/* BFD back-end for Hitachi H8/300 COFF binaries. Copyright 1990, 91, 92, 93, 94, 95, 1996 Free Software Foundation, Inc. Written by Steve Chamberlain, . This file is part of BFD, the Binary File Descriptor library. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ #include "bfd.h" #include "sysdep.h" #include "obstack.h" #include "libbfd.h" #include "bfdlink.h" #include "genlink.h" #include "coff/h8300.h" #include "coff/internal.h" #include "libcoff.h" #define COFF_DEFAULT_SECTION_ALIGNMENT_POWER (1) /* We derive a hash table from the basic BFD hash table to hold entries in the function vector. Aside from the info stored by the basic hash table, we need the offset of a particular entry within the hash table as well as the offset where we'll add the next entry. */ struct funcvec_hash_entry { /* The basic hash table entry. */ struct bfd_hash_entry root; /* The offset within the vectors section where this entry lives. */ bfd_vma offset; }; struct funcvec_hash_table { /* The basic hash table. */ struct bfd_hash_table root; bfd *abfd; /* Offset at which we'll add the next entry. */ unsigned int offset; }; static struct bfd_hash_entry * funcvec_hash_newfunc PARAMS ((struct bfd_hash_entry *, struct bfd_hash_table *, const char *)); static boolean funcvec_hash_table_init PARAMS ((struct funcvec_hash_table *, bfd *, struct bfd_hash_entry *(*) PARAMS ((struct bfd_hash_entry *, struct bfd_hash_table *, const char *)))); /* To lookup a value in the function vector hash table. */ #define funcvec_hash_lookup(table, string, create, copy) \ ((struct funcvec_hash_entry *) \ bfd_hash_lookup (&(table)->root, (string), (create), (copy))) /* The derived h8300 COFF linker table. Note it's derived from the generic linker hash table, not the COFF backend linker hash table! We use this to attach additional data structures we need while linking on the h8300. */ struct h8300_coff_link_hash_table { /* The main hash table. */ struct generic_link_hash_table root; /* Section for the vectors table. This gets attached to a random input bfd, we keep it here for easy access. */ asection *vectors_sec; /* Hash table of the functions we need to enter into the function vector. */ struct funcvec_hash_table *funcvec_hash_table; }; static struct bfd_link_hash_table *h8300_coff_link_hash_table_create PARAMS ((bfd *)); /* Get the H8/300 COFF linker hash table from a link_info structure. */ #define h8300_coff_hash_table(p) \ ((struct h8300_coff_link_hash_table *) ((coff_hash_table (p)))) /* Initialize fields within a funcvec hash table entry. Called whenever a new entry is added to the funcvec hash table. */ static struct bfd_hash_entry * funcvec_hash_newfunc (entry, gen_table, string) struct bfd_hash_entry *entry; struct bfd_hash_table *gen_table; const char *string; { struct funcvec_hash_entry *ret; struct funcvec_hash_table *table; ret = (struct funcvec_hash_entry *) entry; table = (struct funcvec_hash_table *) gen_table; /* Allocate the structure if it has not already been allocated by a subclass. */ if (ret == NULL) ret = ((struct funcvec_hash_entry *) bfd_hash_allocate (gen_table, sizeof (struct funcvec_hash_entry))); if (ret == NULL) return NULL; /* Call the allocation method of the superclass. */ ret = ((struct funcvec_hash_entry *) bfd_hash_newfunc ((struct bfd_hash_entry *) ret, gen_table, string)); if (ret == NULL) return NULL; /* Note where this entry will reside in the function vector table. */ ret->offset = table->offset; /* Bump the offset at which we store entries in the function vector. We'd like to bump up the size of the vectors section, but it's not easily available here. */ if (bfd_get_mach (table->abfd) == bfd_mach_h8300) table->offset += 2; else if (bfd_get_mach (table->abfd) == bfd_mach_h8300h) table->offset += 4; else return NULL; /* Everything went OK. */ return (struct bfd_hash_entry *) ret; } /* Initialize the function vector hash table. */ static boolean funcvec_hash_table_init (table, abfd, newfunc) struct funcvec_hash_table *table; bfd *abfd; struct bfd_hash_entry *(*newfunc) PARAMS ((struct bfd_hash_entry *, struct bfd_hash_table *, const char *)); { /* Initialize our local fields, then call the generic initialization routine. */ table->offset = 0; table->abfd = abfd; return (bfd_hash_table_init (&table->root, newfunc)); } /* Create the derived linker hash table. We use a derived hash table basically to hold "static" information during an h8/300 coff link without using static variables. */ static struct bfd_link_hash_table * h8300_coff_link_hash_table_create (abfd) bfd *abfd; { struct h8300_coff_link_hash_table *ret; ret = ((struct h8300_coff_link_hash_table *) bfd_alloc (abfd, sizeof (struct h8300_coff_link_hash_table))); if (ret == NULL) return NULL; if (!_bfd_link_hash_table_init (&ret->root.root, abfd, generic_link_hash_newfunc)) { bfd_release (abfd, ret); return NULL; } /* Initialize our data. */ ret->vectors_sec = NULL; ret->funcvec_hash_table = NULL; /* OK. Everything's intialized, return the base pointer. */ return &ret->root.root; } /* special handling for H8/300 relocs. We only come here for pcrel stuff and return normally if not an -r link. When doing -r, we can't do any arithmetic for the pcrel stuff, because the code in reloc.c assumes that we can manipulate the targets of the pcrel branches. This isn't so, since the H8/300 can do relaxing, which means that the gap after the instruction may not be enough to contain the offset required for the branch, so we have to use the only the addend until the final link */ static bfd_reloc_status_type special (abfd, reloc_entry, symbol, data, input_section, output_bfd, error_message) bfd *abfd; arelent *reloc_entry; asymbol *symbol; PTR data; asection *input_section; bfd *output_bfd; char **error_message; { if (output_bfd == (bfd *) NULL) return bfd_reloc_continue; return bfd_reloc_ok; } static reloc_howto_type howto_table[] = { HOWTO (R_RELBYTE, 0, 0, 8, false, 0, complain_overflow_bitfield, special, "8", false, 0x000000ff, 0x000000ff, false), HOWTO (R_RELWORD, 0, 1, 16, false, 0, complain_overflow_bitfield, special, "16", false, 0x0000ffff, 0x0000ffff, false), HOWTO (R_RELLONG, 0, 2, 32, false, 0, complain_overflow_bitfield, special, "32", false, 0xffffffff, 0xffffffff, false), HOWTO (R_PCRBYTE, 0, 0, 8, true, 0, complain_overflow_signed, special, "DISP8", false, 0x000000ff, 0x000000ff, true), HOWTO (R_PCRWORD, 0, 1, 16, true, 0, complain_overflow_signed, special, "DISP16", false, 0x0000ffff, 0x0000ffff, true), HOWTO (R_PCRLONG, 0, 2, 32, true, 0, complain_overflow_signed, special, "DISP32", false, 0xffffffff, 0xffffffff, true), HOWTO (R_MOVB1, 0, 1, 16, false, 0, complain_overflow_bitfield, special, "16/8", false, 0x0000ffff, 0x0000ffff, false), HOWTO (R_MOVB2, 0, 1, 16, false, 0, complain_overflow_bitfield, special, "8/16", false, 0x0000ffff, 0x0000ffff, false), HOWTO (R_JMP1, 0, 1, 16, false, 0, complain_overflow_bitfield, special, "16/pcrel", false, 0x0000ffff, 0x0000ffff, false), HOWTO (R_JMP2, 0, 0, 8, false, 0, complain_overflow_bitfield, special, "pcrecl/16", false, 0x000000ff, 0x000000ff, false), HOWTO (R_JMPL1, 0, 2, 32, false, 0, complain_overflow_bitfield, special, "24/pcrell", false, 0x00ffffff, 0x00ffffff, false), HOWTO (R_JMPL_B8, 0, 0, 8, false, 0, complain_overflow_bitfield, special, "pc8/24", false, 0x000000ff, 0x000000ff, false), HOWTO (R_MOVLB1, 0, 1, 16, false, 0, complain_overflow_bitfield,special, "24/8", false, 0x0000ffff, 0x0000ffff, false), HOWTO (R_MOVLB2, 0, 1, 16, false, 0, complain_overflow_bitfield, special, "8/24", false, 0x0000ffff, 0x0000ffff, false), /* An indirect reference to a function. This causes the function's address to be added to the function vector in lo-mem and puts the address of the function vector's entry in the jsr instruction. */ HOWTO (R_MEM_INDIRECT, 0, 0, 8, false, 0, complain_overflow_bitfield, special, "8/indirect", false, 0x000000ff, 0x000000ff, false), }; /* Turn a howto into a reloc number */ #define SELECT_RELOC(x,howto) \ { x.r_type = select_reloc(howto); } #define BADMAG(x) (H8300BADMAG(x)&& H8300HBADMAG(x)) #define H8300 1 /* Customize coffcode.h */ #define __A_MAGIC_SET__ /* Code to swap in the reloc */ #define SWAP_IN_RELOC_OFFSET bfd_h_get_32 #define SWAP_OUT_RELOC_OFFSET bfd_h_put_32 #define SWAP_OUT_RELOC_EXTRA(abfd, src, dst) \ dst->r_stuff[0] = 'S'; \ dst->r_stuff[1] = 'C'; static int select_reloc (howto) reloc_howto_type *howto; { return howto->type; } /* Code to turn a r_type into a howto ptr, uses the above howto table */ static void rtype2howto (internal, dst) arelent *internal; struct internal_reloc *dst; { switch (dst->r_type) { case R_RELBYTE: internal->howto = howto_table + 0; break; case R_RELWORD: internal->howto = howto_table + 1; break; case R_RELLONG: internal->howto = howto_table + 2; break; case R_PCRBYTE: internal->howto = howto_table + 3; break; case R_PCRWORD: internal->howto = howto_table + 4; break; case R_PCRLONG: internal->howto = howto_table + 5; break; case R_MOVB1: internal->howto = howto_table + 6; break; case R_MOVB2: internal->howto = howto_table + 7; break; case R_JMP1: internal->howto = howto_table + 8; break; case R_JMP2: internal->howto = howto_table + 9; break; case R_JMPL1: internal->howto = howto_table + 10; break; case R_JMPL_B8: internal->howto = howto_table + 11; break; case R_MOVLB1: internal->howto = howto_table + 12; break; case R_MOVLB2: internal->howto = howto_table + 13; break; case R_MEM_INDIRECT: internal->howto = howto_table + 14; break; default: abort (); break; } } #define RTYPE2HOWTO(internal, relocentry) rtype2howto(internal,relocentry) /* Perform any necessaru magic to the addend in a reloc entry */ #define CALC_ADDEND(abfd, symbol, ext_reloc, cache_ptr) \ cache_ptr->addend = ext_reloc.r_offset; #define RELOC_PROCESSING(relent,reloc,symbols,abfd,section) \ reloc_processing(relent, reloc, symbols, abfd, section) static void reloc_processing (relent, reloc, symbols, abfd, section) arelent * relent; struct internal_reloc *reloc; asymbol ** symbols; bfd * abfd; asection * section; { relent->address = reloc->r_vaddr; rtype2howto (relent, reloc); if (((int) reloc->r_symndx) > 0) { relent->sym_ptr_ptr = symbols + obj_convert (abfd)[reloc->r_symndx]; } else { relent->sym_ptr_ptr = bfd_abs_section_ptr->symbol_ptr_ptr; } relent->addend = reloc->r_offset; relent->address -= section->vma; /* relent->section = 0;*/ } static int h8300_reloc16_estimate(abfd, input_section, reloc, shrink, link_info) bfd *abfd; asection *input_section; arelent *reloc; unsigned int shrink; struct bfd_link_info *link_info; { bfd_vma value; bfd_vma dot; bfd_vma gap; /* The address of the thing to be relocated will have moved back by the size of the shrink - but we don't change reloc->address here, since we need it to know where the relocation lives in the source uncooked section */ /* reloc->address -= shrink; conceptual */ bfd_vma address = reloc->address - shrink; switch (reloc->howto->type) { case R_MOVB2: case R_JMP2: shrink+=2; break; /* Thing is a move one byte */ case R_MOVB1: value = bfd_coff_reloc16_get_value(reloc, link_info, input_section); if (value >= 0xff00) { /* Change the reloc type from 16bit, possible 8 to 8bit possible 16 */ reloc->howto = reloc->howto + 1; /* The place to relc moves back by one */ /* This will be two bytes smaller in the long run */ shrink +=2 ; bfd_perform_slip(abfd, 2, input_section, address); } break; /* This is the 24 bit branch which could become an 8 bitter, the relocation points to the first byte of the insn, not the actual data */ case R_JMPL1: value = bfd_coff_reloc16_get_value(reloc, link_info, input_section); dot = input_section->output_section->vma + input_section->output_offset + address; /* See if the address we're looking at within 127 bytes of where we are, if so then we can use a small branch rather than the jump we were going to */ gap = value - dot ; if (-120 < (long)gap && (long)gap < 120 ) { /* Change the reloc type from 24bit, possible 8 to 8bit possible 32 */ reloc->howto = reloc->howto + 1; /* This will be two bytes smaller in the long run */ shrink +=2 ; bfd_perform_slip(abfd, 2, input_section, address); } break; case R_JMP1: value = bfd_coff_reloc16_get_value(reloc, link_info, input_section); dot = input_section->output_section->vma + input_section->output_offset + address; /* See if the address we're looking at within 127 bytes of where we are, if so then we can use a small branch rather than the jump we were going to */ gap = value - (dot - shrink); if (-120 < (long)gap && (long)gap < 120 ) { /* Change the reloc type from 16bit, possible 8 to 8bit possible 16 */ reloc->howto = reloc->howto + 1; /* The place to relc moves back by one */ /* This will be two bytes smaller in the long run */ shrink +=2 ; bfd_perform_slip(abfd, 2, input_section, address); } break; } return shrink; } /* First phase of a relaxing link */ /* Reloc types large small R_MOVB1 R_MOVB2 mov.b with 16bit or 8 bit address R_JMP1 R_JMP2 jmp or pcrel branch R_JMPL1 R_JMPL_B8 24jmp or pcrel branch R_MOVLB1 R_MOVLB2 24 or 8 bit reloc for mov.b */ static void h8300_reloc16_extra_cases (abfd, link_info, link_order, reloc, data, src_ptr, dst_ptr) bfd *abfd; struct bfd_link_info *link_info; struct bfd_link_order *link_order; arelent *reloc; bfd_byte *data; unsigned int *src_ptr; unsigned int *dst_ptr; { unsigned int src_address = *src_ptr; unsigned int dst_address = *dst_ptr; asection *input_section = link_order->u.indirect.section; switch (reloc->howto->type) { /* A 24 bit branch which could be a 8 bit pcrel, really pointing to the byte before the 24bit hole, so we can treat it as a 32bit pointer */ case R_PCRBYTE: { bfd_vma dot = link_order->offset + dst_address + link_order->u.indirect.section->output_section->vma; int gap = (bfd_coff_reloc16_get_value (reloc, link_info, input_section) - dot); if (gap > 127 || gap < -128) { if (! ((*link_info->callbacks->reloc_overflow) (link_info, bfd_asymbol_name (*reloc->sym_ptr_ptr), reloc->howto->name, reloc->addend, input_section->owner, input_section, reloc->address))) abort (); } gap &= ~1; bfd_put_8 (abfd, gap, data + dst_address); dst_address++; src_address++; break; } case R_PCRWORD: { bfd_vma dot = link_order->offset + dst_address + link_order->u.indirect.section->output_section->vma; int gap = (bfd_coff_reloc16_get_value (reloc, link_info, input_section) - dot) - 1; if (gap > 32767 || gap < -32768) { if (! ((*link_info->callbacks->reloc_overflow) (link_info, bfd_asymbol_name (*reloc->sym_ptr_ptr), reloc->howto->name, reloc->addend, input_section->owner, input_section, reloc->address))) abort (); } bfd_put_16 (abfd, gap, data + dst_address); dst_address+=2; src_address+=2; break; } case R_RELBYTE: { unsigned int gap = bfd_coff_reloc16_get_value (reloc, link_info, input_section); if (gap < 0xff || (gap >= 0x0000ff00 && gap <= 0x0000ffff) || ( gap >= 0x00ffff00 && gap <= 0x00ffffff) || ( gap >= 0xffffff00 && gap <= 0xffffffff)) { bfd_put_8 (abfd, gap, data + dst_address); dst_address += 1; src_address += 1; } else { if (! ((*link_info->callbacks->reloc_overflow) (link_info, bfd_asymbol_name (*reloc->sym_ptr_ptr), reloc->howto->name, reloc->addend, input_section->owner, input_section, reloc->address))) abort (); } } break; case R_JMP1: /* A relword which would have like to have been a pcrel */ case R_MOVB1: /* A relword which would like to have been modified but didn't make it */ case R_RELWORD: bfd_put_16 (abfd, bfd_coff_reloc16_get_value (reloc, link_info, input_section), data + dst_address); dst_address += 2; src_address += 2; break; case R_RELLONG: bfd_put_32 (abfd, bfd_coff_reloc16_get_value (reloc, link_info, input_section), data + dst_address); dst_address += 4; src_address += 4; break; case R_MOVB2: /* Special relaxed type, there will be a gap between where we get stuff from and where we put stuff to now for a mov.b @aa:16 -> mov.b @aa:8 opcode 0x6a 0x0y offset -> 0x2y off */ if (data[dst_address - 1] != 0x6a) abort (); switch (data[src_address] & 0xf0) { case 0x00: /* Src is memory */ data[dst_address - 1] = (data[src_address] & 0xf) | 0x20; break; case 0x80: /* Src is reg */ data[dst_address - 1] = (data[src_address] & 0xf) | 0x30; break; default: abort (); } /* the offset must fit ! after all, what was all the relaxing about ? */ bfd_put_8 (abfd, bfd_coff_reloc16_get_value (reloc, link_info, input_section), data + dst_address); /* Note the magic - src goes up by two bytes, but dst by only one */ dst_address += 1; src_address += 3; break; case R_JMP2: /* Speciial relaxed type */ { bfd_vma dot = link_order->offset + dst_address + link_order->u.indirect.section->output_section->vma; int gap = (bfd_coff_reloc16_get_value (reloc, link_info, input_section) - dot - 1); if ((gap & ~0xff) != 0 && ((gap & 0xff00) != 0xff00)) abort (); bfd_put_8 (abfd, gap, data + dst_address); switch (data[dst_address - 1]) { case 0x5e: /* jsr -> bsr */ bfd_put_8 (abfd, 0x55, data + dst_address - 1); break; case 0x5a: /* jmp ->bra */ bfd_put_8 (abfd, 0x40, data + dst_address - 1); break; default: abort (); } dst_address++; src_address += 3; break; } break; case R_JMPL_B8: /* 24 bit branch which is now 8 bits */ /* Speciial relaxed type */ { bfd_vma dot = link_order->offset + dst_address + link_order->u.indirect.section->output_section->vma; int gap = (bfd_coff_reloc16_get_value (reloc, link_info, input_section) - dot - 2); if ((gap & ~0xff) != 0 && ((gap & 0xff00) != 0xff00)) abort (); switch (data[src_address]) { case 0x5e: /* jsr -> bsr */ bfd_put_8 (abfd, 0x55, data + dst_address); break; case 0x5a: /* jmp ->bra */ bfd_put_8 (abfd, 0x40, data + dst_address); break; default: bfd_put_8 (abfd, 0xde, data + dst_address); break; } bfd_put_8 (abfd, gap, data + dst_address + 1); dst_address += 2; src_address += 4; break; } case R_JMPL1: { int v = bfd_coff_reloc16_get_value (reloc, link_info, input_section); int o = bfd_get_32 (abfd, data + src_address); v = (v & 0x00ffffff) | (o & 0xff000000); bfd_put_32 (abfd, v, data + dst_address); dst_address += 4; src_address += 4; } break; /* A 24 bit mov which could be an 8 bit move, really pointing to the byte before the 24bit hole, so we can treat it as a 32bit pointer */ case R_MOVLB1: { int v = bfd_coff_reloc16_get_value (reloc, link_info, input_section); int o = bfd_get_32 (abfd, data + dst_address); v = (v & 0x00ffffff) | (o & 0xff000000); bfd_put_32 (abfd, v, data + dst_address); dst_address += 4; src_address += 4; } break; /* An 8bit memory indirect instruction (jmp/jsr). There's several things that need to be done to handle this relocation. If this is a reloc against the absolute symbol, then we should handle it just R_RELBYTE. Likewise if it's for a symbol with a value ge 0 and le 0xff. Otherwise it's a jump/call through the function vector, and the linker is expected to set up the function vector and put the right value into the jump/call instruction. */ case R_MEM_INDIRECT: { /* We need to find the symbol so we can determine it's address in the function vector table. */ asymbol *symbol; bfd_vma value; char *name; struct funcvec_hash_entry *h; asection *vectors_sec = h8300_coff_hash_table (link_info)->vectors_sec; /* First see if this is a reloc against the absolute symbol or against a symbol with a nonnegative value <= 0xff. */ symbol = *(reloc->sym_ptr_ptr); value = bfd_coff_reloc16_get_value (reloc, link_info, input_section); if (symbol == bfd_abs_section_ptr->symbol || (value >= 0 && value <= 0xff)) { /* This should be handled in a manner very similar to R_RELBYTES. If the value is in range, then just slam the value into the right location. Else trigger a reloc overflow callback. */ if (value >= 0 && value <= 0xff) { bfd_put_8 (abfd, value, data + dst_address); dst_address += 1; src_address += 1; } else { if (! ((*link_info->callbacks->reloc_overflow) (link_info, bfd_asymbol_name (*reloc->sym_ptr_ptr), reloc->howto->name, reloc->addend, input_section->owner, input_section, reloc->address))) abort (); } break; } /* This is a jump/call through a function vector, and we're expected to create the function vector ourselves. First look up this symbol in the linker hash table -- we need the derived linker symbol which holds this symbol's index in the function vector. */ name = symbol->name; if (symbol->flags & BSF_LOCAL) { char *new_name = bfd_malloc (strlen (name) + 9); if (new_name == NULL) abort (); strcpy (new_name, name); sprintf (new_name + strlen (name), "_%08x", (int)symbol->section); name = new_name; } h = funcvec_hash_lookup (h8300_coff_hash_table (link_info)->funcvec_hash_table, name, false, false); /* This shouldn't ever happen. If it does that means we've got data corruption of some kind. Aborting seems like a reasonable think to do here. */ if (h == NULL || vectors_sec == NULL) abort (); /* Place the address of the function vector entry into the reloc's address. */ bfd_put_8 (abfd, vectors_sec->output_offset + h->offset, data + dst_address); dst_address++; src_address++; /* Now create an entry in the function vector itself. */ if (bfd_get_mach (input_section->owner) == bfd_mach_h8300) bfd_put_16 (abfd, bfd_coff_reloc16_get_value (reloc, link_info, input_section), vectors_sec->contents + h->offset); else if (bfd_get_mach (input_section->owner) == bfd_mach_h8300h) bfd_put_32 (abfd, bfd_coff_reloc16_get_value (reloc, link_info, input_section), vectors_sec->contents + h->offset); else abort (); /* Gross. We've already written the contents of the vector section before we get here... So we write it again with the new data. */ bfd_set_section_contents (vectors_sec->output_section->owner, vectors_sec->output_section, vectors_sec->contents, vectors_sec->output_offset, vectors_sec->_raw_size); break; } default: abort (); break; } *src_ptr = src_address; *dst_ptr = dst_address; } /* Routine for the h8300 linker. This routine is necessary to handle the special R_MEM_INDIRECT relocs on the h8300. It's responsible for generating a vectors section and attaching it to an input bfd as well as sizing the vectors section. It also creates our vectors hash table. It uses the generic linker routines to actually add the symbols. from this BFD to the bfd linker hash table. It may add a few selected static symbols to the bfd linker hash table. */ static boolean h8300_bfd_link_add_symbols(abfd, info) bfd *abfd; struct bfd_link_info *info; { asection *sec; struct funcvec_hash_table *funcvec_hash_table; /* If we haven't created a vectors section, do so now. */ if (!h8300_coff_hash_table (info)->vectors_sec) { flagword flags; /* Make sure the appropriate flags are set, including SEC_IN_MEMORY. */ flags = (SEC_ALLOC | SEC_LOAD | SEC_HAS_CONTENTS | SEC_IN_MEMORY | SEC_READONLY); h8300_coff_hash_table (info)->vectors_sec = bfd_make_section (abfd, ".vectors"); /* If the section wasn't created, or we couldn't set the flags, quit quickly now, rather than dieing a painful death later. */ if (! h8300_coff_hash_table (info)->vectors_sec || ! bfd_set_section_flags (abfd, h8300_coff_hash_table(info)->vectors_sec, flags)) return false; /* Also create the vector hash table. */ funcvec_hash_table = ((struct funcvec_hash_table *) bfd_alloc (abfd, sizeof (struct funcvec_hash_table))); if (!funcvec_hash_table) return false; /* And initialize the funcvec hash table. */ if (!funcvec_hash_table_init (funcvec_hash_table, abfd, funcvec_hash_newfunc)) { bfd_release (abfd, funcvec_hash_table); return false; } /* Store away a pointer to the funcvec hash table. */ h8300_coff_hash_table (info)->funcvec_hash_table = funcvec_hash_table; } /* Load up the function vector hash table. */ funcvec_hash_table = h8300_coff_hash_table (info)->funcvec_hash_table; /* Add the symbols using the generic code. */ _bfd_generic_link_add_symbols (abfd, info); /* Now scan the relocs for all the sections in this bfd; create additional space in the .vectors section as needed. */ for (sec = abfd->sections; sec; sec = sec->next) { unsigned long reloc_size, reloc_count, i; asymbol **symbols; arelent **relocs; /* Suck in the relocs, symbols & canonicalize them. */ reloc_size = bfd_get_reloc_upper_bound (abfd, sec); if (reloc_size <= 0) continue; relocs = (arelent **)bfd_malloc ((size_t)reloc_size); if (!relocs) return false; /* The symbols should have been read in by _bfd_generic link_add_symbols call abovec, so we can cheat and use the pointer to them that was saved in the above call. */ symbols = _bfd_generic_link_get_symbols(abfd); reloc_count = bfd_canonicalize_reloc (abfd, sec, relocs, symbols); /* Now walk through all the relocations in this section. */ for (i = 0; i < reloc_count; i++) { arelent *reloc = relocs[i]; asymbol *symbol = *(reloc->sym_ptr_ptr); char *name; /* We've got an indirect reloc. See if we need to add it to the function vector table. At this point, we have to add a new entry for each unique symbol referenced by an R_MEM_INDIRECT relocation except for a reloc against the absolute section symbol. */ if (reloc->howto->type == R_MEM_INDIRECT && symbol != bfd_abs_section_ptr->symbol) { struct funcvec_hash_entry *h; name = symbol->name; if (symbol->flags & BSF_LOCAL) { char *new_name = bfd_malloc (strlen (name) + 9); if (new_name == NULL) abort (); strcpy (new_name, name); sprintf (new_name + strlen (name), "_%08x", (int)symbol->section); name = new_name; } /* Look this symbol up in the function vector hash table. */ h = funcvec_hash_lookup (h8300_coff_hash_table (info)->funcvec_hash_table, name, false, false); /* If this symbol isn't already in the hash table, add it and bump up the size of the hash table. */ if (h == NULL) { h = funcvec_hash_lookup (h8300_coff_hash_table (info)->funcvec_hash_table, name, true, true); if (h == NULL) { free (relocs); return false; } /* Bump the size of the vectors section. Each vector takes 2 bytes on the h8300 and 4 bytes on the h8300h. */ if (bfd_get_mach (abfd) == bfd_mach_h8300) h8300_coff_hash_table (info)->vectors_sec->_raw_size += 2; else if (bfd_get_mach (abfd) == bfd_mach_h8300h) h8300_coff_hash_table (info)->vectors_sec->_raw_size += 4; } } } /* We're done with the relocations, release them. */ free (relocs); } /* Now actually allocate some space for the function vector. It's wasteful to do this more than once, but this is easier. */ if (h8300_coff_hash_table (info)->vectors_sec->_raw_size != 0) { /* Free the old contents. */ if (h8300_coff_hash_table (info)->vectors_sec->contents) free (h8300_coff_hash_table (info)->vectors_sec->contents); /* Allocate new contents. */ h8300_coff_hash_table (info)->vectors_sec->contents = bfd_malloc (h8300_coff_hash_table (info)->vectors_sec->_raw_size); } return true; } #define coff_reloc16_extra_cases h8300_reloc16_extra_cases #define coff_reloc16_estimate h8300_reloc16_estimate #define coff_bfd_link_add_symbols h8300_bfd_link_add_symbols #define coff_bfd_link_hash_table_create h8300_coff_link_hash_table_create #define COFF_LONG_FILENAMES #include "coffcode.h" #undef coff_bfd_get_relocated_section_contents #undef coff_bfd_relax_section #define coff_bfd_get_relocated_section_contents \ bfd_coff_reloc16_get_relocated_section_contents #define coff_bfd_relax_section bfd_coff_reloc16_relax_section const bfd_target h8300coff_vec = { "coff-h8300", /* name */ bfd_target_coff_flavour, BFD_ENDIAN_BIG, /* data byte order is big */ BFD_ENDIAN_BIG, /* header byte order is big */ (HAS_RELOC | EXEC_P | /* object flags */ HAS_LINENO | HAS_DEBUG | HAS_SYMS | HAS_LOCALS | WP_TEXT | BFD_IS_RELAXABLE ), (SEC_HAS_CONTENTS | SEC_ALLOC | SEC_LOAD | SEC_RELOC), /* section flags */ '_', /* leading char */ '/', /* ar_pad_char */ 15, /* ar_max_namelen */ bfd_getb64, bfd_getb_signed_64, bfd_putb64, bfd_getb32, bfd_getb_signed_32, bfd_putb32, bfd_getb16, bfd_getb_signed_16, bfd_putb16, /* data */ bfd_getb64, bfd_getb_signed_64, bfd_putb64, bfd_getb32, bfd_getb_signed_32, bfd_putb32, bfd_getb16, bfd_getb_signed_16, bfd_putb16, /* hdrs */ {_bfd_dummy_target, coff_object_p, /* bfd_check_format */ bfd_generic_archive_p, _bfd_dummy_target}, {bfd_false, coff_mkobject, _bfd_generic_mkarchive, /* bfd_set_format */ bfd_false}, {bfd_false, coff_write_object_contents, /* bfd_write_contents */ _bfd_write_archive_contents, bfd_false}, BFD_JUMP_TABLE_GENERIC (coff), BFD_JUMP_TABLE_COPY (coff), BFD_JUMP_TABLE_CORE (_bfd_nocore), BFD_JUMP_TABLE_ARCHIVE (_bfd_archive_coff), BFD_JUMP_TABLE_SYMBOLS (coff), BFD_JUMP_TABLE_RELOCS (coff), BFD_JUMP_TABLE_WRITE (coff), BFD_JUMP_TABLE_LINK (coff), BFD_JUMP_TABLE_DYNAMIC (_bfd_nodynamic), COFF_SWAP_TABLE, };