147d994bcd
PR ld/20401 * coffgen.c (fini_reloc_cookie_rels): Check for the extistence of the coff_section_data before using it.
3066 lines
87 KiB
C
3066 lines
87 KiB
C
/* Support for the generic parts of COFF, for BFD.
|
|
Copyright (C) 1990-2016 Free Software Foundation, Inc.
|
|
Written by 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. */
|
|
|
|
/* Most of this hacked by Steve Chamberlain, sac@cygnus.com.
|
|
Split out of coffcode.h by Ian Taylor, ian@cygnus.com. */
|
|
|
|
/* This file contains COFF code that is not dependent on any
|
|
particular COFF target. There is only one version of this file in
|
|
libbfd.a, so no target specific code may be put in here. Or, to
|
|
put it another way,
|
|
|
|
********** DO NOT PUT TARGET SPECIFIC CODE IN THIS FILE **********
|
|
|
|
If you need to add some target specific behaviour, add a new hook
|
|
function to bfd_coff_backend_data.
|
|
|
|
Some of these functions are also called by the ECOFF routines.
|
|
Those functions may not use any COFF specific information, such as
|
|
coff_data (abfd). */
|
|
|
|
#include "sysdep.h"
|
|
#include "bfd.h"
|
|
#include "libbfd.h"
|
|
#include "coff/internal.h"
|
|
#include "libcoff.h"
|
|
|
|
/* Take a section header read from a coff file (in HOST byte order),
|
|
and make a BFD "section" out of it. This is used by ECOFF. */
|
|
|
|
static bfd_boolean
|
|
make_a_section_from_file (bfd *abfd,
|
|
struct internal_scnhdr *hdr,
|
|
unsigned int target_index)
|
|
{
|
|
asection *return_section;
|
|
char *name;
|
|
bfd_boolean result = TRUE;
|
|
flagword flags;
|
|
|
|
name = NULL;
|
|
|
|
/* Handle long section names as in PE. On reading, we want to
|
|
accept long names if the format permits them at all, regardless
|
|
of the current state of the flag that dictates if we would generate
|
|
them in outputs; this construct checks if that is the case by
|
|
attempting to set the flag, without changing its state; the call
|
|
will fail for formats that do not support long names at all. */
|
|
if (bfd_coff_set_long_section_names (abfd, bfd_coff_long_section_names (abfd))
|
|
&& hdr->s_name[0] == '/')
|
|
{
|
|
char buf[SCNNMLEN];
|
|
long strindex;
|
|
char *p;
|
|
const char *strings;
|
|
|
|
/* Flag that this BFD uses long names, even though the format might
|
|
expect them to be off by default. This won't directly affect the
|
|
format of any output BFD created from this one, but the information
|
|
can be used to decide what to do. */
|
|
bfd_coff_set_long_section_names (abfd, TRUE);
|
|
memcpy (buf, hdr->s_name + 1, SCNNMLEN - 1);
|
|
buf[SCNNMLEN - 1] = '\0';
|
|
strindex = strtol (buf, &p, 10);
|
|
if (*p == '\0' && strindex >= 0)
|
|
{
|
|
strings = _bfd_coff_read_string_table (abfd);
|
|
if (strings == NULL)
|
|
return FALSE;
|
|
if ((bfd_size_type)(strindex + 2) >= obj_coff_strings_len (abfd))
|
|
return FALSE;
|
|
strings += strindex;
|
|
name = (char *) bfd_alloc (abfd,
|
|
(bfd_size_type) strlen (strings) + 1 + 1);
|
|
if (name == NULL)
|
|
return FALSE;
|
|
strcpy (name, strings);
|
|
}
|
|
}
|
|
|
|
if (name == NULL)
|
|
{
|
|
/* Assorted wastage to null-terminate the name, thanks AT&T! */
|
|
name = (char *) bfd_alloc (abfd,
|
|
(bfd_size_type) sizeof (hdr->s_name) + 1 + 1);
|
|
if (name == NULL)
|
|
return FALSE;
|
|
strncpy (name, (char *) &hdr->s_name[0], sizeof (hdr->s_name));
|
|
name[sizeof (hdr->s_name)] = 0;
|
|
}
|
|
|
|
return_section = bfd_make_section_anyway (abfd, name);
|
|
if (return_section == NULL)
|
|
return FALSE;
|
|
|
|
return_section->vma = hdr->s_vaddr;
|
|
return_section->lma = hdr->s_paddr;
|
|
return_section->size = hdr->s_size;
|
|
return_section->filepos = hdr->s_scnptr;
|
|
return_section->rel_filepos = hdr->s_relptr;
|
|
return_section->reloc_count = hdr->s_nreloc;
|
|
|
|
bfd_coff_set_alignment_hook (abfd, return_section, hdr);
|
|
|
|
return_section->line_filepos = hdr->s_lnnoptr;
|
|
|
|
return_section->lineno_count = hdr->s_nlnno;
|
|
return_section->userdata = NULL;
|
|
return_section->next = NULL;
|
|
return_section->target_index = target_index;
|
|
|
|
if (! bfd_coff_styp_to_sec_flags_hook (abfd, hdr, name, return_section,
|
|
& flags))
|
|
result = FALSE;
|
|
|
|
return_section->flags = flags;
|
|
|
|
/* At least on i386-coff, the line number count for a shared library
|
|
section must be ignored. */
|
|
if ((return_section->flags & SEC_COFF_SHARED_LIBRARY) != 0)
|
|
return_section->lineno_count = 0;
|
|
|
|
if (hdr->s_nreloc != 0)
|
|
return_section->flags |= SEC_RELOC;
|
|
/* FIXME: should this check 'hdr->s_size > 0'. */
|
|
if (hdr->s_scnptr != 0)
|
|
return_section->flags |= SEC_HAS_CONTENTS;
|
|
|
|
/* Compress/decompress DWARF debug sections with names: .debug_* and
|
|
.zdebug_*, after the section flags is set. */
|
|
if ((flags & SEC_DEBUGGING)
|
|
&& strlen (name) > 7
|
|
&& ((name[1] == 'd' && name[6] == '_')
|
|
|| (strlen (name) > 8 && name[1] == 'z' && name[7] == '_')))
|
|
{
|
|
enum { nothing, compress, decompress } action = nothing;
|
|
char *new_name = NULL;
|
|
|
|
if (bfd_is_section_compressed (abfd, return_section))
|
|
{
|
|
/* Compressed section. Check if we should decompress. */
|
|
if ((abfd->flags & BFD_DECOMPRESS))
|
|
action = decompress;
|
|
}
|
|
else if (!bfd_is_section_compressed (abfd, return_section))
|
|
{
|
|
/* Normal section. Check if we should compress. */
|
|
if ((abfd->flags & BFD_COMPRESS) && return_section->size != 0)
|
|
action = compress;
|
|
}
|
|
|
|
switch (action)
|
|
{
|
|
case nothing:
|
|
break;
|
|
case compress:
|
|
if (!bfd_init_section_compress_status (abfd, return_section))
|
|
{
|
|
(*_bfd_error_handler)
|
|
(_("%B: unable to initialize compress status for section %s"),
|
|
abfd, name);
|
|
return FALSE;
|
|
}
|
|
if (return_section->compress_status == COMPRESS_SECTION_DONE)
|
|
{
|
|
if (name[1] != 'z')
|
|
{
|
|
unsigned int len = strlen (name);
|
|
|
|
new_name = bfd_alloc (abfd, len + 2);
|
|
if (new_name == NULL)
|
|
return FALSE;
|
|
new_name[0] = '.';
|
|
new_name[1] = 'z';
|
|
memcpy (new_name + 2, name + 1, len);
|
|
}
|
|
}
|
|
break;
|
|
case decompress:
|
|
if (!bfd_init_section_decompress_status (abfd, return_section))
|
|
{
|
|
(*_bfd_error_handler)
|
|
(_("%B: unable to initialize decompress status for section %s"),
|
|
abfd, name);
|
|
return FALSE;
|
|
}
|
|
if (name[1] == 'z')
|
|
{
|
|
unsigned int len = strlen (name);
|
|
|
|
new_name = bfd_alloc (abfd, len);
|
|
if (new_name == NULL)
|
|
return FALSE;
|
|
new_name[0] = '.';
|
|
memcpy (new_name + 1, name + 2, len - 1);
|
|
}
|
|
break;
|
|
}
|
|
if (new_name != NULL)
|
|
bfd_rename_section (abfd, return_section, new_name);
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
/* Read in a COFF object and make it into a BFD. This is used by
|
|
ECOFF as well. */
|
|
const bfd_target *
|
|
coff_real_object_p (bfd *,
|
|
unsigned,
|
|
struct internal_filehdr *,
|
|
struct internal_aouthdr *);
|
|
const bfd_target *
|
|
coff_real_object_p (bfd *abfd,
|
|
unsigned nscns,
|
|
struct internal_filehdr *internal_f,
|
|
struct internal_aouthdr *internal_a)
|
|
{
|
|
flagword oflags = abfd->flags;
|
|
bfd_vma ostart = bfd_get_start_address (abfd);
|
|
void * tdata;
|
|
void * tdata_save;
|
|
bfd_size_type readsize; /* Length of file_info. */
|
|
unsigned int scnhsz;
|
|
char *external_sections;
|
|
|
|
if (!(internal_f->f_flags & F_RELFLG))
|
|
abfd->flags |= HAS_RELOC;
|
|
if ((internal_f->f_flags & F_EXEC))
|
|
abfd->flags |= EXEC_P;
|
|
if (!(internal_f->f_flags & F_LNNO))
|
|
abfd->flags |= HAS_LINENO;
|
|
if (!(internal_f->f_flags & F_LSYMS))
|
|
abfd->flags |= HAS_LOCALS;
|
|
|
|
/* FIXME: How can we set D_PAGED correctly? */
|
|
if ((internal_f->f_flags & F_EXEC) != 0)
|
|
abfd->flags |= D_PAGED;
|
|
|
|
bfd_get_symcount (abfd) = internal_f->f_nsyms;
|
|
if (internal_f->f_nsyms)
|
|
abfd->flags |= HAS_SYMS;
|
|
|
|
if (internal_a != (struct internal_aouthdr *) NULL)
|
|
bfd_get_start_address (abfd) = internal_a->entry;
|
|
else
|
|
bfd_get_start_address (abfd) = 0;
|
|
|
|
/* Set up the tdata area. ECOFF uses its own routine, and overrides
|
|
abfd->flags. */
|
|
tdata_save = abfd->tdata.any;
|
|
tdata = bfd_coff_mkobject_hook (abfd, (void *) internal_f, (void *) internal_a);
|
|
if (tdata == NULL)
|
|
goto fail2;
|
|
|
|
scnhsz = bfd_coff_scnhsz (abfd);
|
|
readsize = (bfd_size_type) nscns * scnhsz;
|
|
external_sections = (char *) bfd_alloc (abfd, readsize);
|
|
if (!external_sections)
|
|
goto fail;
|
|
|
|
if (bfd_bread ((void *) external_sections, readsize, abfd) != readsize)
|
|
goto fail;
|
|
|
|
/* Set the arch/mach *before* swapping in sections; section header swapping
|
|
may depend on arch/mach info. */
|
|
if (! bfd_coff_set_arch_mach_hook (abfd, (void *) internal_f))
|
|
goto fail;
|
|
|
|
/* Now copy data as required; construct all asections etc. */
|
|
if (nscns != 0)
|
|
{
|
|
unsigned int i;
|
|
for (i = 0; i < nscns; i++)
|
|
{
|
|
struct internal_scnhdr tmp;
|
|
bfd_coff_swap_scnhdr_in (abfd,
|
|
(void *) (external_sections + i * scnhsz),
|
|
(void *) & tmp);
|
|
if (! make_a_section_from_file (abfd, &tmp, i + 1))
|
|
goto fail;
|
|
}
|
|
}
|
|
|
|
return abfd->xvec;
|
|
|
|
fail:
|
|
bfd_release (abfd, tdata);
|
|
fail2:
|
|
abfd->tdata.any = tdata_save;
|
|
abfd->flags = oflags;
|
|
bfd_get_start_address (abfd) = ostart;
|
|
return (const bfd_target *) NULL;
|
|
}
|
|
|
|
/* Turn a COFF file into a BFD, but fail with bfd_error_wrong_format if it is
|
|
not a COFF file. This is also used by ECOFF. */
|
|
|
|
const bfd_target *
|
|
coff_object_p (bfd *abfd)
|
|
{
|
|
bfd_size_type filhsz;
|
|
bfd_size_type aoutsz;
|
|
unsigned int nscns;
|
|
void * filehdr;
|
|
struct internal_filehdr internal_f;
|
|
struct internal_aouthdr internal_a;
|
|
|
|
/* Figure out how much to read. */
|
|
filhsz = bfd_coff_filhsz (abfd);
|
|
aoutsz = bfd_coff_aoutsz (abfd);
|
|
|
|
filehdr = bfd_alloc (abfd, filhsz);
|
|
if (filehdr == NULL)
|
|
return NULL;
|
|
if (bfd_bread (filehdr, filhsz, abfd) != filhsz)
|
|
{
|
|
if (bfd_get_error () != bfd_error_system_call)
|
|
bfd_set_error (bfd_error_wrong_format);
|
|
bfd_release (abfd, filehdr);
|
|
return NULL;
|
|
}
|
|
bfd_coff_swap_filehdr_in (abfd, filehdr, &internal_f);
|
|
bfd_release (abfd, filehdr);
|
|
|
|
/* The XCOFF format has two sizes for the f_opthdr. SMALL_AOUTSZ
|
|
(less than aoutsz) used in object files and AOUTSZ (equal to
|
|
aoutsz) in executables. The bfd_coff_swap_aouthdr_in function
|
|
expects this header to be aoutsz bytes in length, so we use that
|
|
value in the call to bfd_alloc below. But we must be careful to
|
|
only read in f_opthdr bytes in the call to bfd_bread. We should
|
|
also attempt to catch corrupt or non-COFF binaries with a strange
|
|
value for f_opthdr. */
|
|
if (! bfd_coff_bad_format_hook (abfd, &internal_f)
|
|
|| internal_f.f_opthdr > aoutsz)
|
|
{
|
|
bfd_set_error (bfd_error_wrong_format);
|
|
return NULL;
|
|
}
|
|
nscns = internal_f.f_nscns;
|
|
|
|
if (internal_f.f_opthdr)
|
|
{
|
|
void * opthdr;
|
|
|
|
opthdr = bfd_alloc (abfd, aoutsz);
|
|
if (opthdr == NULL)
|
|
return NULL;
|
|
if (bfd_bread (opthdr, (bfd_size_type) internal_f.f_opthdr, abfd)
|
|
!= internal_f.f_opthdr)
|
|
{
|
|
bfd_release (abfd, opthdr);
|
|
return NULL;
|
|
}
|
|
/* PR 17512: file: 11056-1136-0.004. */
|
|
if (internal_f.f_opthdr < aoutsz)
|
|
memset (((char *) opthdr) + internal_f.f_opthdr, 0, aoutsz - internal_f.f_opthdr);
|
|
|
|
bfd_coff_swap_aouthdr_in (abfd, opthdr, (void *) &internal_a);
|
|
bfd_release (abfd, opthdr);
|
|
}
|
|
|
|
return coff_real_object_p (abfd, nscns, &internal_f,
|
|
(internal_f.f_opthdr != 0
|
|
? &internal_a
|
|
: (struct internal_aouthdr *) NULL));
|
|
}
|
|
|
|
/* Get the BFD section from a COFF symbol section number. */
|
|
|
|
asection *
|
|
coff_section_from_bfd_index (bfd *abfd, int section_index)
|
|
{
|
|
struct bfd_section *answer = abfd->sections;
|
|
|
|
if (section_index == N_ABS)
|
|
return bfd_abs_section_ptr;
|
|
if (section_index == N_UNDEF)
|
|
return bfd_und_section_ptr;
|
|
if (section_index == N_DEBUG)
|
|
return bfd_abs_section_ptr;
|
|
|
|
while (answer)
|
|
{
|
|
if (answer->target_index == section_index)
|
|
return answer;
|
|
answer = answer->next;
|
|
}
|
|
|
|
/* We should not reach this point, but the SCO 3.2v4 /lib/libc_s.a
|
|
has a bad symbol table in biglitpow.o. */
|
|
return bfd_und_section_ptr;
|
|
}
|
|
|
|
/* Get the upper bound of a COFF symbol table. */
|
|
|
|
long
|
|
coff_get_symtab_upper_bound (bfd *abfd)
|
|
{
|
|
if (!bfd_coff_slurp_symbol_table (abfd))
|
|
return -1;
|
|
|
|
return (bfd_get_symcount (abfd) + 1) * (sizeof (coff_symbol_type *));
|
|
}
|
|
|
|
/* Canonicalize a COFF symbol table. */
|
|
|
|
long
|
|
coff_canonicalize_symtab (bfd *abfd, asymbol **alocation)
|
|
{
|
|
unsigned int counter;
|
|
coff_symbol_type *symbase;
|
|
coff_symbol_type **location = (coff_symbol_type **) alocation;
|
|
|
|
if (!bfd_coff_slurp_symbol_table (abfd))
|
|
return -1;
|
|
|
|
symbase = obj_symbols (abfd);
|
|
counter = bfd_get_symcount (abfd);
|
|
while (counter-- > 0)
|
|
*location++ = symbase++;
|
|
|
|
*location = NULL;
|
|
|
|
return bfd_get_symcount (abfd);
|
|
}
|
|
|
|
/* Get the name of a symbol. The caller must pass in a buffer of size
|
|
>= SYMNMLEN + 1. */
|
|
|
|
const char *
|
|
_bfd_coff_internal_syment_name (bfd *abfd,
|
|
const struct internal_syment *sym,
|
|
char *buf)
|
|
{
|
|
/* FIXME: It's not clear this will work correctly if sizeof
|
|
(_n_zeroes) != 4. */
|
|
if (sym->_n._n_n._n_zeroes != 0
|
|
|| sym->_n._n_n._n_offset == 0)
|
|
{
|
|
memcpy (buf, sym->_n._n_name, SYMNMLEN);
|
|
buf[SYMNMLEN] = '\0';
|
|
return buf;
|
|
}
|
|
else
|
|
{
|
|
const char *strings;
|
|
|
|
BFD_ASSERT (sym->_n._n_n._n_offset >= STRING_SIZE_SIZE);
|
|
strings = obj_coff_strings (abfd);
|
|
if (strings == NULL)
|
|
{
|
|
strings = _bfd_coff_read_string_table (abfd);
|
|
if (strings == NULL)
|
|
return NULL;
|
|
}
|
|
/* PR 17910: Only check for string overflow if the length has been set.
|
|
Some DLLs, eg those produced by Visual Studio, may not set the length field. */
|
|
if (obj_coff_strings_len (abfd) > 0
|
|
&& sym->_n._n_n._n_offset >= obj_coff_strings_len (abfd))
|
|
return NULL;
|
|
return strings + sym->_n._n_n._n_offset;
|
|
}
|
|
}
|
|
|
|
/* Read in and swap the relocs. This returns a buffer holding the
|
|
relocs for section SEC in file ABFD. If CACHE is TRUE and
|
|
INTERNAL_RELOCS is NULL, the relocs read in will be saved in case
|
|
the function is called again. If EXTERNAL_RELOCS is not NULL, it
|
|
is a buffer large enough to hold the unswapped relocs. If
|
|
INTERNAL_RELOCS is not NULL, it is a buffer large enough to hold
|
|
the swapped relocs. If REQUIRE_INTERNAL is TRUE, then the return
|
|
value must be INTERNAL_RELOCS. The function returns NULL on error. */
|
|
|
|
struct internal_reloc *
|
|
_bfd_coff_read_internal_relocs (bfd *abfd,
|
|
asection *sec,
|
|
bfd_boolean cache,
|
|
bfd_byte *external_relocs,
|
|
bfd_boolean require_internal,
|
|
struct internal_reloc *internal_relocs)
|
|
{
|
|
bfd_size_type relsz;
|
|
bfd_byte *free_external = NULL;
|
|
struct internal_reloc *free_internal = NULL;
|
|
bfd_byte *erel;
|
|
bfd_byte *erel_end;
|
|
struct internal_reloc *irel;
|
|
bfd_size_type amt;
|
|
|
|
if (sec->reloc_count == 0)
|
|
return internal_relocs; /* Nothing to do. */
|
|
|
|
if (coff_section_data (abfd, sec) != NULL
|
|
&& coff_section_data (abfd, sec)->relocs != NULL)
|
|
{
|
|
if (! require_internal)
|
|
return coff_section_data (abfd, sec)->relocs;
|
|
memcpy (internal_relocs, coff_section_data (abfd, sec)->relocs,
|
|
sec->reloc_count * sizeof (struct internal_reloc));
|
|
return internal_relocs;
|
|
}
|
|
|
|
relsz = bfd_coff_relsz (abfd);
|
|
|
|
amt = sec->reloc_count * relsz;
|
|
if (external_relocs == NULL)
|
|
{
|
|
free_external = (bfd_byte *) bfd_malloc (amt);
|
|
if (free_external == NULL)
|
|
goto error_return;
|
|
external_relocs = free_external;
|
|
}
|
|
|
|
if (bfd_seek (abfd, sec->rel_filepos, SEEK_SET) != 0
|
|
|| bfd_bread (external_relocs, amt, abfd) != amt)
|
|
goto error_return;
|
|
|
|
if (internal_relocs == NULL)
|
|
{
|
|
amt = sec->reloc_count;
|
|
amt *= sizeof (struct internal_reloc);
|
|
free_internal = (struct internal_reloc *) bfd_malloc (amt);
|
|
if (free_internal == NULL)
|
|
goto error_return;
|
|
internal_relocs = free_internal;
|
|
}
|
|
|
|
/* Swap in the relocs. */
|
|
erel = external_relocs;
|
|
erel_end = erel + relsz * sec->reloc_count;
|
|
irel = internal_relocs;
|
|
for (; erel < erel_end; erel += relsz, irel++)
|
|
bfd_coff_swap_reloc_in (abfd, (void *) erel, (void *) irel);
|
|
|
|
if (free_external != NULL)
|
|
{
|
|
free (free_external);
|
|
free_external = NULL;
|
|
}
|
|
|
|
if (cache && free_internal != NULL)
|
|
{
|
|
if (coff_section_data (abfd, sec) == NULL)
|
|
{
|
|
amt = sizeof (struct coff_section_tdata);
|
|
sec->used_by_bfd = bfd_zalloc (abfd, amt);
|
|
if (sec->used_by_bfd == NULL)
|
|
goto error_return;
|
|
coff_section_data (abfd, sec)->contents = NULL;
|
|
}
|
|
coff_section_data (abfd, sec)->relocs = free_internal;
|
|
}
|
|
|
|
return internal_relocs;
|
|
|
|
error_return:
|
|
if (free_external != NULL)
|
|
free (free_external);
|
|
if (free_internal != NULL)
|
|
free (free_internal);
|
|
return NULL;
|
|
}
|
|
|
|
/* Set lineno_count for the output sections of a COFF file. */
|
|
|
|
int
|
|
coff_count_linenumbers (bfd *abfd)
|
|
{
|
|
unsigned int limit = bfd_get_symcount (abfd);
|
|
unsigned int i;
|
|
int total = 0;
|
|
asymbol **p;
|
|
asection *s;
|
|
|
|
if (limit == 0)
|
|
{
|
|
/* This may be from the backend linker, in which case the
|
|
lineno_count in the sections is correct. */
|
|
for (s = abfd->sections; s != NULL; s = s->next)
|
|
total += s->lineno_count;
|
|
return total;
|
|
}
|
|
|
|
for (s = abfd->sections; s != NULL; s = s->next)
|
|
BFD_ASSERT (s->lineno_count == 0);
|
|
|
|
for (p = abfd->outsymbols, i = 0; i < limit; i++, p++)
|
|
{
|
|
asymbol *q_maybe = *p;
|
|
|
|
if (bfd_family_coff (bfd_asymbol_bfd (q_maybe)))
|
|
{
|
|
coff_symbol_type *q = coffsymbol (q_maybe);
|
|
|
|
/* The AIX 4.1 compiler can sometimes generate line numbers
|
|
attached to debugging symbols. We try to simply ignore
|
|
those here. */
|
|
if (q->lineno != NULL
|
|
&& q->symbol.section->owner != NULL)
|
|
{
|
|
/* This symbol has line numbers. Increment the owning
|
|
section's linenumber count. */
|
|
alent *l = q->lineno;
|
|
|
|
do
|
|
{
|
|
asection * sec = q->symbol.section->output_section;
|
|
|
|
/* Do not try to update fields in read-only sections. */
|
|
if (! bfd_is_const_section (sec))
|
|
sec->lineno_count ++;
|
|
|
|
++total;
|
|
++l;
|
|
}
|
|
while (l->line_number != 0);
|
|
}
|
|
}
|
|
}
|
|
|
|
return total;
|
|
}
|
|
|
|
static void
|
|
fixup_symbol_value (bfd *abfd,
|
|
coff_symbol_type *coff_symbol_ptr,
|
|
struct internal_syment *syment)
|
|
{
|
|
/* Normalize the symbol flags. */
|
|
if (coff_symbol_ptr->symbol.section
|
|
&& bfd_is_com_section (coff_symbol_ptr->symbol.section))
|
|
{
|
|
/* A common symbol is undefined with a value. */
|
|
syment->n_scnum = N_UNDEF;
|
|
syment->n_value = coff_symbol_ptr->symbol.value;
|
|
}
|
|
else if ((coff_symbol_ptr->symbol.flags & BSF_DEBUGGING) != 0
|
|
&& (coff_symbol_ptr->symbol.flags & BSF_DEBUGGING_RELOC) == 0)
|
|
{
|
|
syment->n_value = coff_symbol_ptr->symbol.value;
|
|
}
|
|
else if (bfd_is_und_section (coff_symbol_ptr->symbol.section))
|
|
{
|
|
syment->n_scnum = N_UNDEF;
|
|
syment->n_value = 0;
|
|
}
|
|
/* FIXME: Do we need to handle the absolute section here? */
|
|
else
|
|
{
|
|
if (coff_symbol_ptr->symbol.section)
|
|
{
|
|
syment->n_scnum =
|
|
coff_symbol_ptr->symbol.section->output_section->target_index;
|
|
|
|
syment->n_value = (coff_symbol_ptr->symbol.value
|
|
+ coff_symbol_ptr->symbol.section->output_offset);
|
|
if (! obj_pe (abfd))
|
|
{
|
|
syment->n_value += (syment->n_sclass == C_STATLAB)
|
|
? coff_symbol_ptr->symbol.section->output_section->lma
|
|
: coff_symbol_ptr->symbol.section->output_section->vma;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
BFD_ASSERT (0);
|
|
/* This can happen, but I don't know why yet (steve@cygnus.com) */
|
|
syment->n_scnum = N_ABS;
|
|
syment->n_value = coff_symbol_ptr->symbol.value;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Run through all the symbols in the symbol table and work out what
|
|
their indexes into the symbol table will be when output.
|
|
|
|
Coff requires that each C_FILE symbol points to the next one in the
|
|
chain, and that the last one points to the first external symbol. We
|
|
do that here too. */
|
|
|
|
bfd_boolean
|
|
coff_renumber_symbols (bfd *bfd_ptr, int *first_undef)
|
|
{
|
|
unsigned int symbol_count = bfd_get_symcount (bfd_ptr);
|
|
asymbol **symbol_ptr_ptr = bfd_ptr->outsymbols;
|
|
unsigned int native_index = 0;
|
|
struct internal_syment *last_file = NULL;
|
|
unsigned int symbol_index;
|
|
|
|
/* COFF demands that undefined symbols come after all other symbols.
|
|
Since we don't need to impose this extra knowledge on all our
|
|
client programs, deal with that here. Sort the symbol table;
|
|
just move the undefined symbols to the end, leaving the rest
|
|
alone. The O'Reilly book says that defined global symbols come
|
|
at the end before the undefined symbols, so we do that here as
|
|
well. */
|
|
/* @@ Do we have some condition we could test for, so we don't always
|
|
have to do this? I don't think relocatability is quite right, but
|
|
I'm not certain. [raeburn:19920508.1711EST] */
|
|
{
|
|
asymbol **newsyms;
|
|
unsigned int i;
|
|
bfd_size_type amt;
|
|
|
|
amt = sizeof (asymbol *) * ((bfd_size_type) symbol_count + 1);
|
|
newsyms = (asymbol **) bfd_alloc (bfd_ptr, amt);
|
|
if (!newsyms)
|
|
return FALSE;
|
|
bfd_ptr->outsymbols = newsyms;
|
|
for (i = 0; i < symbol_count; i++)
|
|
if ((symbol_ptr_ptr[i]->flags & BSF_NOT_AT_END) != 0
|
|
|| (!bfd_is_und_section (symbol_ptr_ptr[i]->section)
|
|
&& !bfd_is_com_section (symbol_ptr_ptr[i]->section)
|
|
&& ((symbol_ptr_ptr[i]->flags & BSF_FUNCTION) != 0
|
|
|| ((symbol_ptr_ptr[i]->flags & (BSF_GLOBAL | BSF_WEAK))
|
|
== 0))))
|
|
*newsyms++ = symbol_ptr_ptr[i];
|
|
|
|
for (i = 0; i < symbol_count; i++)
|
|
if ((symbol_ptr_ptr[i]->flags & BSF_NOT_AT_END) == 0
|
|
&& !bfd_is_und_section (symbol_ptr_ptr[i]->section)
|
|
&& (bfd_is_com_section (symbol_ptr_ptr[i]->section)
|
|
|| ((symbol_ptr_ptr[i]->flags & BSF_FUNCTION) == 0
|
|
&& ((symbol_ptr_ptr[i]->flags & (BSF_GLOBAL | BSF_WEAK))
|
|
!= 0))))
|
|
*newsyms++ = symbol_ptr_ptr[i];
|
|
|
|
*first_undef = newsyms - bfd_ptr->outsymbols;
|
|
|
|
for (i = 0; i < symbol_count; i++)
|
|
if ((symbol_ptr_ptr[i]->flags & BSF_NOT_AT_END) == 0
|
|
&& bfd_is_und_section (symbol_ptr_ptr[i]->section))
|
|
*newsyms++ = symbol_ptr_ptr[i];
|
|
*newsyms = (asymbol *) NULL;
|
|
symbol_ptr_ptr = bfd_ptr->outsymbols;
|
|
}
|
|
|
|
for (symbol_index = 0; symbol_index < symbol_count; symbol_index++)
|
|
{
|
|
coff_symbol_type *coff_symbol_ptr;
|
|
|
|
coff_symbol_ptr = coff_symbol_from (symbol_ptr_ptr[symbol_index]);
|
|
symbol_ptr_ptr[symbol_index]->udata.i = symbol_index;
|
|
if (coff_symbol_ptr && coff_symbol_ptr->native)
|
|
{
|
|
combined_entry_type *s = coff_symbol_ptr->native;
|
|
int i;
|
|
|
|
BFD_ASSERT (s->is_sym);
|
|
if (s->u.syment.n_sclass == C_FILE)
|
|
{
|
|
if (last_file != NULL)
|
|
last_file->n_value = native_index;
|
|
last_file = &(s->u.syment);
|
|
}
|
|
else
|
|
/* Modify the symbol values according to their section and
|
|
type. */
|
|
fixup_symbol_value (bfd_ptr, coff_symbol_ptr, &(s->u.syment));
|
|
|
|
for (i = 0; i < s->u.syment.n_numaux + 1; i++)
|
|
s[i].offset = native_index++;
|
|
}
|
|
else
|
|
native_index++;
|
|
}
|
|
|
|
obj_conv_table_size (bfd_ptr) = native_index;
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
/* Run thorough the symbol table again, and fix it so that all
|
|
pointers to entries are changed to the entries' index in the output
|
|
symbol table. */
|
|
|
|
void
|
|
coff_mangle_symbols (bfd *bfd_ptr)
|
|
{
|
|
unsigned int symbol_count = bfd_get_symcount (bfd_ptr);
|
|
asymbol **symbol_ptr_ptr = bfd_ptr->outsymbols;
|
|
unsigned int symbol_index;
|
|
|
|
for (symbol_index = 0; symbol_index < symbol_count; symbol_index++)
|
|
{
|
|
coff_symbol_type *coff_symbol_ptr;
|
|
|
|
coff_symbol_ptr = coff_symbol_from (symbol_ptr_ptr[symbol_index]);
|
|
if (coff_symbol_ptr && coff_symbol_ptr->native)
|
|
{
|
|
int i;
|
|
combined_entry_type *s = coff_symbol_ptr->native;
|
|
|
|
BFD_ASSERT (s->is_sym);
|
|
if (s->fix_value)
|
|
{
|
|
/* FIXME: We should use a union here. */
|
|
s->u.syment.n_value =
|
|
(bfd_hostptr_t) ((combined_entry_type *)
|
|
((bfd_hostptr_t) s->u.syment.n_value))->offset;
|
|
s->fix_value = 0;
|
|
}
|
|
if (s->fix_line)
|
|
{
|
|
/* The value is the offset into the line number entries
|
|
for the symbol's section. On output, the symbol's
|
|
section should be N_DEBUG. */
|
|
s->u.syment.n_value =
|
|
(coff_symbol_ptr->symbol.section->output_section->line_filepos
|
|
+ s->u.syment.n_value * bfd_coff_linesz (bfd_ptr));
|
|
coff_symbol_ptr->symbol.section =
|
|
coff_section_from_bfd_index (bfd_ptr, N_DEBUG);
|
|
BFD_ASSERT (coff_symbol_ptr->symbol.flags & BSF_DEBUGGING);
|
|
}
|
|
for (i = 0; i < s->u.syment.n_numaux; i++)
|
|
{
|
|
combined_entry_type *a = s + i + 1;
|
|
|
|
BFD_ASSERT (! a->is_sym);
|
|
if (a->fix_tag)
|
|
{
|
|
a->u.auxent.x_sym.x_tagndx.l =
|
|
a->u.auxent.x_sym.x_tagndx.p->offset;
|
|
a->fix_tag = 0;
|
|
}
|
|
if (a->fix_end)
|
|
{
|
|
a->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.l =
|
|
a->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.p->offset;
|
|
a->fix_end = 0;
|
|
}
|
|
if (a->fix_scnlen)
|
|
{
|
|
a->u.auxent.x_csect.x_scnlen.l =
|
|
a->u.auxent.x_csect.x_scnlen.p->offset;
|
|
a->fix_scnlen = 0;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
static void
|
|
coff_fix_symbol_name (bfd *abfd,
|
|
asymbol *symbol,
|
|
combined_entry_type *native,
|
|
bfd_size_type *string_size_p,
|
|
asection **debug_string_section_p,
|
|
bfd_size_type *debug_string_size_p)
|
|
{
|
|
unsigned int name_length;
|
|
union internal_auxent *auxent;
|
|
char *name = (char *) (symbol->name);
|
|
|
|
if (name == NULL)
|
|
{
|
|
/* COFF symbols always have names, so we'll make one up. */
|
|
symbol->name = "strange";
|
|
name = (char *) symbol->name;
|
|
}
|
|
name_length = strlen (name);
|
|
|
|
BFD_ASSERT (native->is_sym);
|
|
if (native->u.syment.n_sclass == C_FILE
|
|
&& native->u.syment.n_numaux > 0)
|
|
{
|
|
unsigned int filnmlen;
|
|
|
|
if (bfd_coff_force_symnames_in_strings (abfd))
|
|
{
|
|
native->u.syment._n._n_n._n_offset =
|
|
(*string_size_p + STRING_SIZE_SIZE);
|
|
native->u.syment._n._n_n._n_zeroes = 0;
|
|
*string_size_p += 6; /* strlen(".file") + 1 */
|
|
}
|
|
else
|
|
strncpy (native->u.syment._n._n_name, ".file", SYMNMLEN);
|
|
|
|
BFD_ASSERT (! (native + 1)->is_sym);
|
|
auxent = &(native + 1)->u.auxent;
|
|
|
|
filnmlen = bfd_coff_filnmlen (abfd);
|
|
|
|
if (bfd_coff_long_filenames (abfd))
|
|
{
|
|
if (name_length <= filnmlen)
|
|
strncpy (auxent->x_file.x_fname, name, filnmlen);
|
|
else
|
|
{
|
|
auxent->x_file.x_n.x_offset = *string_size_p + STRING_SIZE_SIZE;
|
|
auxent->x_file.x_n.x_zeroes = 0;
|
|
*string_size_p += name_length + 1;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
strncpy (auxent->x_file.x_fname, name, filnmlen);
|
|
if (name_length > filnmlen)
|
|
name[filnmlen] = '\0';
|
|
}
|
|
}
|
|
else
|
|
{
|
|
if (name_length <= SYMNMLEN && !bfd_coff_force_symnames_in_strings (abfd))
|
|
/* This name will fit into the symbol neatly. */
|
|
strncpy (native->u.syment._n._n_name, symbol->name, SYMNMLEN);
|
|
|
|
else if (!bfd_coff_symname_in_debug (abfd, &native->u.syment))
|
|
{
|
|
native->u.syment._n._n_n._n_offset = (*string_size_p
|
|
+ STRING_SIZE_SIZE);
|
|
native->u.syment._n._n_n._n_zeroes = 0;
|
|
*string_size_p += name_length + 1;
|
|
}
|
|
else
|
|
{
|
|
file_ptr filepos;
|
|
bfd_byte buf[4];
|
|
int prefix_len = bfd_coff_debug_string_prefix_length (abfd);
|
|
|
|
/* This name should be written into the .debug section. For
|
|
some reason each name is preceded by a two byte length
|
|
and also followed by a null byte. FIXME: We assume that
|
|
the .debug section has already been created, and that it
|
|
is large enough. */
|
|
if (*debug_string_section_p == (asection *) NULL)
|
|
*debug_string_section_p = bfd_get_section_by_name (abfd, ".debug");
|
|
filepos = bfd_tell (abfd);
|
|
if (prefix_len == 4)
|
|
bfd_put_32 (abfd, (bfd_vma) (name_length + 1), buf);
|
|
else
|
|
bfd_put_16 (abfd, (bfd_vma) (name_length + 1), buf);
|
|
|
|
if (!bfd_set_section_contents (abfd,
|
|
*debug_string_section_p,
|
|
(void *) buf,
|
|
(file_ptr) *debug_string_size_p,
|
|
(bfd_size_type) prefix_len)
|
|
|| !bfd_set_section_contents (abfd,
|
|
*debug_string_section_p,
|
|
(void *) symbol->name,
|
|
(file_ptr) (*debug_string_size_p
|
|
+ prefix_len),
|
|
(bfd_size_type) name_length + 1))
|
|
abort ();
|
|
if (bfd_seek (abfd, filepos, SEEK_SET) != 0)
|
|
abort ();
|
|
native->u.syment._n._n_n._n_offset =
|
|
*debug_string_size_p + prefix_len;
|
|
native->u.syment._n._n_n._n_zeroes = 0;
|
|
*debug_string_size_p += name_length + 1 + prefix_len;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* We need to keep track of the symbol index so that when we write out
|
|
the relocs we can get the index for a symbol. This method is a
|
|
hack. FIXME. */
|
|
|
|
#define set_index(symbol, idx) ((symbol)->udata.i = (idx))
|
|
|
|
/* Write a symbol out to a COFF file. */
|
|
|
|
static bfd_boolean
|
|
coff_write_symbol (bfd *abfd,
|
|
asymbol *symbol,
|
|
combined_entry_type *native,
|
|
bfd_vma *written,
|
|
bfd_size_type *string_size_p,
|
|
asection **debug_string_section_p,
|
|
bfd_size_type *debug_string_size_p)
|
|
{
|
|
unsigned int numaux = native->u.syment.n_numaux;
|
|
int type = native->u.syment.n_type;
|
|
int n_sclass = (int) native->u.syment.n_sclass;
|
|
asection *output_section = symbol->section->output_section
|
|
? symbol->section->output_section
|
|
: symbol->section;
|
|
void * buf;
|
|
bfd_size_type symesz;
|
|
|
|
BFD_ASSERT (native->is_sym);
|
|
|
|
if (native->u.syment.n_sclass == C_FILE)
|
|
symbol->flags |= BSF_DEBUGGING;
|
|
|
|
if (symbol->flags & BSF_DEBUGGING
|
|
&& bfd_is_abs_section (symbol->section))
|
|
native->u.syment.n_scnum = N_DEBUG;
|
|
|
|
else if (bfd_is_abs_section (symbol->section))
|
|
native->u.syment.n_scnum = N_ABS;
|
|
|
|
else if (bfd_is_und_section (symbol->section))
|
|
native->u.syment.n_scnum = N_UNDEF;
|
|
|
|
else
|
|
native->u.syment.n_scnum =
|
|
output_section->target_index;
|
|
|
|
coff_fix_symbol_name (abfd, symbol, native, string_size_p,
|
|
debug_string_section_p, debug_string_size_p);
|
|
|
|
symesz = bfd_coff_symesz (abfd);
|
|
buf = bfd_alloc (abfd, symesz);
|
|
if (!buf)
|
|
return FALSE;
|
|
bfd_coff_swap_sym_out (abfd, &native->u.syment, buf);
|
|
if (bfd_bwrite (buf, symesz, abfd) != symesz)
|
|
return FALSE;
|
|
bfd_release (abfd, buf);
|
|
|
|
if (native->u.syment.n_numaux > 0)
|
|
{
|
|
bfd_size_type auxesz;
|
|
unsigned int j;
|
|
|
|
auxesz = bfd_coff_auxesz (abfd);
|
|
buf = bfd_alloc (abfd, auxesz);
|
|
if (!buf)
|
|
return FALSE;
|
|
for (j = 0; j < native->u.syment.n_numaux; j++)
|
|
{
|
|
BFD_ASSERT (! (native + j + 1)->is_sym);
|
|
bfd_coff_swap_aux_out (abfd,
|
|
&((native + j + 1)->u.auxent),
|
|
type, n_sclass, (int) j,
|
|
native->u.syment.n_numaux,
|
|
buf);
|
|
if (bfd_bwrite (buf, auxesz, abfd) != auxesz)
|
|
return FALSE;
|
|
}
|
|
bfd_release (abfd, buf);
|
|
}
|
|
|
|
/* Store the index for use when we write out the relocs. */
|
|
set_index (symbol, *written);
|
|
|
|
*written += numaux + 1;
|
|
return TRUE;
|
|
}
|
|
|
|
/* Write out a symbol to a COFF file that does not come from a COFF
|
|
file originally. This symbol may have been created by the linker,
|
|
or we may be linking a non COFF file to a COFF file. */
|
|
|
|
bfd_boolean
|
|
coff_write_alien_symbol (bfd *abfd,
|
|
asymbol *symbol,
|
|
struct internal_syment *isym,
|
|
union internal_auxent *iaux,
|
|
bfd_vma *written,
|
|
bfd_size_type *string_size_p,
|
|
asection **debug_string_section_p,
|
|
bfd_size_type *debug_string_size_p)
|
|
{
|
|
combined_entry_type *native;
|
|
combined_entry_type dummy[2];
|
|
asection *output_section = symbol->section->output_section
|
|
? symbol->section->output_section
|
|
: symbol->section;
|
|
struct bfd_link_info *link_info = coff_data (abfd)->link_info;
|
|
bfd_boolean ret;
|
|
|
|
if ((!link_info || link_info->strip_discarded)
|
|
&& !bfd_is_abs_section (symbol->section)
|
|
&& symbol->section->output_section == bfd_abs_section_ptr)
|
|
{
|
|
symbol->name = "";
|
|
if (isym != NULL)
|
|
memset (isym, 0, sizeof (*isym));
|
|
return TRUE;
|
|
}
|
|
native = dummy;
|
|
native->is_sym = TRUE;
|
|
native[1].is_sym = FALSE;
|
|
native->u.syment.n_type = T_NULL;
|
|
native->u.syment.n_flags = 0;
|
|
native->u.syment.n_numaux = 0;
|
|
if (bfd_is_und_section (symbol->section))
|
|
{
|
|
native->u.syment.n_scnum = N_UNDEF;
|
|
native->u.syment.n_value = symbol->value;
|
|
}
|
|
else if (bfd_is_com_section (symbol->section))
|
|
{
|
|
native->u.syment.n_scnum = N_UNDEF;
|
|
native->u.syment.n_value = symbol->value;
|
|
}
|
|
else if (symbol->flags & BSF_FILE)
|
|
{
|
|
native->u.syment.n_scnum = N_DEBUG;
|
|
native->u.syment.n_numaux = 1;
|
|
}
|
|
else if (symbol->flags & BSF_DEBUGGING)
|
|
{
|
|
/* There isn't much point to writing out a debugging symbol
|
|
unless we are prepared to convert it into COFF debugging
|
|
format. So, we just ignore them. We must clobber the symbol
|
|
name to keep it from being put in the string table. */
|
|
symbol->name = "";
|
|
if (isym != NULL)
|
|
memset (isym, 0, sizeof (*isym));
|
|
return TRUE;
|
|
}
|
|
else
|
|
{
|
|
native->u.syment.n_scnum = output_section->target_index;
|
|
native->u.syment.n_value = (symbol->value
|
|
+ symbol->section->output_offset);
|
|
if (! obj_pe (abfd))
|
|
native->u.syment.n_value += output_section->vma;
|
|
|
|
/* Copy the any flags from the file header into the symbol.
|
|
FIXME: Why? */
|
|
{
|
|
coff_symbol_type *c = coff_symbol_from (symbol);
|
|
if (c != (coff_symbol_type *) NULL)
|
|
native->u.syment.n_flags = bfd_asymbol_bfd (&c->symbol)->flags;
|
|
}
|
|
}
|
|
|
|
native->u.syment.n_type = 0;
|
|
if (symbol->flags & BSF_FILE)
|
|
native->u.syment.n_sclass = C_FILE;
|
|
else if (symbol->flags & BSF_LOCAL)
|
|
native->u.syment.n_sclass = C_STAT;
|
|
else if (symbol->flags & BSF_WEAK)
|
|
native->u.syment.n_sclass = obj_pe (abfd) ? C_NT_WEAK : C_WEAKEXT;
|
|
else
|
|
native->u.syment.n_sclass = C_EXT;
|
|
|
|
ret = coff_write_symbol (abfd, symbol, native, written, string_size_p,
|
|
debug_string_section_p, debug_string_size_p);
|
|
if (isym != NULL)
|
|
*isym = native->u.syment;
|
|
if (iaux != NULL && native->u.syment.n_numaux)
|
|
*iaux = native[1].u.auxent;
|
|
return ret;
|
|
}
|
|
|
|
/* Write a native symbol to a COFF file. */
|
|
|
|
static bfd_boolean
|
|
coff_write_native_symbol (bfd *abfd,
|
|
coff_symbol_type *symbol,
|
|
bfd_vma *written,
|
|
bfd_size_type *string_size_p,
|
|
asection **debug_string_section_p,
|
|
bfd_size_type *debug_string_size_p)
|
|
{
|
|
combined_entry_type *native = symbol->native;
|
|
alent *lineno = symbol->lineno;
|
|
struct bfd_link_info *link_info = coff_data (abfd)->link_info;
|
|
|
|
if ((!link_info || link_info->strip_discarded)
|
|
&& !bfd_is_abs_section (symbol->symbol.section)
|
|
&& symbol->symbol.section->output_section == bfd_abs_section_ptr)
|
|
{
|
|
symbol->symbol.name = "";
|
|
return TRUE;
|
|
}
|
|
|
|
BFD_ASSERT (native->is_sym);
|
|
/* If this symbol has an associated line number, we must store the
|
|
symbol index in the line number field. We also tag the auxent to
|
|
point to the right place in the lineno table. */
|
|
if (lineno && !symbol->done_lineno && symbol->symbol.section->owner != NULL)
|
|
{
|
|
unsigned int count = 0;
|
|
|
|
lineno[count].u.offset = *written;
|
|
if (native->u.syment.n_numaux)
|
|
{
|
|
union internal_auxent *a = &((native + 1)->u.auxent);
|
|
|
|
a->x_sym.x_fcnary.x_fcn.x_lnnoptr =
|
|
symbol->symbol.section->output_section->moving_line_filepos;
|
|
}
|
|
|
|
/* Count and relocate all other linenumbers. */
|
|
count++;
|
|
while (lineno[count].line_number != 0)
|
|
{
|
|
lineno[count].u.offset +=
|
|
(symbol->symbol.section->output_section->vma
|
|
+ symbol->symbol.section->output_offset);
|
|
count++;
|
|
}
|
|
symbol->done_lineno = TRUE;
|
|
|
|
if (! bfd_is_const_section (symbol->symbol.section->output_section))
|
|
symbol->symbol.section->output_section->moving_line_filepos +=
|
|
count * bfd_coff_linesz (abfd);
|
|
}
|
|
|
|
return coff_write_symbol (abfd, &(symbol->symbol), native, written,
|
|
string_size_p, debug_string_section_p,
|
|
debug_string_size_p);
|
|
}
|
|
|
|
static void
|
|
null_error_handler (const char * fmt ATTRIBUTE_UNUSED, ...)
|
|
{
|
|
}
|
|
|
|
/* Write out the COFF symbols. */
|
|
|
|
bfd_boolean
|
|
coff_write_symbols (bfd *abfd)
|
|
{
|
|
bfd_size_type string_size;
|
|
asection *debug_string_section;
|
|
bfd_size_type debug_string_size;
|
|
unsigned int i;
|
|
unsigned int limit = bfd_get_symcount (abfd);
|
|
bfd_vma written = 0;
|
|
asymbol **p;
|
|
|
|
string_size = 0;
|
|
debug_string_section = NULL;
|
|
debug_string_size = 0;
|
|
|
|
/* If this target supports long section names, they must be put into
|
|
the string table. This is supported by PE. This code must
|
|
handle section names just as they are handled in
|
|
coff_write_object_contents. */
|
|
if (bfd_coff_long_section_names (abfd))
|
|
{
|
|
asection *o;
|
|
|
|
for (o = abfd->sections; o != NULL; o = o->next)
|
|
{
|
|
size_t len;
|
|
|
|
len = strlen (o->name);
|
|
if (len > SCNNMLEN)
|
|
string_size += len + 1;
|
|
}
|
|
}
|
|
|
|
/* Seek to the right place. */
|
|
if (bfd_seek (abfd, obj_sym_filepos (abfd), SEEK_SET) != 0)
|
|
return FALSE;
|
|
|
|
/* Output all the symbols we have. */
|
|
written = 0;
|
|
for (p = abfd->outsymbols, i = 0; i < limit; i++, p++)
|
|
{
|
|
asymbol *symbol = *p;
|
|
coff_symbol_type *c_symbol = coff_symbol_from (symbol);
|
|
|
|
if (c_symbol == (coff_symbol_type *) NULL
|
|
|| c_symbol->native == (combined_entry_type *) NULL)
|
|
{
|
|
if (!coff_write_alien_symbol (abfd, symbol, NULL, NULL, &written,
|
|
&string_size, &debug_string_section,
|
|
&debug_string_size))
|
|
return FALSE;
|
|
}
|
|
else
|
|
{
|
|
if (coff_backend_info (abfd)->_bfd_coff_classify_symbol != NULL)
|
|
{
|
|
bfd_error_handler_type current_error_handler;
|
|
enum coff_symbol_classification sym_class;
|
|
unsigned char *n_sclass;
|
|
|
|
/* Suppress error reporting by bfd_coff_classify_symbol.
|
|
Error messages can be generated when we are processing a local
|
|
symbol which has no associated section and we do not have to
|
|
worry about this, all we need to know is that it is local. */
|
|
current_error_handler = bfd_set_error_handler (null_error_handler);
|
|
BFD_ASSERT (c_symbol->native->is_sym);
|
|
sym_class = bfd_coff_classify_symbol (abfd,
|
|
&c_symbol->native->u.syment);
|
|
(void) bfd_set_error_handler (current_error_handler);
|
|
|
|
n_sclass = &c_symbol->native->u.syment.n_sclass;
|
|
|
|
/* If the symbol class has been changed (eg objcopy/ld script/etc)
|
|
we cannot retain the existing sclass from the original symbol.
|
|
Weak symbols only have one valid sclass, so just set it always.
|
|
If it is not local class and should be, set it C_STAT.
|
|
If it is global and not classified as global, or if it is
|
|
weak (which is also classified as global), set it C_EXT. */
|
|
|
|
if (symbol->flags & BSF_WEAK)
|
|
*n_sclass = obj_pe (abfd) ? C_NT_WEAK : C_WEAKEXT;
|
|
else if (symbol->flags & BSF_LOCAL && sym_class != COFF_SYMBOL_LOCAL)
|
|
*n_sclass = C_STAT;
|
|
else if (symbol->flags & BSF_GLOBAL
|
|
&& (sym_class != COFF_SYMBOL_GLOBAL
|
|
#ifdef COFF_WITH_PE
|
|
|| *n_sclass == C_NT_WEAK
|
|
#endif
|
|
|| *n_sclass == C_WEAKEXT))
|
|
c_symbol->native->u.syment.n_sclass = C_EXT;
|
|
}
|
|
|
|
if (!coff_write_native_symbol (abfd, c_symbol, &written,
|
|
&string_size, &debug_string_section,
|
|
&debug_string_size))
|
|
return FALSE;
|
|
}
|
|
}
|
|
|
|
obj_raw_syment_count (abfd) = written;
|
|
|
|
/* Now write out strings. */
|
|
if (string_size != 0)
|
|
{
|
|
unsigned int size = string_size + STRING_SIZE_SIZE;
|
|
bfd_byte buffer[STRING_SIZE_SIZE];
|
|
|
|
#if STRING_SIZE_SIZE == 4
|
|
H_PUT_32 (abfd, size, buffer);
|
|
#else
|
|
#error Change H_PUT_32
|
|
#endif
|
|
if (bfd_bwrite ((void *) buffer, (bfd_size_type) sizeof (buffer), abfd)
|
|
!= sizeof (buffer))
|
|
return FALSE;
|
|
|
|
/* Handle long section names. This code must handle section
|
|
names just as they are handled in coff_write_object_contents. */
|
|
if (bfd_coff_long_section_names (abfd))
|
|
{
|
|
asection *o;
|
|
|
|
for (o = abfd->sections; o != NULL; o = o->next)
|
|
{
|
|
size_t len;
|
|
|
|
len = strlen (o->name);
|
|
if (len > SCNNMLEN)
|
|
{
|
|
if (bfd_bwrite (o->name, (bfd_size_type) (len + 1), abfd)
|
|
!= len + 1)
|
|
return FALSE;
|
|
}
|
|
}
|
|
}
|
|
|
|
for (p = abfd->outsymbols, i = 0;
|
|
i < limit;
|
|
i++, p++)
|
|
{
|
|
asymbol *q = *p;
|
|
size_t name_length = strlen (q->name);
|
|
coff_symbol_type *c_symbol = coff_symbol_from (q);
|
|
size_t maxlen;
|
|
|
|
/* Figure out whether the symbol name should go in the string
|
|
table. Symbol names that are short enough are stored
|
|
directly in the syment structure. File names permit a
|
|
different, longer, length in the syment structure. On
|
|
XCOFF, some symbol names are stored in the .debug section
|
|
rather than in the string table. */
|
|
|
|
if (c_symbol == NULL
|
|
|| c_symbol->native == NULL)
|
|
/* This is not a COFF symbol, so it certainly is not a
|
|
file name, nor does it go in the .debug section. */
|
|
maxlen = bfd_coff_force_symnames_in_strings (abfd) ? 0 : SYMNMLEN;
|
|
|
|
else if (! c_symbol->native->is_sym)
|
|
maxlen = bfd_coff_force_symnames_in_strings (abfd) ? 0 : SYMNMLEN;
|
|
|
|
else if (bfd_coff_symname_in_debug (abfd,
|
|
&c_symbol->native->u.syment))
|
|
/* This symbol name is in the XCOFF .debug section.
|
|
Don't write it into the string table. */
|
|
maxlen = name_length;
|
|
|
|
else if (c_symbol->native->u.syment.n_sclass == C_FILE
|
|
&& c_symbol->native->u.syment.n_numaux > 0)
|
|
{
|
|
if (bfd_coff_force_symnames_in_strings (abfd))
|
|
{
|
|
if (bfd_bwrite (".file", (bfd_size_type) 6, abfd) != 6)
|
|
return FALSE;
|
|
}
|
|
maxlen = bfd_coff_filnmlen (abfd);
|
|
}
|
|
else
|
|
maxlen = bfd_coff_force_symnames_in_strings (abfd) ? 0 : SYMNMLEN;
|
|
|
|
if (name_length > maxlen)
|
|
{
|
|
if (bfd_bwrite ((void *) (q->name), (bfd_size_type) name_length + 1,
|
|
abfd) != name_length + 1)
|
|
return FALSE;
|
|
}
|
|
}
|
|
}
|
|
else
|
|
{
|
|
/* We would normally not write anything here, but we'll write
|
|
out 4 so that any stupid coff reader which tries to read the
|
|
string table even when there isn't one won't croak. */
|
|
unsigned int size = STRING_SIZE_SIZE;
|
|
bfd_byte buffer[STRING_SIZE_SIZE];
|
|
|
|
#if STRING_SIZE_SIZE == 4
|
|
H_PUT_32 (abfd, size, buffer);
|
|
#else
|
|
#error Change H_PUT_32
|
|
#endif
|
|
if (bfd_bwrite ((void *) buffer, (bfd_size_type) STRING_SIZE_SIZE, abfd)
|
|
!= STRING_SIZE_SIZE)
|
|
return FALSE;
|
|
}
|
|
|
|
/* Make sure the .debug section was created to be the correct size.
|
|
We should create it ourselves on the fly, but we don't because
|
|
BFD won't let us write to any section until we know how large all
|
|
the sections are. We could still do it by making another pass
|
|
over the symbols. FIXME. */
|
|
BFD_ASSERT (debug_string_size == 0
|
|
|| (debug_string_section != (asection *) NULL
|
|
&& (BFD_ALIGN (debug_string_size,
|
|
1 << debug_string_section->alignment_power)
|
|
== debug_string_section->size)));
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
bfd_boolean
|
|
coff_write_linenumbers (bfd *abfd)
|
|
{
|
|
asection *s;
|
|
bfd_size_type linesz;
|
|
void * buff;
|
|
|
|
linesz = bfd_coff_linesz (abfd);
|
|
buff = bfd_alloc (abfd, linesz);
|
|
if (!buff)
|
|
return FALSE;
|
|
for (s = abfd->sections; s != (asection *) NULL; s = s->next)
|
|
{
|
|
if (s->lineno_count)
|
|
{
|
|
asymbol **q = abfd->outsymbols;
|
|
if (bfd_seek (abfd, s->line_filepos, SEEK_SET) != 0)
|
|
return FALSE;
|
|
/* Find all the linenumbers in this section. */
|
|
while (*q)
|
|
{
|
|
asymbol *p = *q;
|
|
if (p->section->output_section == s)
|
|
{
|
|
alent *l =
|
|
BFD_SEND (bfd_asymbol_bfd (p), _get_lineno,
|
|
(bfd_asymbol_bfd (p), p));
|
|
if (l)
|
|
{
|
|
/* Found a linenumber entry, output. */
|
|
struct internal_lineno out;
|
|
|
|
memset ((void *) & out, 0, sizeof (out));
|
|
out.l_lnno = 0;
|
|
out.l_addr.l_symndx = l->u.offset;
|
|
bfd_coff_swap_lineno_out (abfd, &out, buff);
|
|
if (bfd_bwrite (buff, (bfd_size_type) linesz, abfd)
|
|
!= linesz)
|
|
return FALSE;
|
|
l++;
|
|
while (l->line_number)
|
|
{
|
|
out.l_lnno = l->line_number;
|
|
out.l_addr.l_symndx = l->u.offset;
|
|
bfd_coff_swap_lineno_out (abfd, &out, buff);
|
|
if (bfd_bwrite (buff, (bfd_size_type) linesz, abfd)
|
|
!= linesz)
|
|
return FALSE;
|
|
l++;
|
|
}
|
|
}
|
|
}
|
|
q++;
|
|
}
|
|
}
|
|
}
|
|
bfd_release (abfd, buff);
|
|
return TRUE;
|
|
}
|
|
|
|
alent *
|
|
coff_get_lineno (bfd *ignore_abfd ATTRIBUTE_UNUSED, asymbol *symbol)
|
|
{
|
|
return coffsymbol (symbol)->lineno;
|
|
}
|
|
|
|
/* This function transforms the offsets into the symbol table into
|
|
pointers to syments. */
|
|
|
|
static void
|
|
coff_pointerize_aux (bfd *abfd,
|
|
combined_entry_type *table_base,
|
|
combined_entry_type *symbol,
|
|
unsigned int indaux,
|
|
combined_entry_type *auxent)
|
|
{
|
|
unsigned int type = symbol->u.syment.n_type;
|
|
unsigned int n_sclass = symbol->u.syment.n_sclass;
|
|
|
|
BFD_ASSERT (symbol->is_sym);
|
|
if (coff_backend_info (abfd)->_bfd_coff_pointerize_aux_hook)
|
|
{
|
|
if ((*coff_backend_info (abfd)->_bfd_coff_pointerize_aux_hook)
|
|
(abfd, table_base, symbol, indaux, auxent))
|
|
return;
|
|
}
|
|
|
|
/* Don't bother if this is a file or a section. */
|
|
if (n_sclass == C_STAT && type == T_NULL)
|
|
return;
|
|
if (n_sclass == C_FILE)
|
|
return;
|
|
|
|
BFD_ASSERT (! auxent->is_sym);
|
|
/* Otherwise patch up. */
|
|
#define N_TMASK coff_data (abfd)->local_n_tmask
|
|
#define N_BTSHFT coff_data (abfd)->local_n_btshft
|
|
|
|
if ((ISFCN (type) || ISTAG (n_sclass) || n_sclass == C_BLOCK
|
|
|| n_sclass == C_FCN)
|
|
&& auxent->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.l > 0)
|
|
{
|
|
auxent->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.p =
|
|
table_base + auxent->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.l;
|
|
auxent->fix_end = 1;
|
|
}
|
|
/* A negative tagndx is meaningless, but the SCO 3.2v4 cc can
|
|
generate one, so we must be careful to ignore it. */
|
|
if (auxent->u.auxent.x_sym.x_tagndx.l > 0)
|
|
{
|
|
auxent->u.auxent.x_sym.x_tagndx.p =
|
|
table_base + auxent->u.auxent.x_sym.x_tagndx.l;
|
|
auxent->fix_tag = 1;
|
|
}
|
|
}
|
|
|
|
/* Allocate space for the ".debug" section, and read it.
|
|
We did not read the debug section until now, because
|
|
we didn't want to go to the trouble until someone needed it. */
|
|
|
|
static char *
|
|
build_debug_section (bfd *abfd, asection ** sect_return)
|
|
{
|
|
char *debug_section;
|
|
file_ptr position;
|
|
bfd_size_type sec_size;
|
|
|
|
asection *sect = bfd_get_section_by_name (abfd, ".debug");
|
|
|
|
if (!sect)
|
|
{
|
|
bfd_set_error (bfd_error_no_debug_section);
|
|
return NULL;
|
|
}
|
|
|
|
sec_size = sect->size;
|
|
debug_section = (char *) bfd_alloc (abfd, sec_size);
|
|
if (debug_section == NULL)
|
|
return NULL;
|
|
|
|
/* Seek to the beginning of the `.debug' section and read it.
|
|
Save the current position first; it is needed by our caller.
|
|
Then read debug section and reset the file pointer. */
|
|
|
|
position = bfd_tell (abfd);
|
|
if (bfd_seek (abfd, sect->filepos, SEEK_SET) != 0
|
|
|| bfd_bread (debug_section, sec_size, abfd) != sec_size
|
|
|| bfd_seek (abfd, position, SEEK_SET) != 0)
|
|
return NULL;
|
|
|
|
* sect_return = sect;
|
|
return debug_section;
|
|
}
|
|
|
|
/* Return a pointer to a malloc'd copy of 'name'. 'name' may not be
|
|
\0-terminated, but will not exceed 'maxlen' characters. The copy *will*
|
|
be \0-terminated. */
|
|
|
|
static char *
|
|
copy_name (bfd *abfd, char *name, size_t maxlen)
|
|
{
|
|
size_t len;
|
|
char *newname;
|
|
|
|
for (len = 0; len < maxlen; ++len)
|
|
if (name[len] == '\0')
|
|
break;
|
|
|
|
if ((newname = (char *) bfd_alloc (abfd, (bfd_size_type) len + 1)) == NULL)
|
|
return NULL;
|
|
|
|
strncpy (newname, name, len);
|
|
newname[len] = '\0';
|
|
return newname;
|
|
}
|
|
|
|
/* Read in the external symbols. */
|
|
|
|
bfd_boolean
|
|
_bfd_coff_get_external_symbols (bfd *abfd)
|
|
{
|
|
bfd_size_type symesz;
|
|
bfd_size_type size;
|
|
void * syms;
|
|
|
|
if (obj_coff_external_syms (abfd) != NULL)
|
|
return TRUE;
|
|
|
|
symesz = bfd_coff_symesz (abfd);
|
|
|
|
size = obj_raw_syment_count (abfd) * symesz;
|
|
if (size == 0)
|
|
return TRUE;
|
|
|
|
syms = bfd_malloc (size);
|
|
if (syms == NULL)
|
|
return FALSE;
|
|
|
|
if (bfd_seek (abfd, obj_sym_filepos (abfd), SEEK_SET) != 0
|
|
|| bfd_bread (syms, size, abfd) != size)
|
|
{
|
|
if (syms != NULL)
|
|
free (syms);
|
|
return FALSE;
|
|
}
|
|
|
|
obj_coff_external_syms (abfd) = syms;
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
/* Read in the external strings. The strings are not loaded until
|
|
they are needed. This is because we have no simple way of
|
|
detecting a missing string table in an archive. If the strings
|
|
are loaded then the STRINGS and STRINGS_LEN fields in the
|
|
coff_tdata structure will be set. */
|
|
|
|
const char *
|
|
_bfd_coff_read_string_table (bfd *abfd)
|
|
{
|
|
char extstrsize[STRING_SIZE_SIZE];
|
|
bfd_size_type strsize;
|
|
char *strings;
|
|
file_ptr pos;
|
|
|
|
if (obj_coff_strings (abfd) != NULL)
|
|
return obj_coff_strings (abfd);
|
|
|
|
if (obj_sym_filepos (abfd) == 0)
|
|
{
|
|
bfd_set_error (bfd_error_no_symbols);
|
|
return NULL;
|
|
}
|
|
|
|
pos = obj_sym_filepos (abfd);
|
|
pos += obj_raw_syment_count (abfd) * bfd_coff_symesz (abfd);
|
|
if (bfd_seek (abfd, pos, SEEK_SET) != 0)
|
|
return NULL;
|
|
|
|
if (bfd_bread (extstrsize, (bfd_size_type) sizeof extstrsize, abfd)
|
|
!= sizeof extstrsize)
|
|
{
|
|
if (bfd_get_error () != bfd_error_file_truncated)
|
|
return NULL;
|
|
|
|
/* There is no string table. */
|
|
strsize = STRING_SIZE_SIZE;
|
|
}
|
|
else
|
|
{
|
|
#if STRING_SIZE_SIZE == 4
|
|
strsize = H_GET_32 (abfd, extstrsize);
|
|
#else
|
|
#error Change H_GET_32
|
|
#endif
|
|
}
|
|
|
|
if (strsize < STRING_SIZE_SIZE)
|
|
{
|
|
(*_bfd_error_handler)
|
|
(_("%B: bad string table size %lu"), abfd, (unsigned long) strsize);
|
|
bfd_set_error (bfd_error_bad_value);
|
|
return NULL;
|
|
}
|
|
|
|
strings = (char *) bfd_malloc (strsize + 1);
|
|
if (strings == NULL)
|
|
return NULL;
|
|
|
|
/* PR 17521 file: 079-54929-0.004.
|
|
A corrupt file could contain an index that points into the first
|
|
STRING_SIZE_SIZE bytes of the string table, so make sure that
|
|
they are zero. */
|
|
memset (strings, 0, STRING_SIZE_SIZE);
|
|
|
|
if (bfd_bread (strings + STRING_SIZE_SIZE, strsize - STRING_SIZE_SIZE, abfd)
|
|
!= strsize - STRING_SIZE_SIZE)
|
|
{
|
|
free (strings);
|
|
return NULL;
|
|
}
|
|
|
|
obj_coff_strings (abfd) = strings;
|
|
obj_coff_strings_len (abfd) = strsize;
|
|
/* Terminate the string table, just in case. */
|
|
strings[strsize] = 0;
|
|
return strings;
|
|
}
|
|
|
|
/* Free up the external symbols and strings read from a COFF file. */
|
|
|
|
bfd_boolean
|
|
_bfd_coff_free_symbols (bfd *abfd)
|
|
{
|
|
if (obj_coff_external_syms (abfd) != NULL
|
|
&& ! obj_coff_keep_syms (abfd))
|
|
{
|
|
free (obj_coff_external_syms (abfd));
|
|
obj_coff_external_syms (abfd) = NULL;
|
|
}
|
|
if (obj_coff_strings (abfd) != NULL
|
|
&& ! obj_coff_keep_strings (abfd))
|
|
{
|
|
free (obj_coff_strings (abfd));
|
|
obj_coff_strings (abfd) = NULL;
|
|
obj_coff_strings_len (abfd) = 0;
|
|
}
|
|
return TRUE;
|
|
}
|
|
|
|
/* Read a symbol table into freshly bfd_allocated memory, swap it, and
|
|
knit the symbol names into a normalized form. By normalized here I
|
|
mean that all symbols have an n_offset pointer that points to a null-
|
|
terminated string. */
|
|
|
|
combined_entry_type *
|
|
coff_get_normalized_symtab (bfd *abfd)
|
|
{
|
|
combined_entry_type *internal;
|
|
combined_entry_type *internal_ptr;
|
|
combined_entry_type *symbol_ptr;
|
|
combined_entry_type *internal_end;
|
|
size_t symesz;
|
|
char *raw_src;
|
|
char *raw_end;
|
|
const char *string_table = NULL;
|
|
asection * debug_sec = NULL;
|
|
char *debug_sec_data = NULL;
|
|
bfd_size_type size;
|
|
|
|
if (obj_raw_syments (abfd) != NULL)
|
|
return obj_raw_syments (abfd);
|
|
|
|
if (! _bfd_coff_get_external_symbols (abfd))
|
|
return NULL;
|
|
|
|
size = obj_raw_syment_count (abfd) * sizeof (combined_entry_type);
|
|
internal = (combined_entry_type *) bfd_zalloc (abfd, size);
|
|
if (internal == NULL && size != 0)
|
|
return NULL;
|
|
internal_end = internal + obj_raw_syment_count (abfd);
|
|
|
|
raw_src = (char *) obj_coff_external_syms (abfd);
|
|
|
|
/* Mark the end of the symbols. */
|
|
symesz = bfd_coff_symesz (abfd);
|
|
raw_end = (char *) raw_src + obj_raw_syment_count (abfd) * symesz;
|
|
|
|
/* FIXME SOMEDAY. A string table size of zero is very weird, but
|
|
probably possible. If one shows up, it will probably kill us. */
|
|
|
|
/* Swap all the raw entries. */
|
|
for (internal_ptr = internal;
|
|
raw_src < raw_end;
|
|
raw_src += symesz, internal_ptr++)
|
|
{
|
|
unsigned int i;
|
|
|
|
bfd_coff_swap_sym_in (abfd, (void *) raw_src,
|
|
(void *) & internal_ptr->u.syment);
|
|
symbol_ptr = internal_ptr;
|
|
internal_ptr->is_sym = TRUE;
|
|
|
|
/* PR 17512: file: 1353-1166-0.004. */
|
|
if (symbol_ptr->u.syment.n_sclass == C_FILE
|
|
&& symbol_ptr->u.syment.n_numaux > 0
|
|
&& raw_src + symesz + symbol_ptr->u.syment.n_numaux
|
|
* symesz > raw_end)
|
|
{
|
|
bfd_release (abfd, internal);
|
|
return NULL;
|
|
}
|
|
|
|
for (i = 0;
|
|
i < symbol_ptr->u.syment.n_numaux;
|
|
i++)
|
|
{
|
|
internal_ptr++;
|
|
/* PR 17512: Prevent buffer overrun. */
|
|
if (internal_ptr >= internal_end)
|
|
{
|
|
bfd_release (abfd, internal);
|
|
return NULL;
|
|
}
|
|
|
|
raw_src += symesz;
|
|
bfd_coff_swap_aux_in (abfd, (void *) raw_src,
|
|
symbol_ptr->u.syment.n_type,
|
|
symbol_ptr->u.syment.n_sclass,
|
|
(int) i, symbol_ptr->u.syment.n_numaux,
|
|
&(internal_ptr->u.auxent));
|
|
|
|
internal_ptr->is_sym = FALSE;
|
|
coff_pointerize_aux (abfd, internal, symbol_ptr, i,
|
|
internal_ptr);
|
|
}
|
|
}
|
|
|
|
/* Free the raw symbols, but not the strings (if we have them). */
|
|
obj_coff_keep_strings (abfd) = TRUE;
|
|
if (! _bfd_coff_free_symbols (abfd))
|
|
return NULL;
|
|
|
|
for (internal_ptr = internal; internal_ptr < internal_end;
|
|
internal_ptr++)
|
|
{
|
|
BFD_ASSERT (internal_ptr->is_sym);
|
|
|
|
if (internal_ptr->u.syment.n_sclass == C_FILE
|
|
&& internal_ptr->u.syment.n_numaux > 0)
|
|
{
|
|
combined_entry_type * aux = internal_ptr + 1;
|
|
|
|
/* Make a file symbol point to the name in the auxent, since
|
|
the text ".file" is redundant. */
|
|
BFD_ASSERT (! aux->is_sym);
|
|
|
|
if (aux->u.auxent.x_file.x_n.x_zeroes == 0)
|
|
{
|
|
/* The filename is a long one, point into the string table. */
|
|
if (string_table == NULL)
|
|
{
|
|
string_table = _bfd_coff_read_string_table (abfd);
|
|
if (string_table == NULL)
|
|
return NULL;
|
|
}
|
|
|
|
if ((bfd_size_type)(aux->u.auxent.x_file.x_n.x_offset)
|
|
>= obj_coff_strings_len (abfd))
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t) _("<corrupt>");
|
|
else
|
|
internal_ptr->u.syment._n._n_n._n_offset =
|
|
(bfd_hostptr_t) (string_table + (aux->u.auxent.x_file.x_n.x_offset));
|
|
}
|
|
else
|
|
{
|
|
/* Ordinary short filename, put into memory anyway. The
|
|
Microsoft PE tools sometimes store a filename in
|
|
multiple AUX entries. */
|
|
if (internal_ptr->u.syment.n_numaux > 1
|
|
&& coff_data (abfd)->pe)
|
|
internal_ptr->u.syment._n._n_n._n_offset =
|
|
(bfd_hostptr_t)
|
|
copy_name (abfd,
|
|
aux->u.auxent.x_file.x_fname,
|
|
internal_ptr->u.syment.n_numaux * symesz);
|
|
else
|
|
internal_ptr->u.syment._n._n_n._n_offset =
|
|
((bfd_hostptr_t)
|
|
copy_name (abfd,
|
|
aux->u.auxent.x_file.x_fname,
|
|
(size_t) bfd_coff_filnmlen (abfd)));
|
|
}
|
|
}
|
|
else
|
|
{
|
|
if (internal_ptr->u.syment._n._n_n._n_zeroes != 0)
|
|
{
|
|
/* This is a "short" name. Make it long. */
|
|
size_t i;
|
|
char *newstring;
|
|
|
|
/* Find the length of this string without walking into memory
|
|
that isn't ours. */
|
|
for (i = 0; i < 8; ++i)
|
|
if (internal_ptr->u.syment._n._n_name[i] == '\0')
|
|
break;
|
|
|
|
newstring = (char *) bfd_zalloc (abfd, (bfd_size_type) (i + 1));
|
|
if (newstring == NULL)
|
|
return NULL;
|
|
strncpy (newstring, internal_ptr->u.syment._n._n_name, i);
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t) newstring;
|
|
internal_ptr->u.syment._n._n_n._n_zeroes = 0;
|
|
}
|
|
else if (internal_ptr->u.syment._n._n_n._n_offset == 0)
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t) "";
|
|
else if (!bfd_coff_symname_in_debug (abfd, &internal_ptr->u.syment))
|
|
{
|
|
/* Long name already. Point symbol at the string in the
|
|
table. */
|
|
if (string_table == NULL)
|
|
{
|
|
string_table = _bfd_coff_read_string_table (abfd);
|
|
if (string_table == NULL)
|
|
return NULL;
|
|
}
|
|
if (internal_ptr->u.syment._n._n_n._n_offset >= obj_coff_strings_len (abfd)
|
|
|| string_table + internal_ptr->u.syment._n._n_n._n_offset < string_table)
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t) _("<corrupt>");
|
|
else
|
|
internal_ptr->u.syment._n._n_n._n_offset =
|
|
((bfd_hostptr_t)
|
|
(string_table
|
|
+ internal_ptr->u.syment._n._n_n._n_offset));
|
|
}
|
|
else
|
|
{
|
|
/* Long name in debug section. Very similar. */
|
|
if (debug_sec_data == NULL)
|
|
debug_sec_data = build_debug_section (abfd, & debug_sec);
|
|
if (debug_sec_data != NULL)
|
|
{
|
|
BFD_ASSERT (debug_sec != NULL);
|
|
/* PR binutils/17512: Catch out of range offsets into the debug data. */
|
|
if (internal_ptr->u.syment._n._n_n._n_offset > debug_sec->size
|
|
|| debug_sec_data + internal_ptr->u.syment._n._n_n._n_offset < debug_sec_data)
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t) _("<corrupt>");
|
|
else
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t)
|
|
(debug_sec_data + internal_ptr->u.syment._n._n_n._n_offset);
|
|
}
|
|
else
|
|
internal_ptr->u.syment._n._n_n._n_offset = (bfd_hostptr_t) "";
|
|
}
|
|
}
|
|
internal_ptr += internal_ptr->u.syment.n_numaux;
|
|
}
|
|
|
|
obj_raw_syments (abfd) = internal;
|
|
BFD_ASSERT (obj_raw_syment_count (abfd)
|
|
== (unsigned int) (internal_ptr - internal));
|
|
|
|
return internal;
|
|
}
|
|
|
|
long
|
|
coff_get_reloc_upper_bound (bfd *abfd, sec_ptr asect)
|
|
{
|
|
if (bfd_get_format (abfd) != bfd_object)
|
|
{
|
|
bfd_set_error (bfd_error_invalid_operation);
|
|
return -1;
|
|
}
|
|
return (asect->reloc_count + 1) * sizeof (arelent *);
|
|
}
|
|
|
|
asymbol *
|
|
coff_make_empty_symbol (bfd *abfd)
|
|
{
|
|
bfd_size_type amt = sizeof (coff_symbol_type);
|
|
coff_symbol_type *new_symbol = (coff_symbol_type *) bfd_zalloc (abfd, amt);
|
|
|
|
if (new_symbol == NULL)
|
|
return NULL;
|
|
new_symbol->symbol.section = 0;
|
|
new_symbol->native = NULL;
|
|
new_symbol->lineno = NULL;
|
|
new_symbol->done_lineno = FALSE;
|
|
new_symbol->symbol.the_bfd = abfd;
|
|
|
|
return & new_symbol->symbol;
|
|
}
|
|
|
|
/* Make a debugging symbol. */
|
|
|
|
asymbol *
|
|
coff_bfd_make_debug_symbol (bfd *abfd,
|
|
void * ptr ATTRIBUTE_UNUSED,
|
|
unsigned long sz ATTRIBUTE_UNUSED)
|
|
{
|
|
bfd_size_type amt = sizeof (coff_symbol_type);
|
|
coff_symbol_type *new_symbol = (coff_symbol_type *) bfd_alloc (abfd, amt);
|
|
|
|
if (new_symbol == NULL)
|
|
return NULL;
|
|
/* @@ The 10 is a guess at a plausible maximum number of aux entries
|
|
(but shouldn't be a constant). */
|
|
amt = sizeof (combined_entry_type) * 10;
|
|
new_symbol->native = (combined_entry_type *) bfd_zalloc (abfd, amt);
|
|
if (!new_symbol->native)
|
|
return NULL;
|
|
new_symbol->native->is_sym = TRUE;
|
|
new_symbol->symbol.section = bfd_abs_section_ptr;
|
|
new_symbol->symbol.flags = BSF_DEBUGGING;
|
|
new_symbol->lineno = NULL;
|
|
new_symbol->done_lineno = FALSE;
|
|
new_symbol->symbol.the_bfd = abfd;
|
|
|
|
return & new_symbol->symbol;
|
|
}
|
|
|
|
void
|
|
coff_get_symbol_info (bfd *abfd, asymbol *symbol, symbol_info *ret)
|
|
{
|
|
bfd_symbol_info (symbol, ret);
|
|
|
|
if (coffsymbol (symbol)->native != NULL
|
|
&& coffsymbol (symbol)->native->fix_value
|
|
&& coffsymbol (symbol)->native->is_sym)
|
|
ret->value = coffsymbol (symbol)->native->u.syment.n_value -
|
|
(bfd_hostptr_t) obj_raw_syments (abfd);
|
|
}
|
|
|
|
/* Print out information about COFF symbol. */
|
|
|
|
void
|
|
coff_print_symbol (bfd *abfd,
|
|
void * filep,
|
|
asymbol *symbol,
|
|
bfd_print_symbol_type how)
|
|
{
|
|
FILE * file = (FILE *) filep;
|
|
|
|
switch (how)
|
|
{
|
|
case bfd_print_symbol_name:
|
|
fprintf (file, "%s", symbol->name);
|
|
break;
|
|
|
|
case bfd_print_symbol_more:
|
|
fprintf (file, "coff %s %s",
|
|
coffsymbol (symbol)->native ? "n" : "g",
|
|
coffsymbol (symbol)->lineno ? "l" : " ");
|
|
break;
|
|
|
|
case bfd_print_symbol_all:
|
|
if (coffsymbol (symbol)->native)
|
|
{
|
|
bfd_vma val;
|
|
unsigned int aux;
|
|
combined_entry_type *combined = coffsymbol (symbol)->native;
|
|
combined_entry_type *root = obj_raw_syments (abfd);
|
|
struct lineno_cache_entry *l = coffsymbol (symbol)->lineno;
|
|
|
|
fprintf (file, "[%3ld]", (long) (combined - root));
|
|
|
|
/* PR 17512: file: 079-33786-0.001:0.1. */
|
|
if (combined < obj_raw_syments (abfd)
|
|
|| combined >= obj_raw_syments (abfd) + obj_raw_syment_count (abfd))
|
|
{
|
|
fprintf (file, _("<corrupt info> %s"), symbol->name);
|
|
break;
|
|
}
|
|
|
|
BFD_ASSERT (combined->is_sym);
|
|
if (! combined->fix_value)
|
|
val = (bfd_vma) combined->u.syment.n_value;
|
|
else
|
|
val = combined->u.syment.n_value - (bfd_hostptr_t) root;
|
|
|
|
fprintf (file, "(sec %2d)(fl 0x%02x)(ty %3x)(scl %3d) (nx %d) 0x",
|
|
combined->u.syment.n_scnum,
|
|
combined->u.syment.n_flags,
|
|
combined->u.syment.n_type,
|
|
combined->u.syment.n_sclass,
|
|
combined->u.syment.n_numaux);
|
|
bfd_fprintf_vma (abfd, file, val);
|
|
fprintf (file, " %s", symbol->name);
|
|
|
|
for (aux = 0; aux < combined->u.syment.n_numaux; aux++)
|
|
{
|
|
combined_entry_type *auxp = combined + aux + 1;
|
|
long tagndx;
|
|
|
|
BFD_ASSERT (! auxp->is_sym);
|
|
if (auxp->fix_tag)
|
|
tagndx = auxp->u.auxent.x_sym.x_tagndx.p - root;
|
|
else
|
|
tagndx = auxp->u.auxent.x_sym.x_tagndx.l;
|
|
|
|
fprintf (file, "\n");
|
|
|
|
if (bfd_coff_print_aux (abfd, file, root, combined, auxp, aux))
|
|
continue;
|
|
|
|
switch (combined->u.syment.n_sclass)
|
|
{
|
|
case C_FILE:
|
|
fprintf (file, "File ");
|
|
break;
|
|
|
|
case C_STAT:
|
|
if (combined->u.syment.n_type == T_NULL)
|
|
/* Probably a section symbol ? */
|
|
{
|
|
fprintf (file, "AUX scnlen 0x%lx nreloc %d nlnno %d",
|
|
(unsigned long) auxp->u.auxent.x_scn.x_scnlen,
|
|
auxp->u.auxent.x_scn.x_nreloc,
|
|
auxp->u.auxent.x_scn.x_nlinno);
|
|
if (auxp->u.auxent.x_scn.x_checksum != 0
|
|
|| auxp->u.auxent.x_scn.x_associated != 0
|
|
|| auxp->u.auxent.x_scn.x_comdat != 0)
|
|
fprintf (file, " checksum 0x%lx assoc %d comdat %d",
|
|
auxp->u.auxent.x_scn.x_checksum,
|
|
auxp->u.auxent.x_scn.x_associated,
|
|
auxp->u.auxent.x_scn.x_comdat);
|
|
break;
|
|
}
|
|
/* Otherwise fall through. */
|
|
case C_EXT:
|
|
case C_AIX_WEAKEXT:
|
|
if (ISFCN (combined->u.syment.n_type))
|
|
{
|
|
long next, llnos;
|
|
|
|
if (auxp->fix_end)
|
|
next = (auxp->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.p
|
|
- root);
|
|
else
|
|
next = auxp->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.l;
|
|
llnos = auxp->u.auxent.x_sym.x_fcnary.x_fcn.x_lnnoptr;
|
|
fprintf (file,
|
|
"AUX tagndx %ld ttlsiz 0x%lx lnnos %ld next %ld",
|
|
tagndx,
|
|
(unsigned long) auxp->u.auxent.x_sym.x_misc.x_fsize,
|
|
llnos, next);
|
|
break;
|
|
}
|
|
/* Otherwise fall through. */
|
|
default:
|
|
fprintf (file, "AUX lnno %d size 0x%x tagndx %ld",
|
|
auxp->u.auxent.x_sym.x_misc.x_lnsz.x_lnno,
|
|
auxp->u.auxent.x_sym.x_misc.x_lnsz.x_size,
|
|
tagndx);
|
|
if (auxp->fix_end)
|
|
fprintf (file, " endndx %ld",
|
|
((long)
|
|
(auxp->u.auxent.x_sym.x_fcnary.x_fcn.x_endndx.p
|
|
- root)));
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (l)
|
|
{
|
|
fprintf (file, "\n%s :", l->u.sym->name);
|
|
l++;
|
|
while (l->line_number)
|
|
{
|
|
if (l->line_number > 0)
|
|
{
|
|
fprintf (file, "\n%4d : ", l->line_number);
|
|
bfd_fprintf_vma (abfd, file, l->u.offset + symbol->section->vma);
|
|
}
|
|
l++;
|
|
}
|
|
}
|
|
}
|
|
else
|
|
{
|
|
bfd_print_symbol_vandf (abfd, (void *) file, symbol);
|
|
fprintf (file, " %-5s %s %s %s",
|
|
symbol->section->name,
|
|
coffsymbol (symbol)->native ? "n" : "g",
|
|
coffsymbol (symbol)->lineno ? "l" : " ",
|
|
symbol->name);
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Return whether a symbol name implies a local symbol. In COFF,
|
|
local symbols generally start with ``.L''. Most targets use this
|
|
function for the is_local_label_name entry point, but some may
|
|
override it. */
|
|
|
|
bfd_boolean
|
|
_bfd_coff_is_local_label_name (bfd *abfd ATTRIBUTE_UNUSED,
|
|
const char *name)
|
|
{
|
|
return name[0] == '.' && name[1] == 'L';
|
|
}
|
|
|
|
/* Provided a BFD, a section and an offset (in bytes, not octets) into the
|
|
section, calculate and return the name of the source file and the line
|
|
nearest to the wanted location. */
|
|
|
|
bfd_boolean
|
|
coff_find_nearest_line_with_names (bfd *abfd,
|
|
asymbol **symbols,
|
|
asection *section,
|
|
bfd_vma offset,
|
|
const char **filename_ptr,
|
|
const char **functionname_ptr,
|
|
unsigned int *line_ptr,
|
|
const struct dwarf_debug_section *debug_sections)
|
|
{
|
|
bfd_boolean found;
|
|
unsigned int i;
|
|
unsigned int line_base;
|
|
coff_data_type *cof = coff_data (abfd);
|
|
/* Run through the raw syments if available. */
|
|
combined_entry_type *p;
|
|
combined_entry_type *pend;
|
|
alent *l;
|
|
struct coff_section_tdata *sec_data;
|
|
bfd_size_type amt;
|
|
|
|
/* Before looking through the symbol table, try to use a .stab
|
|
section to find the information. */
|
|
if (! _bfd_stab_section_find_nearest_line (abfd, symbols, section, offset,
|
|
&found, filename_ptr,
|
|
functionname_ptr, line_ptr,
|
|
&coff_data(abfd)->line_info))
|
|
return FALSE;
|
|
|
|
if (found)
|
|
return TRUE;
|
|
|
|
/* Also try examining DWARF2 debugging information. */
|
|
if (_bfd_dwarf2_find_nearest_line (abfd, symbols, NULL, section, offset,
|
|
filename_ptr, functionname_ptr,
|
|
line_ptr, NULL, debug_sections, 0,
|
|
&coff_data(abfd)->dwarf2_find_line_info))
|
|
return TRUE;
|
|
|
|
/* If the DWARF lookup failed, but there is DWARF information available
|
|
then the problem might be that the file has been rebased. This tool
|
|
changes the VMAs of all the sections, but it does not update the DWARF
|
|
information. So try again, using a bias against the address sought. */
|
|
if (coff_data (abfd)->dwarf2_find_line_info != NULL)
|
|
{
|
|
bfd_signed_vma bias;
|
|
|
|
bias = _bfd_dwarf2_find_symbol_bias (symbols,
|
|
& coff_data (abfd)->dwarf2_find_line_info);
|
|
|
|
if (bias
|
|
&& _bfd_dwarf2_find_nearest_line (abfd, symbols, NULL, section,
|
|
offset + bias,
|
|
filename_ptr, functionname_ptr,
|
|
line_ptr, NULL, debug_sections, 0,
|
|
&coff_data(abfd)->dwarf2_find_line_info))
|
|
return TRUE;
|
|
}
|
|
|
|
*filename_ptr = 0;
|
|
*functionname_ptr = 0;
|
|
*line_ptr = 0;
|
|
|
|
/* Don't try and find line numbers in a non coff file. */
|
|
if (!bfd_family_coff (abfd))
|
|
return FALSE;
|
|
|
|
if (cof == NULL)
|
|
return FALSE;
|
|
|
|
/* Find the first C_FILE symbol. */
|
|
p = cof->raw_syments;
|
|
if (!p)
|
|
return FALSE;
|
|
|
|
pend = p + cof->raw_syment_count;
|
|
while (p < pend)
|
|
{
|
|
BFD_ASSERT (p->is_sym);
|
|
if (p->u.syment.n_sclass == C_FILE)
|
|
break;
|
|
p += 1 + p->u.syment.n_numaux;
|
|
}
|
|
|
|
if (p < pend)
|
|
{
|
|
bfd_vma sec_vma;
|
|
bfd_vma maxdiff;
|
|
|
|
/* Look through the C_FILE symbols to find the best one. */
|
|
sec_vma = bfd_get_section_vma (abfd, section);
|
|
*filename_ptr = (char *) p->u.syment._n._n_n._n_offset;
|
|
maxdiff = (bfd_vma) 0 - (bfd_vma) 1;
|
|
while (1)
|
|
{
|
|
bfd_vma file_addr;
|
|
combined_entry_type *p2;
|
|
|
|
for (p2 = p + 1 + p->u.syment.n_numaux;
|
|
p2 < pend;
|
|
p2 += 1 + p2->u.syment.n_numaux)
|
|
{
|
|
BFD_ASSERT (p2->is_sym);
|
|
if (p2->u.syment.n_scnum > 0
|
|
&& (section
|
|
== coff_section_from_bfd_index (abfd,
|
|
p2->u.syment.n_scnum)))
|
|
break;
|
|
if (p2->u.syment.n_sclass == C_FILE)
|
|
{
|
|
p2 = pend;
|
|
break;
|
|
}
|
|
}
|
|
if (p2 >= pend)
|
|
break;
|
|
|
|
file_addr = (bfd_vma) p2->u.syment.n_value;
|
|
/* PR 11512: Include the section address of the function name symbol. */
|
|
if (p2->u.syment.n_scnum > 0)
|
|
file_addr += coff_section_from_bfd_index (abfd,
|
|
p2->u.syment.n_scnum)->vma;
|
|
/* We use <= MAXDIFF here so that if we get a zero length
|
|
file, we actually use the next file entry. */
|
|
if (p2 < pend
|
|
&& offset + sec_vma >= file_addr
|
|
&& offset + sec_vma - file_addr <= maxdiff)
|
|
{
|
|
*filename_ptr = (char *) p->u.syment._n._n_n._n_offset;
|
|
maxdiff = offset + sec_vma - p2->u.syment.n_value;
|
|
}
|
|
|
|
/* Avoid endless loops on erroneous files by ensuring that
|
|
we always move forward in the file. */
|
|
if (p >= cof->raw_syments + p->u.syment.n_value)
|
|
break;
|
|
|
|
p = cof->raw_syments + p->u.syment.n_value;
|
|
if (p > pend || p->u.syment.n_sclass != C_FILE)
|
|
break;
|
|
}
|
|
}
|
|
|
|
/* Now wander though the raw linenumbers of the section. */
|
|
/* If we have been called on this section before, and the offset we
|
|
want is further down then we can prime the lookup loop. */
|
|
sec_data = coff_section_data (abfd, section);
|
|
if (sec_data != NULL
|
|
&& sec_data->i > 0
|
|
&& offset >= sec_data->offset)
|
|
{
|
|
i = sec_data->i;
|
|
*functionname_ptr = sec_data->function;
|
|
line_base = sec_data->line_base;
|
|
}
|
|
else
|
|
{
|
|
i = 0;
|
|
line_base = 0;
|
|
}
|
|
|
|
if (section->lineno != NULL)
|
|
{
|
|
bfd_vma last_value = 0;
|
|
|
|
l = §ion->lineno[i];
|
|
|
|
for (; i < section->lineno_count; i++)
|
|
{
|
|
if (l->line_number == 0)
|
|
{
|
|
/* Get the symbol this line number points at. */
|
|
coff_symbol_type *coff = (coff_symbol_type *) (l->u.sym);
|
|
if (coff->symbol.value > offset)
|
|
break;
|
|
*functionname_ptr = coff->symbol.name;
|
|
last_value = coff->symbol.value;
|
|
if (coff->native)
|
|
{
|
|
combined_entry_type *s = coff->native;
|
|
|
|
BFD_ASSERT (s->is_sym);
|
|
s = s + 1 + s->u.syment.n_numaux;
|
|
|
|
/* In XCOFF a debugging symbol can follow the
|
|
function symbol. */
|
|
if (s->u.syment.n_scnum == N_DEBUG)
|
|
s = s + 1 + s->u.syment.n_numaux;
|
|
|
|
/* S should now point to the .bf of the function. */
|
|
if (s->u.syment.n_numaux)
|
|
{
|
|
/* The linenumber is stored in the auxent. */
|
|
union internal_auxent *a = &((s + 1)->u.auxent);
|
|
|
|
line_base = a->x_sym.x_misc.x_lnsz.x_lnno;
|
|
*line_ptr = line_base;
|
|
}
|
|
}
|
|
}
|
|
else
|
|
{
|
|
if (l->u.offset > offset)
|
|
break;
|
|
*line_ptr = l->line_number + line_base - 1;
|
|
}
|
|
l++;
|
|
}
|
|
|
|
/* If we fell off the end of the loop, then assume that this
|
|
symbol has no line number info. Otherwise, symbols with no
|
|
line number info get reported with the line number of the
|
|
last line of the last symbol which does have line number
|
|
info. We use 0x100 as a slop to account for cases where the
|
|
last line has executable code. */
|
|
if (i >= section->lineno_count
|
|
&& last_value != 0
|
|
&& offset - last_value > 0x100)
|
|
{
|
|
*functionname_ptr = NULL;
|
|
*line_ptr = 0;
|
|
}
|
|
}
|
|
|
|
/* Cache the results for the next call. */
|
|
if (sec_data == NULL && section->owner == abfd)
|
|
{
|
|
amt = sizeof (struct coff_section_tdata);
|
|
section->used_by_bfd = bfd_zalloc (abfd, amt);
|
|
sec_data = (struct coff_section_tdata *) section->used_by_bfd;
|
|
}
|
|
if (sec_data != NULL)
|
|
{
|
|
sec_data->offset = offset;
|
|
sec_data->i = i - 1;
|
|
sec_data->function = *functionname_ptr;
|
|
sec_data->line_base = line_base;
|
|
}
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
bfd_boolean
|
|
coff_find_nearest_line (bfd *abfd,
|
|
asymbol **symbols,
|
|
asection *section,
|
|
bfd_vma offset,
|
|
const char **filename_ptr,
|
|
const char **functionname_ptr,
|
|
unsigned int *line_ptr,
|
|
unsigned int *discriminator_ptr)
|
|
{
|
|
if (discriminator_ptr)
|
|
*discriminator_ptr = 0;
|
|
return coff_find_nearest_line_with_names (abfd, symbols, section, offset,
|
|
filename_ptr, functionname_ptr,
|
|
line_ptr, dwarf_debug_sections);
|
|
}
|
|
|
|
bfd_boolean
|
|
coff_find_inliner_info (bfd *abfd,
|
|
const char **filename_ptr,
|
|
const char **functionname_ptr,
|
|
unsigned int *line_ptr)
|
|
{
|
|
bfd_boolean found;
|
|
|
|
found = _bfd_dwarf2_find_inliner_info (abfd, filename_ptr,
|
|
functionname_ptr, line_ptr,
|
|
&coff_data(abfd)->dwarf2_find_line_info);
|
|
return (found);
|
|
}
|
|
|
|
int
|
|
coff_sizeof_headers (bfd *abfd, struct bfd_link_info *info)
|
|
{
|
|
size_t size;
|
|
|
|
if (!bfd_link_relocatable (info))
|
|
size = bfd_coff_filhsz (abfd) + bfd_coff_aoutsz (abfd);
|
|
else
|
|
size = bfd_coff_filhsz (abfd);
|
|
|
|
size += abfd->section_count * bfd_coff_scnhsz (abfd);
|
|
return size;
|
|
}
|
|
|
|
/* Change the class of a coff symbol held by BFD. */
|
|
|
|
bfd_boolean
|
|
bfd_coff_set_symbol_class (bfd * abfd,
|
|
asymbol * symbol,
|
|
unsigned int symbol_class)
|
|
{
|
|
coff_symbol_type * csym;
|
|
|
|
csym = coff_symbol_from (symbol);
|
|
if (csym == NULL)
|
|
{
|
|
bfd_set_error (bfd_error_invalid_operation);
|
|
return FALSE;
|
|
}
|
|
else if (csym->native == NULL)
|
|
{
|
|
/* This is an alien symbol which no native coff backend data.
|
|
We cheat here by creating a fake native entry for it and
|
|
then filling in the class. This code is based on that in
|
|
coff_write_alien_symbol(). */
|
|
|
|
combined_entry_type * native;
|
|
bfd_size_type amt = sizeof (* native);
|
|
|
|
native = (combined_entry_type *) bfd_zalloc (abfd, amt);
|
|
if (native == NULL)
|
|
return FALSE;
|
|
|
|
native->is_sym = TRUE;
|
|
native->u.syment.n_type = T_NULL;
|
|
native->u.syment.n_sclass = symbol_class;
|
|
|
|
if (bfd_is_und_section (symbol->section))
|
|
{
|
|
native->u.syment.n_scnum = N_UNDEF;
|
|
native->u.syment.n_value = symbol->value;
|
|
}
|
|
else if (bfd_is_com_section (symbol->section))
|
|
{
|
|
native->u.syment.n_scnum = N_UNDEF;
|
|
native->u.syment.n_value = symbol->value;
|
|
}
|
|
else
|
|
{
|
|
native->u.syment.n_scnum =
|
|
symbol->section->output_section->target_index;
|
|
native->u.syment.n_value = (symbol->value
|
|
+ symbol->section->output_offset);
|
|
if (! obj_pe (abfd))
|
|
native->u.syment.n_value += symbol->section->output_section->vma;
|
|
|
|
/* Copy the any flags from the file header into the symbol.
|
|
FIXME: Why? */
|
|
native->u.syment.n_flags = bfd_asymbol_bfd (& csym->symbol)->flags;
|
|
}
|
|
|
|
csym->native = native;
|
|
}
|
|
else
|
|
csym->native->u.syment.n_sclass = symbol_class;
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
bfd_boolean
|
|
_bfd_coff_section_already_linked (bfd *abfd,
|
|
asection *sec,
|
|
struct bfd_link_info *info)
|
|
{
|
|
flagword flags;
|
|
const char *name, *key;
|
|
struct bfd_section_already_linked *l;
|
|
struct bfd_section_already_linked_hash_entry *already_linked_list;
|
|
struct coff_comdat_info *s_comdat;
|
|
|
|
flags = sec->flags;
|
|
if ((flags & SEC_LINK_ONCE) == 0)
|
|
return FALSE;
|
|
|
|
/* The COFF backend linker doesn't support group sections. */
|
|
if ((flags & SEC_GROUP) != 0)
|
|
return FALSE;
|
|
|
|
name = bfd_get_section_name (abfd, sec);
|
|
s_comdat = bfd_coff_get_comdat_section (abfd, sec);
|
|
|
|
if (s_comdat != NULL)
|
|
key = s_comdat->name;
|
|
else
|
|
{
|
|
if (CONST_STRNEQ (name, ".gnu.linkonce.")
|
|
&& (key = strchr (name + sizeof (".gnu.linkonce.") - 1, '.')) != NULL)
|
|
key++;
|
|
else
|
|
/* FIXME: gcc as of 2011-09 emits sections like .text$<key>,
|
|
.xdata$<key> and .pdata$<key> only the first of which has a
|
|
comdat key. Should these all match the LTO IR key? */
|
|
key = name;
|
|
}
|
|
|
|
already_linked_list = bfd_section_already_linked_table_lookup (key);
|
|
|
|
for (l = already_linked_list->entry; l != NULL; l = l->next)
|
|
{
|
|
struct coff_comdat_info *l_comdat;
|
|
|
|
l_comdat = bfd_coff_get_comdat_section (l->sec->owner, l->sec);
|
|
|
|
/* The section names must match, and both sections must be
|
|
comdat and have the same comdat name, or both sections must
|
|
be non-comdat. LTO IR plugin sections are an exception. They
|
|
are always named .gnu.linkonce.t.<key> (<key> is some string)
|
|
and match any comdat section with comdat name of <key>, and
|
|
any linkonce section with the same suffix, ie.
|
|
.gnu.linkonce.*.<key>. */
|
|
if (((s_comdat != NULL) == (l_comdat != NULL)
|
|
&& strcmp (name, l->sec->name) == 0)
|
|
|| (l->sec->owner->flags & BFD_PLUGIN) != 0)
|
|
{
|
|
/* The section has already been linked. See if we should
|
|
issue a warning. */
|
|
return _bfd_handle_already_linked (sec, l, info);
|
|
}
|
|
}
|
|
|
|
/* This is the first section with this name. Record it. */
|
|
if (!bfd_section_already_linked_table_insert (already_linked_list, sec))
|
|
info->callbacks->einfo (_("%F%P: already_linked_table: %E\n"));
|
|
return FALSE;
|
|
}
|
|
|
|
/* Initialize COOKIE for input bfd ABFD. */
|
|
|
|
static bfd_boolean
|
|
init_reloc_cookie (struct coff_reloc_cookie *cookie,
|
|
struct bfd_link_info *info ATTRIBUTE_UNUSED,
|
|
bfd *abfd)
|
|
{
|
|
/* Sometimes the symbol table does not yet have been loaded here. */
|
|
bfd_coff_slurp_symbol_table (abfd);
|
|
|
|
cookie->abfd = abfd;
|
|
cookie->sym_hashes = obj_coff_sym_hashes (abfd);
|
|
|
|
cookie->symbols = obj_symbols (abfd);
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
/* Free the memory allocated by init_reloc_cookie, if appropriate. */
|
|
|
|
static void
|
|
fini_reloc_cookie (struct coff_reloc_cookie *cookie ATTRIBUTE_UNUSED,
|
|
bfd *abfd ATTRIBUTE_UNUSED)
|
|
{
|
|
/* Nothing to do. */
|
|
}
|
|
|
|
/* Initialize the relocation information in COOKIE for input section SEC
|
|
of input bfd ABFD. */
|
|
|
|
static bfd_boolean
|
|
init_reloc_cookie_rels (struct coff_reloc_cookie *cookie,
|
|
struct bfd_link_info *info ATTRIBUTE_UNUSED,
|
|
bfd *abfd,
|
|
asection *sec)
|
|
{
|
|
if (sec->reloc_count == 0)
|
|
{
|
|
cookie->rels = NULL;
|
|
cookie->relend = NULL;
|
|
cookie->rel = NULL;
|
|
return TRUE;
|
|
}
|
|
|
|
cookie->rels = _bfd_coff_read_internal_relocs (abfd, sec, FALSE, NULL, 0, NULL);
|
|
|
|
if (cookie->rels == NULL)
|
|
return FALSE;
|
|
|
|
cookie->rel = cookie->rels;
|
|
cookie->relend = (cookie->rels + sec->reloc_count);
|
|
return TRUE;
|
|
}
|
|
|
|
/* Free the memory allocated by init_reloc_cookie_rels,
|
|
if appropriate. */
|
|
|
|
static void
|
|
fini_reloc_cookie_rels (struct coff_reloc_cookie *cookie,
|
|
asection *sec)
|
|
{
|
|
if (cookie->rels
|
|
/* PR 20401. The relocs may not have been cached, so check first.
|
|
If the relocs were loaded by init_reloc_cookie_rels() then this
|
|
will be the case. FIXME: Would performance be improved if the
|
|
relocs *were* cached ? */
|
|
&& coff_section_data (NULL, sec)
|
|
&& coff_section_data (NULL, sec)->relocs != cookie->rels)
|
|
free (cookie->rels);
|
|
}
|
|
|
|
/* Initialize the whole of COOKIE for input section SEC. */
|
|
|
|
static bfd_boolean
|
|
init_reloc_cookie_for_section (struct coff_reloc_cookie *cookie,
|
|
struct bfd_link_info *info,
|
|
asection *sec)
|
|
{
|
|
if (!init_reloc_cookie (cookie, info, sec->owner))
|
|
return FALSE;
|
|
|
|
if (!init_reloc_cookie_rels (cookie, info, sec->owner, sec))
|
|
{
|
|
fini_reloc_cookie (cookie, sec->owner);
|
|
return FALSE;
|
|
}
|
|
return TRUE;
|
|
}
|
|
|
|
/* Free the memory allocated by init_reloc_cookie_for_section,
|
|
if appropriate. */
|
|
|
|
static void
|
|
fini_reloc_cookie_for_section (struct coff_reloc_cookie *cookie,
|
|
asection *sec)
|
|
{
|
|
fini_reloc_cookie_rels (cookie, sec);
|
|
fini_reloc_cookie (cookie, sec->owner);
|
|
}
|
|
|
|
static asection *
|
|
_bfd_coff_gc_mark_hook (asection *sec,
|
|
struct bfd_link_info *info ATTRIBUTE_UNUSED,
|
|
struct internal_reloc *rel ATTRIBUTE_UNUSED,
|
|
struct coff_link_hash_entry *h,
|
|
struct internal_syment *sym)
|
|
{
|
|
if (h != NULL)
|
|
{
|
|
switch (h->root.type)
|
|
{
|
|
case bfd_link_hash_defined:
|
|
case bfd_link_hash_defweak:
|
|
return h->root.u.def.section;
|
|
|
|
case bfd_link_hash_common:
|
|
return h->root.u.c.p->section;
|
|
|
|
case bfd_link_hash_undefined:
|
|
case bfd_link_hash_undefweak:
|
|
default:
|
|
break;
|
|
}
|
|
return NULL;
|
|
}
|
|
|
|
return coff_section_from_bfd_index (sec->owner, sym->n_scnum);
|
|
}
|
|
|
|
/* COOKIE->rel describes a relocation against section SEC, which is
|
|
a section we've decided to keep. Return the section that contains
|
|
the relocation symbol, or NULL if no section contains it. */
|
|
|
|
static asection *
|
|
_bfd_coff_gc_mark_rsec (struct bfd_link_info *info, asection *sec,
|
|
coff_gc_mark_hook_fn gc_mark_hook,
|
|
struct coff_reloc_cookie *cookie)
|
|
{
|
|
struct coff_link_hash_entry *h;
|
|
|
|
h = cookie->sym_hashes[cookie->rel->r_symndx];
|
|
if (h != NULL)
|
|
{
|
|
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;
|
|
|
|
return (*gc_mark_hook) (sec, info, cookie->rel, h, NULL);
|
|
}
|
|
|
|
return (*gc_mark_hook) (sec, info, cookie->rel, NULL,
|
|
&(cookie->symbols
|
|
+ obj_convert (sec->owner)[cookie->rel->r_symndx])->native->u.syment);
|
|
}
|
|
|
|
static bfd_boolean _bfd_coff_gc_mark
|
|
(struct bfd_link_info *, asection *, coff_gc_mark_hook_fn);
|
|
|
|
/* COOKIE->rel describes a relocation against section SEC, which is
|
|
a section we've decided to keep. Mark the section that contains
|
|
the relocation symbol. */
|
|
|
|
static bfd_boolean
|
|
_bfd_coff_gc_mark_reloc (struct bfd_link_info *info,
|
|
asection *sec,
|
|
coff_gc_mark_hook_fn gc_mark_hook,
|
|
struct coff_reloc_cookie *cookie)
|
|
{
|
|
asection *rsec;
|
|
|
|
rsec = _bfd_coff_gc_mark_rsec (info, sec, gc_mark_hook, cookie);
|
|
if (rsec && !rsec->gc_mark)
|
|
{
|
|
if (bfd_get_flavour (rsec->owner) != bfd_target_coff_flavour)
|
|
rsec->gc_mark = 1;
|
|
else if (!_bfd_coff_gc_mark (info, rsec, gc_mark_hook))
|
|
return FALSE;
|
|
}
|
|
return TRUE;
|
|
}
|
|
|
|
/* The mark phase of garbage collection. For a given section, mark
|
|
it and any sections in this section's group, and all the sections
|
|
which define symbols to which it refers. */
|
|
|
|
static bfd_boolean
|
|
_bfd_coff_gc_mark (struct bfd_link_info *info,
|
|
asection *sec,
|
|
coff_gc_mark_hook_fn gc_mark_hook)
|
|
{
|
|
bfd_boolean ret = TRUE;
|
|
|
|
sec->gc_mark = 1;
|
|
|
|
/* Look through the section relocs. */
|
|
if ((sec->flags & SEC_RELOC) != 0
|
|
&& sec->reloc_count > 0)
|
|
{
|
|
struct coff_reloc_cookie cookie;
|
|
|
|
if (!init_reloc_cookie_for_section (&cookie, info, sec))
|
|
ret = FALSE;
|
|
else
|
|
{
|
|
for (; cookie.rel < cookie.relend; cookie.rel++)
|
|
{
|
|
if (!_bfd_coff_gc_mark_reloc (info, sec, gc_mark_hook, &cookie))
|
|
{
|
|
ret = FALSE;
|
|
break;
|
|
}
|
|
}
|
|
fini_reloc_cookie_for_section (&cookie, sec);
|
|
}
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
static bfd_boolean
|
|
_bfd_coff_gc_mark_extra_sections (struct bfd_link_info *info,
|
|
coff_gc_mark_hook_fn mark_hook ATTRIBUTE_UNUSED)
|
|
{
|
|
bfd *ibfd;
|
|
|
|
for (ibfd = info->input_bfds; ibfd != NULL; ibfd = ibfd->link.next)
|
|
{
|
|
asection *isec;
|
|
bfd_boolean some_kept;
|
|
|
|
if (bfd_get_flavour (ibfd) != bfd_target_elf_flavour)
|
|
continue;
|
|
|
|
/* Ensure all linker created sections are kept, and see whether
|
|
any other section is already marked. */
|
|
some_kept = FALSE;
|
|
for (isec = ibfd->sections; isec != NULL; isec = isec->next)
|
|
{
|
|
if ((isec->flags & SEC_LINKER_CREATED) != 0)
|
|
isec->gc_mark = 1;
|
|
else if (isec->gc_mark)
|
|
some_kept = TRUE;
|
|
}
|
|
|
|
/* If no section in this file will be kept, then we can
|
|
toss out debug sections. */
|
|
if (!some_kept)
|
|
continue;
|
|
|
|
/* Keep debug and special sections like .comment when they are
|
|
not part of a group, or when we have single-member groups. */
|
|
for (isec = ibfd->sections; isec != NULL; isec = isec->next)
|
|
if ((isec->flags & SEC_DEBUGGING) != 0
|
|
|| (isec->flags & (SEC_ALLOC | SEC_LOAD | SEC_RELOC)) == 0)
|
|
isec->gc_mark = 1;
|
|
}
|
|
return TRUE;
|
|
}
|
|
|
|
/* Sweep symbols in swept sections. Called via coff_link_hash_traverse. */
|
|
|
|
static bfd_boolean
|
|
coff_gc_sweep_symbol (struct coff_link_hash_entry *h,
|
|
void *data ATTRIBUTE_UNUSED)
|
|
{
|
|
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_defined
|
|
|| h->root.type == bfd_link_hash_defweak)
|
|
&& !h->root.u.def.section->gc_mark
|
|
&& !(h->root.u.def.section->owner->flags & DYNAMIC))
|
|
{
|
|
/* Do our best to hide the symbol. */
|
|
h->root.u.def.section = bfd_und_section_ptr;
|
|
h->symbol_class = C_HIDDEN;
|
|
}
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
/* The sweep phase of garbage collection. Remove all garbage sections. */
|
|
|
|
typedef bfd_boolean (*gc_sweep_hook_fn)
|
|
(bfd *, struct bfd_link_info *, asection *, const struct internal_reloc *);
|
|
|
|
static bfd_boolean
|
|
coff_gc_sweep (bfd *abfd ATTRIBUTE_UNUSED, struct bfd_link_info *info)
|
|
{
|
|
bfd *sub;
|
|
|
|
for (sub = info->input_bfds; sub != NULL; sub = sub->link.next)
|
|
{
|
|
asection *o;
|
|
|
|
if (bfd_get_flavour (sub) != bfd_target_coff_flavour)
|
|
continue;
|
|
|
|
for (o = sub->sections; o != NULL; o = o->next)
|
|
{
|
|
/* Keep debug and special sections. */
|
|
if ((o->flags & (SEC_DEBUGGING | SEC_LINKER_CREATED)) != 0
|
|
|| (o->flags & (SEC_ALLOC | SEC_LOAD | SEC_RELOC)) == 0)
|
|
o->gc_mark = 1;
|
|
else if (CONST_STRNEQ (o->name, ".idata")
|
|
|| CONST_STRNEQ (o->name, ".pdata")
|
|
|| CONST_STRNEQ (o->name, ".xdata")
|
|
|| CONST_STRNEQ (o->name, ".rsrc"))
|
|
o->gc_mark = 1;
|
|
|
|
if (o->gc_mark)
|
|
continue;
|
|
|
|
/* Skip sweeping sections already excluded. */
|
|
if (o->flags & SEC_EXCLUDE)
|
|
continue;
|
|
|
|
/* Since this is early in the link process, it is simple
|
|
to remove a section from the output. */
|
|
o->flags |= SEC_EXCLUDE;
|
|
|
|
if (info->print_gc_sections && o->size != 0)
|
|
_bfd_error_handler (_("Removing unused section '%s' in file '%B'"), sub, o->name);
|
|
|
|
#if 0
|
|
/* But we also have to update some of the relocation
|
|
info we collected before. */
|
|
if (gc_sweep_hook
|
|
&& (o->flags & SEC_RELOC) != 0
|
|
&& o->reloc_count > 0
|
|
&& !bfd_is_abs_section (o->output_section))
|
|
{
|
|
struct internal_reloc *internal_relocs;
|
|
bfd_boolean r;
|
|
|
|
internal_relocs
|
|
= _bfd_coff_link_read_relocs (o->owner, o, NULL, NULL,
|
|
info->keep_memory);
|
|
if (internal_relocs == NULL)
|
|
return FALSE;
|
|
|
|
r = (*gc_sweep_hook) (o->owner, info, o, internal_relocs);
|
|
|
|
if (coff_section_data (o)->relocs != internal_relocs)
|
|
free (internal_relocs);
|
|
|
|
if (!r)
|
|
return FALSE;
|
|
}
|
|
#endif
|
|
}
|
|
}
|
|
|
|
/* Remove the symbols that were in the swept sections from the dynamic
|
|
symbol table. */
|
|
coff_link_hash_traverse (coff_hash_table (info), coff_gc_sweep_symbol,
|
|
NULL);
|
|
|
|
return TRUE;
|
|
}
|
|
|
|
/* Keep all sections containing symbols undefined on the command-line,
|
|
and the section containing the entry symbol. */
|
|
|
|
static void
|
|
_bfd_coff_gc_keep (struct bfd_link_info *info)
|
|
{
|
|
struct bfd_sym_chain *sym;
|
|
|
|
for (sym = info->gc_sym_list; sym != NULL; sym = sym->next)
|
|
{
|
|
struct coff_link_hash_entry *h;
|
|
|
|
h = coff_link_hash_lookup (coff_hash_table (info), sym->name,
|
|
FALSE, FALSE, FALSE);
|
|
|
|
if (h != NULL
|
|
&& (h->root.type == bfd_link_hash_defined
|
|
|| h->root.type == bfd_link_hash_defweak)
|
|
&& !bfd_is_abs_section (h->root.u.def.section))
|
|
h->root.u.def.section->flags |= SEC_KEEP;
|
|
}
|
|
}
|
|
|
|
/* Do mark and sweep of unused sections. */
|
|
|
|
bfd_boolean
|
|
bfd_coff_gc_sections (bfd *abfd ATTRIBUTE_UNUSED, struct bfd_link_info *info)
|
|
{
|
|
bfd *sub;
|
|
|
|
/* FIXME: Should we implement this? */
|
|
#if 0
|
|
const bfd_coff_backend_data *bed = coff_backend_info (abfd);
|
|
|
|
if (!bed->can_gc_sections
|
|
|| !is_coff_hash_table (info->hash))
|
|
{
|
|
(*_bfd_error_handler)(_("Warning: gc-sections option ignored"));
|
|
return TRUE;
|
|
}
|
|
#endif
|
|
|
|
_bfd_coff_gc_keep (info);
|
|
|
|
/* Grovel through relocs to find out who stays ... */
|
|
for (sub = info->input_bfds; sub != NULL; sub = sub->link.next)
|
|
{
|
|
asection *o;
|
|
|
|
if (bfd_get_flavour (sub) != bfd_target_coff_flavour)
|
|
continue;
|
|
|
|
for (o = sub->sections; o != NULL; o = o->next)
|
|
{
|
|
if (((o->flags & (SEC_EXCLUDE | SEC_KEEP)) == SEC_KEEP
|
|
|| CONST_STRNEQ (o->name, ".vectors")
|
|
|| CONST_STRNEQ (o->name, ".ctors")
|
|
|| CONST_STRNEQ (o->name, ".dtors"))
|
|
&& !o->gc_mark)
|
|
{
|
|
if (!_bfd_coff_gc_mark (info, o, _bfd_coff_gc_mark_hook))
|
|
return FALSE;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Allow the backend to mark additional target specific sections. */
|
|
_bfd_coff_gc_mark_extra_sections (info, _bfd_coff_gc_mark_hook);
|
|
|
|
/* ... and mark SEC_EXCLUDE for those that go. */
|
|
return coff_gc_sweep (abfd, info);
|
|
}
|