cmd/ld: darwin support for host linking

R=ken2
CC=golang-dev
https://codereview.appspot.com/7626045

11 files changed, 341 insertions, 105 deletions

diff --git a/src/cmd/6l/asm.c b/src/cmd/6l/asm.c
index e1d114642..de6ea3a63 100644
--- a/src/cmd/6l/asm.c
+++ b/src/cmd/6l/asm.c
@@ -317,6 +317,63 @@ elfreloc1(Reloc *r, vlong sectoff)
 }
 
 int
+machoreloc1(Reloc *r, vlong sectoff)
+{
+	uint32 v;
+	Sym *rs;
+	
+	rs = r->xsym;
+
+	if(rs->type == SHOSTOBJ) {
+		if(rs->dynid < 0) {
+			diag("reloc %d to non-macho symbol %s type=%d", r->type, rs->name, rs->type);
+			return -1;
+		}
+		v = rs->dynid;			
+		v |= 1<<27; // external relocation
+	} else {
+		v = rs->sect->extnum;
+		if(v == 0) {
+			diag("reloc %d to symbol %s in non-macho section %s type=%d", r->type, rs->name, rs->sect->name, rs->type);
+			return -1;
+		}
+	}
+
+	switch(r->type) {
+	default:
+		return -1;
+	case D_ADDR:
+		v |= MACHO_X86_64_RELOC_UNSIGNED<<28;
+		break;
+	case D_PCREL:
+		v |= 1<<24; // pc-relative bit
+		v |= MACHO_X86_64_RELOC_BRANCH<<28;
+		break;
+	}
+	
+	switch(r->siz) {
+	default:
+		return -1;
+	case 1:
+		v |= 0<<25;
+		break;
+	case 2:
+		v |= 1<<25;
+		break;
+	case 4:
+		v |= 2<<25;
+		break;
+	case 8:
+		v |= 3<<25;
+		break;
+	}
+
+	LPUT(sectoff);
+	LPUT(v);
+	return 0;
+}
+
+int
 archreloc(Reloc *r, Sym *s, vlong *val)
 {
 	USED(r);
@@ -677,6 +734,10 @@ asmb(void)
 
 			dwarfemitdebugsections();
 			break;
+		case Hdarwin:
+			if(isobj)
+				machoemitreloc();
+			break;
 		}
 	}
 
diff --git a/src/cmd/6l/obj.c b/src/cmd/6l/obj.c
index 3a1f862ed..6ea88de27 100644
--- a/src/cmd/6l/obj.c
+++ b/src/cmd/6l/obj.c
@@ -141,8 +141,11 @@ main(int argc, char *argv[])
 		switch(HEADTYPE) {
 		default:
 			sysfatal("cannot use -hostobj with -H %s", headstr(HEADTYPE));
-		case Hlinux:
+		case Hdarwin:
 		case Hfreebsd:
+		case Hlinux:
+		case Hnetbsd:
+		case Hopenbsd:
 			break;
 		}
 	}
diff --git a/src/cmd/8l/asm.c b/src/cmd/8l/asm.c
index 9bd04ff1a..402360d70 100644
--- a/src/cmd/8l/asm.c
+++ b/src/cmd/8l/asm.c
@@ -299,6 +299,78 @@ elfreloc1(Reloc *r, vlong sectoff)
 	return 0;
 }
 
+int
+machoreloc1(Reloc *r, vlong sectoff)
+{
+	uint32 v;
+	Sym *rs;
+	
+	rs = r->xsym;
+
+	if(rs->type == SHOSTOBJ) {
+		if(rs->dynid < 0) {
+			diag("reloc %d to non-macho symbol %s type=%d", r->type, rs->name, rs->type);
+			return -1;
+		}
+		v = rs->dynid;			
+		v |= 1<<27; // external relocation
+	} else {
+		v = rs->sect->extnum;
+		if(v == 0) {
+			diag("reloc %d to symbol %s in non-macho section %s type=%d", r->type, rs->name, rs->sect->name, rs->type);
+			return -1;
+		}
+	}
+
+	switch(r->type) {
+	default:
+		return -1;
+	case D_ADDR:
+		v |= MACHO_GENERIC_RELOC_VANILLA<<28;
+		break;
+	case D_PCREL:
+		v |= 1<<24; // pc-relative bit
+		v |= MACHO_GENERIC_RELOC_VANILLA<<28;
+		break;
+	}
+	
+	switch(r->siz) {
+	default:
+		return -1;
+	case 1:
+		v |= 0<<25;
+		break;
+	case 2:
+		v |= 1<<25;
+		break;
+	case 4:
+		v |= 2<<25;
+		break;
+	case 8:
+		v |= 3<<25;
+		break;
+	}
+
+	LPUT(sectoff);
+	LPUT(v);
+	return 0;
+}
+
+int
+archreloc(Reloc *r, Sym *s, vlong *val)
+{
+	USED(s);
+	switch(r->type) {
+	case D_CONST:
+		*val = r->add;
+		return 0;
+	case D_GOTOFF:
+		*val = symaddr(r->sym) + r->add - symaddr(lookup(".got", 0));
+		return 0;
+	}
+	return -1;
+}
+
 void
 elfsetupplt(void)
 {
@@ -327,21 +399,6 @@ elfsetupplt(void)
 	}
 }
 
-int
-archreloc(Reloc *r, Sym *s, vlong *val)
-{
-	USED(s);
-	switch(r->type) {
-	case D_CONST:
-		*val = r->add;
-		return 0;
-	case D_GOTOFF:
-		*val = symaddr(r->sym) + r->add - symaddr(lookup(".got", 0));
-		return 0;
-	}
-	return -1;
-}
-
 static void
 addpltsym(Sym *s)
 {
@@ -636,6 +693,10 @@ asmb(void)
 				Bprint(&bso, "%5.2f dwarf\n", cputime());
 			dwarfemitdebugsections();
 			break;
+		case Hdarwin:
+			if(isobj)
+				machoemitreloc();
+			break;
 		}
 	}
 	if(debug['v'])
diff --git a/src/cmd/8l/obj.c b/src/cmd/8l/obj.c
index f926652a6..ad453064c 100644
--- a/src/cmd/8l/obj.c
+++ b/src/cmd/8l/obj.c
@@ -147,8 +147,11 @@ main(int argc, char *argv[])
 		switch(HEADTYPE) {
 		default:
 			sysfatal("cannot use -hostobj with -H %s", headstr(HEADTYPE));
+		case Hdarwin:
+		case Hfreebsd:
 		case Hlinux:
 		case Hnetbsd:
+		case Hopenbsd:
 			break;
 		}
 	}
diff --git a/src/cmd/ld/data.c b/src/cmd/ld/data.c
index ca6c5300b..fdf4d043d 100644
--- a/src/cmd/ld/data.c
+++ b/src/cmd/ld/data.c
@@ -192,12 +192,20 @@ relocsym(Sym *s)
 					r->xadd += symaddr(rs) - symaddr(rs->outer);
 					rs = rs->outer;
 				}
+				if(rs->type != SHOSTOBJ && rs->sect == nil)
+					diag("missing section for %s", rs->name);
 				r->xsym = rs;
 
-				if(thechar == '6')
-					o = 0;
-				else
-					o = r->xadd;
+				o = r->xadd;
+				if(iself) {
+					if(thechar == '6')
+						o = 0;
+				} else if(HEADTYPE == Hdarwin) {
+					if(rs->type != SHOSTOBJ)
+						o += symaddr(rs);
+				} else {
+					diag("unhandled pcrel relocation for %s", headtype);
+				}
 				break;
 			}
 			o = symaddr(r->sym) + r->add;
@@ -214,13 +222,22 @@ relocsym(Sym *s)
 					r->xadd += symaddr(rs) - symaddr(rs->outer);
 					rs = rs->outer;
 				}
+				r->xadd -= r->siz; // relative to address after the relocated chunk
+				if(rs->type != SHOSTOBJ && rs->sect == nil)
+					diag("missing section for %s", rs->name);
 				r->xsym = rs;
-				r->xadd -= r->siz;
 
-				if(thechar == '6')
-					o = 0;
-				else
-					o = r->xadd;
+				o = r->xadd;
+				if(iself) {
+					if(thechar == '6')
+						o = 0;
+				} else if(HEADTYPE == Hdarwin) {
+					if(rs->type != SHOSTOBJ)
+						o += symaddr(rs) - rs->sect->vaddr;
+					o -= r->off; // WTF?
+				} else {
+					diag("unhandled pcrel relocation for %s", headtype);
+				}
 				break;
 			}
 			o = 0;
diff --git a/src/cmd/ld/go.c b/src/cmd/ld/go.c
index 66eddd5d8..27307819f 100644
--- a/src/cmd/ld/go.c
+++ b/src/cmd/ld/go.c
@@ -486,11 +486,8 @@ loadcgo(char *file, char *pkg, char *p, int n)
 		}
 
 		if(strcmp(f[0], "cgo_export_static") == 0 || strcmp(f[0], "cgo_export_dynamic") == 0) {
-			// TODO: Make Mach-O code happier. Right now it sees the dynimpname and
-			// includes CgoExportStatic symbols in the dynamic table, and then dyld
-			// cannot find them when we run the binary. Disabling Windows too
-			// because it probably has the same issue.
-			if(strcmp(f[0], "cgo_export_static") == 0 && (HEADTYPE == Hdarwin || HEADTYPE == Hwindows))
+			// TODO: Remove once we know Windows is okay.
+			if(strcmp(f[0], "cgo_export_static") == 0 && HEADTYPE == Hwindows)
 				continue;
 
 			if(nf < 2 || nf > 3)
diff --git a/src/cmd/ld/lib.c b/src/cmd/ld/lib.c
index 82a7df341..b895e5088 100644
--- a/src/cmd/ld/lib.c
+++ b/src/cmd/ld/lib.c
@@ -614,6 +614,8 @@ hostlink(void)
 	}
 	if(!debug['s'])
 		argv[argc++] = "-ggdb"; 
+	if(HEADTYPE == Hdarwin)
+		argv[argc++] = "-Wl,-no_pie,-pagezero_size,4000000";
 	argv[argc++] = "-o";
 	argv[argc++] = outfile;
 	
diff --git a/src/cmd/ld/macho.c b/src/cmd/ld/macho.c
index b85b7d6d8..0053e1095 100644
--- a/src/cmd/ld/macho.c
+++ b/src/cmd/ld/macho.c
@@ -150,7 +150,10 @@ machowrite(void)
 		LPUT(0xfeedface);
 	LPUT(hdr.cpu);
 	LPUT(hdr.subcpu);
-	LPUT(2);	/* file type - mach executable */
+	if(isobj)
+		LPUT(1);	/* file type - mach object */
+	else
+		LPUT(2);	/* file type - mach executable */
 	LPUT(nload+nseg+ndebug);
 	LPUT(loadsize);
 	LPUT(1);	/* flags - no undefines */
@@ -245,22 +248,24 @@ domacho(void)
 	s->type = SMACHOSYMTAB;
 	s->reachable = 1;
 	
-	s = lookup(".plt", 0);	// will be __symbol_stub
-	s->type = SMACHOPLT;
-	s->reachable = 1;
+	if(!isobj) {
+		s = lookup(".plt", 0);	// will be __symbol_stub
+		s->type = SMACHOPLT;
+		s->reachable = 1;
 	
-	s = lookup(".got", 0);	// will be __nl_symbol_ptr
-	s->type = SMACHOGOT;
-	s->reachable = 1;
-	s->align = 4;
+		s = lookup(".got", 0);	// will be __nl_symbol_ptr
+		s->type = SMACHOGOT;
+		s->reachable = 1;
+		s->align = 4;
 	
-	s = lookup(".linkedit.plt", 0);	// indirect table for .plt
-	s->type = SMACHOINDIRECTPLT;
-	s->reachable = 1;
+		s = lookup(".linkedit.plt", 0);	// indirect table for .plt
+		s->type = SMACHOINDIRECTPLT;
+		s->reachable = 1;
 	
-	s = lookup(".linkedit.got", 0);	// indirect table for .got
-	s->type = SMACHOINDIRECTGOT;
-	s->reachable = 1;
+		s = lookup(".linkedit.got", 0);	// indirect table for .got
+		s->type = SMACHOINDIRECTGOT;
+		s->reachable = 1;
+	}
 }
 
 void
@@ -295,7 +300,11 @@ machoshbits(MachoSeg *mseg, Section *sect, char *segname)
 			*p = '_';
 
 	msect = newMachoSect(mseg, estrdup(buf), segname);
-	
+	if(sect->rellen > 0) {
+		msect->reloc = sect->reloff;
+		msect->nreloc = sect->rellen / 8;
+	}
+
 	while(1<<msect->align < sect->align)
 		msect->align++;
 	msect->addr = sect->vaddr;
@@ -356,54 +365,70 @@ asmbmacho(void)
 		mh->subcpu = MACHO_SUBCPU_X86;
 		break;
 	}
+	
+	ms = nil;
+	if(isobj) {
+		/* segment for entire file */
+		ms = newMachoSeg("", 40);
+		ms->fileoffset = segtext.fileoff;
+		ms->filesize = segdata.fileoff + segdata.filelen - segtext.fileoff;
+	}
 
 	/* segment for zero page */
-	ms = newMachoSeg("__PAGEZERO", 0);
-	ms->vsize = va;
+	if(!isobj) {
+		ms = newMachoSeg("__PAGEZERO", 0);
+		ms->vsize = va;
+	}
 
 	/* text */
 	v = rnd(HEADR+segtext.len, INITRND);
-	ms = newMachoSeg("__TEXT", 20);
-	ms->vaddr = va;
-	ms->vsize = v;
-	ms->fileoffset = 0;
-	ms->filesize = v;
-	ms->prot1 = 7;
-	ms->prot2 = 5;
-	
+	if(!isobj) {
+		ms = newMachoSeg("__TEXT", 20);
+		ms->vaddr = va;
+		ms->vsize = v;
+		ms->fileoffset = 0;
+		ms->filesize = v;
+		ms->prot1 = 7;
+		ms->prot2 = 5;
+	}
+
 	for(sect=segtext.sect; sect!=nil; sect=sect->next)
 		machoshbits(ms, sect, "__TEXT");
 
 	/* data */
-	w = segdata.len;
-	ms = newMachoSeg("__DATA", 20);
-	ms->vaddr = va+v;
-	ms->vsize = w;
-	ms->fileoffset = v;
-	ms->filesize = segdata.filelen;
-	ms->prot1 = 3;
-	ms->prot2 = 3;
-	
+	if(!isobj) {
+		w = segdata.len;
+		ms = newMachoSeg("__DATA", 20);
+		ms->vaddr = va+v;
+		ms->vsize = w;
+		ms->fileoffset = v;
+		ms->filesize = segdata.filelen;
+		ms->prot1 = 3;
+		ms->prot2 = 3;
+	}
+
 	for(sect=segdata.sect; sect!=nil; sect=sect->next)
 		machoshbits(ms, sect, "__DATA");
 
-	switch(thechar) {
-	default:
-		diag("unknown macho architecture");
-		errorexit();
-	case '6':
-		ml = newMachoLoad(5, 42+2);	/* unix thread */
-		ml->data[0] = 4;	/* thread type */
-		ml->data[1] = 42;	/* word count */
-		ml->data[2+32] = entryvalue();	/* start pc */
-		ml->data[2+32+1] = entryvalue()>>16>>16;	// hide >>32 for 8l
-		break;
-	case '8':
-		ml = newMachoLoad(5, 16+2);	/* unix thread */
-		ml->data[0] = 1;	/* thread type */
-		ml->data[1] = 16;	/* word count */
-		ml->data[2+10] = entryvalue();	/* start pc */
-		break;
+	if(!isobj) {
+		switch(thechar) {
+		default:
+			diag("unknown macho architecture");
+			errorexit();
+		case '6':
+			ml = newMachoLoad(5, 42+2);	/* unix thread */
+			ml->data[0] = 4;	/* thread type */
+			ml->data[1] = 42;	/* word count */
+			ml->data[2+32] = entryvalue();	/* start pc */
+			ml->data[2+32+1] = entryvalue()>>16>>16;	// hide >>32 for 8l
+			break;
+		case '8':
+			ml = newMachoLoad(5, 16+2);	/* unix thread */
+			ml->data[0] = 1;	/* thread type */
+			ml->data[1] = 16;	/* word count */
+			ml->data[2+10] = entryvalue();	/* start pc */
+			break;
+		}
 	}
 	
 	if(!debug['d']) {
@@ -415,13 +440,15 @@ asmbmacho(void)
 		s3 = lookup(".linkedit.got", 0);
 		s4 = lookup(".machosymstr", 0);
 
-		ms = newMachoSeg("__LINKEDIT", 0);
-		ms->vaddr = va+v+rnd(segdata.len, INITRND);
-		ms->vsize = s1->size + s2->size + s3->size + s4->size;
-		ms->fileoffset = linkoff;
-		ms->filesize = ms->vsize;
-		ms->prot1 = 7;
-		ms->prot2 = 3;
+		if(!isobj) {
+			ms = newMachoSeg("__LINKEDIT", 0);
+			ms->vaddr = va+v+rnd(segdata.len, INITRND);
+			ms->vsize = s1->size + s2->size + s3->size + s4->size;
+			ms->fileoffset = linkoff;
+			ms->filesize = ms->vsize;
+			ms->prot1 = 7;
+			ms->prot2 = 3;
+		}
 
 		ml = newMachoLoad(2, 4);	/* LC_SYMTAB */
 		ml->data[0] = linkoff;	/* symoff */
@@ -431,21 +458,24 @@ asmbmacho(void)
 
 		machodysymtab();
 
-		ml = newMachoLoad(14, 6);	/* LC_LOAD_DYLINKER */
-		ml->data[0] = 12;	/* offset to string */
-		strcpy((char*)&ml->data[1], "/usr/lib/dyld");
-
-		for(i=0; i<ndylib; i++) {
-			ml = newMachoLoad(12, 4+(strlen(dylib[i])+1+7)/8*2);	/* LC_LOAD_DYLIB */
-			ml->data[0] = 24;	/* offset of string from beginning of load */
-			ml->data[1] = 0;	/* time stamp */
-			ml->data[2] = 0;	/* version */
-			ml->data[3] = 0;	/* compatibility version */
-			strcpy((char*)&ml->data[4], dylib[i]);
+		if(!isobj) {
+			ml = newMachoLoad(14, 6);	/* LC_LOAD_DYLINKER */
+			ml->data[0] = 12;	/* offset to string */
+			strcpy((char*)&ml->data[1], "/usr/lib/dyld");
+	
+			for(i=0; i<ndylib; i++) {
+				ml = newMachoLoad(12, 4+(strlen(dylib[i])+1+7)/8*2);	/* LC_LOAD_DYLIB */
+				ml->data[0] = 24;	/* offset of string from beginning of load */
+				ml->data[1] = 0;	/* time stamp */
+				ml->data[2] = 0;	/* version */
+				ml->data[3] = 0;	/* compatibility version */
+				strcpy((char*)&ml->data[4], dylib[i]);
+			}
 		}
 	}
 
-	if(!debug['s'])
+	// TODO: dwarf headers go in ms too
+	if(!debug['s'] && !isobj)
 		dwarfaddmachoheaders();
 
 	a = machowrite();
@@ -515,7 +545,8 @@ machogenasmsym(void (*put)(Sym*, char*, int, vlong, vlong, int, Sym*))
 
 	genasmsym(put);
 	for(s=allsym; s; s=s->allsym)
-		if(s->type == SDYNIMPORT)
+		if(s->type == SDYNIMPORT || s->type == SHOSTOBJ)
+		if(s->reachable)
 			put(s, nil, 'D', 0, 0, 0, nil);
 }
 			
@@ -552,13 +583,16 @@ machosymtab(void)
 		adduint32(symtab, symstr->size);
 		adduint8(symstr, '_');
 		addstring(symstr, s->extname);
-		if(s->type == SDYNIMPORT) {
+		if(s->type == SDYNIMPORT || s->type == SHOSTOBJ) {
 			adduint8(symtab, 0x01); // type N_EXT, external symbol
 			adduint8(symtab, 0); // no section
 			adduint16(symtab, 0); // desc
 			adduintxx(symtab, 0, PtrSize); // no value
 		} else {
-			adduint8(symtab, 0x0f);
+			if(s->cgoexport)
+				adduint8(symtab, 0x0f);
+			else
+				adduint8(symtab, 0x0e);
 			o = s;
 			while(o->outer != nil)
 				o = o->outer;
@@ -663,3 +697,56 @@ domacholink(void)
 	return rnd(size, INITRND);
 }
 
+
+void
+machorelocsect(Section *sect, Sym *first)
+{
+	Sym *sym;
+	int32 eaddr;
+	Reloc *r;
+
+	// If main section has no bits, nothing to relocate.
+	if(sect->vaddr >= sect->seg->vaddr + sect->seg->filelen)
+		return;
+	
+	sect->reloff = cpos();
+	for(sym = first; sym != nil; sym = sym->next) {
+		if(!sym->reachable)
+			continue;
+		if(sym->value >= sect->vaddr)
+			break;
+	}
+	
+	eaddr = sect->vaddr + sect->len;
+	for(; sym != nil; sym = sym->next) {
+		if(!sym->reachable)
+			continue;
+		if(sym->value >= eaddr)
+			break;
+		cursym = sym;
+		
+		for(r = sym->r; r < sym->r+sym->nr; r++) {
+			if(r->done)
+				continue;
+			if(machoreloc1(r, sym->value+r->off - sect->vaddr) < 0)
+				diag("unsupported obj reloc %d/%d to %s", r->type, r->siz, r->sym->name);
+		}
+	}
+		
+	sect->rellen = cpos() - sect->reloff;
+}
+
+void
+machoemitreloc(void)
+{
+	Section *sect;
+
+	while(cpos()&7)
+		cput(0);
+
+	machorelocsect(segtext.sect, textp);
+	for(sect=segtext.sect->next; sect!=nil; sect=sect->next)
+		machorelocsect(sect, datap);	
+	for(sect=segdata.sect; sect!=nil; sect=sect->next)
+		machorelocsect(sect, datap);	
+}
diff --git a/src/cmd/ld/macho.h b/src/cmd/ld/macho.h
index 59900c940..d759f4b0f 100644
--- a/src/cmd/ld/macho.h
+++ b/src/cmd/ld/macho.h
@@ -52,6 +52,8 @@ MachoLoad*	newMachoLoad(uint32, uint32);
 int	machowrite(void);
 void	machoinit(void);
 void	machosymorder(void);
+void	machoemitreloc(void);
+int	machoreloc1(Reloc*, vlong);
 
 /*
  * Total amount of space to reserve at the start of the file
diff --git a/src/pkg/runtime/asm_386.s b/src/pkg/runtime/asm_386.s
index 375274e0b..15f1ce804 100644
--- a/src/pkg/runtime/asm_386.s
+++ b/src/pkg/runtime/asm_386.s
@@ -26,9 +26,8 @@ TEXT _rt0_386(SB),7,$0
 	MOVL	_cgo_init(SB), AX
 	TESTL	AX, AX
 	JZ	needtls
-	PUSHL	BP
+	MOVL	BP, 0(SP)
 	CALL	AX
-	POPL	BP
 	// skip runtime·ldt0setup(SB) and tls test after _cgo_init for non-windows
 	CMPL runtime·iswindows(SB), $0
 	JEQ ok
diff --git a/src/run.bash b/src/run.bash
index a026b459c..cabe745b6 100755
--- a/src/run.bash
+++ b/src/run.bash
@@ -77,6 +77,10 @@ go run $GOROOT/test/run.go - .
 [ "$GOHOSTOS" == openbsd ] || # issue 4878
 (xcd ../misc/cgo/test
 go test
+case "$GOHOSTOS-$GOARCH" in
+darwin-386 | darwin-amd64 | linux-386 | linux-amd64)
+	go test -ldflags '-w -hostobj'
+esac
 ) || exit $?
 
 [ "$CGO_ENABLED" != 1 ] ||