obj.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	"l.h"
#include	<ar.h>

#ifndef	DEFAULT
#define	DEFAULT	'9'
#endif

char	*noname		= "<none>";
char	symname[]	= SYMDEF;
char	thechar		= '8';
char	*thestring 	= "386";

/*
 *	-H0 -T0x40004C -D0x10000000	is garbage unix
 *	-H1 -T0xd0 -R4			is unix coff
 *	-H2 -T4128 -R4096		is plan9 format
 *	-H3 -Tx -Rx			is MS-DOS .COM
 *	-H4 -Tx -Rx			is fake MS-DOS .EXE
 *	-H5 -T0x80100020 -R4096		is ELF
 */

static int
isobjfile(char *f)
{
	int n, v;
	Biobuf *b;
	char buf1[5], buf2[SARMAG];

	b = Bopen(f, OREAD);
	if(b == nil)
		return 0;
	n = Bread(b, buf1, 5);
	if(n == 5 && (buf1[2] == 1 && buf1[3] == '<' || buf1[3] == 1 && buf1[4] == '<'))
		v = 1;	/* good enough for our purposes */
	else{
		Bseek(b, 0, 0);
		n = Bread(b, buf2, SARMAG);
		v = n == SARMAG && strncmp(buf2, ARMAG, SARMAG) == 0;
	}
	Bterm(b);
	return v;
}

void
main(int argc, char *argv[])
{
	int i, c;
	char *a;

	Binit(&bso, 1, OWRITE);
	cout = -1;
	listinit();
	memset(debug, 0, sizeof(debug));
	nerrors = 0;
	outfile = "8.out";
	HEADTYPE = -1;
	INITTEXT = -1;
	INITDAT = -1;
	INITRND = -1;
	INITENTRY = 0;
	ARGBEGIN {
	default:
		c = ARGC();
		if(c >= 0 && c < sizeof(debug))
			debug[c]++;
		break;
	case 'o': /* output to (next arg) */
		outfile = ARGF();
		break;
	case 'E':
		a = ARGF();
		if(a)
			INITENTRY = a;
		break;
	case 'H':
		a = ARGF();
		if(a)
			HEADTYPE = atolwhex(a);
		break;
	case 'T':
		a = ARGF();
		if(a)
			INITTEXT = atolwhex(a);
		break;
	case 'D':
		a = ARGF();
		if(a)
			INITDAT = atolwhex(a);
		break;
	case 'R':
		a = ARGF();
		if(a)
			INITRND = atolwhex(a);
		break;
	case 'x':	/* produce export table */
		doexp = 1;
		if(argv[1] != nil && argv[1][0] != '-' && !isobjfile(argv[1]))
			readundefs(ARGF(), SEXPORT);
		break;
	case 'u':	/* produce dynamically loadable module */
		dlm = 1;
		debug['l']++;
		if(argv[1] != nil && argv[1][0] != '-' && !isobjfile(argv[1]))
			readundefs(ARGF(), SIMPORT);
		break;
	} ARGEND
	USED(argc);
	if(*argv == 0) {
		diag("usage: 8l [-options] objects");
		errorexit();
	}
	if(!debug['9'] && !debug['U'] && !debug['B'])
		debug[DEFAULT] = 1;
	if(HEADTYPE == -1) {
		if(debug['U'])
			HEADTYPE = 1;
		if(debug['B'])
			HEADTYPE = 2;
		if(debug['9'])
			HEADTYPE = 2;
	}
	switch(HEADTYPE) {
	default:
		diag("unknown -H option");
		errorexit();

	case 0:	/* this is garbage */
		HEADR = 20L+56L;
		if(INITTEXT == -1)
			INITTEXT = 0x40004CL;
		if(INITDAT == -1)
			INITDAT = 0x10000000L;
		if(INITRND == -1)
			INITRND = 0;
		break;
	case 1:	/* is unix coff */
		HEADR = 0xd0L;
		if(INITTEXT == -1)
			INITTEXT = 0xd0;
		if(INITDAT == -1)
			INITDAT = 0x400000;
		if(INITRND == -1)
			INITRND = 0;
		break;
	case 2:	/* plan 9 */
		HEADR = 32L;
		if(INITTEXT == -1)
			INITTEXT = 4096+32;
		if(INITDAT == -1)
			INITDAT = 0;
		if(INITRND == -1)
			INITRND = 4096;
		break;
	case 3:	/* MS-DOS .COM */
		HEADR = 0;
		if(INITTEXT == -1)
			INITTEXT = 0x0100;
		if(INITDAT == -1)
			INITDAT = 0;
		if(INITRND == -1)
			INITRND = 4;
		break;
	case 4:	/* fake MS-DOS .EXE */
		HEADR = 0x200;
		if(INITTEXT == -1)
			INITTEXT = 0x0100;
		if(INITDAT == -1)
			INITDAT = 0;
		if(INITRND == -1)
			INITRND = 4;
		HEADR += (INITTEXT & 0xFFFF);
		if(debug['v'])
			Bprint(&bso, "HEADR = 0x%ld\n", HEADR);
		break;
	case 5:	/* elf executable */
		HEADR = rnd(52L+3*32L, 16);
		if(INITTEXT == -1)
			INITTEXT = 0x80100020L;
		if(INITDAT == -1)
			INITDAT = 0;
		if(INITRND == -1)
			INITRND = 4096;
		break;
	}
	if(INITDAT != 0 && INITRND != 0)
		print("warning: -D0x%lux is ignored because of -R0x%lux\n",
			INITDAT, INITRND);
	if(debug['v'])
		Bprint(&bso, "HEADER = -H0x%ld -T0x%lux -D0x%lux -R0x%lux\n",
			HEADTYPE, INITTEXT, INITDAT, INITRND);
	Bflush(&bso);
	for(i=1; optab[i].as; i++)
		if(i != optab[i].as) {
			diag("phase error in optab: %d", i);
			errorexit();
		}

	for(i=0; i<Ymax; i++)
		ycover[i*Ymax + i] = 1;

	ycover[Yi0*Ymax + Yi8] = 1;
	ycover[Yi1*Ymax + Yi8] = 1;

	ycover[Yi0*Ymax + Yi32] = 1;
	ycover[Yi1*Ymax + Yi32] = 1;
	ycover[Yi8*Ymax + Yi32] = 1;

	ycover[Yal*Ymax + Yrb] = 1;
	ycover[Ycl*Ymax + Yrb] = 1;
	ycover[Yax*Ymax + Yrb] = 1;
	ycover[Ycx*Ymax + Yrb] = 1;
	ycover[Yrx*Ymax + Yrb] = 1;

	ycover[Yax*Ymax + Yrx] = 1;
	ycover[Ycx*Ymax + Yrx] = 1;

	ycover[Yax*Ymax + Yrl] = 1;
	ycover[Ycx*Ymax + Yrl] = 1;
	ycover[Yrx*Ymax + Yrl] = 1;

	ycover[Yf0*Ymax + Yrf] = 1;

	ycover[Yal*Ymax + Ymb] = 1;
	ycover[Ycl*Ymax + Ymb] = 1;
	ycover[Yax*Ymax + Ymb] = 1;
	ycover[Ycx*Ymax + Ymb] = 1;
	ycover[Yrx*Ymax + Ymb] = 1;
	ycover[Yrb*Ymax + Ymb] = 1;
	ycover[Ym*Ymax + Ymb] = 1;

	ycover[Yax*Ymax + Yml] = 1;
	ycover[Ycx*Ymax + Yml] = 1;
	ycover[Yrx*Ymax + Yml] = 1;
	ycover[Yrl*Ymax + Yml] = 1;
	ycover[Ym*Ymax + Yml] = 1;

	for(i=0; i<D_NONE; i++) {
		reg[i] = -1;
		if(i >= D_AL && i <= D_BH)
			reg[i] = (i-D_AL) & 7;
		if(i >= D_AX && i <= D_DI)
			reg[i] = (i-D_AX) & 7;
		if(i >= D_F0 && i <= D_F0+7)
			reg[i] = (i-D_F0) & 7;
	}

	zprg.link = P;
	zprg.pcond = P;
	zprg.back = 2;
	zprg.as = AGOK;
	zprg.from.type = D_NONE;
	zprg.from.index = D_NONE;
	zprg.from.scale = 1;
	zprg.to = zprg.from;

	pcstr = "%.6lux ";
	nuxiinit();
	histgen = 0;
	textp = P;
	datap = P;
	edatap = P;
	pc = 0;
	dtype = 4;
	cout = create(outfile, 1, 0775);
	if(cout < 0) {
		diag("cannot create %s", outfile);
		errorexit();
	}
	version = 0;
	cbp = buf.cbuf;
	cbc = sizeof(buf.cbuf);
	firstp = prg();
	lastp = firstp;

	if(INITENTRY == 0) {
		INITENTRY = "_main";
		if(debug['p'])
			INITENTRY = "_mainp";
		if(!debug['l'])
			lookup(INITENTRY, 0)->type = SXREF;
	} else
		lookup(INITENTRY, 0)->type = SXREF;

	while(*argv)
		objfile(*argv++);
	if(!debug['l'])
		loadlib();
	firstp = firstp->link;
	if(firstp == P)
		errorexit();
	if(doexp || dlm){
		EXPTAB = "_exporttab";
		zerosig(EXPTAB);
		zerosig("etext");
		zerosig("edata");
		zerosig("end");
		if(dlm){
			import();
			HEADTYPE = 2;
			INITTEXT = INITDAT = 0;
			INITRND = 8;
			INITENTRY = EXPTAB;
		}
		export();
	}
	patch();
	follow();
	dodata();
	dostkoff();
	if(debug['p'])
		if(debug['1'])
			doprof1();
		else
			doprof2();
	span();
	doinit();
	asmb();
	undef();
	if(debug['v']) {
		Bprint(&bso, "%5.2f cpu time\n", cputime());
		Bprint(&bso, "%ld symbols\n", nsymbol);
		Bprint(&bso, "%ld memory used\n", thunk);
		Bprint(&bso, "%d sizeof adr\n", sizeof(Adr));
		Bprint(&bso, "%d sizeof prog\n", sizeof(Prog));
	}
	Bflush(&bso);

	errorexit();
}

void
loadlib(void)
{
	int i;
	long h;
	Sym *s;

loop:
	xrefresolv = 0;
	for(i=0; i<libraryp; i++) {
		if(debug['v'])
			Bprint(&bso, "%5.2f autolib: %s (from %s)\n", cputime(), library[i], libraryobj[i]);
		objfile(library[i]);
	}
	if(xrefresolv)
	for(h=0; h<nelem(hash); h++)
	for(s = hash[h]; s != S; s = s->link)
		if(s->type == SXREF)
			goto loop;
}

void
errorexit(void)
{

	if(nerrors) {
		if(cout >= 0)
			remove(outfile);
		exits("error");
	}
	exits(0);
}

void
objfile(char *file)
{
	long off, esym, cnt, l;
	int f, work;
	Sym *s;
	char magbuf[SARMAG];
	char name[100], pname[150];
	struct ar_hdr arhdr;
	char *e, *start, *stop;

	if(file[0] == '-' && file[1] == 'l') {
		if(debug['9'])
			sprint(name, "/%s/lib/lib", thestring);
		else
			sprint(name, "/usr/%clib/lib", thechar);
		strcat(name, file+2);
		strcat(name, ".a");
		file = name;
	}
	if(debug['v'])
		Bprint(&bso, "%5.2f ldobj: %s\n", cputime(), file);
	Bflush(&bso);
	f = open(file, 0);
	if(f < 0) {
		diag("cannot open file: %s", file);
		errorexit();
	}
	l = read(f, magbuf, SARMAG);
	if(l != SARMAG || strncmp(magbuf, ARMAG, SARMAG)){
		/* load it as a regular file */
		l = seek(f, 0L, 2);
		seek(f, 0L, 0);
		ldobj(f, l, file);
		close(f);
		return;
	}

	l = read(f, &arhdr, SAR_HDR);
	if(l != SAR_HDR) {
		diag("%s: short read on archive file symbol header", file);
		goto out;
	}
	if(strncmp(arhdr.name, symname, strlen(symname))) {
		diag("%s: first entry not symbol header", file);
		goto out;
	}

	esym = SARMAG + SAR_HDR + atolwhex(arhdr.size);
	off = SARMAG + SAR_HDR;

	/*
	 * just bang the whole symbol file into memory
	 */
	seek(f, off, 0);
	cnt = esym - off;
	start = malloc(cnt + 10);
	cnt = read(f, start, cnt);
	if(cnt <= 0){
		close(f);
		return;
	}
	stop = &start[cnt];
	memset(stop, 0, 10);

	work = 1;
	while(work) {
		if(debug['v'])
			Bprint(&bso, "%5.2f library pass: %s\n", cputime(), file);
		Bflush(&bso);
		work = 0;
		for(e = start; e < stop; e = strchr(e+5, 0) + 1) {
			s = lookup(e+5, 0);
			if(s->type != SXREF)
				continue;
			sprint(pname, "%s(%s)", file, s->name);
			if(debug['v'])
				Bprint(&bso, "%5.2f library: %s\n", cputime(), pname);
			Bflush(&bso);
			l = e[1] & 0xff;
			l |= (e[2] & 0xff) << 8;
			l |= (e[3] & 0xff) << 16;
			l |= (e[4] & 0xff) << 24;
			seek(f, l, 0);
			l = read(f, &arhdr, SAR_HDR);
			if(l != SAR_HDR)
				goto bad;
			if(strncmp(arhdr.fmag, ARFMAG, sizeof(arhdr.fmag)))
				goto bad;
			l = atolwhex(arhdr.size);
			ldobj(f, l, pname);
			if(s->type == SXREF) {
				diag("%s: failed to load: %s", file, s->name);
				errorexit();
			}
			work = 1;
			xrefresolv = 1;
		}
	}
	return;

bad:
	diag("%s: bad or out of date archive", file);
out:
	close(f);
}

int
zaddr(uchar *p, Adr *a, Sym *h[])
{
	int c, t, i;
	long l;
	Sym *s;
	Auto *u;

	t = p[0];

	c = 1;
	if(t & T_INDEX) {
		a->index = p[c];
		a->scale = p[c+1];
		c += 2;
	} else {
		a->index = D_NONE;
		a->scale = 0;
	}
	a->offset = 0;
	if(t & T_OFFSET) {
		a->offset = p[c] | (p[c+1]<<8) | (p[c+2]<<16) | (p[c+3]<<24);
		c += 4;
	}
	a->sym = S;
	if(t & T_SYM) {
		a->sym = h[p[c]];
		c++;
	}
	a->type = D_NONE;
	if(t & T_FCONST) {
		a->ieee.l = p[c] | (p[c+1]<<8) | (p[c+2]<<16) | (p[c+3]<<24);
		a->ieee.h = p[c+4] | (p[c+5]<<8) | (p[c+6]<<16) | (p[c+7]<<24);
		c += 8;
		a->type = D_FCONST;
	} else
	if(t & T_SCONST) {
		for(i=0; i<NSNAME; i++)
			a->scon[i] = p[c+i];
		c += NSNAME;
		a->type = D_SCONST;
	}
	if(t & T_TYPE) {
		a->type = p[c];
		c++;
	}
	s = a->sym;
	if(s == S)
		return c;

	t = a->type;
	if(t != D_AUTO && t != D_PARAM)
		return c;
	l = a->offset;
	for(u=curauto; u; u=u->link) {
		if(u->asym == s)
		if(u->type == t) {
			if(u->aoffset > l)
				u->aoffset = l;
			return c;
		}
	}

	while(nhunk < sizeof(Auto))
		gethunk();
	u = (Auto*)hunk;
	nhunk -= sizeof(Auto);
	hunk += sizeof(Auto);

	u->link = curauto;
	curauto = u;
	u->asym = s;
	u->aoffset = l;
	u->type = t;
	return c;
}

void
addlib(char *obj)
{
	char name[1024], comp[256], *p;
	int i;

	if(histfrogp <= 0)
		return;

	if(histfrog[0]->name[1] == '/') {
		sprint(name, "");
		i = 1;
	} else
	if(histfrog[0]->name[1] == '.') {
		sprint(name, ".");
		i = 0;
	} else {
		if(debug['9'])
			sprint(name, "/%s/lib", thestring);
		else
			sprint(name, "/usr/%clib", thechar);
		i = 0;
	}

	for(; i<histfrogp; i++) {
		snprint(comp, sizeof comp, histfrog[i]->name+1);
		for(;;) {
			p = strstr(comp, "$O");
			if(p == 0)
				break;
			memmove(p+1, p+2, strlen(p+2)+1);
			p[0] = thechar;
		}
		for(;;) {
			p = strstr(comp, "$M");
			if(p == 0)
				break;
			if(strlen(comp)+strlen(thestring)-2+1 >= sizeof comp) {
				diag("library component too long");
				return;
			}
			memmove(p+strlen(thestring), p+2, strlen(p+2)+1);
			memmove(p, thestring, strlen(thestring));
		}
		if(strlen(name) + strlen(comp) + 3 >= sizeof(name)) {
			diag("library component too long");
			return;
		}
		strcat(name, "/");
		strcat(name, comp);
	}
	for(i=0; i<libraryp; i++)
		if(strcmp(name, library[i]) == 0)
			return;
	if(libraryp == nelem(library)){
		diag("too many autolibs; skipping %s", name);
		return;
	}

	p = malloc(strlen(name) + 1);
	strcpy(p, name);
	library[libraryp] = p;
	p = malloc(strlen(obj) + 1);
	strcpy(p, obj);
	libraryobj[libraryp] = p;
	libraryp++;
}

void
addhist(long line, int type)
{
	Auto *u;
	Sym *s;
	int i, j, k;

	u = malloc(sizeof(Auto));
	s = malloc(sizeof(Sym));
	s->name = malloc(2*(histfrogp+1) + 1);

	u->asym = s;
	u->type = type;
	u->aoffset = line;
	u->link = curhist;
	curhist = u;

	j = 1;
	for(i=0; i<histfrogp; i++) {
		k = histfrog[i]->value;
		s->name[j+0] = k>>8;
		s->name[j+1] = k;
		j += 2;
	}
}

void
histtoauto(void)
{
	Auto *l;

	while(l = curhist) {
		curhist = l->link;
		l->link = curauto;
		curauto = l;
	}
}

void
collapsefrog(Sym *s)
{
	int i;

	/*
	 * bad encoding of path components only allows
	 * MAXHIST components. if there is an overflow,
	 * first try to collapse xxx/..
	 */
	for(i=1; i<histfrogp; i++)
		if(strcmp(histfrog[i]->name+1, "..") == 0) {
			memmove(histfrog+i-1, histfrog+i+1,
				(histfrogp-i-1)*sizeof(histfrog[0]));
			histfrogp--;
			goto out;
		}

	/*
	 * next try to collapse .
	 */
	for(i=0; i<histfrogp; i++)
		if(strcmp(histfrog[i]->name+1, ".") == 0) {
			memmove(histfrog+i, histfrog+i+1,
				(histfrogp-i-1)*sizeof(histfrog[0]));
			goto out;
		}

	/*
	 * last chance, just truncate from front
	 */
	memmove(histfrog+0, histfrog+1,
		(histfrogp-1)*sizeof(histfrog[0]));

out:
	histfrog[histfrogp-1] = s;
}

void
nopout(Prog *p)
{
	p->as = ANOP;
	p->from.type = D_NONE;
	p->to.type = D_NONE;
}

uchar*
readsome(int f, uchar *buf, uchar *good, uchar *stop, int max)
{
	int n;

	n = stop - good;
	memmove(buf, good, stop - good);
	stop = buf + n;
	n = MAXIO - n;
	if(n > max)
		n = max;
	n = read(f, stop, n);
	if(n <= 0)
		return 0;
	return stop + n;
}

void
ldobj(int f, long c, char *pn)
{
	long ipc;
	Prog *p, *t;
	uchar *bloc, *bsize, *stop;
	int v, o, r, skip;
	Sym *h[NSYM], *s, *di;
	ulong sig;
	static int files;
	static char **filen;
	char **nfilen;

	if((files&15) == 0){
		nfilen = malloc((files+16)*sizeof(char*));
		memmove(nfilen, filen, files*sizeof(char*));
		free(filen);
		filen = nfilen;
	}
	filen[files++] = strdup(pn);

	bsize = buf.xbuf;
	bloc = buf.xbuf;
	di = S;

newloop:
	memset(h, 0, sizeof(h));
	version++;
	histfrogp = 0;
	ipc = pc;
	skip = 0;

loop:
	if(c <= 0)