2
0
mirror of https://sourceware.org/git/binutils-gdb.git synced 2024-12-21 04:42:53 +08:00
binutils-gdb/ld/ldwrite.c
Alan Modra 22216541c1 PR13616, linker should pad executable sections with nops, not zeros
This implements padding of orphan executable sections for PowerPC.
Of course, the simple implementation of bfd_arch_ppc_nop_fill and
removing the NOP definition didn't work, with powerpc64 hitting a
testsuite failure linking to S-records.  That's because the srec
target is BFD_ENDIAN_UNKNOWN so the test of bfd_big_endian (abfd) in
default_data_link_order therefore returned false, resulting in a
little-endian nop pattern.  The rest of the patch fixes that problem
by adding a new field to bfd_link_info that can be used to determine
actual endianness on targets like srec.

	PR 13616
include/
	* bfdlink.h (struct bfd_link_info <big_endian>): New field.
bfd/
	* cpu-powerpc.c (bfd_arch_ppc_nop_fill): New function, use it
	for all ppc arch info.
	* linker.c (default_data_link_order): Pass info->big_endian to
	arch_info->fill function.
ld/
	* emulparams/elf64lppc.sh (NOP): Don't define.
	* emulparams/elf64ppc.sh (NOP): Don't define.
	* ldwrite.c (build_link_order): Use link_info.big_endian.  Move
	code determining endian to use for data_statement to..
	* ldemul.c (after_open_default): ..here.  Set link_info.big_endian.
2019-10-16 23:07:27 +10:30

557 lines
15 KiB
C

/* ldwrite.c -- write out the linked file
Copyright (C) 1991-2019 Free Software Foundation, Inc.
Written by Steve Chamberlain sac@cygnus.com
This file is part of the GNU Binutils.
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. */
#include "sysdep.h"
#include "bfd.h"
#include "bfdlink.h"
#include "libiberty.h"
#include "ctf-api.h"
#include "safe-ctype.h"
#include "ld.h"
#include "ldexp.h"
#include "ldlang.h"
#include "ldwrite.h"
#include "ldmisc.h"
#include <ldgram.h>
#include "ldmain.h"
/* Build link_order structures for the BFD linker. */
static void
build_link_order (lang_statement_union_type *statement)
{
switch (statement->header.type)
{
case lang_data_statement_enum:
{
asection *output_section;
struct bfd_link_order *link_order;
bfd_vma value;
output_section = statement->data_statement.output_section;
ASSERT (output_section->owner == link_info.output_bfd);
if (!((output_section->flags & SEC_HAS_CONTENTS) != 0
|| ((output_section->flags & SEC_LOAD) != 0
&& (output_section->flags & SEC_THREAD_LOCAL))))
break;
link_order = bfd_new_link_order (link_info.output_bfd, output_section);
if (link_order == NULL)
einfo (_("%F%P: bfd_new_link_order failed\n"));
link_order->type = bfd_data_link_order;
link_order->offset = statement->data_statement.output_offset;
link_order->u.data.contents = (bfd_byte *) xmalloc (QUAD_SIZE);
value = statement->data_statement.value;
/* By convention, the bfd_put routines for an unknown
endianness are big endian, so we must swap here if the
input is little endian. */
if (!bfd_big_endian (link_info.output_bfd)
&& !bfd_little_endian (link_info.output_bfd)
&& !link_info.big_endian)
{
bfd_byte buffer[8];
switch (statement->data_statement.type)
{
case QUAD:
case SQUAD:
if (sizeof (bfd_vma) >= QUAD_SIZE)
{
bfd_putl64 (value, buffer);
value = bfd_getb64 (buffer);
break;
}
/* Fall through. */
case LONG:
bfd_putl32 (value, buffer);
value = bfd_getb32 (buffer);
break;
case SHORT:
bfd_putl16 (value, buffer);
value = bfd_getb16 (buffer);
break;
case BYTE:
break;
default:
abort ();
}
}
ASSERT (output_section->owner == link_info.output_bfd);
switch (statement->data_statement.type)
{
case QUAD:
case SQUAD:
if (sizeof (bfd_vma) >= QUAD_SIZE)
bfd_put_64 (link_info.output_bfd, value,
link_order->u.data.contents);
else
{
bfd_vma high;
if (statement->data_statement.type == QUAD)
high = 0;
else if ((value & 0x80000000) == 0)
high = 0;
else
high = (bfd_vma) -1;
bfd_put_32 (link_info.output_bfd, high,
(link_order->u.data.contents
+ (link_info.big_endian ? 0 : 4)));
bfd_put_32 (link_info.output_bfd, value,
(link_order->u.data.contents
+ (link_info.big_endian ? 4 : 0)));
}
link_order->size = QUAD_SIZE;
break;
case LONG:
bfd_put_32 (link_info.output_bfd, value,
link_order->u.data.contents);
link_order->size = LONG_SIZE;
break;
case SHORT:
bfd_put_16 (link_info.output_bfd, value,
link_order->u.data.contents);
link_order->size = SHORT_SIZE;
break;
case BYTE:
bfd_put_8 (link_info.output_bfd, value,
link_order->u.data.contents);
link_order->size = BYTE_SIZE;
break;
default:
abort ();
}
link_order->u.data.size = link_order->size;
}
break;
case lang_reloc_statement_enum:
{
lang_reloc_statement_type *rs;
asection *output_section;
struct bfd_link_order *link_order;
rs = &statement->reloc_statement;
output_section = rs->output_section;
ASSERT (output_section->owner == link_info.output_bfd);
if (!((output_section->flags & SEC_HAS_CONTENTS) != 0
|| ((output_section->flags & SEC_LOAD) != 0
&& (output_section->flags & SEC_THREAD_LOCAL))))
break;
link_order = bfd_new_link_order (link_info.output_bfd, output_section);
if (link_order == NULL)
einfo (_("%F%P: bfd_new_link_order failed\n"));
link_order->offset = rs->output_offset;
link_order->size = bfd_get_reloc_size (rs->howto);
link_order->u.reloc.p = (struct bfd_link_order_reloc *)
xmalloc (sizeof (struct bfd_link_order_reloc));
link_order->u.reloc.p->reloc = rs->reloc;
link_order->u.reloc.p->addend = rs->addend_value;
if (rs->name == NULL)
{
link_order->type = bfd_section_reloc_link_order;
if (rs->section->owner == link_info.output_bfd)
link_order->u.reloc.p->u.section = rs->section;
else
{
link_order->u.reloc.p->u.section = rs->section->output_section;
link_order->u.reloc.p->addend += rs->section->output_offset;
}
}
else
{
link_order->type = bfd_symbol_reloc_link_order;
link_order->u.reloc.p->u.name = rs->name;
}
}
break;
case lang_input_section_enum:
{
/* Create a new link_order in the output section with this
attached */
asection *i = statement->input_section.section;
if (i->sec_info_type != SEC_INFO_TYPE_JUST_SYMS
&& (i->flags & SEC_EXCLUDE) == 0)
{
asection *output_section = i->output_section;
struct bfd_link_order *link_order;
ASSERT (output_section->owner == link_info.output_bfd);
if (!((output_section->flags & SEC_HAS_CONTENTS) != 0
|| ((output_section->flags & SEC_LOAD) != 0
&& (output_section->flags & SEC_THREAD_LOCAL))))
break;
link_order = bfd_new_link_order (link_info.output_bfd,
output_section);
if (link_order == NULL)
einfo (_("%F%P: bfd_new_link_order failed\n"));
if ((i->flags & SEC_NEVER_LOAD) != 0
&& (i->flags & SEC_DEBUGGING) == 0)
{
/* We've got a never load section inside one which is
going to be output, we'll change it into a fill. */
link_order->type = bfd_data_link_order;
link_order->u.data.contents = (unsigned char *) "";
link_order->u.data.size = 1;
}
else
{
link_order->type = bfd_indirect_link_order;
link_order->u.indirect.section = i;
ASSERT (i->output_section == output_section);
}
link_order->size = i->size;
link_order->offset = i->output_offset;
}
}
break;
case lang_padding_statement_enum:
/* Make a new link_order with the right filler */
{
asection *output_section;
struct bfd_link_order *link_order;
output_section = statement->padding_statement.output_section;
ASSERT (statement->padding_statement.output_section->owner
== link_info.output_bfd);
if (!((output_section->flags & SEC_HAS_CONTENTS) != 0
|| ((output_section->flags & SEC_LOAD) != 0
&& (output_section->flags & SEC_THREAD_LOCAL))))
break;
link_order = bfd_new_link_order (link_info.output_bfd,
output_section);
if (link_order == NULL)
einfo (_("%F%P: bfd_new_link_order failed\n"));
link_order->type = bfd_data_link_order;
link_order->size = statement->padding_statement.size;
link_order->offset = statement->padding_statement.output_offset;
link_order->u.data.contents = statement->padding_statement.fill->data;
link_order->u.data.size = statement->padding_statement.fill->size;
}
break;
default:
/* All the other ones fall through */
break;
}
}
/* Return true if NAME is the name of an unsplittable section. These
are the stabs strings, dwarf strings. */
static bfd_boolean
unsplittable_name (const char *name)
{
if (CONST_STRNEQ (name, ".stab"))
{
/* There are several stab like string sections. We pattern match on
".stab...str" */
unsigned len = strlen (name);
if (strcmp (&name[len-3], "str") == 0)
return TRUE;
}
else if (strcmp (name, "$GDB_STRINGS$") == 0)
return TRUE;
return FALSE;
}
/* Wander around the input sections, make sure that
we'll never try and create an output section with more relocs
than will fit.. Do this by always assuming the worst case, and
creating new output sections with all the right bits. */
#define TESTIT 1
static asection *
clone_section (bfd *abfd, asection *s, const char *name, int *count)
{
char *tname;
char *sname;
unsigned int len;
asection *n;
struct bfd_link_hash_entry *h;
/* Invent a section name from the section name and a dotted numeric
suffix. */
len = strlen (name);
tname = (char *) xmalloc (len + 1);
memcpy (tname, name, len + 1);
/* Remove a dotted number suffix, from a previous split link. */
while (len && ISDIGIT (tname[len-1]))
len--;
if (len > 1 && tname[len-1] == '.')
/* It was a dotted number. */
tname[len-1] = 0;
/* We want to use the whole of the original section name for the
split name, but coff can be restricted to 8 character names. */
if (bfd_family_coff (abfd) && strlen (tname) > 5)
{
/* Some section names cannot be truncated, as the name is
used to locate some other section. */
if (CONST_STRNEQ (name, ".stab")
|| strcmp (name, "$GDB_SYMBOLS$") == 0)
{
einfo (_ ("%F%P: cannot create split section name for %s\n"), name);
/* Silence gcc warnings. einfo exits, so we never reach here. */
return NULL;
}
tname[5] = 0;
}
if ((sname = bfd_get_unique_section_name (abfd, tname, count)) == NULL
|| (n = bfd_make_section_anyway (abfd, sname)) == NULL
|| (h = bfd_link_hash_lookup (link_info.hash,
sname, TRUE, TRUE, FALSE)) == NULL)
{
einfo (_("%F%P: clone section failed: %E\n"));
/* Silence gcc warnings. einfo exits, so we never reach here. */
return NULL;
}
free (tname);
/* Set up section symbol. */
h->type = bfd_link_hash_defined;
h->u.def.value = 0;
h->u.def.section = n;
n->flags = s->flags;
n->vma = s->vma;
n->user_set_vma = s->user_set_vma;
n->lma = s->lma;
n->size = 0;
n->output_offset = s->output_offset;
n->output_section = n;
n->orelocation = 0;
n->reloc_count = 0;
n->alignment_power = s->alignment_power;
bfd_copy_private_section_data (abfd, s, abfd, n);
return n;
}
#if TESTING
static void
ds (asection *s)
{
struct bfd_link_order *l = s->map_head.link_order;
printf ("vma %x size %x\n", s->vma, s->size);
while (l)
{
if (l->type == bfd_indirect_link_order)
printf ("%8x %s\n", l->offset, l->u.indirect.section->owner->filename);
else
printf (_("%8x something else\n"), l->offset);
l = l->next;
}
printf ("\n");
}
dump (char *s, asection *a1, asection *a2)
{
printf ("%s\n", s);
ds (a1);
ds (a2);
}
static void
sanity_check (bfd *abfd)
{
asection *s;
for (s = abfd->sections; s; s = s->next)
{
struct bfd_link_order *p;
bfd_vma prev = 0;
for (p = s->map_head.link_order; p; p = p->next)
{
if (p->offset > 100000)
abort ();
if (p->offset < prev)
abort ();
prev = p->offset;
}
}
}
#else
#define sanity_check(a)
#define dump(a, b, c)
#endif
static void
split_sections (bfd *abfd, struct bfd_link_info *info)
{
asection *original_sec;
int nsecs = abfd->section_count;
sanity_check (abfd);
/* Look through all the original sections. */
for (original_sec = abfd->sections;
original_sec && nsecs;
original_sec = original_sec->next, nsecs--)
{
int count = 0;
unsigned int lines = 0;
unsigned int relocs = 0;
bfd_size_type sec_size = 0;
struct bfd_link_order *l;
struct bfd_link_order *p;
bfd_vma vma = original_sec->vma;
asection *cursor = original_sec;
/* Count up the relocations and line entries to see if anything
would be too big to fit. Accumulate section size too. */
for (l = NULL, p = cursor->map_head.link_order; p != NULL; p = l->next)
{
unsigned int thislines = 0;
unsigned int thisrelocs = 0;
bfd_size_type thissize = 0;
if (p->type == bfd_indirect_link_order)
{
asection *sec;
sec = p->u.indirect.section;
if (info->strip == strip_none
|| info->strip == strip_some)
thislines = sec->lineno_count;
if (bfd_link_relocatable (info))
thisrelocs = sec->reloc_count;
thissize = sec->size;
}
else if (bfd_link_relocatable (info)
&& (p->type == bfd_section_reloc_link_order
|| p->type == bfd_symbol_reloc_link_order))
thisrelocs++;
if (l != NULL
&& (thisrelocs + relocs >= config.split_by_reloc
|| thislines + lines >= config.split_by_reloc
|| (thissize + sec_size >= config.split_by_file))
&& !unsplittable_name (cursor->name))
{
/* Create a new section and put this link order and the
following link orders into it. */
bfd_vma shift_offset;
asection *n;
n = clone_section (abfd, cursor, original_sec->name, &count);
/* Attach the link orders to the new section and snip
them off from the old section. */
n->map_head.link_order = p;
n->map_tail.link_order = cursor->map_tail.link_order;
cursor->map_tail.link_order = l;
l->next = NULL;
l = p;
/* Change the size of the original section and
update the vma of the new one. */
dump ("before snip", cursor, n);
shift_offset = p->offset;
n->size = cursor->size - shift_offset;
cursor->size = shift_offset;
vma += shift_offset;
n->lma = n->vma = vma;
/* Run down the chain and change the output section to
the right one, update the offsets too. */
do
{
p->offset -= shift_offset;
if (p->type == bfd_indirect_link_order)
{
p->u.indirect.section->output_section = n;
p->u.indirect.section->output_offset = p->offset;
}
p = p->next;
}
while (p);
dump ("after snip", cursor, n);
cursor = n;
relocs = thisrelocs;
lines = thislines;
sec_size = thissize;
}
else
{
l = p;
relocs += thisrelocs;
lines += thislines;
sec_size += thissize;
}
}
}
sanity_check (abfd);
}
/* Call BFD to write out the linked file. */
void
ldwrite (void)
{
/* Reset error indicator, which can typically something like invalid
format from opening up the .o files. */
bfd_set_error (bfd_error_no_error);
lang_clear_os_map ();
lang_for_each_statement (build_link_order);
if (config.split_by_reloc != (unsigned) -1
|| config.split_by_file != (bfd_size_type) -1)
split_sections (link_info.output_bfd, &link_info);
if (!bfd_final_link (link_info.output_bfd, &link_info))
{
/* If there was an error recorded, print it out. Otherwise assume
an appropriate error message like unknown symbol was printed
out. */
if (bfd_get_error () != bfd_error_no_error)
einfo (_("%F%P: final link failed: %E\n"));
else
xexit (1);
}
}