",	LABS,	AEXTSB,
	"EXTSBCC",	LABS,	AEXTSBCC,
	"EXTSH",	LABS, AEXTSH,
	"EXTSHCC",	LABS, AEXTSHCC,

	"CNTLZW",	LABS, ACNTLZW,
	"CNTLZWCC",	LABS, ACNTLZWCC,

	"RLWMI",	LRLWM, ARLWMI,
	"RLWMICC",	LRLWM, ARLWMICC,
	"RLWNM",	LRLWM, ARLWNM,
	"RLWNMCC", LRLWM, ARLWNMCC,

	"SLW",		LSHW, ASLW,
	"SLWCC",	LSHW, ASLWCC,
	"SRW",		LSHW, ASRW,
	"SRWCC",	LSHW, ASRWCC,
	"SRAW",		LSHW, ASRAW,
	"SRAWCC",	LSHW, ASRAWCC,

	"BR",		LBRA, ABR,
	"BC",		LBRA, ABC,
	"BCL",		LBRA, ABC,
	"BL",		LBRA, ABL,
	"BEQ",		LBRA, ABEQ,
	"BNE",		LBRA, ABNE,
	"BGT",		LBRA, ABGT,
	"BGE",		LBRA, ABGE,
	"BLT",		LBRA, ABLT,
	"BLE",		LBRA, ABLE,
	"BVC",		LBRA, ABVC,
	"BVS",		LBRA, ABVS,

	"CMP",		LCMP, ACMP,
	"CMPU",		LCMP, ACMPU,

	"DIVW",		LLOGW, ADIVW,
	"DIVWV",	LLOGW, ADIVWV,
	"DIVWCC",	LLOGW, ADIVWCC,
	"DIVWVCC",	LLOGW, ADIVWVCC,
	"DIVWU",	LLOGW, ADIVWU,
	"DIVWUV",	LLOGW, ADIVWUV,
	"DIVWUCC",	LLOGW, ADIVWUCC,
	"DIVWUVCC",	LLOGW, ADIVWUVCC,

	"FABS",		LFCONV,	AFABS,
	"FABSCC",	LFCONV,	AFABSCC,
	"FNEG",		LFCONV,	AFNEG,
	"FNEGCC",	LFCONV,	AFNEGCC,
	"FNABS",	LFCONV,	AFNABS,
	"FNABSCC",	LFCONV,	AFNABSCC,

	"FADD",		LFADD,	AFADD,
	"FADDCC",	LFADD,	AFADDCC,
	"FSUB",		LFADD,  AFSUB,
	"FSUBCC",	LFADD,	AFSUBCC,
	"FMUL",		LFADD,	AFMUL,
	"FMULCC",	LFADD,	AFMULCC,
	"FDIV",		LFADD,	AFDIV,
	"FDIVCC",	LFADD,	AFDIVCC,
	"FRSP",		LFCONV,	AFRSP,
	"FRSPCC",	LFCONV,	AFRSPCC,

	"FMADD",	LFMA, AFMADD,
	"FMADDCC",	LFMA, AFMADDCC,
	"FMSUB",	LFMA, AFMSUB,
	"FMSUBCC",	LFMA, AFMSUBCC,
	"FNMADD",	LFMA, AFNMADD,
	"FNMADDCC",	LFMA, AFNMADDCC,
	"FNMSUB",	LFMA, AFNMSUB,
	"FNMSUBCC",	LFMA, AFNMSUBCC,
	"FMADDS",	LFMA, AFMADDS,
	"FMADDSCC",	LFMA, AFMADDSCC,
	"FMSUBS",	LFMA, AFMSUBS,
	"FMSUBSCC",	LFMA, AFMSUBSCC,
	"FNMADDS",	LFMA, AFNMADDS,
	"FNMADDSCC",	LFMA, AFNMADDSCC,
	"FNMSUBS",	LFMA, AFNMSUBS,
	"FNMSUBSCC",	LFMA, AFNMSUBSCC,

	"FCMPU",	LFCMP, AFCMPU,
	"FCMPO",	LFCMP, AFCMPO,
	"MTFSB0",	LMTFSB, AMTFSB0,
	"MTFSB1",	LMTFSB,	AMTFSB1,

	"FMOVD",	LFMOV, AFMOVD,
	"FMOVS",	LFMOV, AFMOVS,
	"FMOVDCC",	LFCONV,	AFMOVDCC,	/* fmr. */

	"GLOBL",	LTEXT, AGLOBL,

	"MOVB",		LMOVB, AMOVB,
	"MOVBZ",	LMOVB, AMOVBZ,
	"MOVBU",	LMOVB, AMOVBU,
	"MOVBZU", LMOVB, AMOVBZU,
	"MOVH",		LMOVB, AMOVH,
	"MOVHZ",	LMOVB, AMOVHZ,
	"MOVHU",	LMOVB, AMOVHU,
	"MOVHZU", LMOVB, AMOVHZU,
	"MOVHBR", 	LXMV, AMOVHBR,
	"MOVWBR",	LXMV, AMOVWBR,
	"MOVW",		LMOVW, AMOVW,
	"MOVWU",	LMOVW, AMOVWU,
	"MOVMW",	LMOVMW, AMOVMW,
	"MOVFL",	LMOVW,	AMOVFL,

	"MULLW",	LADDW, AMULLW,		/* includes multiply immediate 10-139 */