diff --git a/scripts/recordmcount.c b/scripts/recordmcount.c
index 34f32be..7f7f718 100644
--- a/scripts/recordmcount.c
+++ b/scripts/recordmcount.c
@@ -212,550 +212,10 @@
 		0 == strcmp(".text.unlikely", txtname);
 }
 
-/* Append the new shstrtab, Elf32_Shdr[], __mcount_loc and its relocations. */
-static void append32(Elf32_Ehdr *const ehdr,
-		     Elf32_Shdr *const shstr,
-		     uint32_t const *const mloc0,
-		     uint32_t const *const mlocp,
-		     Elf32_Rel const *const mrel0,
-		     Elf32_Rel const *const mrelp,
-		     unsigned int const rel_entsize,
-		     unsigned int const symsec_sh_link)
-{
-	/* Begin constructing output file */
-	Elf32_Shdr mcsec;
-	char const *mc_name = (sizeof(Elf32_Rela) == rel_entsize)
-		? ".rela__mcount_loc"
-		:  ".rel__mcount_loc";
-	unsigned const old_shnum = w2(ehdr->e_shnum);
-	uint32_t const old_shoff = w(ehdr->e_shoff);
-	uint32_t const old_shstr_sh_size   = w(shstr->sh_size);
-	uint32_t const old_shstr_sh_offset = w(shstr->sh_offset);
-	uint32_t t = 1 + strlen(mc_name) + w(shstr->sh_size);
-	uint32_t new_e_shoff;
-
-	shstr->sh_size = w(t);
-	shstr->sh_offset = w(sb.st_size);
-	t += sb.st_size;
-	t += (3u & -t);  /* 4-byte align */
-	new_e_shoff = t;
-
-	/* body for new shstrtab */
-	ulseek(fd_map, sb.st_size, SEEK_SET);
-	uwrite(fd_map, old_shstr_sh_offset + (void *)ehdr, old_shstr_sh_size);
-	uwrite(fd_map, mc_name, 1 + strlen(mc_name));
-
-	/* old(modified) Elf32_Shdr table, 4-byte aligned */
-	ulseek(fd_map, t, SEEK_SET);
-	t += sizeof(Elf32_Shdr) * old_shnum;
-	uwrite(fd_map, old_shoff + (void *)ehdr,
-	       sizeof(Elf32_Shdr) * old_shnum);
-
-	/* new sections __mcount_loc and .rel__mcount_loc */
-	t += 2*sizeof(mcsec);
-	mcsec.sh_name = w((sizeof(Elf32_Rela) == rel_entsize) + strlen(".rel")
-		+ old_shstr_sh_size);
-	mcsec.sh_type = w(SHT_PROGBITS);
-	mcsec.sh_flags = w(SHF_ALLOC);
-	mcsec.sh_addr = 0;
-	mcsec.sh_offset = w(t);
-	mcsec.sh_size = w((void *)mlocp - (void *)mloc0);
-	mcsec.sh_link = 0;
-	mcsec.sh_info = 0;
-	mcsec.sh_addralign = w(4);
-	mcsec.sh_entsize = w(4);
-	uwrite(fd_map, &mcsec, sizeof(mcsec));
-
-	mcsec.sh_name = w(old_shstr_sh_size);
-	mcsec.sh_type = (sizeof(Elf32_Rela) == rel_entsize)
-		? w(SHT_RELA)
-		: w(SHT_REL);
-	mcsec.sh_flags = 0;
-	mcsec.sh_addr = 0;
-	mcsec.sh_offset = w((void *)mlocp - (void *)mloc0 + t);
-	mcsec.sh_size   = w((void *)mrelp - (void *)mrel0);
-	mcsec.sh_link = w(symsec_sh_link);
-	mcsec.sh_info = w(old_shnum);
-	mcsec.sh_addralign = w(4);
-	mcsec.sh_entsize = w(rel_entsize);
-	uwrite(fd_map, &mcsec, sizeof(mcsec));
-
-	uwrite(fd_map, mloc0, (void *)mlocp - (void *)mloc0);
-	uwrite(fd_map, mrel0, (void *)mrelp - (void *)mrel0);
-
-	ehdr->e_shoff = w(new_e_shoff);
-	ehdr->e_shnum = w2(2 + w2(ehdr->e_shnum));  /* {.rel,}__mcount_loc */
-	ulseek(fd_map, 0, SEEK_SET);
-	uwrite(fd_map, ehdr, sizeof(*ehdr));
-}
-
-/*
- * append64 and append32 (and other analogous pairs) could be templated
- * using C++, but the complexity is high.  (For an example, look at p_elf.h
- * in the source for UPX, http://upx.sourceforge.net)  So: remember to make
- * the corresponding change in the routine for the other size.
- */
-static void append64(Elf64_Ehdr *const ehdr,
-		     Elf64_Shdr *const shstr,
-		     uint64_t const *const mloc0,
-		     uint64_t const *const mlocp,
-		     Elf64_Rel const *const mrel0,
-		     Elf64_Rel const *const mrelp,
-		     unsigned int const rel_entsize,
-		     unsigned int const symsec_sh_link)
-{
-	/* Begin constructing output file */
-	Elf64_Shdr mcsec;
-	char const *mc_name = (sizeof(Elf64_Rela) == rel_entsize)
-		? ".rela__mcount_loc"
-		:  ".rel__mcount_loc";
-	unsigned const old_shnum = w2(ehdr->e_shnum);
-	uint64_t const old_shoff = w8(ehdr->e_shoff);
-	uint64_t const old_shstr_sh_size   = w8(shstr->sh_size);
-	uint64_t const old_shstr_sh_offset = w8(shstr->sh_offset);
-	uint64_t t = 1 + strlen(mc_name) + w8(shstr->sh_size);
-	uint64_t new_e_shoff;
-
-	shstr->sh_size = w8(t);
-	shstr->sh_offset = w8(sb.st_size);
-	t += sb.st_size;
-	t += (7u & -t);  /* 8-byte align */
-	new_e_shoff = t;
-
-	/* body for new shstrtab */
-	ulseek(fd_map, sb.st_size, SEEK_SET);
-	uwrite(fd_map, old_shstr_sh_offset + (void *)ehdr, old_shstr_sh_size);
-	uwrite(fd_map, mc_name, 1 + strlen(mc_name));
-
-	/* old(modified) Elf64_Shdr table, 8-byte aligned */
-	ulseek(fd_map, t, SEEK_SET);
-	t += sizeof(Elf64_Shdr) * old_shnum;
-	uwrite(fd_map, old_shoff + (void *)ehdr,
-		sizeof(Elf64_Shdr) * old_shnum);
-
-	/* new sections __mcount_loc and .rel__mcount_loc */
-	t += 2*sizeof(mcsec);
-	mcsec.sh_name = w((sizeof(Elf64_Rela) == rel_entsize) + strlen(".rel")
-		+ old_shstr_sh_size);
-	mcsec.sh_type = w(SHT_PROGBITS);
-	mcsec.sh_flags = w8(SHF_ALLOC);
-	mcsec.sh_addr = 0;
-	mcsec.sh_offset = w8(t);
-	mcsec.sh_size = w8((void *)mlocp - (void *)mloc0);
-	mcsec.sh_link = 0;
-	mcsec.sh_info = 0;
-	mcsec.sh_addralign = w8(8);
-	mcsec.sh_entsize = w8(8);
-	uwrite(fd_map, &mcsec, sizeof(mcsec));
-
-	mcsec.sh_name = w(old_shstr_sh_size);
-	mcsec.sh_type = (sizeof(Elf64_Rela) == rel_entsize)
-		? w(SHT_RELA)
-		: w(SHT_REL);
-	mcsec.sh_flags = 0;
-	mcsec.sh_addr = 0;
-	mcsec.sh_offset = w8((void *)mlocp - (void *)mloc0 + t);
-	mcsec.sh_size   = w8((void *)mrelp - (void *)mrel0);
-	mcsec.sh_link = w(symsec_sh_link);
-	mcsec.sh_info = w(old_shnum);
-	mcsec.sh_addralign = w8(8);
-	mcsec.sh_entsize = w8(rel_entsize);
-	uwrite(fd_map, &mcsec, sizeof(mcsec));
-
-	uwrite(fd_map, mloc0, (void *)mlocp - (void *)mloc0);
-	uwrite(fd_map, mrel0, (void *)mrelp - (void *)mrel0);
-
-	ehdr->e_shoff = w8(new_e_shoff);
-	ehdr->e_shnum = w2(2 + w2(ehdr->e_shnum));  /* {.rel,}__mcount_loc */
-	ulseek(fd_map, 0, SEEK_SET);
-	uwrite(fd_map, ehdr, sizeof(*ehdr));
-}
-
-/*
- * Look at the relocations in order to find the calls to mcount.
- * Accumulate the section offsets that are found, and their relocation info,
- * onto the end of the existing arrays.
- */
-static uint32_t *sift32_rel_mcount(uint32_t *mlocp,
-				   unsigned const offbase,
-				   Elf32_Rel **const mrelpp,
-				   Elf32_Shdr const *const relhdr,
-				   Elf32_Ehdr const *const ehdr,
-				   unsigned const recsym,
-				   uint32_t const recval,
-				   unsigned const reltype)
-{
-	uint32_t *const mloc0 = mlocp;
-	Elf32_Rel *mrelp = *mrelpp;
-	Elf32_Shdr *const shdr0 = (Elf32_Shdr *)(w(ehdr->e_shoff)
-		+ (void *)ehdr);
-	unsigned const symsec_sh_link = w(relhdr->sh_link);
-	Elf32_Shdr const *const symsec = &shdr0[symsec_sh_link];
-	Elf32_Sym const *const sym0 = (Elf32_Sym const *)(w(symsec->sh_offset)
-		+ (void *)ehdr);
-
-	Elf32_Shdr const *const strsec = &shdr0[w(symsec->sh_link)];
-	char const *const str0 = (char const *)(w(strsec->sh_offset)
-		+ (void *)ehdr);
-
-	Elf32_Rel const *const rel0 = (Elf32_Rel const *)(w(relhdr->sh_offset)
-		+ (void *)ehdr);
-	unsigned rel_entsize = w(relhdr->sh_entsize);
-	unsigned const nrel = w(relhdr->sh_size) / rel_entsize;
-	Elf32_Rel const *relp = rel0;
-
-	unsigned mcountsym = 0;
-	unsigned t;
-
-	for (t = nrel; t; --t) {
-		if (!mcountsym) {
-			Elf32_Sym const *const symp =
-				&sym0[ELF32_R_SYM(w(relp->r_info))];
-
-			if (0 == strcmp((('_' == gpfx) ? "_mcount" : "mcount"),
-					&str0[w(symp->st_name)]))
-				mcountsym = ELF32_R_SYM(w(relp->r_info));
-		}
-		if (mcountsym == ELF32_R_SYM(w(relp->r_info))) {
-			uint32_t const addend = w(w(relp->r_offset) - recval);
-			mrelp->r_offset = w(offbase
-				+ ((void *)mlocp - (void *)mloc0));
-			mrelp->r_info = w(ELF32_R_INFO(recsym, reltype));
-			if (sizeof(Elf32_Rela) == rel_entsize) {
-				((Elf32_Rela *)mrelp)->r_addend = addend;
-				*mlocp++ = 0;
-			} else
-				*mlocp++ = addend;
-
-			mrelp = (Elf32_Rel *)(rel_entsize + (void *)mrelp);
-		}
-		relp = (Elf32_Rel const *)(rel_entsize + (void *)relp);
-	}
-	*mrelpp = mrelp;
-	return mlocp;
-}
-
-static uint64_t *sift64_rel_mcount(uint64_t *mlocp,
-				   unsigned const offbase,
-				   Elf64_Rel **const mrelpp,
-				   Elf64_Shdr const *const relhdr,
-				   Elf64_Ehdr const *const ehdr,
-				   unsigned const recsym,
-				   uint64_t const recval,
-				   unsigned const reltype)
-{
-	uint64_t *const mloc0 = mlocp;
-	Elf64_Rel *mrelp = *mrelpp;
-	Elf64_Shdr *const shdr0 = (Elf64_Shdr *)(w8(ehdr->e_shoff)
-		+ (void *)ehdr);
-	unsigned const symsec_sh_link = w(relhdr->sh_link);
-	Elf64_Shdr const *const symsec = &shdr0[symsec_sh_link];
-	Elf64_Sym const *const sym0 = (Elf64_Sym const *)(w8(symsec->sh_offset)
-		+ (void *)ehdr);
-
-	Elf64_Shdr const *const strsec = &shdr0[w(symsec->sh_link)];
-	char const *const str0 = (char const *)(w8(strsec->sh_offset)
-		+ (void *)ehdr);
-
-	Elf64_Rel const *const rel0 = (Elf64_Rel const *)(w8(relhdr->sh_offset)
-		+ (void *)ehdr);
-	unsigned rel_entsize = w8(relhdr->sh_entsize);
-	unsigned const nrel = w8(relhdr->sh_size) / rel_entsize;
-	Elf64_Rel const *relp = rel0;
-
-	unsigned mcountsym = 0;
-	unsigned t;
-
-	for (t = nrel; 0 != t; --t) {
-		if (!mcountsym) {
-			Elf64_Sym const *const symp =
-				&sym0[ELF64_R_SYM(w8(relp->r_info))];
-			char const *symname = &str0[w(symp->st_name)];
-
-			if ('.' == symname[0])
-				++symname;  /* ppc64 hack */
-			if (0 == strcmp((('_' == gpfx) ? "_mcount" : "mcount"),
-					symname))
-				mcountsym = ELF64_R_SYM(w8(relp->r_info));
-		}
-
-		if (mcountsym == ELF64_R_SYM(w8(relp->r_info))) {
-			uint64_t const addend = w8(w8(relp->r_offset) - recval);
-
-			mrelp->r_offset = w8(offbase
-				+ ((void *)mlocp - (void *)mloc0));
-			mrelp->r_info = w8(ELF64_R_INFO(recsym, reltype));
-			if (sizeof(Elf64_Rela) == rel_entsize) {
-				((Elf64_Rela *)mrelp)->r_addend = addend;
-				*mlocp++ = 0;
-			} else
-				*mlocp++ = addend;
-
-			mrelp = (Elf64_Rel *)(rel_entsize + (void *)mrelp);
-		}
-		relp = (Elf64_Rel const *)(rel_entsize + (void *)relp);
-	}
-	*mrelpp = mrelp;
-
-	return mlocp;
-}
-
-/*
- * Find a symbol in the given section, to be used as the base for relocating
- * the table of offsets of calls to mcount.  A local or global symbol suffices,
- * but avoid a Weak symbol because it may be overridden; the change in value
- * would invalidate the relocations of the offsets of the calls to mcount.
- * Often the found symbol will be the unnamed local symbol generated by
- * GNU 'as' for the start of each section.  For example:
- *    Num:    Value  Size Type    Bind   Vis      Ndx Name
- *      2: 00000000     0 SECTION LOCAL  DEFAULT    1
- */
-static unsigned find32_secsym_ndx(unsigned const txtndx,
-				  char const *const txtname,
-				  uint32_t *const recvalp,
-				  Elf32_Shdr const *const symhdr,
-				  Elf32_Ehdr const *const ehdr)
-{
-	Elf32_Sym const *const sym0 = (Elf32_Sym const *)(w(symhdr->sh_offset)
-		+ (void *)ehdr);
-	unsigned const nsym = w(symhdr->sh_size) / w(symhdr->sh_entsize);
-	Elf32_Sym const *symp;
-	unsigned t;
-
-	for (symp = sym0, t = nsym; t; --t, ++symp) {
-		unsigned int const st_bind = ELF32_ST_BIND(symp->st_info);
-
-		if (txtndx == w2(symp->st_shndx)
-			/* avoid STB_WEAK */
-		    && (STB_LOCAL == st_bind || STB_GLOBAL == st_bind)) {
-			*recvalp = w(symp->st_value);
-			return symp - sym0;
-		}
-	}
-	fprintf(stderr, "Cannot find symbol for section %d: %s.\n",
-		txtndx, txtname);
-	fail_file();
-}
-
-static unsigned find64_secsym_ndx(unsigned const txtndx,
-				  char const *const txtname,
-				  uint64_t *const recvalp,
-				  Elf64_Shdr const *const symhdr,
-				  Elf64_Ehdr const *const ehdr)
-{
-	Elf64_Sym const *const sym0 = (Elf64_Sym const *)(w8(symhdr->sh_offset)
-		+ (void *)ehdr);
-	unsigned const nsym = w8(symhdr->sh_size) / w8(symhdr->sh_entsize);
-	Elf64_Sym const *symp;
-	unsigned t;
-
-	for (symp = sym0, t = nsym; t; --t, ++symp) {
-		unsigned int const st_bind = ELF64_ST_BIND(symp->st_info);
-
-		if (txtndx == w2(symp->st_shndx)
-			/* avoid STB_WEAK */
-		    && (STB_LOCAL == st_bind || STB_GLOBAL == st_bind)) {
-			*recvalp = w8(symp->st_value);
-			return symp - sym0;
-		}
-	}
-	fprintf(stderr, "Cannot find symbol for section %d: %s.\n",
-		txtndx, txtname);
-	fail_file();
-}
-
-/*
- * Evade ISO C restriction: no declaration after statement in
- * has32_rel_mcount.
- */
-static char const *
-__has32_rel_mcount(Elf32_Shdr const *const relhdr,  /* is SHT_REL or SHT_RELA */
-		   Elf32_Shdr const *const shdr0,
-		   char const *const shstrtab,
-		   char const *const fname)
-{
-	/* .sh_info depends on .sh_type == SHT_REL[,A] */
-	Elf32_Shdr const *const txthdr = &shdr0[w(relhdr->sh_info)];
-	char const *const txtname = &shstrtab[w(txthdr->sh_name)];
-
-	if (0 == strcmp("__mcount_loc", txtname)) {
-		fprintf(stderr, "warning: __mcount_loc already exists: %s\n",
-			fname);
-		succeed_file();
-	}
-	if (SHT_PROGBITS != w(txthdr->sh_type) ||
-	    !is_mcounted_section_name(txtname))
-		return NULL;
-	return txtname;
-}
-
-static char const *has32_rel_mcount(Elf32_Shdr const *const relhdr,
-				    Elf32_Shdr const *const shdr0,
-				    char const *const shstrtab,
-				    char const *const fname)
-{
-	if (SHT_REL  != w(relhdr->sh_type) && SHT_RELA != w(relhdr->sh_type))
-		return NULL;
-	return __has32_rel_mcount(relhdr, shdr0, shstrtab, fname);
-}
-
-static char const *__has64_rel_mcount(Elf64_Shdr const *const relhdr,
-				      Elf64_Shdr const *const shdr0,
-				      char const *const shstrtab,
-				      char const *const fname)
-{
-	/* .sh_info depends on .sh_type == SHT_REL[,A] */
-	Elf64_Shdr const *const txthdr = &shdr0[w(relhdr->sh_info)];
-	char const *const txtname = &shstrtab[w(txthdr->sh_name)];
-
-	if (0 == strcmp("__mcount_loc", txtname)) {
-		fprintf(stderr, "warning: __mcount_loc already exists: %s\n",
-			fname);
-		succeed_file();
-	}
-	if (SHT_PROGBITS != w(txthdr->sh_type) ||
-	    !is_mcounted_section_name(txtname))
-		return NULL;
-	return txtname;
-}
-
-static char const *has64_rel_mcount(Elf64_Shdr const *const relhdr,
-				    Elf64_Shdr const *const shdr0,
-				    char const *const shstrtab,
-				    char const *const fname)
-{
-	if (SHT_REL  != w(relhdr->sh_type) && SHT_RELA != w(relhdr->sh_type))
-		return NULL;
-	return __has64_rel_mcount(relhdr, shdr0, shstrtab, fname);
-}
-
-static unsigned tot32_relsize(Elf32_Shdr const *const shdr0,
-			      unsigned nhdr,
-			      const char *const shstrtab,
-			      const char *const fname)
-{
-	unsigned totrelsz = 0;
-	Elf32_Shdr const *shdrp = shdr0;
-	for (; 0 != nhdr; --nhdr, ++shdrp) {
-		if (has32_rel_mcount(shdrp, shdr0, shstrtab, fname))
-			totrelsz += w(shdrp->sh_size);
-	}
-	return totrelsz;
-}
-
-static unsigned tot64_relsize(Elf64_Shdr const *const shdr0,
-			      unsigned nhdr,
-			      const char *const shstrtab,
-			      const char *const fname)
-{
-	unsigned totrelsz = 0;
-	Elf64_Shdr const *shdrp = shdr0;
-
-	for (; nhdr; --nhdr, ++shdrp) {
-		if (has64_rel_mcount(shdrp, shdr0, shstrtab, fname))
-			totrelsz += w8(shdrp->sh_size);
-	}
-	return totrelsz;
-}
-
-/* Overall supervision for Elf32 ET_REL file. */
-static void
-do32(Elf32_Ehdr *const ehdr, char const *const fname, unsigned const reltype)
-{
-	Elf32_Shdr *const shdr0 = (Elf32_Shdr *)(w(ehdr->e_shoff)
-		+ (void *)ehdr);
-	unsigned const nhdr = w2(ehdr->e_shnum);
-	Elf32_Shdr *const shstr = &shdr0[w2(ehdr->e_shstrndx)];
-	char const *const shstrtab = (char const *)(w(shstr->sh_offset)
-		+ (void *)ehdr);
-
-	Elf32_Shdr const *relhdr;
-	unsigned k;
-
-	/* Upper bound on space: assume all relevant relocs are for mcount. */
-	unsigned const totrelsz = tot32_relsize(shdr0, nhdr, shstrtab, fname);
-	Elf32_Rel *const mrel0 = umalloc(totrelsz);
-	Elf32_Rel *      mrelp = mrel0;
-
-	/* 2*sizeof(address) <= sizeof(Elf32_Rel) */
-	uint32_t *const mloc0 = umalloc(totrelsz>>1);
-	uint32_t *      mlocp = mloc0;
-
-	unsigned rel_entsize = 0;
-	unsigned symsec_sh_link = 0;
-
-	for (relhdr = shdr0, k = nhdr; k; --k, ++relhdr) {
-		char const *const txtname = has32_rel_mcount(relhdr, shdr0,
-			shstrtab, fname);
-		if (txtname) {
-			uint32_t recval = 0;
-			unsigned const recsym = find32_secsym_ndx(
-				w(relhdr->sh_info), txtname, &recval,
-				&shdr0[symsec_sh_link = w(relhdr->sh_link)],
-				ehdr);
-
-			rel_entsize = w(relhdr->sh_entsize);
-			mlocp = sift32_rel_mcount(mlocp,
-				(void *)mlocp - (void *)mloc0, &mrelp,
-				relhdr, ehdr, recsym, recval, reltype);
-		}
-	}
-	if (mloc0 != mlocp) {
-		append32(ehdr, shstr, mloc0, mlocp, mrel0, mrelp,
-			rel_entsize, symsec_sh_link);
-	}
-	free(mrel0);
-	free(mloc0);
-}
-
-static void
-do64(Elf64_Ehdr *const ehdr, char const *const fname, unsigned const reltype)
-{
-	Elf64_Shdr *const shdr0 = (Elf64_Shdr *)(w8(ehdr->e_shoff)
-		+ (void *)ehdr);
-	unsigned const nhdr = w2(ehdr->e_shnum);
-	Elf64_Shdr *const shstr = &shdr0[w2(ehdr->e_shstrndx)];
-	char const *const shstrtab = (char const *)(w8(shstr->sh_offset)
-		+ (void *)ehdr);
-
-	Elf64_Shdr const *relhdr;
-	unsigned k;
-
-	/* Upper bound on space: assume all relevant relocs are for mcount. */
-	unsigned const totrelsz = tot64_relsize(shdr0, nhdr, shstrtab, fname);
-	Elf64_Rel *const mrel0 = umalloc(totrelsz);
-	Elf64_Rel *      mrelp = mrel0;
-
-	/* 2*sizeof(address) <= sizeof(Elf64_Rel) */
-	uint64_t *const mloc0 = umalloc(totrelsz>>1);
-	uint64_t *      mlocp = mloc0;
-
-	unsigned rel_entsize = 0;
-	unsigned symsec_sh_link = 0;
-
-	for ((relhdr = shdr0), k = nhdr; k; --k, ++relhdr) {
-		char const *const txtname = has64_rel_mcount(relhdr, shdr0,
-			shstrtab, fname);
-		if (txtname) {
-			uint64_t recval = 0;
-			unsigned const recsym = find64_secsym_ndx(
-				w(relhdr->sh_info), txtname, &recval,
-				&shdr0[symsec_sh_link = w(relhdr->sh_link)],
-				ehdr);
-
-			rel_entsize = w8(relhdr->sh_entsize);
-			mlocp = sift64_rel_mcount(mlocp,
-				(void *)mlocp - (void *)mloc0, &mrelp,
-				relhdr, ehdr, recsym, recval, reltype);
-		}
-	}
-	if (mloc0 != mlocp) {
-		append64(ehdr, shstr, mloc0, mlocp, mrel0, mrelp,
-			rel_entsize, symsec_sh_link);
-	}
-	free(mrel0);
-	free(mloc0);
-}
+/* 32 bit and 64 bit are very similar */
+#include "recordmcount.h"
+#define RECORD_MCOUNT_64
+#include "recordmcount.h"
 
 static void
 do_file(char const *const fname)
diff --git a/scripts/recordmcount.h b/scripts/recordmcount.h
new file mode 100644
index 0000000..7f39d09
--- /dev/null
+++ b/scripts/recordmcount.h
@@ -0,0 +1,366 @@
+/*
+ * recordmcount.h
+ *
+ * This code was taken out of recordmcount.c written by
+ * Copyright 2009 John F. Reiser <jreiser@BitWagon.com>.  All rights reserved.
+ *
+ * The original code had the same algorithms for both 32bit
+ * and 64bit ELF files, but the code was duplicated to support
+ * the difference in structures that were used. This
+ * file creates a macro of everything that is different between
+ * the 64 and 32 bit code, such that by including this header
+ * twice we can create both sets of functions by including this
+ * header once with RECORD_MCOUNT_64 undefined, and again with
+ * it defined.
+ *
+ * This conversion to macros was done by:
+ * Copyright 2010 Steven Rostedt <srostedt@redhat.com>, Red Hat Inc.
+ *
+ * Licensed under the GNU General Public License, version 2 (GPLv2).
+ */
+#undef append_func
+#undef sift_rel_mcount
+#undef find_secsym_ndx
+#undef __has_rel_mcount
+#undef has_rel_mcount
+#undef tot_relsize
+#undef do_func
+#undef Elf_Ehdr
+#undef Elf_Shdr
+#undef Elf_Rel
+#undef Elf_Rela
+#undef Elf_Sym
+#undef ELF_R_SYM
+#undef ELF_R_INFO
+#undef ELF_ST_BIND
+#undef uint_t
+#undef _w
+#undef _align
+#undef _size
+
+#ifdef RECORD_MCOUNT_64
+# define append_func		append64
+# define sift_rel_mcount	sift64_rel_mcount
+# define find_secsym_ndx	find64_secsym_ndx
+# define __has_rel_mcount	__has64_rel_mcount
+# define has_rel_mcount		has64_rel_mcount
+# define tot_relsize		tot64_relsize
+# define do_func		do64
+# define Elf_Ehdr		Elf64_Ehdr
+# define Elf_Shdr		Elf64_Shdr
+# define Elf_Rel		Elf64_Rel
+# define Elf_Rela		Elf64_Rela
+# define Elf_Sym		Elf64_Sym
+# define ELF_R_SYM		ELF64_R_SYM
+# define ELF_R_INFO		ELF64_R_INFO
+# define ELF_ST_BIND		ELF64_ST_BIND
+# define uint_t			uint64_t
+# define _w			w8
+# define _align			7u
+# define _size			8
+#else
+# define append_func		append32
+# define sift_rel_mcount	sift32_rel_mcount
+# define find_secsym_ndx	find32_secsym_ndx
+# define __has_rel_mcount	__has32_rel_mcount
+# define has_rel_mcount		has32_rel_mcount
+# define tot_relsize		tot32_relsize
+# define do_func		do32
+# define Elf_Ehdr		Elf32_Ehdr
+# define Elf_Shdr		Elf32_Shdr
+# define Elf_Rel		Elf32_Rel
+# define Elf_Rela		Elf32_Rela
+# define Elf_Sym		Elf32_Sym
+# define ELF_R_SYM		ELF32_R_SYM
+# define ELF_R_INFO		ELF32_R_INFO
+# define ELF_ST_BIND		ELF32_ST_BIND
+# define uint_t			uint32_t
+# define _w			w
+# define _align			3u
+# define _size			4
+#endif
+
+/* Append the new shstrtab, Elf_Shdr[], __mcount_loc and its relocations. */
+static void append_func(Elf_Ehdr *const ehdr,
+			Elf_Shdr *const shstr,
+			uint_t const *const mloc0,
+			uint_t const *const mlocp,
+			Elf_Rel const *const mrel0,
+			Elf_Rel const *const mrelp,
+			unsigned int const rel_entsize,
+			unsigned int const symsec_sh_link)
+{
+	/* Begin constructing output file */
+	Elf_Shdr mcsec;
+	char const *mc_name = (sizeof(Elf_Rela) == rel_entsize)
+		? ".rela__mcount_loc"
+		:  ".rel__mcount_loc";
+	unsigned const old_shnum = w2(ehdr->e_shnum);
+	uint_t const old_shoff = _w(ehdr->e_shoff);
+	uint_t const old_shstr_sh_size   = _w(shstr->sh_size);
+	uint_t const old_shstr_sh_offset = _w(shstr->sh_offset);
+	uint_t t = 1 + strlen(mc_name) + _w(shstr->sh_size);
+	uint_t new_e_shoff;
+
+	shstr->sh_size = _w(t);
+	shstr->sh_offset = _w(sb.st_size);
+	t += sb.st_size;
+	t += (_align & -t);  /* word-byte align */
+	new_e_shoff = t;
+
+	/* body for new shstrtab */
+	ulseek(fd_map, sb.st_size, SEEK_SET);
+	uwrite(fd_map, old_shstr_sh_offset + (void *)ehdr, old_shstr_sh_size);
+	uwrite(fd_map, mc_name, 1 + strlen(mc_name));
+
+	/* old(modified) Elf_Shdr table, word-byte aligned */
+	ulseek(fd_map, t, SEEK_SET);
+	t += sizeof(Elf_Shdr) * old_shnum;
+	uwrite(fd_map, old_shoff + (void *)ehdr,
+	       sizeof(Elf_Shdr) * old_shnum);
+
+	/* new sections __mcount_loc and .rel__mcount_loc */
+	t += 2*sizeof(mcsec);
+	mcsec.sh_name = w((sizeof(Elf_Rela) == rel_entsize) + strlen(".rel")
+		+ old_shstr_sh_size);
+	mcsec.sh_type = w(SHT_PROGBITS);
+	mcsec.sh_flags = _w(SHF_ALLOC);
+	mcsec.sh_addr = 0;
+	mcsec.sh_offset = _w(t);
+	mcsec.sh_size = _w((void *)mlocp - (void *)mloc0);
+	mcsec.sh_link = 0;
+	mcsec.sh_info = 0;
+	mcsec.sh_addralign = _w(_size);
+	mcsec.sh_entsize = _w(_size);
+	uwrite(fd_map, &mcsec, sizeof(mcsec));
+
+	mcsec.sh_name = w(old_shstr_sh_size);
+	mcsec.sh_type = (sizeof(Elf_Rela) == rel_entsize)
+		? w(SHT_RELA)
+		: w(SHT_REL);
+	mcsec.sh_flags = 0;
+	mcsec.sh_addr = 0;
+	mcsec.sh_offset = _w((void *)mlocp - (void *)mloc0 + t);
+	mcsec.sh_size   = _w((void *)mrelp - (void *)mrel0);
+	mcsec.sh_link = w(symsec_sh_link);
+	mcsec.sh_info = w(old_shnum);
+	mcsec.sh_addralign = _w(_size);
+	mcsec.sh_entsize = _w(rel_entsize);
+	uwrite(fd_map, &mcsec, sizeof(mcsec));
+
+	uwrite(fd_map, mloc0, (void *)mlocp - (void *)mloc0);
+	uwrite(fd_map, mrel0, (void *)mrelp - (void *)mrel0);
+
+	ehdr->e_shoff = _w(new_e_shoff);
+	ehdr->e_shnum = w2(2 + w2(ehdr->e_shnum));  /* {.rel,}__mcount_loc */
+	ulseek(fd_map, 0, SEEK_SET);
+	uwrite(fd_map, ehdr, sizeof(*ehdr));
+}
+
+
+/*
+ * Look at the relocations in order to find the calls to mcount.
+ * Accumulate the section offsets that are found, and their relocation info,
+ * onto the end of the existing arrays.
+ */
+static uint_t *sift_rel_mcount(uint_t *mlocp,
+			       unsigned const offbase,
+			       Elf_Rel **const mrelpp,
+			       Elf_Shdr const *const relhdr,
+			       Elf_Ehdr const *const ehdr,
+			       unsigned const recsym,
+			       uint_t const recval,
+			       unsigned const reltype)
+{
+	uint_t *const mloc0 = mlocp;
+	Elf_Rel *mrelp = *mrelpp;
+	Elf_Shdr *const shdr0 = (Elf_Shdr *)(_w(ehdr->e_shoff)
+		+ (void *)ehdr);
+	unsigned const symsec_sh_link = w(relhdr->sh_link);
+	Elf_Shdr const *const symsec = &shdr0[symsec_sh_link];
+	Elf_Sym const *const sym0 = (Elf_Sym const *)(_w(symsec->sh_offset)
+		+ (void *)ehdr);
+
+	Elf_Shdr const *const strsec = &shdr0[w(symsec->sh_link)];
+	char const *const str0 = (char const *)(_w(strsec->sh_offset)
+		+ (void *)ehdr);
+
+	Elf_Rel const *const rel0 = (Elf_Rel const *)(_w(relhdr->sh_offset)
+		+ (void *)ehdr);
+	unsigned rel_entsize = _w(relhdr->sh_entsize);
+	unsigned const nrel = _w(relhdr->sh_size) / rel_entsize;
+	Elf_Rel const *relp = rel0;
+
+	unsigned mcountsym = 0;
+	unsigned t;
+
+	for (t = nrel; t; --t) {
+		if (!mcountsym) {
+			Elf_Sym const *const symp =
+				&sym0[ELF_R_SYM(_w(relp->r_info))];
+			char const *symname = &str0[w(symp->st_name)];
+
+			if ('.' == symname[0])
+				++symname;  /* ppc64 hack */
+			if (0 == strcmp((('_' == gpfx) ? "_mcount" : "mcount"),
+					symname))
+				mcountsym = ELF_R_SYM(_w(relp->r_info));
+		}
+
+		if (mcountsym == ELF_R_SYM(_w(relp->r_info))) {
+			uint_t const addend = _w(_w(relp->r_offset) - recval);
+
+			mrelp->r_offset = _w(offbase
+				+ ((void *)mlocp - (void *)mloc0));
+			mrelp->r_info = _w(ELF_R_INFO(recsym, reltype));
+			if (sizeof(Elf_Rela) == rel_entsize) {
+				((Elf_Rela *)mrelp)->r_addend = addend;
+				*mlocp++ = 0;
+			} else
+				*mlocp++ = addend;
+
+			mrelp = (Elf_Rel *)(rel_entsize + (void *)mrelp);
+		}
+		relp = (Elf_Rel const *)(rel_entsize + (void *)relp);
+	}
+	*mrelpp = mrelp;
+	return mlocp;
+}
+
+
+/*
+ * Find a symbol in the given section, to be used as the base for relocating
+ * the table of offsets of calls to mcount.  A local or global symbol suffices,
+ * but avoid a Weak symbol because it may be overridden; the change in value
+ * would invalidate the relocations of the offsets of the calls to mcount.
+ * Often the found symbol will be the unnamed local symbol generated by
+ * GNU 'as' for the start of each section.  For example:
+ *    Num:    Value  Size Type    Bind   Vis      Ndx Name
+ *      2: 00000000     0 SECTION LOCAL  DEFAULT    1
+ */
+static unsigned find_secsym_ndx(unsigned const txtndx,
+				char const *const txtname,
+				uint_t *const recvalp,
+				Elf_Shdr const *const symhdr,
+				Elf_Ehdr const *const ehdr)
+{
+	Elf_Sym const *const sym0 = (Elf_Sym const *)(_w(symhdr->sh_offset)
+		+ (void *)ehdr);
+	unsigned const nsym = _w(symhdr->sh_size) / _w(symhdr->sh_entsize);
+	Elf_Sym const *symp;
+	unsigned t;
+
+	for (symp = sym0, t = nsym; t; --t, ++symp) {
+		unsigned int const st_bind = ELF_ST_BIND(symp->st_info);
+
+		if (txtndx == w2(symp->st_shndx)
+			/* avoid STB_WEAK */
+		    && (STB_LOCAL == st_bind || STB_GLOBAL == st_bind)) {
+			*recvalp = _w(symp->st_value);
+			return symp - sym0;
+		}
+	}
+	fprintf(stderr, "Cannot find symbol for section %d: %s.\n",
+		txtndx, txtname);
+	fail_file();
+}
+
+
+/* Evade ISO C restriction: no declaration after statement in has_rel_mcount. */
+static char const *
+__has_rel_mcount(Elf_Shdr const *const relhdr,  /* is SHT_REL or SHT_RELA */
+		 Elf_Shdr const *const shdr0,
+		 char const *const shstrtab,
+		 char const *const fname)
+{
+	/* .sh_info depends on .sh_type == SHT_REL[,A] */
+	Elf_Shdr const *const txthdr = &shdr0[w(relhdr->sh_info)];
+	char const *const txtname = &shstrtab[w(txthdr->sh_name)];
+
+	if (0 == strcmp("__mcount_loc", txtname)) {
+		fprintf(stderr, "warning: __mcount_loc already exists: %s\n",
+			fname);
+		succeed_file();
+	}
+	if (SHT_PROGBITS != w(txthdr->sh_type) ||
+	    !is_mcounted_section_name(txtname))
+		return NULL;
+	return txtname;
+}
+
+static char const *has_rel_mcount(Elf_Shdr const *const relhdr,
+				  Elf_Shdr const *const shdr0,
+				  char const *const shstrtab,
+				  char const *const fname)
+{
+	if (SHT_REL  != w(relhdr->sh_type) && SHT_RELA != w(relhdr->sh_type))
+		return NULL;
+	return __has_rel_mcount(relhdr, shdr0, shstrtab, fname);
+}
+
+
+static unsigned tot_relsize(Elf_Shdr const *const shdr0,
+			    unsigned nhdr,
+			    const char *const shstrtab,
+			    const char *const fname)
+{
+	unsigned totrelsz = 0;
+	Elf_Shdr const *shdrp = shdr0;
+
+	for (; nhdr; --nhdr, ++shdrp) {
+		if (has_rel_mcount(shdrp, shdr0, shstrtab, fname))
+			totrelsz += _w(shdrp->sh_size);
+	}
+	return totrelsz;
+}
+
+
+/* Overall supervision for Elf32 ET_REL file. */
+static void
+do_func(Elf_Ehdr *const ehdr, char const *const fname, unsigned const reltype)
+{
+	Elf_Shdr *const shdr0 = (Elf_Shdr *)(_w(ehdr->e_shoff)
+		+ (void *)ehdr);
+	unsigned const nhdr = w2(ehdr->e_shnum);
+	Elf_Shdr *const shstr = &shdr0[w2(ehdr->e_shstrndx)];
+	char const *const shstrtab = (char const *)(_w(shstr->sh_offset)
+		+ (void *)ehdr);
+
+	Elf_Shdr const *relhdr;
+	unsigned k;
+
+	/* Upper bound on space: assume all relevant relocs are for mcount. */
+	unsigned const totrelsz = tot_relsize(shdr0, nhdr, shstrtab, fname);
+	Elf_Rel *const mrel0 = umalloc(totrelsz);
+	Elf_Rel *      mrelp = mrel0;
+
+	/* 2*sizeof(address) <= sizeof(Elf_Rel) */
+	uint_t *const mloc0 = umalloc(totrelsz>>1);
+	uint_t *      mlocp = mloc0;
+
+	unsigned rel_entsize = 0;
+	unsigned symsec_sh_link = 0;
+
+	for (relhdr = shdr0, k = nhdr; k; --k, ++relhdr) {
+		char const *const txtname = has_rel_mcount(relhdr, shdr0,
+			shstrtab, fname);
+		if (txtname) {
+			uint_t recval = 0;
+			unsigned const recsym = find_secsym_ndx(
+				w(relhdr->sh_info), txtname, &recval,
+				&shdr0[symsec_sh_link = w(relhdr->sh_link)],
+				ehdr);
+
+			rel_entsize = _w(relhdr->sh_entsize);
+			mlocp = sift_rel_mcount(mlocp,
+				(void *)mlocp - (void *)mloc0, &mrelp,
+				relhdr, ehdr, recsym, recval, reltype);
+		}
+	}
+	if (mloc0 != mlocp) {
+		append_func(ehdr, shstr, mloc0, mlocp, mrel0, mrelp,
+			    rel_entsize, symsec_sh_link);
+	}
+	free(mrel0);
+	free(mloc0);
+}