liblink, cmd/ld, cmd/5l: darwin/arm support

liblink:
 - set dummy value for ctxt->tlsoffset.
cmd/ld:
 - always do external linking when using cgo on darwin/arm,
   as our linker might not generate codesign-compatible binary.
cmd/5l:
 - support generate ARM Mach-O binaries
 - add machoreloc1() that translate our internal relocation to
   macho relocations used by external linking.

Change-Id: Ic5454aeb87009aaf8f1453ec7fe33e6da55d5f06
Reviewed-on: https://go-review.googlesource.com/3273Reviewed-by: David Crawshaw <crawshaw@golang.org>

src/cmd/5l/asm.c

@@ -33,9 +33,9 @@
 #include	"l.h"
 #include	"../ld/lib.h"
 #include	"../ld/elf.h"
+#include	"../ld/macho.h"
 #include	"../ld/dwarf.h"
 
-
 char linuxdynld[] = "/lib/ld-linux.so.3"; // 2 for OABI, 3 for EABI
 char freebsddynld[] = "/usr/libexec/ld-elf.so.1";
 char openbsddynld[] = "XXX";
@@ -301,10 +301,58 @@ elfsetupplt(void)
 int
 machoreloc1(Reloc *r, vlong sectoff)
 {
-	USED(r);
-	USED(sectoff);
+	uint32 v;
+	LSym *rs;
 
-	return -1;
+	rs = r->xsym;
+
+	if(rs->type == SHOSTOBJ || r->type == R_CALLARM) {
+		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 R_ADDR:
+		v |= MACHO_GENERIC_RELOC_VANILLA<<28;
+		break;
+	case R_CALLARM:
+		v |= 1<<24; // pc-relative bit
+		v |= MACHO_ARM_RELOC_BR24<<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;
 }
 
 
@@ -333,6 +381,14 @@ archreloc(Reloc *r, LSym *s, vlong *val)
 				diag("missing section for %s", rs->name);
 			r->xsym = rs;
 
+			// ld64 for arm seems to want the symbol table to contain offset
+			// into the section rather than pseudo virtual address that contains
+			// the section load address.
+			// we need to compensate that by removing the instruction's address
+			// from addend.
+			if(HEADTYPE == Hdarwin)
+				r->xadd -= symaddr(s) + r->off;
+
 			*val = braddoff((0xff000000U & (uint32)r->add), 
 							(0xffffff & (uint32)(r->xadd / 4)));
 			return 0;
@@ -539,6 +595,8 @@ adddynlib(char *lib)
 		if(s->size == 0)
 			addstring(s, "");
 		elfwritedynent(linklookup(ctxt, ".dynamic", 0), DT_NEEDED, addstring(s, lib));
+	} else if(HEADTYPE == Hdarwin) {
+		machoadddynlib(lib);
 	} else {
 		diag("adddynlib: unsupported binary format");
 	}
@@ -547,7 +605,7 @@ adddynlib(char *lib)
 void
 asmb(void)
 {
-	uint32 symo;
+	uint32 symo, dwarfoff, machlink;
 	Section *sect;
 	LSym *sym;
 	int i;
@@ -583,6 +641,22 @@ asmb(void)
 	cseek(segdata.fileoff);
 	datblk(segdata.vaddr, segdata.filelen);
 
+	machlink = 0;
+	if(HEADTYPE == Hdarwin) {
+		if(debug['v'])
+			Bprint(&bso, "%5.2f dwarf\n", cputime());
+
+		if(!debug['w']) { // TODO(minux): enable DWARF Support
+			dwarfoff = rnd(HEADR+segtext.len, INITRND) + rnd(segdata.filelen, INITRND);
+			cseek(dwarfoff);
+
+			segdwarf.fileoff = cpos();
+			dwarfemitdebugsections();
+			segdwarf.filelen = cpos() - segdwarf.fileoff;
+		}
+		machlink = domacholink();
+	}
+
 	/* output symbol table */
 	symsize = 0;
 	lcsize = 0;
@@ -599,6 +673,9 @@ asmb(void)
 		case Hplan9:
 			symo = segdata.fileoff+segdata.filelen;
 			break;
+		case Hdarwin:
+			symo = rnd(HEADR+segtext.filelen, INITRND)+rnd(segdata.filelen, INITRND)+machlink;
+			break;
 		ElfSym:
 			symo = segdata.fileoff+segdata.filelen;
 			symo = rnd(symo, INITRND);
@@ -635,6 +712,10 @@ asmb(void)
 				cflush();
 			}
 			break;
+		case Hdarwin:
+			if(linkmode == LinkExternal)
+				machoemitreloc();
+			break;
 		}
 	}
 
@@ -662,6 +743,9 @@ asmb(void)
 	case Hnacl:
 		asmbelf(symo);
 		break;
+	case Hdarwin:
+		asmbmacho();
+		break;
 	}
 	cflush();
 	if(debug['c']){

src/cmd/5l/obj.c

@@ -33,6 +33,7 @@
 #include	"l.h"
 #include	"../ld/lib.h"
 #include	"../ld/elf.h"
+#include	"../ld/macho.h"
 #include	"../ld/dwarf.h"
 #include	<ar.h>
 
@@ -64,6 +65,7 @@ archinit(void)
 	case Hlinux:
 	case Hfreebsd:
 	case Hnacl:
+	case Hdarwin:
 		break;
 	}
 
@@ -104,6 +106,17 @@ archinit(void)
 		if(INITRND == -1)
 			INITRND = 0x10000;
 		break;
+	case Hdarwin:   /* apple MACH */
+		debug['w'] = 1; // disable DWARF generataion
+		machoinit();
+		HEADR = INITIAL_MACHO_HEADR;
+		if(INITTEXT == -1)
+			INITTEXT = 4096+HEADR;
+		if(INITDAT == -1)
+			INITDAT = 0;
+		if(INITRND == -1)
+			INITRND = 4096;
+		break;
 	}
 	if(INITDAT != 0 && INITRND != 0)
 		print("warning: -D0x%ux is ignored because of -R0x%ux\n",

src/cmd/ld/lib.c

@@ -212,6 +212,14 @@ loadlib(void)
 		// Force external linking for android.
 		if(strcmp(goos, "android") == 0)
 			linkmode = LinkExternal;
+
+		// cgo on Darwin must use external linking
+		// we can always use external linking, but then there will be circular
+		// dependency problems when compiling natively (external linking requires
+		// runtime/cgo, runtime/cgo requires cmd/cgo, but cmd/cgo needs to be
+		// compiled using external linking.)
+		if(thechar == '5' && HEADTYPE == Hdarwin && iscgo)
+			linkmode = LinkExternal;
 	}
 
 	if(linkmode == LinkExternal && !iscgo) {

src/cmd/ld/macho.c

@@ -357,6 +357,10 @@ asmbmacho(void)
 	default:
 		diag("unknown mach architecture");
 		errorexit();
+	case '5':
+		mh->cpu = MACHO_CPU_ARM;
+		mh->subcpu = MACHO_SUBCPU_ARMV7;
+		break;
 	case '6':
 		mh->cpu = MACHO_CPU_AMD64;
 		mh->subcpu = MACHO_SUBCPU_X86;
@@ -416,6 +420,12 @@ asmbmacho(void)
 		default:
 			diag("unknown macho architecture");
 			errorexit();
+		case '5':
+			ml = newMachoLoad(5, 17+2);	/* unix thread */
+			ml->data[0] = 1;	/* thread type */
+			ml->data[1] = 17;	/* word count */
+			ml->data[2+15] = entryvalue();	/* start pc */
+			break;
 		case '6':
 			ml = newMachoLoad(5, 42+2);	/* unix thread */
 			ml->data[0] = 4;	/* thread type */

src/cmd/ld/macho.h

@@ -66,6 +66,9 @@ enum {
 	MACHO_CPU_AMD64 = (1<<24)|7,
 	MACHO_CPU_386 = 7,
 	MACHO_SUBCPU_X86 = 3,
+	MACHO_CPU_ARM = 12,
+	MACHO_SUBCPU_ARM = 0,
+	MACHO_SUBCPU_ARMV7 = 9,
 
 	MACHO32SYMSIZE = 12,
 	MACHO64SYMSIZE = 16,
@@ -80,6 +83,9 @@ enum {
 	MACHO_X86_64_RELOC_SIGNED_2 = 7,
 	MACHO_X86_64_RELOC_SIGNED_4 = 8,
 	
+	MACHO_ARM_RELOC_VANILLA = 0,
+	MACHO_ARM_RELOC_BR24 = 5,
+	
 	MACHO_GENERIC_RELOC_VANILLA = 0,
 	
 	MACHO_FAKE_GOTPCREL = 100,

src/liblink/sym.c

@@ -148,15 +148,15 @@ linknew(LinkArch *arch)
 		switch(ctxt->arch->thechar) {
 		default:
 			sysfatal("unknown thread-local storage offset for nacl/%s", ctxt->arch->name);
+		case '5':
+			ctxt->tlsoffset = 0;
+			break;
 		case '6':
 			ctxt->tlsoffset = 0;
 			break;
 		case '8':
 			ctxt->tlsoffset = -8;
 			break;
-		case '5':
-			ctxt->tlsoffset = 0;
-			break;
 		}
 		break;
 
@@ -174,6 +174,9 @@ linknew(LinkArch *arch)
 		case '8':
 			ctxt->tlsoffset = 0x468;
 			break;
+		case '5':
+			ctxt->tlsoffset = 0; // dummy value, not needed
+			break;
 		}
 		break;
 	}