/* COFF specific linker code. Copyright 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2011 Free Software Foundation, Inc. Written by Ian Lance Taylor, Cygnus Support. 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 3 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., 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA. */ /* This file contains the COFF backend linker code. */ #include "sysdep.h" #include "bfd.h" #include "bfdlink.h" #include "libbfd.h" #include "coff/internal.h" #include "libcoff.h" #include "safe-ctype.h" static bfd_boolean coff_link_add_object_symbols (bfd *abfd, struct bfd_link_info *info); static bfd_boolean coff_link_check_archive_element (bfd *abfd, struct bfd_link_info *info, bfd_boolean *pneeded); static bfd_boolean coff_link_add_symbols (bfd *abfd, struct bfd_link_info *info); /* Return TRUE if SYM is a weak, external symbol. */ #define IS_WEAK_EXTERNAL(abfd, sym) \ ((sym).n_sclass == C_WEAKEXT \ || (obj_pe (abfd) && (sym).n_sclass == C_NT_WEAK)) /* Return TRUE if SYM is an external symbol. */ #define IS_EXTERNAL(abfd, sym) \ ((sym).n_sclass == C_EXT || IS_WEAK_EXTERNAL (abfd, sym)) /* Define macros so that the ISFCN, et. al., macros work correctly. These macros are defined in include/coff/internal.h in terms of N_TMASK, etc. These definitions require a user to define local variables with the appropriate names, and with values from the coff_data (abfd) structure. */ #define N_TMASK n_tmask #define N_BTSHFT n_btshft #define N_BTMASK n_btmask /* Create an entry in a COFF linker hash table. */ struct bfd_hash_entry * _bfd_coff_link_hash_newfunc (struct bfd_hash_entry *entry, struct bfd_hash_table *table, const char *string) { struct coff_link_hash_entry *ret = (struct coff_link_hash_entry *) entry; /* Allocate the structure if it has not already been allocated by a subclass. */ if (ret == (struct coff_link_hash_entry *) NULL) ret = ((struct coff_link_hash_entry *) bfd_hash_allocate (table, sizeof (struct coff_link_hash_entry))); if (ret == (struct coff_link_hash_entry *) NULL) return (struct bfd_hash_entry *) ret; /* Call the allocation method of the superclass. */ ret = ((struct coff_link_hash_entry *) _bfd_link_hash_newfunc ((struct bfd_hash_entry *) ret, table, string)); if (ret != (struct coff_link_hash_entry *) NULL) { /* Set local fields. */ ret->indx = -1; ret->type = T_NULL; ret->symbol_class = C_NULL; ret->numaux = 0; ret->auxbfd = NULL; ret->aux = NULL; } return (struct bfd_hash_entry *) ret; } /* Initialize a COFF linker hash table. */ bfd_boolean _bfd_coff_link_hash_table_init (struct coff_link_hash_table *table, bfd *abfd, struct bfd_hash_entry *(*newfunc) (struct bfd_hash_entry *, struct bfd_hash_table *, const char *), unsigned int entsize) { memset (&table->stab_info, 0, sizeof (table->stab_info)); return _bfd_link_hash_table_init (&table->root, abfd, newfunc, entsize); } /* Create a COFF linker hash table. */ struct bfd_link_hash_table * _bfd_coff_link_hash_table_create (bfd *abfd) { struct coff_link_hash_table *ret; bfd_size_type amt = sizeof (struct coff_link_hash_table); ret = (struct coff_link_hash_table *) bfd_malloc (amt); if (ret == NULL) return NULL; if (! _bfd_coff_link_hash_table_init (ret, abfd, _bfd_coff_link_hash_newfunc, sizeof (struct coff_link_hash_entry))) { free (ret); return (struct bfd_link_hash_table *) NULL; } return &ret->root; } /* Create an entry in a COFF debug merge hash table. */ struct bfd_hash_entry * _bfd_coff_debug_merge_hash_newfunc (struct bfd_hash_entry *entry, struct bfd_hash_table *table, const char *string) { struct coff_debug_merge_hash_entry *ret = (struct coff_debug_merge_hash_entry *) entry; /* Allocate the structure if it has not already been allocated by a subclass. */ if (ret == (struct coff_debug_merge_hash_entry *) NULL) ret = ((struct coff_debug_merge_hash_entry *) bfd_hash_allocate (table, sizeof (struct coff_debug_merge_hash_entry))); if (ret == (struct coff_debug_merge_hash_entry *) NULL) return (struct bfd_hash_entry *) ret; /* Call the allocation method of the superclass. */ ret = ((struct coff_debug_merge_hash_entry *) bfd_hash_newfunc ((struct bfd_hash_entry *) ret, table, string)); if (ret != (struct coff_debug_merge_hash_entry *) NULL) { /* Set local fields. */ ret->types = NULL; } return (struct bfd_hash_entry *) ret; } /* Given a COFF BFD, add symbols to the global hash table as appropriate. */ bfd_boolean _bfd_coff_link_add_symbols (bfd *abfd, struct bfd_link_info *info) { switch (bfd_get_format (abfd)) { case bfd_object: return coff_link_add_object_symbols (abfd, info); case bfd_archive: return _bfd_generic_link_add_archive_symbols (abfd, info, coff_link_check_archive_element); default: bfd_set_error (bfd_error_wrong_format); return FALSE; } } /* Add symbols from a COFF object file. */ static bfd_boolean coff_link_add_object_symbols (bfd *abfd, struct bfd_link_info *info) { if (! _bfd_coff_get_external_symbols (abfd)) return FALSE; if (! coff_link_add_symbols (abfd, info)) return FALSE; if (! info->keep_memory && ! _bfd_coff_free_symbols (abfd)) return FALSE; return TRUE; } /* Look through the symbols to see if this object file should be included in the link. */ static bfd_boolean coff_link_check_ar_symbols (bfd *abfd, struct bfd_link_info *info, bfd_boolean *pneeded, bfd **subsbfd) { bfd_size_type symesz; bfd_byte *esym; bfd_byte *esym_end; *pneeded = FALSE; symesz = bfd_coff_symesz (abfd); esym = (bfd_byte *) obj_coff_external_syms (abfd); esym_end = esym + obj_raw_syment_count (abfd) * symesz; while (esym < esym_end) { struct internal_syment sym; enum coff_symbol_classification classification; bfd_coff_swap_sym_in (abfd, esym, &sym); classification = bfd_coff_classify_symbol (abfd, &sym); if (classification == COFF_SYMBOL_GLOBAL || classification == COFF_SYMBOL_COMMON) { const char *name; char buf[SYMNMLEN + 1]; struct bfd_link_hash_entry *h; /* This symbol is externally visible, and is defined by this object file. */ name = _bfd_coff_internal_syment_name (abfd, &sym, buf); if (name == NULL) return FALSE; h = bfd_link_hash_lookup (info->hash, name, FALSE, FALSE, TRUE); /* Auto import. */ if (!h && info->pei386_auto_import && CONST_STRNEQ (name, "__imp_")) h = bfd_link_hash_lookup (info->hash, name + 6, FALSE, FALSE, TRUE); /* We are only interested in symbols that are currently undefined. If a symbol is currently known to be common, COFF linkers do not bring in an object file which defines it. */ if (h != (struct bfd_link_hash_entry *) NULL && h->type == bfd_link_hash_undefined) { if (!(*info->callbacks ->add_archive_element) (info, abfd, name, subsbfd)) return FALSE; *pneeded = TRUE; return TRUE; } } esym += (sym.n_numaux + 1) * symesz; } /* We do not need this object file. */ return TRUE; } /* Check a single archive element to see if we need to include it in the link. *PNEEDED is set according to whether this element is needed in the link or not. This is called via _bfd_generic_link_add_archive_symbols. */ static bfd_boolean coff_link_check_archive_element (bfd *abfd, struct bfd_link_info *info, bfd_boolean *pneeded) { bfd *oldbfd; bfd_boolean needed; if (!_bfd_coff_get_external_symbols (abfd)) return FALSE; oldbfd = abfd; if (!coff_link_check_ar_symbols (abfd, info, pneeded, &abfd)) return FALSE; needed = *pneeded; if (needed) { /* Potentially, the add_archive_element hook may have set a substitute BFD for us. */ if (abfd != oldbfd) { if (!info->keep_memory && !_bfd_coff_free_symbols (oldbfd)) return FALSE; if (!_bfd_coff_get_external_symbols (abfd)) return FALSE; } if (!coff_link_add_symbols (abfd, info)) return FALSE; } if (!info->keep_memory || !needed) { if (!_bfd_coff_free_symbols (abfd)) return FALSE; } return TRUE; } /* Add all the symbols from an object file to the hash table. */ static bfd_boolean coff_link_add_symbols (bfd *abfd, struct bfd_link_info *info) { unsigned int n_tmask = coff_data (abfd)->local_n_tmask; unsigned int n_btshft = coff_data (abfd)->local_n_btshft; unsigned int n_btmask = coff_data (abfd)->local_n_btmask; bfd_boolean keep_syms; bfd_boolean default_copy; bfd_size_type symcount; struct coff_link_hash_entry **sym_hash; bfd_size_type symesz; bfd_byte *esym; bfd_byte *esym_end; bfd_size_type amt; symcount = obj_raw_syment_count (abfd); if (symcount == 0) return TRUE; /* Nothing to do. */ /* Keep the symbols during this function, in case the linker needs to read the generic symbols in order to report an error message. */ keep_syms = obj_coff_keep_syms (abfd); obj_coff_keep_syms (abfd) = TRUE; if (info->keep_memory) default_copy = FALSE; else default_copy = TRUE; /* We keep a list of the linker hash table entries that correspond to particular symbols. */ amt = symcount * sizeof (struct coff_link_hash_entry *); sym_hash = (struct coff_link_hash_entry **) bfd_zalloc (abfd, amt); if (sym_hash == NULL) goto error_return; obj_coff_sym_hashes (abfd) = sym_hash; symesz = bfd_coff_symesz (abfd); BFD_ASSERT (symesz == bfd_coff_auxesz (abfd)); esym = (bfd_byte *) obj_coff_external_syms (abfd); esym_end = esym + symcount * symesz; while (esym < esym_end) { struct internal_syment sym; enum coff_symbol_classification classification; bfd_boolean copy; bfd_coff_swap_sym_in (abfd, esym, &sym); classification = bfd_coff_classify_symbol (abfd, &sym); if (classification != COFF_SYMBOL_LOCAL) { const char *name; char buf[SYMNMLEN + 1]; flagword flags; asection *section; bfd_vma value; bfd_boolean addit; /* This symbol is externally visible. */ name = _bfd_coff_internal_syment_name (abfd, &sym, buf); if (name == NULL) goto error_return; /* We must copy the name into memory if we got it from the syment itself, rather than the string table. */ copy = default_copy; if (sym._n._n_n._n_zeroes != 0 || sym._n._n_n._n_offset == 0) copy = TRUE; value = sym.n_value; switch (classification) { default: abort (); case COFF_SYMBOL_GLOBAL: flags = BSF_EXPORT | BSF_GLOBAL; section = coff_section_from_bfd_index (abfd, sym.n_scnum); if (! obj_pe (abfd)) value -= section->vma; /* Treat a symbol from a discarded section as undefined. */ if (bfd_is_abs_section (section) || !bfd_is_abs_section (section->output_section)) break; /* Fall thru */ case COFF_SYMBOL_UNDEFINED: flags = 0; section = bfd_und_section_ptr; break; case COFF_SYMBOL_COMMON: flags = BSF_GLOBAL; section = bfd_com_section_ptr; break; case COFF_SYMBOL_PE_SECTION: flags = BSF_SECTION_SYM | BSF_GLOBAL; section = coff_section_from_bfd_index (abfd, sym.n_scnum); break; } if (IS_WEAK_EXTERNAL (abfd, sym)) flags = BSF_WEAK; addit = TRUE; /* In the PE format, section symbols actually refer to the start of the output section. We handle them specially here. */ if (obj_pe (abfd) && (flags & BSF_SECTION_SYM) != 0) { *sym_hash = coff_link_hash_lookup (coff_hash_table (info), name, FALSE, copy, FALSE); if (*sym_hash != NULL) { if (((*sym_hash)->coff_link_hash_flags & COFF_LINK_HASH_PE_SECTION_SYMBOL) == 0 && (*sym_hash)->root.type != bfd_link_hash_undefined && (*sym_hash)->root.type != bfd_link_hash_undefweak) (*_bfd_error_handler) ("Warning: symbol `%s' is both section and non-section", name); addit = FALSE; } } /* The Microsoft Visual C compiler does string pooling by hashing the constants to an internal symbol name, and relying on the linker comdat support to discard duplicate names. However, if one string is a literal and one is a data initializer, one will end up in the .data section and one will end up in the .rdata section. The Microsoft linker will combine them into the .data section, which seems to be wrong since it might cause the literal to change. As long as there are no external references to the symbols, which there shouldn't be, we can treat the .data and .rdata instances as separate symbols. The comdat code in the linker will do the appropriate merging. Here we avoid getting a multiple definition error for one of these special symbols. FIXME: I don't think this will work in the case where there are two object files which use the constants as a literal and two object files which use it as a data initializer. One or the other of the second object files is going to wind up with an inappropriate reference. */ if (obj_pe (abfd) && (classification == COFF_SYMBOL_GLOBAL || classification == COFF_SYMBOL_PE_SECTION) && coff_section_data (abfd, section) != NULL && coff_section_data (abfd, section)->comdat != NULL && CONST_STRNEQ (name, "??_") && strcmp (name, coff_section_data (abfd, section)->comdat->name) == 0) { if (*sym_hash == NULL) *sym_hash = coff_link_hash_lookup (coff_hash_table (info), name, FALSE, copy, FALSE); if (*sym_hash != NULL && (*sym_hash)->root.type == bfd_link_hash_defined && coff_section_data (abfd, (*sym_hash)->root.u.def.section)->comdat != NULL && strcmp (coff_section_data (abfd, (*sym_hash)->root.u.def.section)->comdat->name, coff_section_data (abfd, section)->comdat->name) == 0) addit = FALSE; } if (addit) { if (! (bfd_coff_link_add_one_symbol (info, abfd, name, flags, section, value, (const char *) NULL, copy, FALSE, (struct bfd_link_hash_entry **) sym_hash))) goto error_return; } if (obj_pe (abfd) && (flags & BSF_SECTION_SYM) != 0) (*sym_hash)->coff_link_hash_flags |= COFF_LINK_HASH_PE_SECTION_SYMBOL; /* Limit the alignment of a common symbol to the possible alignment of a section. There is no point to permitting a higher alignment for a common symbol: we can not guarantee it, and it may cause us to allocate extra space in the common section. */ if (section == bfd_com_section_ptr && (*sym_hash)->root.type == bfd_link_hash_common && ((*sym_hash)->root.u.c.p->alignment_power > bfd_coff_default_section_alignment_power (abfd))) (*sym_hash)->root.u.c.p->alignment_power = bfd_coff_default_section_alignment_power (abfd); if (bfd_get_flavour (info->output_bfd) == bfd_get_flavour (abfd)) { /* If we don't have any symbol information currently in the hash table, or if we are looking at a symbol definition, then update the symbol class and type in the hash table. */ if (((*sym_hash)->symbol_class == C_NULL && (*sym_hash)->type == T_NULL) || sym.n_scnum != 0 || (sym.n_value != 0 && (*sym_hash)->root.type != bfd_link_hash_defined && (*sym_hash)->root.type != bfd_link_hash_defweak)) { (*sym_hash)->symbol_class = sym.n_sclass; if (sym.n_type != T_NULL) { /* We want to warn if the type changed, but not if it changed from an unspecified type. Testing the whole type byte may work, but the change from (e.g.) a function of unspecified type to function of known type also wants to skip the warning. */ if ((*sym_hash)->type != T_NULL && (*sym_hash)->type != sym.n_type && !(DTYPE ((*sym_hash)->type) == DTYPE (sym.n_type) && (BTYPE ((*sym_hash)->type) == T_NULL || BTYPE (sym.n_type) == T_NULL))) (*_bfd_error_handler) (_("Warning: type of symbol `%s' changed from %d to %d in %B"), abfd, name, (*sym_hash)->type, sym.n_type); /* We don't want to change from a meaningful base type to a null one, but if we know nothing, take what little we might now know. */ if (BTYPE (sym.n_type) != T_NULL || (*sym_hash)->type == T_NULL) (*sym_hash)->type = sym.n_type; } (*sym_hash)->auxbfd = abfd; if (sym.n_numaux != 0) { union internal_auxent *alloc; unsigned int i; bfd_byte *eaux; union internal_auxent *iaux; (*sym_hash)->numaux = sym.n_numaux; alloc = ((union internal_auxent *) bfd_hash_allocate (&info->hash->table, (sym.n_numaux * sizeof (*alloc)))); if (alloc == NULL) goto error_return; for (i = 0, eaux = esym + symesz, iaux = alloc; i < sym.n_numaux; i++, eaux += symesz, iaux++) bfd_coff_swap_aux_in (abfd, eaux, sym.n_type, sym.n_sclass, (int) i, sym.n_numaux, iaux); (*sym_hash)->aux = alloc; } } } if (classification == COFF_SYMBOL_PE_SECTION && (*sym_hash)->numaux != 0) { /* Some PE sections (such as .bss) have a zero size in the section header, but a non-zero size in the AUX record. Correct that here. FIXME: This is not at all the right place to do this. For example, it won't help objdump. This needs to be done when we swap in the section header. */ BFD_ASSERT ((*sym_hash)->numaux == 1); if (section->size == 0) section->size = (*sym_hash)->aux[0].x_scn.x_scnlen; /* FIXME: We could test whether the section sizes matches the size in the aux entry, but apparently that sometimes fails unexpectedly. */ } } esym += (sym.n_numaux + 1) * symesz; sym_hash += sym.n_numaux + 1; } /* If this is a non-traditional, non-relocatable link, try to optimize the handling of any .stab/.stabstr sections. */ if (! info->relocatable && ! info->traditional_format && bfd_get_flavour (info->output_bfd) == bfd_get_flavour (abfd) && (info->strip != strip_all && info->strip != strip_debugger)) { asection *stabstr; stabstr = bfd_get_section_by_name (abfd, ".stabstr"); if (stabstr != NULL) { bfd_size_type string_offset = 0; asection *stab; for (stab = abfd->sections; stab; stab = stab->next) if (CONST_STRNEQ (stab->name, ".stab") && (!stab->name[5] || (stab->name[5] == '.' && ISDIGIT (stab->name[6])))) { struct coff_link_hash_table *table; struct coff_section_tdata *secdata = coff_section_data (abfd, stab); if (secdata == NULL) { amt = sizeof (struct coff_section_tdata); stab->used_by_bfd = bfd_zalloc (abfd, amt); if (stab->used_by_bfd == NULL) goto error_return; secdata = coff_section_data (abfd, stab); } table = coff_hash_table (info); if (! _bfd_link_section_stabs (abfd, &table->stab_info, stab, stabstr, &secdata->stab_info, &string_offset)) goto error_return; } } } obj_coff_keep_syms (abfd) = keep_syms; return TRUE; error_return: obj_coff_keep_syms (abfd) = keep_syms; return FALSE; } /* Do the final link step. */ bfd_boolean _bfd_coff_final_link (bfd *abfd, struct bfd_link_info *info) { bfd_size_type symesz; struct coff_final_link_info finfo; bfd_boolean debug_merge_allocated; bfd_boolean long_section_names; asection *o; struct bfd_link_order *p; bfd_size_type max_sym_count; bfd_size_type max_lineno_count; bfd_size_type max_reloc_count; bfd_size_type max_output_reloc_count; bfd_size_type max_contents_size; file_ptr rel_filepos; unsigned int relsz; file_ptr line_filepos; unsigned int linesz; bfd *sub; bfd_byte *external_relocs = NULL; char strbuf[STRING_SIZE_SIZE]; bfd_size_type amt; symesz = bfd_coff_symesz (abfd); finfo.info = info; finfo.output_bfd = abfd; finfo.strtab = NULL; finfo.section_info = NULL; finfo.last_file_index = -1; finfo.last_bf_index = -1; finfo.internal_syms = NULL; finfo.sec_ptrs = NULL; finfo.sym_indices = NULL; finfo.outsyms = NULL; finfo.linenos = NULL; finfo.contents = NULL; finfo.external_relocs = NULL; finfo.internal_relocs = NULL; finfo.global_to_static = FALSE; debug_merge_allocated = FALSE; coff_data (abfd)->link_info = info; finfo.strtab = _bfd_stringtab_init (); if (finfo.strtab == NULL) goto error_return; if (! coff_debug_merge_hash_table_init (&finfo.debug_merge)) goto error_return; debug_merge_allocated = TRUE; /* Compute the file positions for all the sections. */ if (! abfd->output_has_begun) { if (! bfd_coff_compute_section_file_positions (abfd)) goto error_return; } /* Count the line numbers and relocation entries required for the output file. Set the file positions for the relocs. */ rel_filepos = obj_relocbase (abfd); relsz = bfd_coff_relsz (abfd); max_contents_size = 0; max_lineno_count = 0; max_reloc_count = 0; long_section_names = FALSE; for (o = abfd->sections; o != NULL; o = o->next) { o->reloc_count = 0; o->lineno_count = 0; for (p = o->map_head.link_order; p != NULL; p = p->next) { if (p->type == bfd_indirect_link_order) { asection *sec; sec = p->u.indirect.section; /* Mark all sections which are to be included in the link. This will normally be every section. We need to do this so that we can identify any sections which the linker has decided to not include. */ sec->linker_mark = TRUE; if (info->strip == strip_none || info->strip == strip_some) o->lineno_count += sec->lineno_count; if (info->relocatable) o->reloc_count += sec->reloc_count; if (sec->rawsize > max_contents_size) max_contents_size = sec->rawsize; if (sec->size > max_contents_size) max_contents_size = sec->size; if (sec->lineno_count > max_lineno_count) max_lineno_count = sec->lineno_count; if (sec->reloc_count > max_reloc_count) max_reloc_count = sec->reloc_count; } else if (info->relocatable && (p->type == bfd_section_reloc_link_order || p->type == bfd_symbol_reloc_link_order)) ++o->reloc_count; } if (o->reloc_count == 0) o->rel_filepos = 0; else { o->flags |= SEC_RELOC; o->rel_filepos = rel_filepos; rel_filepos += o->reloc_count * relsz; /* In PE COFF, if there are at least 0xffff relocations an extra relocation will be written out to encode the count. */ if (obj_pe (abfd) && o->reloc_count >= 0xffff) rel_filepos += relsz; } if (bfd_coff_long_section_names (abfd) && strlen (o->name) > SCNNMLEN) { /* This section has a long name which must go in the string table. This must correspond to the code in coff_write_object_contents which puts the string index into the s_name field of the section header. That is why we pass hash as FALSE. */ if (_bfd_stringtab_add (finfo.strtab, o->name, FALSE, FALSE) == (bfd_size_type) -1) goto error_return; long_section_names = TRUE; } } /* If doing a relocatable link, allocate space for the pointers we need to keep. */ if (info->relocatable) { unsigned int i; /* We use section_count + 1, rather than section_count, because the target_index fields are 1 based. */ amt = abfd->section_count + 1; amt *= sizeof (struct coff_link_section_info); finfo.section_info = (struct coff_link_section_info *) bfd_malloc (amt); if (finfo.section_info == NULL) goto error_return; for (i = 0; i <= abfd->section_count; i++) { finfo.section_info[i].relocs = NULL; finfo.section_info[i].rel_hashes = NULL; } } /* We now know the size of the relocs, so we can determine the file positions of the line numbers. */ line_filepos = rel_filepos; linesz = bfd_coff_linesz (abfd); max_output_reloc_count = 0; for (o = abfd->sections; o != NULL; o = o->next) { if (o->lineno_count == 0) o->line_filepos = 0; else { o->line_filepos = line_filepos; line_filepos += o->lineno_count * linesz; } if (o->reloc_count != 0) { /* We don't know the indices of global symbols until we have written out all the local symbols. For each section in the output file, we keep an array of pointers to hash table entries. Each entry in the array corresponds to a reloc. When we find a reloc against a global symbol, we set the corresponding entry in this array so that we can fix up the symbol index after we have written out all the local symbols. Because of this problem, we also keep the relocs in memory until the end of the link. This wastes memory, but only when doing a relocatable link, which is not the common case. */ BFD_ASSERT (info->relocatable); amt = o->reloc_count; amt *= sizeof (struct internal_reloc); finfo.section_info[o->target_index].relocs = (struct internal_reloc *) bfd_malloc (amt); amt = o->reloc_count; amt *= sizeof (struct coff_link_hash_entry *); finfo.section_info[o->target_index].rel_hashes = (struct coff_link_hash_entry **) bfd_malloc (amt); if (finfo.section_info[o->target_index].relocs == NULL || finfo.section_info[o->target_index].rel_hashes == NULL) goto error_return; if (o->reloc_count > max_output_reloc_count) max_output_reloc_count = o->reloc_count; } /* Reset the reloc and lineno counts, so that we can use them to count the number of entries we have output so far. */ o->reloc_count = 0; o->lineno_count = 0; } obj_sym_filepos (abfd) = line_filepos; /* Figure out the largest number of symbols in an input BFD. Take the opportunity to clear the output_has_begun fields of all the input BFD's. */ max_sym_count = 0; for (sub = info->input_bfds; sub != NULL; sub = sub->link_next) { size_t sz; sub->output_has_begun = FALSE; sz = obj_raw_syment_count (sub); if (sz > max_sym_count) max_sym_count = sz; } /* Allocate some buffers used while linking. */ amt = max_sym_count * sizeof (struct internal_syment); finfo.internal_syms = (struct internal_syment *) bfd_malloc (amt); amt = max_sym_count * sizeof (asection *); finfo.sec_ptrs = (asection **) bfd_malloc (amt); amt = max_sym_count * sizeof (long); finfo.sym_indices = (long int *) bfd_malloc (amt); finfo.outsyms = (bfd_byte *) bfd_malloc ((max_sym_count + 1) * symesz); amt = max_lineno_count * bfd_coff_linesz (abfd); finfo.linenos = (bfd_byte *) bfd_malloc (amt); finfo.contents = (bfd_byte *) bfd_malloc (max_contents_size); amt = max_reloc_count * relsz; finfo.external_relocs = (bfd_byte *) bfd_malloc (amt); if (! info->relocatable) { amt = max_reloc_count * sizeof (struct internal_reloc); finfo.internal_relocs = (struct internal_reloc *) bfd_malloc (amt); } if ((finfo.internal_syms == NULL && max_sym_count > 0) || (finfo.sec_ptrs == NULL && max_sym_count > 0) || (finfo.sym_indices == NULL && max_sym_count > 0) || finfo.outsyms == NULL || (finfo.linenos == NULL && max_lineno_count > 0) || (finfo.contents == NULL && max_contents_size > 0) || (finfo.external_relocs == NULL && max_reloc_count > 0) || (! info->relocatable && finfo.internal_relocs == NULL && max_reloc_count > 0)) goto error_return; /* We now know the position of everything in the file, except that we don't know the size of the symbol table and therefore we don't know where the string table starts. We just build the string table in memory as we go along. We process all the relocations for a single input file at once. */ obj_raw_syment_count (abfd) = 0; if (coff_backend_info (abfd)->_bfd_coff_start_final_link) { if (! bfd_coff_start_final_link (abfd, info)) goto error_return; } for (o = abfd->sections; o != NULL; o = o->next) { for (p = o->map_head.link_order; p != NULL; p = p->next) { if (p->type == bfd_indirect_link_order && bfd_family_coff (p->u.indirect.section->owner)) { sub = p->u.indirect.section->owner; if (! bfd_coff_link_output_has_begun (sub, & finfo)) { if (! _bfd_coff_link_input_bfd (&finfo, sub)) goto error_return; sub->output_has_begun = TRUE; } } else if (p->type == bfd_section_reloc_link_order || p->type == bfd_symbol_reloc_link_order) { if (! _bfd_coff_reloc_link_order (abfd, &finfo, o, p)) goto error_return; } else { if (! _bfd_default_link_order (abfd, info, o, p)) goto error_return; } } } if (! bfd_coff_final_link_postscript (abfd, & finfo)) goto error_return; /* Free up the buffers used by _bfd_coff_link_input_bfd. */ coff_debug_merge_hash_table_free (&finfo.debug_merge); debug_merge_allocated = FALSE; if (finfo.internal_syms != NULL) { free (finfo.internal_syms); finfo.internal_syms = NULL; } if (finfo.sec_ptrs != NULL) { free (finfo.sec_ptrs); finfo.sec_ptrs = NULL; } if (finfo.sym_indices != NULL) { free (finfo.sym_indices); finfo.sym_indices = NULL; } if (finfo.linenos != NULL) { free (finfo.linenos); finfo.linenos = NULL; } if (finfo.contents != NULL) { free (finfo.contents); finfo.contents = NULL; } if (finfo.external_relocs != NULL) { free (finfo.external_relocs); finfo.external_relocs = NULL; } if (finfo.internal_relocs != NULL) { free (finfo.internal_relocs); finfo.internal_relocs = NULL; } /* The value of the last C_FILE symbol is supposed to be the symbol index of the first external symbol. Write it out again if necessary. */ if (finfo.last_file_index != -1 && (unsigned int) finfo.last_file.n_value != obj_raw_syment_count (abfd)) { file_ptr pos; finfo.last_file.n_value = obj_raw_syment_count (abfd); bfd_coff_swap_sym_out (abfd, &finfo.last_file, finfo.outsyms); pos = obj_sym_filepos (abfd) + finfo.last_file_index * symesz; if (bfd_seek (abfd, pos, SEEK_SET) != 0 || bfd_bwrite (finfo.outsyms, symesz, abfd) != symesz) return FALSE; } /* If doing task linking (ld --task-link) then make a pass through the global symbols, writing out any that are defined, and making them static. */ if (info->task_link) { finfo.failed = FALSE; coff_link_hash_traverse (coff_hash_table (info), _bfd_coff_write_task_globals, &finfo); if (finfo.failed) goto error_return; } /* Write out the global symbols. */ finfo.failed = FALSE; bfd_hash_traverse (&info->hash->table, _bfd_coff_write_global_sym, &finfo); if (finfo.failed) goto error_return; /* The outsyms buffer is used by _bfd_coff_write_global_sym. */ if (finfo.outsyms != NULL) { free (finfo.outsyms); finfo.outsyms = NULL; } if (info->relocatable && max_output_reloc_count > 0) { /* Now that we have written out all the global symbols, we know the symbol indices to use for relocs against them, and we can finally write out the relocs. */ amt = max_output_reloc_count * relsz; external_relocs = (bfd_byte *) bfd_malloc (amt); if (external_relocs == NULL) goto error_return; for (o = abfd->sections; o != NULL; o = o->next) { struct internal_reloc *irel; struct internal_reloc *irelend; struct coff_link_hash_entry **rel_hash; bfd_byte *erel; if (o->reloc_count == 0) continue; irel = finfo.section_info[o->target_index].relocs; irelend = irel + o->reloc_count; rel_hash = finfo.section_info[o->target_index].rel_hashes; erel = external_relocs; for (; irel < irelend; irel++, rel_hash++, erel += relsz) { if (*rel_hash != NULL) { BFD_ASSERT ((*rel_hash)->indx >= 0); irel->r_symndx = (*rel_hash)->indx; } bfd_coff_swap_reloc_out (abfd, irel, erel); } if (bfd_seek (abfd, o->rel_filepos, SEEK_SET) != 0) goto error_return; if (obj_pe (abfd) && o->reloc_count >= 0xffff) { /* In PE COFF, write the count of relocs as the first reloc. The header overflow bit will be set elsewhere. */ struct internal_reloc incount; bfd_byte *excount = (bfd_byte *)bfd_malloc (relsz); memset (&incount, 0, sizeof (incount)); incount.r_vaddr = o->reloc_count + 1; bfd_coff_swap_reloc_out (abfd, (PTR) &incount, (PTR) excount); if (bfd_bwrite (excount, relsz, abfd) != relsz) /* We'll leak, but it's an error anyway. */ goto error_return; free (excount); } if (bfd_bwrite (external_relocs, (bfd_size_type) relsz * o->reloc_count, abfd) != (bfd_size_type) relsz * o->reloc_count) goto error_return; } free (external_relocs); external_relocs = NULL; } /* Free up the section information. */ if (finfo.section_info != NULL) { unsigned int i; for (i = 0; i < abfd->section_count; i++) { if (finfo.section_info[i].relocs != NULL) free (finfo.section_info[i].relocs); if (finfo.section_info[i].rel_hashes != NULL) free (finfo.section_info[i].rel_hashes); } free (finfo.section_info); finfo.section_info = NULL; } /* If we have optimized stabs strings, output them. */ if (coff_hash_table (info)->stab_info.stabstr != NULL) { if (! _bfd_write_stab_strings (abfd, &coff_hash_table (info)->stab_info)) return FALSE; } /* Write out the string table. */ if (obj_raw_syment_count (abfd) != 0 || long_section_names) { file_ptr pos; pos = obj_sym_filepos (abfd) + obj_raw_syment_count (abfd) * symesz; if (bfd_seek (abfd, pos, SEEK_SET) != 0) return FALSE; #if STRING_SIZE_SIZE == 4 H_PUT_32 (abfd, _bfd_stringtab_size (finfo.strtab) + STRING_SIZE_SIZE, strbuf); #else #error Change H_PUT_32 above #endif if (bfd_bwrite (strbuf, (bfd_size_type) STRING_SIZE_SIZE, abfd) != STRING_SIZE_SIZE) return FALSE; if (! _bfd_stringtab_emit (abfd, finfo.strtab)) return FALSE; obj_coff_strings_written (abfd) = TRUE; } _bfd_stringtab_free (finfo.strtab); /* Setting bfd_get_symcount to 0 will cause write_object_contents to not try to write out the symbols. */ bfd_get_symcount (abfd) = 0; return TRUE; error_return: if (debug_merge_allocated) coff_debug_merge_hash_table_free (&finfo.debug_merge); if (finfo.strtab != NULL) _bfd_stringtab_free (finfo.strtab); if (finfo.section_info != NULL) { unsigned int i; for (i = 0; i < abfd->section_count; i++) { if (finfo.section_info[i].relocs != NULL) free (finfo.section_info[i].relocs); if (finfo.section_info[i].rel_hashes != NULL) free (finfo.section_info[i].rel_hashes); } free (finfo.section_info); } if (finfo.internal_syms != NULL) free (finfo.internal_syms); if (finfo.sec_ptrs != NULL) free (finfo.sec_ptrs); if (finfo.sym_indices != NULL) free (finfo.sym_indices); if (finfo.outsyms != NULL) free (finfo.outsyms); if (finfo.linenos != NULL) free (finfo.linenos); if (finfo.contents != NULL) free (finfo.contents); if (finfo.external_relocs != NULL) free (finfo.external_relocs); if (finfo.internal_relocs != NULL) free (finfo.internal_relocs); if (external_relocs != NULL) free (external_relocs); return FALSE; } /* Parse out a -heap , line. */ static char * dores_com (char *ptr, bfd *output_bfd, int heap) { if (coff_data(output_bfd)->pe) { int val = strtoul (ptr, &ptr, 0); if (heap) pe_data(output_bfd)->pe_opthdr.SizeOfHeapReserve = val; else pe_data(output_bfd)->pe_opthdr.SizeOfStackReserve = val; if (ptr[0] == ',') { val = strtoul (ptr+1, &ptr, 0); if (heap) pe_data(output_bfd)->pe_opthdr.SizeOfHeapCommit = val; else pe_data(output_bfd)->pe_opthdr.SizeOfStackCommit = val; } } return ptr; } static char * get_name (char *ptr, char **dst) { while (*ptr == ' ') ptr++; *dst = ptr; while (*ptr && *ptr != ' ') ptr++; *ptr = 0; return ptr+1; } /* Process any magic embedded commands in a section called .drectve. */ static int process_embedded_commands (bfd *output_bfd, struct bfd_link_info *info ATTRIBUTE_UNUSED, bfd *abfd) { asection *sec = bfd_get_section_by_name (abfd, ".drectve"); char *s; char *e; bfd_byte *copy; if (!sec) return 1; if (!bfd_malloc_and_get_section (abfd, sec, ©)) { if (copy != NULL) free (copy); return 0; } e = (char *) copy + sec->size; for (s = (char *) copy; s < e ; ) { if (s[0] != '-') { s++; continue; } if (CONST_STRNEQ (s, "-attr")) { char *name; char *attribs; asection *asec; int loop = 1; int had_write = 0; int had_exec= 0; s += 5; s = get_name (s, &name); s = get_name (s, &attribs); while (loop) { switch (*attribs++) { case 'W': had_write = 1; break; case 'R': break; case 'S': break; case 'X': had_exec = 1; break; default: loop = 0; } } asec = bfd_get_section_by_name (abfd, name); if (asec) { if (had_exec) asec->flags |= SEC_CODE; if (!had_write) asec->flags |= SEC_READONLY; } } else if (CONST_STRNEQ (s, "-heap")) s = dores_com (s + 5, output_bfd, 1); else if (CONST_STRNEQ (s, "-stack")) s = dores_com (s + 6, output_bfd, 0); /* GNU extension for aligned commons. */ else if (CONST_STRNEQ (s, "-aligncomm:")) { /* Common symbols must be aligned on reading, as it is too late to do anything here, after they have already been allocated, so just skip the directive. */ s += 11; } else s++; } free (copy); return 1; } /* Place a marker against all symbols which are used by relocations. This marker can be picked up by the 'do we skip this symbol ?' loop in _bfd_coff_link_input_bfd() and used to prevent skipping that symbol. */ static void mark_relocs (struct coff_final_link_info *finfo, bfd *input_bfd) { asection * a; if ((bfd_get_file_flags (input_bfd) & HAS_SYMS) == 0) return; for (a = input_bfd->sections; a != (asection *) NULL; a = a->next) { struct internal_reloc * internal_relocs; struct internal_reloc * irel; struct internal_reloc * irelend; if ((a->flags & SEC_RELOC) == 0 || a->reloc_count < 1) continue; /* Don't mark relocs in excluded sections. */ if (a->output_section == bfd_abs_section_ptr) continue; /* Read in the relocs. */ internal_relocs = _bfd_coff_read_internal_relocs (input_bfd, a, FALSE, finfo->external_relocs, finfo->info->relocatable, (finfo->info->relocatable ? (finfo->section_info[ a->output_section->target_index ].relocs + a->output_section->reloc_count) : finfo->internal_relocs) ); if (internal_relocs == NULL) continue; irel = internal_relocs; irelend = irel + a->reloc_count; /* Place a mark in the sym_indices array (whose entries have been initialised to 0) for all of the symbols that are used in the relocation table. This will then be picked up in the skip/don't-skip pass. */ for (; irel < irelend; irel++) finfo->sym_indices[ irel->r_symndx ] = -1; } } /* Link an input file into the linker output file. This function handles all the sections and relocations of the input file at once. */ bfd_boolean _bfd_coff_link_input_bfd (struct coff_final_link_info *finfo, bfd *input_bfd) { unsigned int n_tmask = coff_data (input_bfd)->local_n_tmask; unsigned int n_btshft = coff_data (input_bfd)->local_n_btshft; bfd_boolean (*adjust_symndx) (bfd *, struct bfd_link_info *, bfd *, asection *, struct internal_reloc *, bfd_boolean *); bfd *output_bfd; const char *strings; bfd_size_type syment_base; bfd_boolean copy, hash; bfd_size_type isymesz; bfd_size_type osymesz; bfd_size_type linesz; bfd_byte *esym; bfd_byte *esym_end; struct internal_syment *isymp; asection **secpp; long *indexp; unsigned long output_index; bfd_byte *outsym; struct coff_link_hash_entry **sym_hash; asection *o; /* Move all the symbols to the output file. */ output_bfd = finfo->output_bfd; strings = NULL; syment_base = obj_raw_syment_count (output_bfd); isymesz = bfd_coff_symesz (input_bfd); osymesz = bfd_coff_symesz (output_bfd); linesz = bfd_coff_linesz (input_bfd); BFD_ASSERT (linesz == bfd_coff_linesz (output_bfd)); copy = FALSE; if (! finfo->info->keep_memory) copy = TRUE; hash = TRUE; if ((output_bfd->flags & BFD_TRADITIONAL_FORMAT) != 0) hash = FALSE; if (! _bfd_coff_get_external_symbols (input_bfd)) return FALSE; esym = (bfd_byte *) obj_coff_external_syms (input_bfd); esym_end = esym + obj_raw_syment_count (input_bfd) * isymesz; isymp = finfo->internal_syms; secpp = finfo->sec_ptrs; indexp = finfo->sym_indices; output_index = syment_base; outsym = finfo->outsyms; if (coff_data (output_bfd)->pe && ! process_embedded_commands (output_bfd, finfo->info, input_bfd)) return FALSE; /* If we are going to perform relocations and also strip/discard some symbols then we must make sure that we do not strip/discard those symbols that are going to be involved in the relocations. */ if (( finfo->info->strip != strip_none || finfo->info->discard != discard_none) && finfo->info->relocatable) { /* Mark the symbol array as 'not-used'. */ memset (indexp, 0, obj_raw_syment_count (input_bfd) * sizeof * indexp); mark_relocs (finfo, input_bfd); } while (esym < esym_end) { struct internal_syment isym; enum coff_symbol_classification classification; bfd_boolean skip; bfd_boolean global; bfd_boolean dont_skip_symbol; int add; bfd_coff_swap_sym_in (input_bfd, esym, isymp); /* Make a copy of *isymp so that the relocate_section function always sees the original values. This is more reliable than always recomputing the symbol value even if we are stripping the symbol. */ isym = *isymp; classification = bfd_coff_classify_symbol (input_bfd, &isym); switch (classification) { default: abort (); case COFF_SYMBOL_GLOBAL: case COFF_SYMBOL_PE_SECTION: case COFF_SYMBOL_LOCAL: *secpp = coff_section_from_bfd_index (input_bfd, isym.n_scnum); break; case COFF_SYMBOL_COMMON: *secpp = bfd_com_section_ptr; break; case COFF_SYMBOL_UNDEFINED: *secpp = bfd_und_section_ptr; break; } /* Extract the flag indicating if this symbol is used by a relocation. */ if ((finfo->info->strip != strip_none || finfo->info->discard != discard_none) && finfo->info->relocatable) dont_skip_symbol = *indexp; else dont_skip_symbol = FALSE; *indexp = -1; skip = FALSE; global = FALSE; add = 1 + isym.n_numaux; /* If we are stripping all symbols, we want to skip this one. */ if (finfo->info->strip == strip_all && ! dont_skip_symbol) skip = TRUE; if (! skip) { switch (classification) { default: abort (); case COFF_SYMBOL_GLOBAL: case COFF_SYMBOL_COMMON: case COFF_SYMBOL_PE_SECTION: /* This is a global symbol. Global symbols come at the end of the symbol table, so skip them for now. Locally defined function symbols, however, are an exception, and are not moved to the end. */ global = TRUE; if (! ISFCN (isym.n_type)) skip = TRUE; break; case COFF_SYMBOL_UNDEFINED: /* Undefined symbols are left for the end. */ global = TRUE; skip = TRUE; break; case COFF_SYMBOL_LOCAL: /* This is a local symbol. Skip it if we are discarding local symbols. */ if (finfo->info->discard == discard_all && ! dont_skip_symbol) skip = TRUE; break; } } #ifndef COFF_WITH_PE /* Skip section symbols for sections which are not going to be emitted. */ if (!skip && !dont_skip_symbol && isym.n_sclass == C_STAT && isym.n_type == T_NULL && isym.n_numaux > 0 && ((*secpp)->output_section == bfd_abs_section_ptr || bfd_section_removed_from_list (output_bfd, (*secpp)->output_section))) skip = TRUE; #endif /* If we stripping debugging symbols, and this is a debugging symbol, then skip it. FIXME: gas sets the section to N_ABS for some types of debugging symbols; I don't know if this is a bug or not. In any case, we handle it here. */ if (! skip && finfo->info->strip == strip_debugger && ! dont_skip_symbol && (isym.n_scnum == N_DEBUG || (isym.n_scnum == N_ABS && (isym.n_sclass == C_AUTO || isym.n_sclass == C_REG || isym.n_sclass == C_MOS || isym.n_sclass == C_MOE || isym.n_sclass == C_MOU || isym.n_sclass == C_ARG || isym.n_sclass == C_REGPARM || isym.n_sclass == C_FIELD || isym.n_sclass == C_EOS)))) skip = TRUE; /* If some symbols are stripped based on the name, work out the name and decide whether to skip this symbol. */ if (! skip && (finfo->info->strip == strip_some || finfo->info->discard == discard_l)) { const char *name; char buf[SYMNMLEN + 1]; name = _bfd_coff_internal_syment_name (input_bfd, &isym, buf); if (name == NULL) return FALSE; if (! dont_skip_symbol && ((finfo->info->strip == strip_some && (bfd_hash_lookup (finfo->info->keep_hash, name, FALSE, FALSE) == NULL)) || (! global && finfo->info->discard == discard_l && bfd_is_local_label_name (input_bfd, name)))) skip = TRUE; } /* If this is an enum, struct, or union tag, see if we have already output an identical type. */ if (! skip && (finfo->output_bfd->flags & BFD_TRADITIONAL_FORMAT) == 0 && (isym.n_sclass == C_ENTAG || isym.n_sclass == C_STRTAG || isym.n_sclass == C_UNTAG) && isym.n_numaux == 1) { const char *name; char buf[SYMNMLEN + 1]; struct coff_debug_merge_hash_entry *mh; struct coff_debug_merge_type *mt; union internal_auxent aux; struct coff_debug_merge_element **epp; bfd_byte *esl, *eslend; struct internal_syment *islp; bfd_size_type amt; name = _bfd_coff_internal_syment_name (input_bfd, &isym, buf); if (name == NULL) return FALSE; /* Ignore fake names invented by compiler; treat them all as the same name. */ if (*name == '~' || *name == '.' || *name == '$' || (*name == bfd_get_symbol_leading_char (input_bfd) && (name[1] == '~' || name[1] == '.' || name[1] == '$'))) name = ""; mh = coff_debug_merge_hash_lookup (&finfo->debug_merge, name, TRUE, TRUE); if (mh == NULL) return FALSE; /* Allocate memory to hold type information. If this turns out to be a duplicate, we pass this address to bfd_release. */ amt = sizeof (struct coff_debug_merge_type); mt = (struct coff_debug_merge_type *) bfd_alloc (input_bfd, amt); if (mt == NULL) return FALSE; mt->type_class = isym.n_sclass; /* Pick up the aux entry, which points to the end of the tag entries. */ bfd_coff_swap_aux_in (input_bfd, (esym + isymesz), isym.n_type, isym.n_sclass, 0, isym.n_numaux, &aux); /* Gather the elements. */ epp = &mt->elements; mt->elements = NULL; islp = isymp + 2; esl = esym + 2 * isymesz; eslend = ((bfd_byte *) obj_coff_external_syms (input_bfd) + aux.x_sym.x_fcnary.x_fcn.x_endndx.l * isymesz); while (esl < eslend) { const char *elename; char elebuf[SYMNMLEN + 1]; char *name_copy; bfd_coff_swap_sym_in (input_bfd, esl, islp); amt = sizeof (struct coff_debug_merge_element); *epp = (struct coff_debug_merge_element *) bfd_alloc (input_bfd, amt); if (*epp == NULL) return FALSE; elename = _bfd_coff_internal_syment_name (input_bfd, islp, elebuf); if (elename == NULL) return FALSE; amt = strlen (elename) + 1; name_copy = (char *) bfd_alloc (input_bfd, amt); if (name_copy == NULL) return FALSE; strcpy (name_copy, elename); (*epp)->name = name_copy; (*epp)->type = islp->n_type; (*epp)->tagndx = 0; if (islp->n_numaux >= 1 && islp->n_type != T_NULL && islp->n_sclass != C_EOS) { union internal_auxent eleaux; long indx; bfd_coff_swap_aux_in (input_bfd, (esl + isymesz), islp->n_type, islp->n_sclass, 0, islp->n_numaux, &eleaux); indx = eleaux.x_sym.x_tagndx.l; /* FIXME: If this tagndx entry refers to a symbol defined later in this file, we just ignore it. Handling this correctly would be tedious, and may not be required. */ if (indx > 0 && (indx < ((esym - (bfd_byte *) obj_coff_external_syms (input_bfd)) / (long) isymesz))) { (*epp)->tagndx = finfo->sym_indices[indx]; if ((*epp)->tagndx < 0) (*epp)->tagndx = 0; } } epp = &(*epp)->next; *epp = NULL; esl += (islp->n_numaux + 1) * isymesz; islp += islp->n_numaux + 1; } /* See if we already have a definition which matches this type. We always output the type if it has no elements, for simplicity. */ if (mt->elements == NULL) bfd_release (input_bfd, mt); else { struct coff_debug_merge_type *mtl; for (mtl = mh->types; mtl != NULL; mtl = mtl->next) { struct coff_debug_merge_element *me, *mel; if (mtl->type_class != mt->type_class) continue; for (me = mt->elements, mel = mtl->elements; me != NULL && mel != NULL; me = me->next, mel = mel->next) { if (strcmp (me->name, mel->name) != 0 || me->type != mel->type || me->tagndx != mel->tagndx) break; } if (me == NULL && mel == NULL) break; } if (mtl == NULL || (bfd_size_type) mtl->indx >= syment_base) { /* This is the first definition of this type. */ mt->indx = output_index; mt->next = mh->types; mh->types = mt; } else { /* This is a redefinition which can be merged. */ bfd_release (input_bfd, mt); *indexp = mtl->indx; add = (eslend - esym) / isymesz; skip = TRUE; } } } /* We now know whether we are to skip this symbol or not. */ if (! skip) { /* Adjust the symbol in order to output it. */ if (isym._n._n_n._n_zeroes == 0 && isym._n._n_n._n_offset != 0) { const char *name; bfd_size_type indx; /* This symbol has a long name. Enter it in the string table we are building. Note that we do not check bfd_coff_symname_in_debug. That is only true for XCOFF, and XCOFF requires different linking code anyhow. */ name = _bfd_coff_internal_syment_name (input_bfd, &isym, NULL); if (name == NULL) return FALSE; indx = _bfd_stringtab_add (finfo->strtab, name, hash, copy); if (indx == (bfd_size_type) -1) return FALSE; isym._n._n_n._n_offset = STRING_SIZE_SIZE + indx; } switch (isym.n_sclass) { case C_AUTO: case C_MOS: case C_EOS: case C_MOE: case C_MOU: case C_UNTAG: case C_STRTAG: case C_ENTAG: case C_TPDEF: case C_ARG: case C_USTATIC: case C_REG: case C_REGPARM: case C_FIELD: /* The symbol value should not be modified. */ break; case C_FCN: if (obj_pe (input_bfd) && strcmp (isym.n_name, ".bf") != 0 && isym.n_scnum > 0) { /* For PE, .lf and .ef get their value left alone, while .bf gets relocated. However, they all have "real" section numbers, and need to be moved into the new section. */ isym.n_scnum = (*secpp)->output_section->target_index; break; } /* Fall through. */ default: case C_LABEL: /* Not completely sure about these 2 */ case C_EXTDEF: case C_BLOCK: case C_EFCN: case C_NULL: case C_EXT: case C_STAT: case C_SECTION: case C_NT_WEAK: /* Compute new symbol location. */ if (isym.n_scnum > 0) { isym.n_scnum = (*secpp)->output_section->target_index; isym.n_value += (*secpp)->output_offset; if (! obj_pe (input_bfd)) isym.n_value -= (*secpp)->vma; if (! obj_pe (finfo->output_bfd)) isym.n_value += (*secpp)->output_section->vma; } break; case C_FILE: /* The value of a C_FILE symbol is the symbol index of the next C_FILE symbol. The value of the last C_FILE symbol is the symbol index to the first external symbol (actually, coff_renumber_symbols does not get this right--it just sets the value of the last C_FILE symbol to zero--and nobody has ever complained about it). We try to get this right, below, just before we write the symbols out, but in the general case we may have to write the symbol out twice. */ if (finfo->last_file_index != -1 && finfo->last_file.n_value != (bfd_vma) output_index) { /* We must correct the value of the last C_FILE entry. */ finfo->last_file.n_value = output_index; if ((bfd_size_type) finfo->last_file_index >= syment_base) { /* The last C_FILE symbol is in this input file. */ bfd_coff_swap_sym_out (output_bfd, &finfo->last_file, (finfo->outsyms + ((finfo->last_file_index - syment_base) * osymesz))); } else { file_ptr pos; /* We have already written out the last C_FILE symbol. We need to write it out again. We borrow *outsym temporarily. */ bfd_coff_swap_sym_out (output_bfd, &finfo->last_file, outsym); pos = obj_sym_filepos (output_bfd); pos += finfo->last_file_index * osymesz; if (bfd_seek (output_bfd, pos, SEEK_SET) != 0 || bfd_bwrite (outsym, osymesz, output_bfd) != osymesz) return FALSE; } } finfo->last_file_index = output_index; finfo->last_file = isym; break; } /* If doing task linking, convert normal global function symbols to static functions. */ if (finfo->info->task_link && IS_EXTERNAL (input_bfd, isym)) isym.n_sclass = C_STAT; /* Output the symbol. */ bfd_coff_swap_sym_out (output_bfd, &isym, outsym); *indexp = output_index; if (global) { long indx; struct coff_link_hash_entry *h; indx = ((esym - (bfd_byte *) obj_coff_external_syms (input_bfd)) / isymesz); h = obj_coff_sym_hashes (input_bfd)[indx]; if (h == NULL) { /* This can happen if there were errors earlier in the link. */ bfd_set_error (bfd_error_bad_value); return FALSE; } h->indx = output_index; } output_index += add; outsym += add * osymesz; } esym += add * isymesz; isymp += add; ++secpp; ++indexp; for (--add; add > 0; --add) { *secpp++ = NULL; *indexp++ = -1; } } /* Fix up the aux entries. This must be done in a separate pass, because we don't know the correct symbol indices until we have already decided which symbols we are going to keep. */ esym = (bfd_byte *) obj_coff_external_syms (input_bfd); esym_end = esym + obj_raw_syment_count (input_bfd) * isymesz; isymp = finfo->internal_syms; indexp = finfo->sym_indices; sym_hash = obj_coff_sym_hashes (input_bfd); outsym = finfo->outsyms; while (esym < esym_end) { int add; add = 1 + isymp->n_numaux; if ((*indexp < 0 || (bfd_size_type) *indexp < syment_base) && (*sym_hash == NULL || (*sym_hash)->auxbfd != input_bfd)) esym += add * isymesz; else { struct coff_link_hash_entry *h; int i; h = NULL; if (*indexp < 0) { h = *sym_hash; /* The m68k-motorola-sysv assembler will sometimes generate two symbols with the same name, but only one will have aux entries. */ BFD_ASSERT (isymp->n_numaux == 0 || h->numaux == 0 || h->numaux == isymp->n_numaux); } esym += isymesz; if (h == NULL) outsym += osymesz; /* Handle the aux entries. This handling is based on coff_pointerize_aux. I don't know if it always correct. */ for (i = 0; i < isymp->n_numaux && esym < esym_end; i++) { union internal_auxent aux; union internal_auxent *auxp; if (h != NULL && h->aux != NULL && (h->numaux > i)) auxp = h->aux + i; else { bfd_coff_swap_aux_in (input_bfd, esym, isymp->n_type, isymp->n_sclass, i, isymp->n_numaux, &aux); auxp = &aux; } if (isymp->n_sclass == C_FILE) { /* If this is a long filename, we must put it in the string table. */ if (auxp->x_file.x_n.x_zeroes == 0 && auxp->x_file.x_n.x_offset != 0) { const char *filename; bfd_size_type indx; BFD_ASSERT (auxp->x_file.x_n.x_offset >= STRING_SIZE_SIZE); if (strings == NULL) { strings = _bfd_coff_read_string_table (input_bfd); if (strings == NULL) return FALSE; } filename = strings + auxp->x_file.x_n.x_offset; indx = _bfd_stringtab_add (finfo->strtab, filename, hash, copy); if (indx == (bfd_size_type) -1) return FALSE; auxp->x_file.x_n.x_offset = STRING_SIZE_SIZE + indx; } } else if ((isymp->n_sclass != C_STAT || isymp->n_type != T_NULL) && isymp->n_sclass != C_NT_WEAK) { unsigned long indx; if (ISFCN (isymp->n_type) || ISTAG (isymp->n_sclass) || isymp->n_sclass == C_BLOCK || isymp->n_sclass == C_FCN) { indx = auxp->x_sym.x_fcnary.x_fcn.x_endndx.l; if (indx > 0 && indx < obj_raw_syment_count (input_bfd)) { /* We look forward through the symbol for the index of the next symbol we are going to include. I don't know if this is entirely right. */ while ((finfo->sym_indices[indx] < 0 || ((bfd_size_type) finfo->sym_indices[indx] < syment_base)) && indx < obj_raw_syment_count (input_bfd)) ++indx; if (indx >= obj_raw_syment_count (input_bfd)) indx = output_index; else indx = finfo->sym_indices[indx]; auxp->x_sym.x_fcnary.x_fcn.x_endndx.l = indx; } } indx = auxp->x_sym.x_tagndx.l; if (indx > 0 && indx < obj_raw_syment_count (input_bfd)) { long symindx; symindx = finfo->sym_indices[indx]; if (symindx < 0) auxp->x_sym.x_tagndx.l = 0; else auxp->x_sym.x_tagndx.l = symindx; } /* The .bf symbols are supposed to be linked through the endndx field. We need to carry this list across object files. */ if (i == 0 && h == NULL && isymp->n_sclass == C_FCN && (isymp->_n._n_n._n_zeroes != 0 || isymp->_n._n_n._n_offset == 0) && isymp->_n._n_name[0] == '.' && isymp->_n._n_name[1] == 'b' && isymp->_n._n_name[2] == 'f' && isymp->_n._n_name[3] == '\0') { if (finfo->last_bf_index != -1) { finfo->last_bf.x_sym.x_fcnary.x_fcn.x_endndx.l = *indexp; if ((bfd_size_type) finfo->last_bf_index >= syment_base) { void *auxout; /* The last .bf symbol is in this input file. This will only happen if the assembler did not set up the .bf endndx symbols correctly. */ auxout = (finfo->outsyms + ((finfo->last_bf_index - syment_base) * osymesz)); bfd_coff_swap_aux_out (output_bfd, &finfo->last_bf, isymp->n_type, isymp->n_sclass, 0, isymp->n_numaux, auxout); } else { file_ptr pos; /* We have already written out the last .bf aux entry. We need to write it out again. We borrow *outsym temporarily. FIXME: This case should be made faster. */ bfd_coff_swap_aux_out (output_bfd, &finfo->last_bf, isymp->n_type, isymp->n_sclass, 0, isymp->n_numaux, outsym); pos = obj_sym_filepos (output_bfd); pos += finfo->last_bf_index * osymesz; if (bfd_seek (output_bfd, pos, SEEK_SET) != 0 || (bfd_bwrite (outsym, osymesz, output_bfd) != osymesz)) return FALSE; } } if (auxp->x_sym.x_fcnary.x_fcn.x_endndx.l != 0) finfo->last_bf_index = -1; else { /* The endndx field of this aux entry must be updated with the symbol number of the next .bf symbol. */ finfo->last_bf = *auxp; finfo->last_bf_index = (((outsym - finfo->outsyms) / osymesz) + syment_base); } } } if (h == NULL) { bfd_coff_swap_aux_out (output_bfd, auxp, isymp->n_type, isymp->n_sclass, i, isymp->n_numaux, outsym); outsym += osymesz; } esym += isymesz; } } indexp += add; isymp += add; sym_hash += add; } /* Relocate the line numbers, unless we are stripping them. */ if (finfo->info->strip == strip_none || finfo->info->strip == strip_some) { for (o = input_bfd->sections; o != NULL; o = o->next) { bfd_vma offset; bfd_byte *eline; bfd_byte *elineend; bfd_byte *oeline; bfd_boolean skipping; file_ptr pos; bfd_size_type amt; /* FIXME: If SEC_HAS_CONTENTS is not for the section, then build_link_order in ldwrite.c will not have created a link order, which means that we will not have seen this input section in _bfd_coff_final_link, which means that we will not have allocated space for the line numbers of this section. I don't think line numbers can be meaningful for a section which does not have SEC_HAS_CONTENTS set, but, if they do, this must be changed. */ if (o->lineno_count == 0 || (o->output_section->flags & SEC_HAS_CONTENTS) == 0) continue; if (bfd_seek (input_bfd, o->line_filepos, SEEK_SET) != 0 || bfd_bread (finfo->linenos, linesz * o->lineno_count, input_bfd) != linesz * o->lineno_count) return FALSE; offset = o->output_section->vma + o->output_offset - o->vma; eline = finfo->linenos; oeline = finfo->linenos; elineend = eline + linesz * o->lineno_count; skipping = FALSE; for (; eline < elineend; eline += linesz) { struct internal_lineno iline; bfd_coff_swap_lineno_in (input_bfd, eline, &iline); if (iline.l_lnno != 0) iline.l_addr.l_paddr += offset; else if (iline.l_addr.l_symndx >= 0 && ((unsigned long) iline.l_addr.l_symndx < obj_raw_syment_count (input_bfd))) { long indx; indx = finfo->sym_indices[iline.l_addr.l_symndx]; if (indx < 0) { /* These line numbers are attached to a symbol which we are stripping. We must discard the line numbers because reading them back with no associated symbol (or associating them all with symbol #0) will fail. We can't regain the space in the output file, but at least they're dense. */ skipping = TRUE; } else { struct internal_syment is; union internal_auxent ia; /* Fix up the lnnoptr field in the aux entry of the symbol. It turns out that we can't do this when we modify the symbol aux entries, because gas sometimes screws up the lnnoptr field and makes it an offset from the start of the line numbers rather than an absolute file index. */ bfd_coff_swap_sym_in (output_bfd, (finfo->outsyms + ((indx - syment_base) * osymesz)), &is); if ((ISFCN (is.n_type) || is.n_sclass == C_BLOCK) && is.n_numaux >= 1) { void *auxptr; auxptr = (finfo->outsyms + ((indx - syment_base + 1) * osymesz)); bfd_coff_swap_aux_in (output_bfd, auxptr, is.n_type, is.n_sclass, 0, is.n_numaux, &ia); ia.x_sym.x_fcnary.x_fcn.x_lnnoptr = (o->output_section->line_filepos + o->output_section->lineno_count * linesz + eline - finfo->linenos); bfd_coff_swap_aux_out (output_bfd, &ia, is.n_type, is.n_sclass, 0, is.n_numaux, auxptr); } skipping = FALSE; } iline.l_addr.l_symndx = indx; } if (!skipping) { bfd_coff_swap_lineno_out (output_bfd, &iline, oeline); oeline += linesz; } } pos = o->output_section->line_filepos; pos += o->output_section->lineno_count * linesz; amt = oeline - finfo->linenos; if (bfd_seek (output_bfd, pos, SEEK_SET) != 0 || bfd_bwrite (finfo->linenos, amt, output_bfd) != amt) return FALSE; o->output_section->lineno_count += amt / linesz; } } /* If we swapped out a C_FILE symbol, guess that the next C_FILE symbol will be the first symbol in the next input file. In the normal case, this will save us from writing out the C_FILE symbol again. */ if (finfo->last_file_index != -1 && (bfd_size_type) finfo->last_file_index >= syment_base) { finfo->last_file.n_value = output_index; bfd_coff_swap_sym_out (output_bfd, &finfo->last_file, (finfo->outsyms + ((finfo->last_file_index - syment_base) * osymesz))); } /* Write the modified symbols to the output file. */ if (outsym > finfo->outsyms) { file_ptr pos; bfd_size_type amt; pos = obj_sym_filepos (output_bfd) + syment_base * osymesz; amt = outsym - finfo->outsyms; if (bfd_seek (output_bfd, pos, SEEK_SET) != 0 || bfd_bwrite (finfo->outsyms, amt, output_bfd) != amt) return FALSE; BFD_ASSERT ((obj_raw_syment_count (output_bfd) + (outsym - finfo->outsyms) / osymesz) == output_index); obj_raw_syment_count (output_bfd) = output_index; } /* Relocate the contents of each section. */ adjust_symndx = coff_backend_info (input_bfd)->_bfd_coff_adjust_symndx; for (o = input_bfd->sections; o != NULL; o = o->next) { bfd_byte *contents; struct coff_section_tdata *secdata; if (! o->linker_mark) /* This section was omitted from the link. */ continue; if ((o->flags & SEC_LINKER_CREATED) != 0) continue; if ((o->flags & SEC_HAS_CONTENTS) == 0 || (o->size == 0 && (o->flags & SEC_RELOC) == 0)) { if ((o->flags & SEC_RELOC) != 0 && o->reloc_count != 0) { (*_bfd_error_handler) (_("%B: relocs in section `%A', but it has no contents"), input_bfd, o); bfd_set_error (bfd_error_no_contents); return FALSE; } continue; } secdata = coff_section_data (input_bfd, o); if (secdata != NULL && secdata->contents != NULL) contents = secdata->contents; else { bfd_size_type x = o->rawsize ? o->rawsize : o->size; if (! bfd_get_section_contents (input_bfd, o, finfo->contents, 0, x)) return FALSE; contents = finfo->contents; } if ((o->flags & SEC_RELOC) != 0) { int target_index; struct internal_reloc *internal_relocs; struct internal_reloc *irel; /* Read in the relocs. */ target_index = o->output_section->target_index; internal_relocs = (_bfd_coff_read_internal_relocs (input_bfd, o, FALSE, finfo->external_relocs, finfo->info->relocatable, (finfo->info->relocatable ? (finfo->section_info[target_index].relocs + o->output_section->reloc_count) : finfo->internal_relocs))); if (internal_relocs == NULL) return FALSE; /* Run through the relocs looking for relocs against symbols coming from discarded sections and complain about them. */ irel = internal_relocs; for (; irel < &internal_relocs[o->reloc_count]; irel++) { struct coff_link_hash_entry *h; asection *ps = NULL; long symndx = irel->r_symndx; if (symndx < 0) continue; h = obj_coff_sym_hashes (input_bfd)[symndx]; if (h == NULL) continue; while (h->root.type == bfd_link_hash_indirect || h->root.type == bfd_link_hash_warning) h = (struct coff_link_hash_entry *) h->root.u.i.link; if (h->root.type == bfd_link_hash_defined || h->root.type == bfd_link_hash_defweak) ps = h->root.u.def.section; if (ps == NULL) continue; /* Complain if definition comes from an excluded section. */ if (ps->flags & SEC_EXCLUDE) (*finfo->info->callbacks->einfo) (_("%X`%s' referenced in section `%A' of %B: " "defined in discarded section `%A' of %B\n"), h->root.root.string, o, input_bfd, ps, ps->owner); } /* Call processor specific code to relocate the section contents. */ if (! bfd_coff_relocate_section (output_bfd, finfo->info, input_bfd, o, contents, internal_relocs, finfo->internal_syms, finfo->sec_ptrs)) return FALSE; if (finfo->info->relocatable) { bfd_vma offset; struct internal_reloc *irelend; struct coff_link_hash_entry **rel_hash; offset = o->output_section->vma + o->output_offset - o->vma; irel = internal_relocs; irelend = irel + o->reloc_count; rel_hash = (finfo->section_info[target_index].rel_hashes + o->output_section->reloc_count); for (; irel < irelend; irel++, rel_hash++) { struct coff_link_hash_entry *h; bfd_boolean adjusted; *rel_hash = NULL; /* Adjust the reloc address and symbol index. */ irel->r_vaddr += offset; if (irel->r_symndx == -1) continue; if (adjust_symndx) { if (! (*adjust_symndx) (output_bfd, finfo->info, input_bfd, o, irel, &adjusted)) return FALSE; if (adjusted) continue; } h = obj_coff_sym_hashes (input_bfd)[irel->r_symndx]; if (h != NULL) { /* This is a global symbol. */ if (h->indx >= 0) irel->r_symndx = h->indx; else { /* This symbol is being written at the end of the file, and we do not yet know the symbol index. We save the pointer to the hash table entry in the rel_hash list. We set the indx field to -2 to indicate that this symbol must not be stripped. */ *rel_hash = h; h->indx = -2; } } else { long indx; indx = finfo->sym_indices[irel->r_symndx]; if (indx != -1) irel->r_symndx = indx; else { struct internal_syment *is; const char *name; char buf[SYMNMLEN + 1]; /* This reloc is against a symbol we are stripping. This should have been handled by the 'dont_skip_symbol' code in the while loop at the top of this function. */ is = finfo->internal_syms + irel->r_symndx; name = (_bfd_coff_internal_syment_name (input_bfd, is, buf)); if (name == NULL) return FALSE; if (! ((*finfo->info->callbacks->unattached_reloc) (finfo->info, name, input_bfd, o, irel->r_vaddr))) return FALSE; } } } o->output_section->reloc_count += o->reloc_count; } } /* Write out the modified section contents. */ if (secdata == NULL || secdata->stab_info == NULL) { file_ptr loc = o->output_offset * bfd_octets_per_byte (output_bfd); if (! bfd_set_section_contents (output_bfd, o->output_section, contents, loc, o->size)) return FALSE; } else { if (! (_bfd_write_section_stabs (output_bfd, &coff_hash_table (finfo->info)->stab_info, o, &secdata->stab_info, contents))) return FALSE; } } if (! finfo->info->keep_memory && ! _bfd_coff_free_symbols (input_bfd)) return FALSE; return TRUE; } /* Write out a global symbol. Called via bfd_hash_traverse. */ bfd_boolean _bfd_coff_write_global_sym (struct bfd_hash_entry *bh, void *data) { struct coff_link_hash_entry *h = (struct coff_link_hash_entry *) bh; struct coff_final_link_info *finfo = (struct coff_final_link_info *) data; bfd *output_bfd; struct internal_syment isym; bfd_size_type symesz; unsigned int i; file_ptr pos; output_bfd = finfo->output_bfd; if (h->root.type == bfd_link_hash_warning) { h = (struct coff_link_hash_entry *) h->root.u.i.link; if (h->root.type == bfd_link_hash_new) return TRUE; } if (h->indx >= 0) return TRUE; if (h->indx != -2 && (finfo->info->strip == strip_all || (finfo->info->strip == strip_some && (bfd_hash_lookup (finfo->info->keep_hash, h->root.root.string, FALSE, FALSE) == NULL)))) return TRUE; switch (h->root.type) { default: case bfd_link_hash_new: case bfd_link_hash_warning: abort (); return FALSE; case bfd_link_hash_undefined: case bfd_link_hash_undefweak: isym.n_scnum = N_UNDEF; isym.n_value = 0; break; case bfd_link_hash_defined: case bfd_link_hash_defweak: { asection *sec; sec = h->root.u.def.section->output_section; if (bfd_is_abs_section (sec)) isym.n_scnum = N_ABS; else isym.n_scnum = sec->target_index; isym.n_value = (h->root.u.def.value + h->root.u.def.section->output_offset); if (! obj_pe (finfo->output_bfd)) isym.n_value += sec->vma; } break; case bfd_link_hash_common: isym.n_scnum = N_UNDEF; isym.n_value = h->root.u.c.size; break; case bfd_link_hash_indirect: /* Just ignore these. They can't be handled anyhow. */ return TRUE; } if (strlen (h->root.root.string) <= SYMNMLEN) strncpy (isym._n._n_name, h->root.root.string, SYMNMLEN); else { bfd_boolean hash; bfd_size_type indx; hash = TRUE; if ((output_bfd->flags & BFD_TRADITIONAL_FORMAT) != 0) hash = FALSE; indx = _bfd_stringtab_add (finfo->strtab, h->root.root.string, hash, FALSE); if (indx == (bfd_size_type) -1) { finfo->failed = TRUE; return FALSE; } isym._n._n_n._n_zeroes = 0; isym._n._n_n._n_offset = STRING_SIZE_SIZE + indx; } isym.n_sclass = h->symbol_class; isym.n_type = h->type; if (isym.n_sclass == C_NULL) isym.n_sclass = C_EXT; /* If doing task linking and this is the pass where we convert defined globals to statics, then do that conversion now. If the symbol is not being converted, just ignore it and it will be output during a later pass. */ if (finfo->global_to_static) { if (! IS_EXTERNAL (output_bfd, isym)) return TRUE; isym.n_sclass = C_STAT; } /* When a weak symbol is not overridden by a strong one, turn it into an external symbol when not building a shared or relocatable object. */ if (! finfo->info->shared && ! finfo->info->relocatable && IS_WEAK_EXTERNAL (finfo->output_bfd, isym)) isym.n_sclass = C_EXT; isym.n_numaux = h->numaux; bfd_coff_swap_sym_out (output_bfd, &isym, finfo->outsyms); symesz = bfd_coff_symesz (output_bfd); pos = obj_sym_filepos (output_bfd); pos += obj_raw_syment_count (output_bfd) * symesz; if (bfd_seek (output_bfd, pos, SEEK_SET) != 0 || bfd_bwrite (finfo->outsyms, symesz, output_bfd) != symesz) { finfo->failed = TRUE; return FALSE; } h->indx = obj_raw_syment_count (output_bfd); ++obj_raw_syment_count (output_bfd); /* Write out any associated aux entries. Most of the aux entries will have been modified in _bfd_coff_link_input_bfd. We have to handle section aux entries here, now that we have the final relocation and line number counts. */ for (i = 0; i < isym.n_numaux; i++) { union internal_auxent *auxp; auxp = h->aux + i; /* Look for a section aux entry here using the same tests that coff_swap_aux_out uses. */ if (i == 0 && (isym.n_sclass == C_STAT || isym.n_sclass == C_HIDDEN) && isym.n_type == T_NULL && (h->root.type == bfd_link_hash_defined || h->root.type == bfd_link_hash_defweak)) { asection *sec; sec = h->root.u.def.section->output_section; if (sec != NULL) { auxp->x_scn.x_scnlen = sec->size; /* For PE, an overflow on the final link reportedly does not matter. FIXME: Why not? */ if (sec->reloc_count > 0xffff && (! obj_pe (output_bfd) || finfo->info->relocatable)) (*_bfd_error_handler) (_("%s: %s: reloc overflow: 0x%lx > 0xffff"), bfd_get_filename (output_bfd), bfd_get_section_name (output_bfd, sec), sec->reloc_count); if (sec->lineno_count > 0xffff && (! obj_pe (output_bfd) || finfo->info->relocatable)) (*_bfd_error_handler) (_("%s: warning: %s: line number overflow: 0x%lx > 0xffff"), bfd_get_filename (output_bfd), bfd_get_section_name (output_bfd, sec), sec->lineno_count); auxp->x_scn.x_nreloc = sec->reloc_count; auxp->x_scn.x_nlinno = sec->lineno_count; auxp->x_scn.x_checksum = 0; auxp->x_scn.x_associated = 0; auxp->x_scn.x_comdat = 0; } } bfd_coff_swap_aux_out (output_bfd, auxp, isym.n_type, isym.n_sclass, (int) i, isym.n_numaux, finfo->outsyms); if (bfd_bwrite (finfo->outsyms, symesz, output_bfd) != symesz) { finfo->failed = TRUE; return FALSE; } ++obj_raw_syment_count (output_bfd); } return TRUE; } /* Write out task global symbols, converting them to statics. Called via coff_link_hash_traverse. Calls bfd_coff_write_global_sym to do the dirty work, if the symbol we are processing needs conversion. */ bfd_boolean _bfd_coff_write_task_globals (struct coff_link_hash_entry *h, void *data) { struct coff_final_link_info *finfo = (struct coff_final_link_info *) data; bfd_boolean rtnval = TRUE; bfd_boolean save_global_to_static; if (h->root.type == bfd_link_hash_warning) h = (struct coff_link_hash_entry *) h->root.u.i.link; if (h->indx < 0) { switch (h->root.type) { case bfd_link_hash_defined: case bfd_link_hash_defweak: save_global_to_static = finfo->global_to_static; finfo->global_to_static = TRUE; rtnval = _bfd_coff_write_global_sym (&h->root.root, data); finfo->global_to_static = save_global_to_static; break; default: break; } } return (rtnval); } /* Handle a link order which is supposed to generate a reloc. */ bfd_boolean _bfd_coff_reloc_link_order (bfd *output_bfd, struct coff_final_link_info *finfo, asection *output_section, struct bfd_link_order *link_order) { reloc_howto_type *howto; struct internal_reloc *irel; struct coff_link_hash_entry **rel_hash_ptr; howto = bfd_reloc_type_lookup (output_bfd, link_order->u.reloc.p->reloc); if (howto == NULL) { bfd_set_error (bfd_error_bad_value); return FALSE; } if (link_order->u.reloc.p->addend != 0) { bfd_size_type size; bfd_byte *buf; bfd_reloc_status_type rstat; bfd_boolean ok; file_ptr loc; size = bfd_get_reloc_size (howto); buf = (bfd_byte *) bfd_zmalloc (size); if (buf == NULL) return FALSE; rstat = _bfd_relocate_contents (howto, output_bfd, (bfd_vma) link_order->u.reloc.p->addend,\ buf); switch (rstat) { case bfd_reloc_ok: break; default: case bfd_reloc_outofrange: abort (); case bfd_reloc_overflow: if (! ((*finfo->info->callbacks->reloc_overflow) (finfo->info, NULL, (link_order->type == bfd_section_reloc_link_order ? bfd_section_name (output_bfd, link_order->u.reloc.p->u.section) : link_order->u.reloc.p->u.name), howto->name, link_order->u.reloc.p->addend, (bfd *) NULL, (asection *) NULL, (bfd_vma) 0))) { free (buf); return FALSE; } break; } loc = link_order->offset * bfd_octets_per_byte (output_bfd); ok = bfd_set_section_contents (output_bfd, output_section, buf, loc, size); free (buf); if (! ok) return FALSE; } /* Store the reloc information in the right place. It will get swapped and written out at the end of the final_link routine. */ irel = (finfo->section_info[output_section->target_index].relocs + output_section->reloc_count); rel_hash_ptr = (finfo->section_info[output_section->target_index].rel_hashes + output_section->reloc_count); memset (irel, 0, sizeof (struct internal_reloc)); *rel_hash_ptr = NULL; irel->r_vaddr = output_section->vma + link_order->offset; if (link_order->type == bfd_section_reloc_link_order) { /* We need to somehow locate a symbol in the right section. The symbol must either have a value of zero, or we must adjust the addend by the value of the symbol. FIXME: Write this when we need it. The old linker couldn't handle this anyhow. */ abort (); *rel_hash_ptr = NULL; irel->r_symndx = 0; } else { struct coff_link_hash_entry *h; h = ((struct coff_link_hash_entry *) bfd_wrapped_link_hash_lookup (output_bfd, finfo->info, link_order->u.reloc.p->u.name, FALSE, FALSE, TRUE)); if (h != NULL) { if (h->indx >= 0) irel->r_symndx = h->indx; else { /* Set the index to -2 to force this symbol to get written out. */ h->indx = -2; *rel_hash_ptr = h; irel->r_symndx = 0; } } else { if (! ((*finfo->info->callbacks->unattached_reloc) (finfo->info, link_order->u.reloc.p->u.name, (bfd *) NULL, (asection *) NULL, (bfd_vma) 0))) return FALSE; irel->r_symndx = 0; } } /* FIXME: Is this always right? */ irel->r_type = howto->type; /* r_size is only used on the RS/6000, which needs its own linker routines anyhow. r_extern is only used for ECOFF. */ /* FIXME: What is the right value for r_offset? Is zero OK? */ ++output_section->reloc_count; return TRUE; } /* A basic reloc handling routine which may be used by processors with simple relocs. */ bfd_boolean _bfd_coff_generic_relocate_section (bfd *output_bfd, struct bfd_link_info *info, bfd *input_bfd, asection *input_section, bfd_byte *contents, struct internal_reloc *relocs, struct internal_syment *syms, asection **sections) { struct internal_reloc *rel; struct internal_reloc *relend; rel = relocs; relend = rel + input_section->reloc_count; for (; rel < relend; rel++) { long symndx; struct coff_link_hash_entry *h; struct internal_syment *sym; bfd_vma addend; bfd_vma val; reloc_howto_type *howto; bfd_reloc_status_type rstat; symndx = rel->r_symndx; if (symndx == -1) { h = NULL; sym = NULL; } else if (symndx < 0 || (unsigned long) symndx >= obj_raw_syment_count (input_bfd)) { (*_bfd_error_handler) ("%B: illegal symbol index %ld in relocs", input_bfd, symndx); return FALSE; } else { h = obj_coff_sym_hashes (input_bfd)[symndx]; sym = syms + symndx; } /* COFF treats common symbols in one of two ways. Either the size of the symbol is included in the section contents, or it is not. We assume that the size is not included, and force the rtype_to_howto function to adjust the addend as needed. */ if (sym != NULL && sym->n_scnum != 0) addend = - sym->n_value; else addend = 0; howto = bfd_coff_rtype_to_howto (input_bfd, input_section, rel, h, sym, &addend); if (howto == NULL) return FALSE; /* If we are doing a relocatable link, then we can just ignore a PC relative reloc that is pcrel_offset. It will already have the correct value. If this is not a relocatable link, then we should ignore the symbol value. */ if (howto->pc_relative && howto->pcrel_offset) { if (info->relocatable) continue; if (sym != NULL && sym->n_scnum != 0) addend += sym->n_value; } val = 0; if (h == NULL) { asection *sec; if (symndx == -1) { sec = bfd_abs_section_ptr; val = 0; } else { sec = sections[symndx]; val = (sec->output_section->vma + sec->output_offset + sym->n_value); if (! obj_pe (input_bfd)) val -= sec->vma; } } else { if (h->root.type == bfd_link_hash_defined || h->root.type == bfd_link_hash_defweak) { /* Defined weak symbols are a GNU extension. */ asection *sec; sec = h->root.u.def.section; val = (h->root.u.def.value + sec->output_section->vma + sec->output_offset); } else if (h->root.type == bfd_link_hash_undefweak) { if (h->symbol_class == C_NT_WEAK && h->numaux == 1) { /* See _Microsoft Portable Executable and Common Object File Format Specification_, section 5.5.3. Note that weak symbols without aux records are a GNU extension. FIXME: All weak externals are treated as having characteristic IMAGE_WEAK_EXTERN_SEARCH_NOLIBRARY (1). These behave as per SVR4 ABI: A library member will resolve a weak external only if a normal external causes the library member to be linked. See also linker.c: generic_link_check_archive_element. */ asection *sec; struct coff_link_hash_entry *h2 = h->auxbfd->tdata.coff_obj_data->sym_hashes[ h->aux->x_sym.x_tagndx.l]; if (!h2 || h2->root.type == bfd_link_hash_undefined) { sec = bfd_abs_section_ptr; val = 0; } else { sec = h2->root.u.def.section; val = h2->root.u.def.value + sec->output_section->vma + sec->output_offset; } } else /* This is a GNU extension. */ val = 0; } else if (! info->relocatable) { if (! ((*info->callbacks->undefined_symbol) (info, h->root.root.string, input_bfd, input_section, rel->r_vaddr - input_section->vma, TRUE))) return FALSE; } } if (info->base_file) { /* Emit a reloc if the backend thinks it needs it. */ if (sym && pe_data (output_bfd)->in_reloc_p (output_bfd, howto)) { /* Relocation to a symbol in a section which isn't absolute. We output the address here to a file. This file is then read by dlltool when generating the reloc section. Note that the base file is not portable between systems. We write out a bfd_vma here, and dlltool reads in a bfd_vma. */ bfd_vma addr = (rel->r_vaddr - input_section->vma + input_section->output_offset + input_section->output_section->vma); if (coff_data (output_bfd)->pe) addr -= pe_data(output_bfd)->pe_opthdr.ImageBase; if (fwrite (&addr, 1, sizeof (bfd_vma), (FILE *) info->base_file) != sizeof (bfd_vma)) { bfd_set_error (bfd_error_system_call); return FALSE; } } } rstat = _bfd_final_link_relocate (howto, input_bfd, input_section, contents, rel->r_vaddr - input_section->vma, val, addend); switch (rstat) { default: abort (); case bfd_reloc_ok: break; case bfd_reloc_outofrange: (*_bfd_error_handler) (_("%B: bad reloc address 0x%lx in section `%A'"), input_bfd, input_section, (unsigned long) rel->r_vaddr); return FALSE; case bfd_reloc_overflow: { const char *name; char buf[SYMNMLEN + 1]; if (symndx == -1) name = "*ABS*"; else if (h != NULL) name = NULL; else { name = _bfd_coff_internal_syment_name (input_bfd, sym, buf); if (name == NULL) return FALSE; } if (! ((*info->callbacks->reloc_overflow) (info, (h ? &h->root : NULL), name, howto->name, (bfd_vma) 0, input_bfd, input_section, rel->r_vaddr - input_section->vma))) return FALSE; } } } return TRUE; }