5l, 6l, 8l: introduce sub-symbols

Sub-symbols are laid out inside a larger symbol
but can be addressed directly.

Use to make Mach-O pointer array not a special case.

Will use later to describe ELF sections.

Glimpses of the beginning of ELF loading.

R=ken2
CC=golang-dev
https://golang.org/cl/2623043

7 files changed, 67 insertions, 48 deletions

diff --git a/src/cmd/5l/l.h b/src/cmd/5l/l.h
index 86885b421a..3a54e20f8f 100644
--- a/src/cmd/5l/l.h
+++ b/src/cmd/5l/l.h
@@ -56,11 +56,9 @@ typedef	struct	Optab	Optab;
 typedef	struct	Oprang	Oprang;
 typedef	uchar	Opcross[32][2][32];
 typedef	struct	Count	Count;
-typedef	struct	Use	Use;
 
 #define	P		((Prog*)0)
 #define	S		((Sym*)0)
-#define	U		((Use*)0)
 #define	TNAME		(cursym?cursym->name:noname)
 
 struct	Adr
@@ -139,9 +137,9 @@ struct	Sym
 	uchar	thumb;	// thumb code
 	uchar	foreign;	// called by arm if thumb, by thumb if arm
 	uchar	fnptr;	// used as fn ptr
-	Use*		use;
 	Sym*	hash;	// in hash table
 	Sym*	next;	// in text or data list
+	Sym*	sub;
 	Sym*	gotype;
 	char*	file;
 	char*	dynimpname;
@@ -191,12 +189,6 @@ struct	Count
 	int32	count;
 	int32	outof;
 };
-struct	Use
-{
-	Prog*	p;	/* use */
-	Prog*	ct;	/* curtext */
-	Use*		link;
-};
 
 enum
 {
@@ -213,6 +205,8 @@ enum
 	SFILE,
 	SCONST,
 
+	SSUB	= 1<<8,
+
 	LFROM		= 1<<0,
 	LTO		= 1<<1,
 	LPOOL		= 1<<2,
diff --git a/src/cmd/6l/l.h b/src/cmd/6l/l.h
index 4f56fe983f..39b3866d2d 100644
--- a/src/cmd/6l/l.h
+++ b/src/cmd/6l/l.h
@@ -131,6 +131,7 @@ struct	Sym
 	int32	sig;
 	Sym*	hash;	// in hash table
 	Sym*	next;	// in text or data list
+	Sym*	sub;	// in SSUB list
 	vlong	value;
 	vlong	size;
 	Sym*	gotype;
@@ -175,12 +176,13 @@ enum
 	SELFDATA,
 	SRODATA,
 	SDATA,
+	SMACHO,
 	SBSS,
 
 	SXREF,
-	SMACHO,
 	SFILE,
 	SCONST,
+	SSUB	= 1<<8,
 
 	NHASH		= 10007,
 	NHUNK		= 100000,
@@ -354,7 +356,6 @@ EXTERN	Sym*	fromgotype;	// type symbol on last p->from read
 EXTERN	vlong	textstksiz;
 EXTERN	vlong	textarg;
 extern	char	thechar;
-EXTERN	int	dynptrsize;
 EXTERN	int	elfstrsize;
 EXTERN	char*	elfstrdat;
 EXTERN	int	elftextsh;
diff --git a/src/cmd/6l/span.c b/src/cmd/6l/span.c
index 605a1a627e..6cb3586f5d 100644
--- a/src/cmd/6l/span.c
+++ b/src/cmd/6l/span.c
@@ -46,6 +46,9 @@ span1(Sym *s)
 	int n, m, i;
 
 	cursym = s;
+	
+	if(s->p != nil)
+		return;
 
 	for(p = s->text; p != P; p = p->link) {
 		p->back = 2;	// use short branches first time through
@@ -145,6 +148,8 @@ span(void)
 	// NOTE(rsc): If we get rid of the globals we should
 	// be able to parallelize these iterations.
 	for(cursym = textp; cursym != nil; cursym = cursym->next) {
+		if(cursym->p != nil)
+			continue;
 		// TODO: move into span1
 		for(p = cursym->text; p != P; p = p->link) {
 			n = 0;
@@ -671,15 +676,9 @@ relput8(Prog *p, Adr *a)
 vlong
 symaddr(Sym *s)
 {
-	switch(s->type) {
-	case SMACHO:
-		return segdata.vaddr + segdata.filelen - dynptrsize + s->value;
-
-	default:
-		if(!s->reachable)
-			diag("unreachable symbol in symaddr - %s", s->name);
-		return s->value;
-	}
+	if(!s->reachable)
+		diag("unreachable symbol in symaddr - %s", s->name);
+	return s->value;
 }
 
 static vlong
diff --git a/src/cmd/8l/l.h b/src/cmd/8l/l.h
index 0049c3f1f3..fd0b642448 100644
--- a/src/cmd/8l/l.h
+++ b/src/cmd/8l/l.h
@@ -132,6 +132,7 @@ struct	Sym
 	int32	sig;
 	Sym*	hash;	// in hash table
 	Sym*	next;	// in text or data list
+	Sym*	sub;	// in sub list
 	Sym*	gotype;
 	char*	file;
 	char*	dynimpname;
@@ -164,14 +165,16 @@ enum
 	/* order here is order in output file */
 	STEXT,
 	SELFDATA,
-	SRODATA,	// TODO(rsc): move
+	SRODATA,
 	SDATA,
+	SMACHO,	/* Mach-O __nl_symbol_ptr */
 	SBSS,
 
 	SXREF,
-	SMACHO,	// TODO(rsc): maybe move between DATA1 and BSS?
 	SFILE,
 	SCONST,
+	
+	SSUB = 1<<8,	/* sub-symbol, linked from parent via ->sub list */
 
 	NHASH		= 10007,
 	NHUNK		= 100000,
@@ -287,7 +290,6 @@ EXTERN	Prog*	curp;
 EXTERN	Sym*	cursym;
 EXTERN	Sym*	datap;
 EXTERN	int32	elfdatsize;
-EXTERN	int32	dynptrsize;
 EXTERN	char	debug[128];
 EXTERN	char	literal[32];
 EXTERN	Sym*	etextp;
diff --git a/src/cmd/8l/span.c b/src/cmd/8l/span.c
index 3aab95874a..7f083eebde 100644
--- a/src/cmd/8l/span.c
+++ b/src/cmd/8l/span.c
@@ -503,15 +503,9 @@ relput4(Prog *p, Adr *a)
 int32
 symaddr(Sym *s)
 {
-	switch(s->type) {
-	case SMACHO:
-		return segdata.vaddr + segdata.filelen - dynptrsize + s->value;
-	
-	default:
-		if(!s->reachable)
-			diag("unreachable symbol in symaddr - %s", s->name);
-		return s->value;
-	}
+	if(!s->reachable)
+		diag("unreachable symbol in symaddr - %s", s->name);
+	return s->value;
 }
 
 static int32
diff --git a/src/cmd/ld/data.c b/src/cmd/ld/data.c
index 461a39950b..55925f15c3 100644
--- a/src/cmd/ld/data.c
+++ b/src/cmd/ld/data.c
@@ -393,6 +393,23 @@ codeblk(int32 addr, int32 size)
 			Bprint(&bso, "\n");
 		}
 		p = sym->text;
+		if(p == nil) {
+			Bprint(&bso, "%.6llux\t%-20s | foreign text\n", (vlong)addr, sym->name);
+			n = sym->size;
+			q = sym->p;
+			
+			while(n >= 16) {
+				Bprint(&bso, "%.6ux\t%-20.16I\n",  addr, q);
+				addr += 16;
+				q += 16;
+				n -= 16;
+			}
+			if(n > 0)
+				Bprint(&bso, "%.6ux\t%-20.*I\n", addr, n, q);
+			addr += n;
+			continue;
+		}
+			
 		Bprint(&bso, "%.6llux\t%-20s | %P\n", (vlong)addr, sym->name, p);
 		for(p = p->link; p != P; p = p->link) {
 			if(p->link != P)
@@ -671,7 +688,6 @@ dodata(void)
 		datsize += t;
 	}
 	sect->len = datsize - sect->vaddr;
-	datsize += dynptrsize;
 
 	/* bss */
 	sect = addsection(&segdata, ".bss", 06);
@@ -702,7 +718,7 @@ void
 address(void)
 {
 	Section *s, *text, *data, *rodata, *bss;
-	Sym *sym;
+	Sym *sym, *sub;
 	uvlong va;
 
 	va = INITTEXT;
@@ -723,11 +739,9 @@ address(void)
 	for(s=segdata.sect; s != nil; s=s->next) {
 		s->vaddr = va;
 		va += s->len;
-		if(s == segdata.sect)
-			va += dynptrsize;
 		segdata.len = va - segdata.vaddr;
 	}
-	segdata.filelen = segdata.sect->len + dynptrsize;	// assume .data is first
+	segdata.filelen = segdata.sect->len;	// assume .data is first
 	
 	text = segtext.sect;
 	rodata = segtext.sect->next;
@@ -740,6 +754,8 @@ address(void)
 			sym->value += rodata->vaddr;
 		else
 			sym->value += data->vaddr;
+		for(sub = sym->sub; sub != nil; sub = sub->sub)
+			sub->value += sym->value;
 	}
 	
 	xdefine("text", STEXT, text->vaddr);
diff --git a/src/cmd/ld/macho.c b/src/cmd/ld/macho.c
index 75c3ad865a..67264233c7 100644
--- a/src/cmd/ld/macho.c
+++ b/src/cmd/ld/macho.c
@@ -281,7 +281,7 @@ domacho(void)
 	char *p;
 	uchar *dat;
 	uint32 x;
-	Sym *s;
+	Sym *s, *smacho;
 	Sym **impsym;
 
 	ptrsize = 4;
@@ -352,10 +352,16 @@ domacho(void)
 		}
 	}
 
+	smacho = lookup("__nl_symbol_ptr", 0);
+	smacho->type = SMACHO;
+	smacho->reachable = 1;
 	for(h=0; h<nimpsym; h++) {
 		s = impsym[h];
-		s->type = SMACHO;
+		s->type = SMACHO | SSUB;
+		s->sub = smacho->sub;
+		smacho->sub = s;
 		s->value = (nexpsym+h) * ptrsize;
+		s->reachable = 1;
 
 		/* symbol table entry - darwin still puts _ prefixes on all C symbols */
 		x = nstrtab;
@@ -398,7 +404,9 @@ domacho(void)
 		dat[3] = x>>24;
 	}
 
-	dynptrsize = (nexpsym+nimpsym) * ptrsize;
+	smacho->size = (nexpsym+nimpsym) * ptrsize;
+	if(smacho->size == 0)
+		smacho->reachable = 0;
 }
 
 vlong
@@ -408,10 +416,13 @@ domacholink(void)
 	uchar *p;
 	Sym *s;
 	uint64 val;
+	Sym *smacho;
+	
+	smacho = lookup("__nl_symbol_ptr", 0);
 
 	linkoff = 0;
 	if(nlinkdata > 0 || nstrtab > 0) {
-		linkoff = rnd(HEADR+segtext.len, INITRND) + rnd(segdata.filelen - dynptrsize, INITRND);
+		linkoff = rnd(HEADR+segtext.len, INITRND) + rnd(segdata.filelen - smacho->size, INITRND);
 		seek(cout, linkoff, 0);
 
 		for(i = 0; i<nexpsym; ++i) {
@@ -452,6 +463,7 @@ asmbmacho(void)
 	MachoSeg *ms;
 	MachoDebug *md;
 	MachoLoad *ml;
+	Sym *smacho;
 
 	/* apple MACH */
 	va = INITTEXT - HEADR;
@@ -492,8 +504,9 @@ asmbmacho(void)
 	msect->flag = 0x400;	/* flag - some instructions */
 
 	/* data */
+	smacho = lookup("__nl_symbol_ptr", 0);
 	w = segdata.len;
-	ms = newMachoSeg("__DATA", 2+(dynptrsize>0));
+	ms = newMachoSeg("__DATA", 2+(smacho->size > 0));
 	ms->vaddr = va+v;
 	ms->vsize = w;
 	ms->fileoffset = v;
@@ -503,14 +516,14 @@ asmbmacho(void)
 
 	msect = newMachoSect(ms, "__data");
 	msect->addr = va+v;
-	msect->size = segdata.filelen - dynptrsize;
+	msect->size = segdata.filelen - smacho->size;
 	msect->off = v;
 
-	if(dynptrsize > 0) {
+	if(smacho->size > 0) {
 		msect = newMachoSect(ms, "__nl_symbol_ptr");
-		msect->addr = va+v+segdata.filelen - dynptrsize;
-		msect->size = dynptrsize;
-		msect->off = v+segdata.filelen - dynptrsize;
+		msect->addr = smacho->value;
+		msect->size = smacho->size;
+		msect->off = datoff(msect->addr);
 		msect->align = 2;
 		msect->flag = 6;	/* section with nonlazy symbol pointers */
 		/*
@@ -551,7 +564,7 @@ asmbmacho(void)
 	if(!debug['d']) {
 		int nsym;
 
-		nsym = dynptrsize/ptrsize;
+		nsym = smacho->size/ptrsize;
 
 		ms = newMachoSeg("__LINKEDIT", 0);
 		ms->vaddr = va+v+rnd(segdata.len, INITRND);