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

	"FSINF",	LTYPE1, AFSINF,
	"FSINHB",	LTYPE1, AFSINHB,
	"FSINHD",	LTYPE1, AFSINHD,
	"FSINHF",	LTYPE1, AFSINHF,
	"FSINHL",	LTYPE1, AFSINHL,
	"FSINHW",	LTYPE1, AFSINHW,
	"FSINL",	LTYPE1, AFSINL,
	"FSINW",	LTYPE1, AFSINW,
	"FSLE",		LTYPE1, AFSLE,
	"FSLT",		LTYPE1, AFSLT,
	"FSNE",		LTYPE1, AFSNE,
	"FSQRTB",	LTYPE1, AFSQRTB,
	"FSQRTD",	LTYPE1, AFSQRTD,
	"FSQRTF",	LTYPE1, AFSQRTF,
	"FSQRTL",	LTYPE1, AFSQRTL,
	"FSQRTW",	LTYPE1, AFSQRTW,
	"FST",		LTYPE1, AFST,
	"FSUBB",	LTYPE1, AFSUBB,
	"FSUBD",	LTYPE1, AFSUBD,
	"FSUBF",	LTYPE1, AFSUBF,
	"FSUBL",	LTYPE1, AFSUBL,
	"FSUBW",	LTYPE1, AFSUBW,
	"FTANB",	LTYPE1, AFTANB,
	"FTAND",	LTYPE1, AFTAND,
	"FTANF",	LTYPE1, AFTANF,
	"FTANHB",	LTYPE1, AFTANHB,
	"FTANHD",	LTYPE1, AFTANHD,
	"FTANHF",	LTYPE1, AFTANHF,
	"FTANHL",	LTYPE1, AFTANHL,
	"FTANHW",	LTYPE1, AFTANHW,
	"FTANL",	LTYPE1, AFTANL,
	"FTANW",	LTYPE1, AFTANW,
	"FTENTOXB",	LTYPE1, AFTENTOXB,
	"FTENTOXD",	LTYPE1, AFTENTOXD,
	"FTENTOXF",	LTYPE1, AFTENTOXF,
	"FTENTOXL",	LTYPE1, AFTENTOXL,
	"FTENTOXW",	LTYPE1, AFTENTOXW,
	"FTSTB",	LTYPE1, AFTSTB,
	"FTSTD",	LTYPE1, AFTSTD,
	"FTSTF",	LTYPE1, AFTSTF,
	"FTSTL",	LTYPE1, AFTSTL,
	"FTSTW",	LTYPE1, AFTSTW,
	"FTWOTOXB",	LTYPE1, AFTWOTOXB,
	"FTWOTOXD",	LTYPE1, AFTWOTOXD,
	"FTWOTOXF",	LTYPE1, AFTWOTOXF,
	"FTWOTOXL",	LTYPE1, AFTWOTOXL,
	"FTWOTOXW",	LTYPE1, AFTWOTOXW,
	"FMOVEM",	LTYPE1, AFMOVEM,
	"FMOVEMC",	LTYPE1, AFMOVEMC,
	"FRESTORE",	LTYPE3, AFRESTORE,
	"FSAVE",	LTYPE3, AFSAVE,
	"GLOBL",	LTYPE1, AGLOBL,
	"GOK",		LTYPE2, AGOK,
	"HISTORY",	LTYPE2, AHISTORY,
	"ILLEG",	LTYPE2, AILLEG,
	"INSTR",	LTYPE3, AINSTR,
	"JMP",		LTYPE3, AJMP,
	"JSR",		LTYPE3, AJSR,
	"LEA",		LTYPE1, ALEA,
	"LINKL",	LTYPE1, ALINKL,
	"LINKW",	LTYPE1, ALINKW,
	"LOCATE",	LTYPE1, ALOCATE,
	"LONG",		LTYPE3, ALONG,
	"LSLB",		LTYPE1, ALSLB,
	"LSLL",		LTYPE1, ALSLL,
	"LSLW",		LTYPE1, ALSLW,
	"LSRB",		LTYPE1, ALSRB,
	"LSRL",		LTYPE1, ALSRL,
	"LSRW",		LTYPE1, ALSRW,
	"MOVB",		LTYPE1, AMOVB,
	"MOVEM",	LTYPE1, AMOVEM,
	"MOVEPL",	LTYPE1, AMOVEPL,
	"MOVEPW",	LTYPE1, AMOVEPW,
	"MOVESB",	LTYPE1, AMOVESB,
	"MOVESL",	LTYPE1, AMOVESL,
	"MOVESW",	LTYPE1, AMOVESW,
	"MOVL",		LTYPE1, AMOVL,
	"MOVW",		LTYPE1, AMOVW,
	"MULSL",	LTYPE1, AMULSL,
	"MULSW",	LTYPE1, AMULSW,
	"MULUL",	LTYPE1, AMULUL,
	"MULUW",	LTYPE1, AMULUW,
	"NAME",		LTYPE1, ANAME,
	"NBCD",		LTYPE3, ANBCD,
	"NEGB",		LTYPE3, ANEGB,
	"NEGL",		LTYPE3, ANEGL,
	"NEGW",		LTYPE3, ANEGW,
	"NEGXB",	LTYPE3, ANEGXB,
	"NEGXL",	LTYPE3, ANEGXL,
	"NEGXW",	LTYPE3, ANEGXW,
	"NOP",		LTYPE9, ANOP,
	"NOTB",		LTYPE3, ANOTB,
	"NOTL",		LTYPE3, ANOTL,
	"NOTW",		LTYPE3, ANOTW,
	"ORB",		LTYPE1, AORB,
	"ORL",		LTYPE1, AORL,
	"ORW",		LTYPE1, AORW,
	"PACK",		LTYPE1, APACK,
	"PEA",		LTYPE3, APEA,
	"RESET",	LTYPE2, ARESET,
	"ROTLB",	LTYPE1, AROTLB,
	"ROTLL",	LTYPE1, AROTLL,
	"ROTLW",	LTYPE1, AROTLW,
	"ROTRB",	LTYPE1, AROTRB,
	"ROTRL",	LTYPE1, AROTRL,
	"ROTRW",	LTYPE1, AROTRW,
	"ROXLB",	LTYPE1, AROXLB,
	"ROXLL",	LTYPE1, AROXLL,
	"ROXLW",	LTYPE1, AROXLW,
	"ROXRB",	LTYPE1, AROXRB,
	"ROXRL",	LTYPE1, AROXRL,
	"ROXRW",	LTYPE1, AROXRW,
	"RTD",		LTYPE3, ARTD,
	"RTE",		LTYPE2, ARTE,
	"RTM",		LTYPE3, ARTM,
	"RTR",		LTYPE2, ARTR,
	"RTS",		LTYPE2, ARTS,
	"SBCD",		LTYPE1, ASBCD,
	"SCC",		LTYPE3, ASCC,
	"SCS",		LTYPE3, ASCS,
	"SEQ",		LTYPE3, ASEQ,
	"SF",		LTYPE3, ASF,
	"SGE",		LTYPE3, ASGE,
	"SGT",		LTYPE3, ASGT,
	"SHI",		LTYPE3, ASHI,
	"SLE",		LTYPE3, ASLE,
	"SLS",		LTYPE3, ASLS,
	"SLT",		LTYPE3, ASLT,
	"SMI",		LTYPE3, ASMI,
	"SNE",		LTYPE3, ASNE,
	"SPL",		LTYPE3, ASPL,
	"ST",		LTYPE3, AST,
	"STOP",		LTYPE3, ASTOP,
	"SUBB",		LTYPE1, ASUBB,
	"SUBL",		LTYPE1, ASUBL,
	"SUBW",		LTYPE1, ASUBW,
	"SUBXB",	LTYPE1, ASUBXB,
	"SUBXL",	LTYPE1, ASUBXL,
	"SUBXW",	LTYPE1, ASUBXW,
	"SVC",		LTYPE2, ASVC,
	"SVS",		LTYPE2, ASVS,
	"SWAP",		LTYPE3, ASWAP,
	"SYS",		LTYPE2, ASYS,
	"TAS",		LTYPE3, ATAS,
	"TEXT",		LTYPEB, ATEXT,
	"TRAP",		LTYPE3, ATRAP,
	"TRAPCC",	LTYPE2, ATRAPCC,
	"TRAPCS",	LTYPE2, ATRAPCS,
	"TRAPEQ",	LTYPE2, ATRAPEQ,
	"TRAPF",	LTYPE2, ATRAPF,
	"TRAPGE",	LTYPE2, ATRAPGE,
	"TRAPGT",	LTYPE2, ATRAPGT,
	"TRAPHI",	LTYPE2, ATRAPHI,
	"TRAPLE",	LTYPE2, ATRAPLE,
	"TRAPLS",	LTYPE2, ATRAPLS,
	"TRAPLT",	LTYPE2, ATRAPLT,
	"TRAPMI",	LTYPE2, ATRAPMI,
	"TRAPNE",	LTYPE2, ATRAPNE,
	"TRAPPL",	LTYPE2, ATRAPPL,
	"TRAPT",	LTYPE2, ATRAPT,
	"TRAPV",	LTYPE2, ATRAPV,
	"TRAPVC",	LTYPE2, ATRAPVC,
	"TRAPVS",	LTYPE2, ATRAPVS,
	"TSTB",		LTYPE3, ATSTB,
	"TSTL",		LTYPE3, ATSTL,
	"TSTW",		LTYPE3, ATSTW,
	"UNLK",		LTYPE3, AUNLK,
	"UNPK",		LTYPE1, AUNPK,
	"WORD",		LTYPE3, AWORD,

	0
};

void
cinit(void)
{
	Sym *s;
	int i;

	nullgen.sym = S;
	nullgen.offset = 0;
	nullgen.type = D_NONE;
	if(FPCHIP)
		nullgen.dval = 0;
	for(i=0; i<sizeof(nullgen.sval); i++)
		nullgen.sval[i] = 0;
	nullgen.displace = 0;
	nullgen.type = D_NONE;
	nullgen.index = D_NONE;
	nullgen.scale = 0;
	nullgen.field = 0;

	nerrors = 0;
	iostack = I;
	iofree = I;
	peekc = IGN;
	nhunk = 0;
	for(i=0; i<NHASH; i++)
		hash[i] = S;
	for(i=0; itab[i].name; i++) {
		s = slookup(itab[i].name);
		s->type = itab[i].type;
		s->value = itab[i].value;
	}

	pathname = allocn(pathname, 0, 100);
	if(mygetwd(pathname, 99) == 0) {
		pathname = allocn(pathname, 100, 900);
		if(mygetwd(pathname, 999) == 0)
			strcpy(pathname, "/???");
	}
}

void
syminit(Sym *s)
{

	s->type = LNAME;
	s->value = 0;
}

void
cclean(void)
{
	Gen2 g2;

	g2.from = nullgen;
	g2.to = nullgen;
	outcode(AEND, &g2);
	Bflush(&obuf);
}

void
zname(char *n, int t, int s)
{

	Bputc(&obuf, ANAME);	/* as */
	Bputc(&obuf, ANAME>>8);
	Bputc(&obuf, t);		/* type */
	Bputc(&obuf, s);		/* sym */
	while(*n) {
		Bputc(&obuf, *n);
		n++;
	}
	Bputc(&obuf, 0);
}

void
zaddr(Gen *a, int s)
{
	long l;
	int i, t;
	char *n;
	Ieee e;

	t = 0;
	if(a->field)
		t |= T_FIELD;
	if(a->index != D_NONE || a->displace != 0)
		t |= T_INDEX;
	if(a->offset != 0)
		t |= T_OFFSET;
	if(s != 0)
		t |= T_SYM;

	if(a->type == D_FCONST)
		t |= T_FCONST;
	else
	if(a->type == D_SCONST)
		t |= T_SCONST;
	else
	if(a->type & ~0xff)
		t |= T_TYPE;
	Bputc(&obuf, t);

	if(t & T_FIELD) {	/* implies field */
		i = a->field;
		Bputc(&obuf, i);
		Bputc(&obuf, i>>8);
	}
	if(t & T_INDEX) {	/* implies index, scale, displace */
		i = a->index;
		Bputc(&obuf, i);
		Bputc(&obuf, i>>8);
		Bputc(&obuf, a->scale);
		l = a->displace;
		Bputc(&obuf, l);
		Bputc(&obuf, l>>8);
		Bputc(&obuf, l>>16);
		Bputc(&obuf, l>>24);
	}
	if(t & T_OFFSET) {	/* implies offset */
		l = a->offset;
		Bputc(&obuf, l);
		Bputc(&obuf, l>>8);
		Bputc(&obuf, l>>16);
		Bputc(&obuf, l>>24);
	}
	if(t & T_SYM)		/* implies sym */
		Bputc(&obuf, s);
	if(t & T_FCONST) {
		ieeedtod(&e, a->dval);
		l = e.l;
		Bputc(&obuf, l);
		Bputc(&obuf, l>>8);
		Bputc(&obuf, l>>16);
		Bputc(&obuf, l>>24);
		l = e.h;
		Bputc(&obuf, l);
		Bputc(&obuf, l>>8);
		Bputc(&obuf, l>>16);
		Bputc(&obuf, l>>24);
		return;
	}
	if(t & T_SCONST) {
		n = a->sval;
		for(i=0; i<NSNAME; i++) {
			Bputc(&obuf, *n);
			n++;
		}
		return;
	}
	i = a->type;
	Bputc(&obuf, i);
	if(t & T_TYPE)
		Bputc(&obuf, i>>8);
}

void
outcode(int a, Gen2 *g2)
{
	int sf, st, t;
	Sym *s;

	if(pass == 1)
		goto out;

jackpot:
	sf = 0;
	s = g2->from.sym;
	while(s != S) {
		sf = s->sym;
		if(sf < 0 || sf >= NSYM)
			sf = 0;
		t = g2->from.type & D_MASK;
		if(h[sf].type == t)
		if(h[sf].sym == s)
			break;
		zname(s->name, t, sym);
		s->sym = sym;
		h[sym].sym = s;
		h[sym].type = t;
		sf = sym;
		sym++;
		if(sym >= NSYM)
			sym = 1;
		break;
	}
	st = 0;
	s = g2->to.sym;
	while(s != S) {
		st = s->sym;
		if(st < 0 || st >= NSYM)
			st = 0;
		t = g2->to.type & D_MASK;
		if(h[st].type == t)
		if(h[st].sym == s)
			break;
		zname(s->name, t, sym);
		s->sym = sym;
		h[sym].sym = s;
		h[sym].type = t;
		st = sym;
		sym++;
		if(sym >= NSYM)
			sym = 1;
		if(st == sf)
			goto jackpot;
		break;
	}
	Bputc(&obuf, a);
	Bputc(&obuf, a>>8);
	Bputc(&obuf, lineno);
	Bputc(&obuf, lineno>>8);
	Bputc(&obuf, lineno>>16);
	Bputc(&obuf, lineno>>24);
	zaddr(&g2->from, sf);
	zaddr(&g2->to, st);

out:
	if(a != AGLOBL && a != ADATA)
		pc++;
}

void
outhist(void)
{
	Gen g;
	Hist *h;
	char *p, *q, *op, c;
	int n;

	g = nullgen;
	c = pathchar();
	for(h = hist; h != H; h = h->link) {
		p = h->name;
		op = 0;
		/* on windows skip drive specifier in pathname */
		if(systemtype(Windows) && p && p[1] == ':'){
			p += 2;
			c = *p;
		}
		if(p && p[0] != c && h->offset == 0 && pathname){
			/* on windows skip drive specifier in pathname */
			if(systemtype(Windows) && pathname[1] == ':') {
				op = p;
				p = pathname+2;
				c = *p;
			} else if(pathname[0] == c){
				op = p;
				p = pathname;
			}
		}
		while(p) {
			q = strchr(p, c);
			if(q) {
				n = q-p;
				if(n == 0){
					n = 1;	/* leading "/" */
					*p = '/';	/* don't emit "\" on windows */
				}
				q++;
			} else {
				n = strlen(p);
				q = 0;
			}
			if(n) {
				Bputc(&obuf, ANAME);
				Bputc(&obuf, ANAME>>8);
				Bputc(&obuf, D_FILE);	/* type */
				Bputc(&obuf, 1);	/* sym */
				Bputc(&obuf, '<');
				Bwrite(&obuf, p, n);
				Bputc(&obuf, 0);
			}
			p = q;
			if(p == 0 && op) {
				p = op;
				op = 0;
			}
		}
		g.offset = h->offset;

		Bputc(&obuf, AHISTORY);
		Bputc(&obuf, AHISTORY>>8);
		Bputc(&obuf, h->line);
		Bputc(&obuf, h->line>>8);
		Bputc(&obuf, h->line>>16);
		Bputc(&obuf, h->line>>24);
		zaddr(&nullgen, 0);
		zaddr(&g, 0);
	}
}

#include "../cc/lexbody"
#include "../cc/macbody"
#include "../cc/compat"