cmd/ld: make mach-o sections match internal sections

This brings Mach-O generation more in line with ELF generation.

Having separate sections for the symtab and pclntab mean that we
can find them that way, instead of using the deprecated debug segments.
(And the host linker will keep separate sections for us, but probably
not the debug segments.)

R=ken2
CC=golang-dev
https://golang.org/cl/7688043

src/cmd/ld/data.c

@@ -879,6 +879,9 @@ symalign(Sym *s)
 {
 	int32 align;
 
+	if(s->align != 0)
+		return s->align;
+
 	align = MaxAlign;
 	while(align > s->size && align > 1)
 		align >>= 1;
@@ -1136,9 +1139,10 @@ dodata(void)
 	lookup("end", 0)->sect = sect;
 
 	/* we finished segdata, begin segtext */
+	s = datap;
+	datsize = 0;
 
 	/* read-only data */
-	s = datap;
 	sect = addsection(&segtext, ".rodata", 04);
 	sect->align = maxalign(s, STYPELINK-1);
 	sect->vaddr = 0;
@@ -1202,7 +1206,7 @@ dodata(void)
 	}
 	sect->len = datsize - sect->vaddr;
 
-	/* read-only ELF sections */
+	/* read-only ELF, Mach-O sections */
 	for(; s != nil && s->type < SELFSECT; s = s->next) {
 		sect = addsection(&segtext, s->name, 04);
 		sect->align = symalign(s);
@@ -1266,6 +1270,7 @@ address(void)
 	Section *typelink;
 	Sym *sym, *sub;
 	uvlong va;
+	vlong vlen;
 
 	va = INITTEXT;
 	segtext.rwx = 05;
@@ -1295,11 +1300,11 @@ address(void)
 	noptrbss = nil;
 	datarelro = nil;
 	for(s=segdata.sect; s != nil; s=s->next) {
+		vlen = s->len;
 		if(s->next)
-			s->len = s->next->vaddr - s->vaddr;
+			vlen = s->next->vaddr - s->vaddr;
 		s->vaddr = va;
-		va += s->len;
-		segdata.filelen += s->len;
+		va += vlen;
 		segdata.len = va - segdata.vaddr;
 		if(strcmp(s->name, ".data") == 0)
 			data = s;

src/cmd/ld/dwarf.c

@@ -2298,42 +2298,42 @@ dwarfaddmachoheaders(void)
 	ms->fileoffset = fakestart;
 	ms->filesize = abbrevo-fakestart;
 
-	msect = newMachoSect(ms, "__debug_abbrev");
+	msect = newMachoSect(ms, "__debug_abbrev", "__DWARF");
 	msect->off = abbrevo;
 	msect->size = abbrevsize;
 	ms->filesize += msect->size;
 
-	msect = newMachoSect(ms, "__debug_line");
+	msect = newMachoSect(ms, "__debug_line", "__DWARF");
 	msect->off = lineo;
 	msect->size = linesize;
 	ms->filesize += msect->size;
 
-	msect = newMachoSect(ms, "__debug_frame");
+	msect = newMachoSect(ms, "__debug_frame", "__DWARF");
 	msect->off = frameo;
 	msect->size = framesize;
 	ms->filesize += msect->size;
 
-	msect = newMachoSect(ms, "__debug_info");
+	msect = newMachoSect(ms, "__debug_info", "__DWARF");
 	msect->off = infoo;
 	msect->size = infosize;
 	ms->filesize += msect->size;
 
 	if (pubnamessize > 0) {
-		msect = newMachoSect(ms, "__debug_pubnames");
+		msect = newMachoSect(ms, "__debug_pubnames", "__DWARF");
 		msect->off = pubnameso;
 		msect->size = pubnamessize;
 		ms->filesize += msect->size;
 	}
 
 	if (pubtypessize > 0) {
-		msect = newMachoSect(ms, "__debug_pubtypes");
+		msect = newMachoSect(ms, "__debug_pubtypes", "__DWARF");
 		msect->off = pubtypeso;
 		msect->size = pubtypessize;
 		ms->filesize += msect->size;
 	}
 
 	if (arangessize > 0) {
-		msect = newMachoSect(ms, "__debug_aranges");
+		msect = newMachoSect(ms, "__debug_aranges", "__DWARF");
 		msect->off = arangeso;
 		msect->size = arangessize;
 		ms->filesize += msect->size;
@@ -2341,7 +2341,7 @@ dwarfaddmachoheaders(void)
 
 	// TODO(lvd) fix gdb/python to load MachO (16 char section name limit)
 	if (gdbscriptsize > 0) {
-		msect = newMachoSect(ms, "__debug_gdb_scripts");
+		msect = newMachoSect(ms, "__debug_gdb_scripts", "__DWARF");
 		msect->off = gdbscripto;
 		msect->size = gdbscriptsize;
 		ms->filesize += msect->size;

src/cmd/ld/lib.h

@@ -34,7 +34,6 @@ enum
 
 	/* order here is order in output file */
 	STEXT,
-	SMACHOPLT,
 	STYPE,
 	SSTRING,
 	SGOSTRING,
@@ -43,12 +42,13 @@ enum
 	SSYMTAB,
 	SPCLNTAB,
 	SELFROSECT,
+	SMACHOPLT,
 	SELFSECT,
+	SMACHO,	/* Mach-O __nl_symbol_ptr */
+	SMACHOGOT,
 	SNOPTRDATA,
 	SDATARELRO,
 	SDATA,
-	SMACHO,	/* Mach-O __nl_symbol_ptr */
-	SMACHOGOT,
 	SWINDOWS,
 	SBSS,
 	SNOPTRBSS,

src/cmd/ld/macho.c

@@ -14,7 +14,6 @@ static	int	macho64;
 static	MachoHdr	hdr;
 static	MachoLoad	*load;
 static	MachoSeg	seg[16];
-static	MachoDebug	xdebug[16];
 static	int	nload, mload, nseg, ndebug, nsect;
 
 // Amount of space left for adding load commands
@@ -86,7 +85,7 @@ newMachoSeg(char *name, int msect)
 }
 
 MachoSect*
-newMachoSect(MachoSeg *seg, char *name)
+newMachoSect(MachoSeg *seg, char *name, char *segname)
 {
 	MachoSect *s;
 
@@ -96,21 +95,11 @@ newMachoSect(MachoSeg *seg, char *name)
 	}
 	s = &seg->sect[seg->nsect++];
 	s->name = name;
+	s->segname = segname;
 	nsect++;
 	return s;
 }
 
-MachoDebug*
-newMachoDebug(void)
-{
-	if(ndebug >= nelem(xdebug)) {
-		diag("too many debugs");
-		errorexit();
-	}
-	return &xdebug[ndebug++];
-}
-
-
 // Generic linking code.
 
 static char **dylib;
@@ -126,7 +115,6 @@ machowrite(void)
 	int i, j;
 	MachoSeg *s;
 	MachoSect *t;
-	MachoDebug *d;
 	MachoLoad *l;
 
 	o1 = cpos();
@@ -186,7 +174,7 @@ machowrite(void)
 			t = &s->sect[j];
 			if(macho64) {
 				strnput(t->name, 16);
-				strnput(s->name, 16);
+				strnput(t->segname, 16);
 				VPUT(t->addr);
 				VPUT(t->size);
 				LPUT(t->off);
@@ -199,7 +187,7 @@ machowrite(void)
 				LPUT(0);	/* reserved */
 			} else {
 				strnput(t->name, 16);
-				strnput(s->name, 16);
+				strnput(t->segname, 16);
 				LPUT(t->addr);
 				LPUT(t->size);
 				LPUT(t->off);
@@ -221,14 +209,6 @@ machowrite(void)
 			LPUT(l->data[j]);
 	}
 
-	for(i=0; i<ndebug; i++) {
-		d = &xdebug[i];
-		LPUT(3);	/* obsolete gdb debug info */
-		LPUT(16);	/* size of symseg command */
-		LPUT(d->fileoffset);
-		LPUT(d->filesize);
-	}
-
 	return cpos() - o1;
 }
 
@@ -258,6 +238,7 @@ domacho(void)
 	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;
@@ -287,6 +268,52 @@ machoadddynlib(char *lib)
 	dylib[ndylib++] = lib;
 }
 
+static void
+machoshbits(MachoSeg *mseg, Section *sect, char *segname)
+{
+	MachoSect *msect;
+	char buf[40];
+	char *p;
+	
+	snprint(buf, sizeof buf, "__%s", sect->name+1);
+	for(p=buf; *p; p++)
+		if(*p == '.')
+			*p = '_';
+
+	msect = newMachoSect(mseg, estrdup(buf), segname);
+	
+	while(1<<msect->align < sect->align)
+		msect->align++;
+	msect->addr = sect->vaddr;
+	msect->size = sect->len;
+	msect->off = sect->seg->fileoff + sect->vaddr - sect->seg->vaddr;
+	
+	if(sect->vaddr < sect->seg->vaddr + sect->seg->filelen) {
+		// data in file
+		if(sect->len > sect->seg->vaddr + sect->seg->filelen - sect->vaddr)
+			diag("macho cannot represent section %s crossing data and bss", sect->name);
+	} else {
+		// zero fill
+		msect->flag |= 1;
+	}
+
+	if(sect->rwx & 1)
+		msect->flag |= 0x400; /* has instructions */
+	
+	if(strcmp(sect->name, ".plt") == 0) {
+		msect->name = "__symbol_stub1";
+		msect->flag = 0x80000408; /* only instructions, code, symbol stubs */
+		msect->res1 = 0;
+		msect->res2 = 6;
+	}
+
+	if(strcmp(sect->name, ".got") == 0) {
+		msect->name = "__nl_symbol_ptr";
+		msect->flag = 6;	/* section with nonlazy symbol pointers */
+		msect->res1 = lookup(".linkedit.plt", 0)->size / 4;	/* offset into indirect symbol table */
+	}
+}
+
 void
 asmbmacho(void)
 {
@@ -294,11 +321,9 @@ asmbmacho(void)
 	vlong va;
 	int a, i;
 	MachoHdr *mh;
-	MachoSect *msect;
 	MachoSeg *ms;
-	MachoDebug *md;
 	MachoLoad *ml;
-	Sym *s;
+	Section *sect;
 
 	/* apple MACH */
 	va = INITTEXT - HEADR;
@@ -323,62 +348,29 @@ asmbmacho(void)
 
 	/* text */
 	v = rnd(HEADR+segtext.len, INITRND);
-	ms = newMachoSeg("__TEXT", 2);
+	ms = newMachoSeg("__TEXT", 20);
 	ms->vaddr = va;
 	ms->vsize = v;
+	ms->fileoffset = 0;
 	ms->filesize = v;
 	ms->prot1 = 7;
 	ms->prot2 = 5;
 	
-	msect = newMachoSect(ms, "__text");
-	msect->addr = INITTEXT;
-	msect->size = segtext.sect->len;
-	msect->off = INITTEXT - va;
-	msect->flag = 0x400;	/* flag - some instructions */
-	
-	s = lookup(".plt", 0);
-	if(s->size > 0) {
-		msect = newMachoSect(ms, "__symbol_stub1");
-		msect->addr = symaddr(s);
-		msect->size = s->size;
-		msect->off = ms->fileoffset + msect->addr - ms->vaddr;
-		msect->flag = 0x80000408;	/* flag */
-		msect->res1 = 0;	/* index into indirect symbol table */
-		msect->res2 = 6;	/* size of stubs */
-	}
+	for(sect=segtext.sect; sect!=nil; sect=sect->next)
+		machoshbits(ms, sect, "__TEXT");
 
 	/* data */
 	w = segdata.len;
-	ms = newMachoSeg("__DATA", 3);
+	ms = newMachoSeg("__DATA", 20);
 	ms->vaddr = va+v;
 	ms->vsize = w;
 	ms->fileoffset = v;
 	ms->filesize = segdata.filelen;
-	ms->prot1 = 7;
+	ms->prot1 = 3;
 	ms->prot2 = 3;
 	
-	msect = newMachoSect(ms, "__data");
-	msect->addr = va+v;
-	msect->off = v;
-	msect->size = segdata.filelen;
-
-	s = lookup(".got", 0);
-	if(s->size > 0) {
-		msect->size = symaddr(s) - msect->addr;
-
-		msect = newMachoSect(ms, "__nl_symbol_ptr");
-		msect->addr = symaddr(s);
-		msect->size = s->size;
-		msect->off = datoff(msect->addr);
-		msect->align = 2;
-		msect->flag = 6;	/* section with nonlazy symbol pointers */
-		msect->res1 = lookup(".linkedit.plt", 0)->size / 4;	/* offset into indirect symbol table */
-	}
-
-	msect = newMachoSect(ms, "__bss");
-	msect->addr = va+v+segdata.filelen;
-	msect->size = segdata.len - segdata.filelen;
-	msect->flag = 1;	/* flag - zero fill */
+	for(sect=segdata.sect; sect!=nil; sect=sect->next)
+		machoshbits(ms, sect, "__DATA");
 
 	switch(thechar) {
 	default:
@@ -456,21 +448,8 @@ asmbmacho(void)
 		}
 	}
 
-	if(!debug['s']) {
-		Sym *s;
-
-		md = newMachoDebug();
-		s = lookup("symtab", 0);
-		md->fileoffset = datoff(s->value);
-		md->filesize = s->size;
-
-		md = newMachoDebug();
-		s = lookup("pclntab", 0);
-		md->fileoffset = datoff(s->value);
-		md->filesize = s->size;
-
+	if(!debug['s'])
 		dwarfaddmachoheaders();
-	}
 
 	a = machowrite();
 	if(a > HEADR)

src/cmd/ld/macho.h

@@ -11,6 +11,7 @@ struct MachoHdr {
 typedef struct MachoSect MachoSect;
 struct MachoSect {
 	char*	name;
+	char*	segname;
 	uint64	addr;
 	uint64	size;
 	uint32	off;
@@ -44,17 +45,10 @@ struct MachoLoad {
 	uint32	*data;
 };
 
-typedef struct MachoDebug MachoDebug;
-struct MachoDebug {
-	uint32	fileoffset;
-	uint32	filesize;
-};
-
 MachoHdr*	getMachoHdr(void);
 MachoSeg*	newMachoSeg(char*, int);
-MachoSect*	newMachoSect(MachoSeg*, char*);
+MachoSect*	newMachoSect(MachoSeg*, char*, char*);
 MachoLoad*	newMachoLoad(uint32, uint32);
-MachoDebug*	newMachoDebug(void);
 int	machowrite(void);
 void	machoinit(void);
 

src/libmach/executable.c

@@ -1030,8 +1030,8 @@ machdotout(int fd, Fhdr *fp, ExecHdr *hp)
 	uvlong textsize, datasize, bsssize;
 	uchar *cmdbuf;
 	uchar *cmdp;
-	int i, hdrsize;
-	uint32 textva, textoff, datava, dataoff;
+	int i, j, hdrsize;
+	uint32 textva, textoff, datava, dataoff, symoff, symsize, pclnoff, pclnsize;
 
 	mp = &hp->e.machhdr;
 	if (leswal(mp->filetype) != MACH_EXECUTABLE_TYPE) {
@@ -1141,33 +1141,24 @@ machdotout(int fd, Fhdr *fp, ExecHdr *hp)
 			if (strcmp(seg32->segname, "__TEXT") == 0) {
 				textva = seg32->vmaddr;
 				textoff = seg32->fileoff;
+				textsize = seg32->vmsize;
 				sect32 = (MachSect32*)(cmdp + sizeof(MachSeg32));
-				if (strcmp(sect32->sectname, "__text") == 0) {
-					textsize = swal(sect32->size);
-				} else {
-					werrstr("no text section");
-					goto bad;
+				for(j = 0; j < seg32->nsects; j++, sect32++) {
+					if (strcmp(sect32->sectname, "__gosymtab") == 0) {
+						symoff = swal(sect32->offset);
+						symsize = swal(sect32->size);
+					}
+					if (strcmp(sect32->sectname, "__gopclntab") == 0) {
+						pclnoff = swal(sect32->offset);
+						pclnsize = swal(sect32->size);
+					}
 				}
 			}
 			if (strcmp(seg32->segname, "__DATA") == 0) {
 				datava = seg32->vmaddr;
 				dataoff = seg32->fileoff;
-				sect32 = (MachSect32*)(cmdp + sizeof(MachSeg32));
-				if (strcmp(sect32->sectname, "__data") == 0) {
-					datasize = swal(sect32->size);
-				} else {
-					werrstr("no data section");
-					goto bad;
-				}
-				sect32++;
-				if (strcmp(sect32->sectname, "__nl_symbol_ptr") == 0)
-					sect32++;
-				if (strcmp(sect32->sectname, "__bss") == 0) {
-					bsssize = swal(sect32->size);
-				} else {
-					werrstr("no bss section");
-					goto bad;
-				}
+				datasize = seg32->filesize;
+				bsssize = seg32->vmsize - seg32->filesize;
 			}
 			break;
 
@@ -1188,42 +1179,39 @@ machdotout(int fd, Fhdr *fp, ExecHdr *hp)
 			if (strcmp(seg->segname, "__TEXT") == 0) {
 				textva = seg->vmaddr;
 				textoff = seg->fileoff;
+				textsize = seg->vmsize;
 				sect = (MachSect64*)(cmdp + sizeof(MachSeg64));
-				if (strcmp(sect->sectname, "__text") == 0) {
-					textsize = swav(sect->size);
-				} else {
-					werrstr("no text section");
-					goto bad;
+				for(j = 0; j < seg->nsects; j++, sect++) {
+print("%s %#x %#x\n", sect->sectname, swal(sect->offset), swal(sect->size));
+					if (strcmp(sect->sectname, "__gosymtab") == 0) {
+						symoff = swal(sect->offset);
+						symsize = swal(sect->size);
+					}
+					if (strcmp(sect->sectname, "__gopclntab") == 0) {
+						pclnoff = swal(sect->offset);
+						pclnsize = swal(sect->size);
+					}
 				}
 			}
 			if (strcmp(seg->segname, "__DATA") == 0) {
 				datava = seg->vmaddr;
 				dataoff = seg->fileoff;
-				sect = (MachSect64*)(cmdp + sizeof(MachSeg64));
-				if (strcmp(sect->sectname, "__data") == 0) {
-					datasize = swav(sect->size);
-				} else {
-					werrstr("no data section");
-					goto bad;
-				}
-				sect++;
-				if (strcmp(sect->sectname, "__nl_symbol_ptr") == 0)
-					sect++;
-				if (strcmp(sect->sectname, "__bss") == 0) {
-					bsssize = swav(sect->size);
-				} else {
-					werrstr("no bss section");
-					goto bad;
-				}
+				datasize = seg->filesize;
+				bsssize = seg->vmsize - seg->filesize;
 			}
 			break;
 		case MACH_UNIXTHREAD:
 			break;
 		case MACH_SYMSEG:
-			if (symtab == 0)
+			if (symtab == 0) {
 				symtab = (MachSymSeg*)c;
-			else if (pclntab == 0)
+				symoff = swal(symtab->fileoff);
+				symsize = swal(symtab->filesize);
+			} else if (pclntab == 0) {
 				pclntab = (MachSymSeg*)c;
+				pclnoff = swal(pclntab->fileoff);
+				pclnsize = swal(pclntab->filesize);
+			}
 			break;
 		}
 		cmdp += c->size;
@@ -1236,8 +1224,8 @@ machdotout(int fd, Fhdr *fp, ExecHdr *hp)
 	/* compute entry by taking address after header - weird - BUG? */
 	settext(fp, textva+sizeof(Machhdr) + mp->sizeofcmds, textva, textsize, textoff);
 	setdata(fp, datava, datasize, dataoff, bsssize);
-	if(symtab != 0)
-		setsym(fp, symtab->fileoff, symtab->filesize, 0, 0, 0, pclntab? pclntab->filesize : 0);
+	if(symoff > 0)
+		setsym(fp, symoff, symsize, 0, 0, pclnoff, pclnsize);
 	free(cmd);
 	free(cmdbuf);
 	return 1;