From: Christian Eggers <ceggers@gmx.de>
To: binutils@sourceware.org, Nick Clifton <nickc@redhat.com>
Cc: amodra@gmail.com, Christian Eggers <ceggers@gmx.de>
Subject: [PATCH v2 2/2] Fix several mix up between octets and bytes in ELF program headers
Date: Mon, 02 Mar 2020 20:17:00 -0000 [thread overview]
Message-ID: <20200302201524.7398-1-ceggers@gmx.de> (raw)
In-Reply-To: <bdd71737-fa24-da30-300e-4fd95493aa93@redhat.com>
Hi Nick,
On Monday, March 2nd 2020, Nick Clifton wrote:
> Approved - please apply - but please fixup the comments first.
Comments changed, thanks.
Christian
---
This is a follow-up patch for my patch from december 09 with the same
subject. It fixes additional locations not handled in the first patch.
When converting between addresses in ELF headers [octets] and bfd
LMA/VMA [bytes], the number of octets per byte needs to be incorperated.
Patch changelog:
v2:
- GNU coding style for comments.
include/
* bfdlink.h (struct bfd_link_order): Add unit (bytes/octets) to
offset and size members.
* elf/internal.h (struct elf_internal_phdr): likewise for
p_align member.
(struct elf_segment_map): likewise for p_paddr and p_size
members
bfd/
* bfd.c (bfd_record_phdr): New local "opb". Fix assignment of
"p_paddr" from "at".
* elfcode.h (bfd_from_remote_memory): Add units to several
parameters. New local "opb". Fix usage of p_align. Fix
calculation of "localbase" from "ehdr_vma" and "p_vaddr". Fix
call of target_read_memory().
* elflink.c (elf_fixup_link_order): Fix scope of "s" local. Fix
calculation of "offset" and "output_offset".
(bfd_elf_final_link): New local "opb". Fix calculation of "size"
from "offset" and fix calculation of "end" from "vma+size". Fix
comparison between "sh_addr" and "vma"/"output_offset".
(bfd_elf_discard_info): Fix calculation of "eh_alignment".
* elf-bfd.h (struct elf_link_hash_table): Add unit to tls_size
member.
* elf.c (_bfd_elf_map_sections_to_segments): Add unit (bytes/
octets) to "wrap_to2 and "phdr_size" locals. Fix calculation of
"wrap_to" value. Add unit (bytes) to phdr_lma variable. Fix
assignment of p_paddr from phdr_lma. Fix comparison between "lma
+size" and "next->lma".
(elf_sort_segments): Fix assignment from p_paddr to lma.
(assign_file_positions_for_load_sections): Add unit (bytes) to
local "align". Fix calculation of local "off_adjust". Fix
calculation of local "filehdr_vaddr".
(assign_file_positions_for_non_load_sections): New local "opb".
Fix calculation of "end" from "p_size". Fix comparison between
"vma+SECTION_SIZE()" and "start". Fix calculation of "p_memsz"
from "end" and "p_vaddr".
(rewrite_elf_program_header): Fix comparison between p_vaddr and
vma. Fix assignment to p_paddr from lma. Fix comparison between
p_paddr and lma. Fix assignment to p_paddr from lma.
* merge.c (sec_merge_emit): New local "opb". Convert
"alignment_power" to octets.
(_bfd_add_merge_section): New locals "alignment_power" and
"opb". Fix comparison between "alignment_power" and
"sizeof(align)".
(_bfd_merge_sections): New local "opb". Divide size by opb
before checking align mask.
Signed-off-by: Christian Eggers <ceggers@gmx.de>
---
bfd/bfd.c | 5 +++--
bfd/elf-bfd.h | 2 +-
bfd/elf.c | 54 +++++++++++++++++++++++++-------------------------
bfd/elfcode.h | 28 ++++++++++++++------------
bfd/elflink.c | 31 +++++++++++++++++------------
bfd/merge.c | 22 ++++++++++++--------
include/bfdlink.h | 4 ++--
include/elf/internal.h | 7 ++++---
8 files changed, 84 insertions(+), 69 deletions(-)
diff --git a/bfd/bfd.c b/bfd/bfd.c
index 1c1238c036a..100359ccfe6 100644
--- a/bfd/bfd.c
+++ b/bfd/bfd.c
@@ -2168,7 +2168,7 @@ bfd_record_phdr (bfd *abfd,
bfd_boolean flags_valid,
flagword flags,
bfd_boolean at_valid,
- bfd_vma at,
+ bfd_vma at, /* Bytes. */
bfd_boolean includes_filehdr,
bfd_boolean includes_phdrs,
unsigned int count,
@@ -2176,6 +2176,7 @@ bfd_record_phdr (bfd *abfd,
{
struct elf_segment_map *m, **pm;
size_t amt;
+ unsigned int opb = bfd_octets_per_byte (abfd, NULL);
if (bfd_get_flavour (abfd) != bfd_target_elf_flavour)
return TRUE;
@@ -2188,7 +2189,7 @@ bfd_record_phdr (bfd *abfd,
m->p_type = type;
m->p_flags = flags;
- m->p_paddr = at;
+ m->p_paddr = at * opb;
m->p_flags_valid = flags_valid;
m->p_paddr_valid = at_valid;
m->includes_filehdr = includes_filehdr;
diff --git a/bfd/elf-bfd.h b/bfd/elf-bfd.h
index 38a9aa06019..54c27a1b76c 100644
--- a/bfd/elf-bfd.h
+++ b/bfd/elf-bfd.h
@@ -635,7 +635,7 @@ struct elf_link_hash_table
/* Cached first output tls section and size of PT_TLS segment. */
asection *tls_sec;
- bfd_size_type tls_size;
+ bfd_size_type tls_size; /* Bytes. */
/* A linked list of dynamic BFD's loaded in the link. */
struct elf_link_loaded_list *dyn_loaded;
diff --git a/bfd/elf.c b/bfd/elf.c
index a329e239828..bc75b96adcb 100644
--- a/bfd/elf.c
+++ b/bfd/elf.c
@@ -4665,8 +4665,8 @@ _bfd_elf_map_sections_to_segments (bfd *abfd, struct bfd_link_info *info)
asection *first_mbind = NULL;
asection *dynsec, *eh_frame_hdr;
size_t amt;
- bfd_vma addr_mask, wrap_to = 0;
- bfd_size_type phdr_size;
+ bfd_vma addr_mask, wrap_to = 0; /* Bytes. */
+ bfd_size_type phdr_size; /* Octets/bytes. */
unsigned int opb = bfd_octets_per_byte (abfd, NULL);
/* Select the allocated sections, and sort them. */
@@ -4694,8 +4694,8 @@ _bfd_elf_map_sections_to_segments (bfd *abfd, struct bfd_link_info *info)
sections[i] = s;
++i;
/* A wrapping section potentially clashes with header. */
- if (((s->lma + s->size) & addr_mask) < (s->lma & addr_mask))
- wrap_to = (s->lma + s->size) & addr_mask;
+ if (((s->lma + s->size / opb) & addr_mask) < (s->lma & addr_mask))
+ wrap_to = (s->lma + s->size / opb) & addr_mask;
}
}
BFD_ASSERT (i <= bfd_count_sections (abfd));
@@ -4779,7 +4779,7 @@ _bfd_elf_map_sections_to_segments (bfd *abfd, struct bfd_link_info *info)
program headers we will need. */
if (phdr_in_segment && count > 0)
{
- bfd_vma phdr_lma;
+ bfd_vma phdr_lma; /* Bytes. */
bfd_boolean separate_phdr = FALSE;
phdr_lma = (sections[0]->lma - phdr_size) & addr_mask & -maxpagesize;
@@ -4819,7 +4819,7 @@ _bfd_elf_map_sections_to_segments (bfd *abfd, struct bfd_link_info *info)
m = make_mapping (abfd, sections, 0, 0, phdr_in_segment);
if (m == NULL)
goto error_return;
- m->p_paddr = phdr_lma;
+ m->p_paddr = phdr_lma * opb;
m->p_vaddr_offset
= (sections[0]->vma - phdr_size) & addr_mask & -maxpagesize;
m->p_paddr_valid = 1;
@@ -5007,7 +5007,7 @@ _bfd_elf_map_sections_to_segments (bfd *abfd, struct bfd_link_info *info)
if (s2->next->alignment_power == alignment_power
&& (s2->next->flags & SEC_LOAD) != 0
&& elf_section_type (s2->next) == SHT_NOTE
- && align_power (s2->lma + s2->size,
+ && align_power (s2->lma + s2->size / opb,
alignment_power)
== s2->next->lma)
count++;
@@ -5305,15 +5305,17 @@ elf_sort_segments (const void *arg1, const void *arg2)
return m1->no_sort_lma ? -1 : 1;
if (m1->p_type == PT_LOAD && !m1->no_sort_lma)
{
- bfd_vma lma1, lma2;
+ unsigned int opb = bfd_octets_per_byte (m1->sections[0]->owner,
+ m1->sections[0]);
+ bfd_vma lma1, lma2; /* Bytes. */
lma1 = 0;
if (m1->p_paddr_valid)
- lma1 = m1->p_paddr;
+ lma1 = m1->p_paddr / opb;
else if (m1->count != 0)
lma1 = m1->sections[0]->lma + m1->p_vaddr_offset;
lma2 = 0;
if (m2->p_paddr_valid)
- lma2 = m2->p_paddr;
+ lma2 = m2->p_paddr / opb;
else if (m2->count != 0)
lma2 = m2->sections[0]->lma + m2->p_vaddr_offset;
if (lma1 != lma2)
@@ -5584,7 +5586,7 @@ assign_file_positions_for_load_sections (bfd *abfd,
if (p->p_type == PT_LOAD
&& m->count > 0)
{
- bfd_size_type align;
+ bfd_size_type align; /* Bytes. */
unsigned int align_power = 0;
if (m->p_align_valid)
@@ -5621,7 +5623,7 @@ assign_file_positions_for_load_sections (bfd *abfd,
break;
}
- off_adjust = vma_page_aligned_bias (p->p_vaddr, off, align);
+ off_adjust = vma_page_aligned_bias (p->p_vaddr, off, align * opb);
/* Broken hardware and/or kernel require that files do not
map the same page with different permissions on some hppa
@@ -5988,7 +5990,7 @@ assign_file_positions_for_load_sections (bfd *abfd,
|| hash->root.type == bfd_link_hash_common))
{
asection *s = NULL;
- bfd_vma filehdr_vaddr = phdrs[phdr_load_seg->idx].p_vaddr;
+ bfd_vma filehdr_vaddr = phdrs[phdr_load_seg->idx].p_vaddr / opb;
if (phdr_load_seg->count != 0)
/* The segment contains sections, so use the first one. */
@@ -6065,6 +6067,7 @@ assign_file_positions_for_non_load_sections (bfd *abfd,
Elf_Internal_Phdr *p;
struct elf_segment_map *m;
file_ptr off;
+ unsigned int opb = bfd_octets_per_byte (abfd, NULL);
i_shdrpp = elf_elfsections (abfd);
end_hdrpp = i_shdrpp + elf_numsections (abfd);
@@ -6131,7 +6134,7 @@ assign_file_positions_for_non_load_sections (bfd *abfd,
{
if (p->p_type == PT_GNU_RELRO)
{
- bfd_vma start, end;
+ bfd_vma start, end; /* Bytes. */
bfd_boolean ok;
if (link_info != NULL)
@@ -6147,7 +6150,7 @@ assign_file_positions_for_non_load_sections (bfd *abfd,
if (!m->p_size_valid)
abort ();
start = m->sections[0]->vma;
- end = start + m->p_size;
+ end = start + m->p_size / opb;
}
else
{
@@ -6172,7 +6175,7 @@ assign_file_positions_for_non_load_sections (bfd *abfd,
&& lm->count != 0
&& (lm->sections[lm->count - 1]->vma
+ (!IS_TBSS (lm->sections[lm->count - 1])
- ? lm->sections[lm->count - 1]->size
+ ? lm->sections[lm->count - 1]->size / opb
: 0)) > start
&& lm->sections[0]->vma < end)
break;
@@ -6192,13 +6195,10 @@ assign_file_positions_for_non_load_sections (bfd *abfd,
if (i < lm->count)
{
- unsigned int opb = bfd_octets_per_byte (abfd,
- lm->sections[i]);
-
p->p_vaddr = lm->sections[i]->vma * opb;
p->p_paddr = lm->sections[i]->lma * opb;
p->p_offset = lm->sections[i]->filepos;
- p->p_memsz = end - p->p_vaddr;
+ p->p_memsz = end * opb - p->p_vaddr;
p->p_filesz = p->p_memsz;
/* The RELRO segment typically ends a few bytes
@@ -7181,8 +7181,8 @@ rewrite_elf_program_header (bfd *ibfd, bfd *obfd)
+ (map->includes_phdrs
? iehdr->e_phnum * iehdr->e_phentsize
: 0),
- output_section->alignment_power)
- == output_section->vma))
+ output_section->alignment_power * opb)
+ == (output_section->vma * opb)))
map->p_paddr = segment->p_vaddr;
/* Match up the physical address of the segment with the
@@ -7250,7 +7250,7 @@ rewrite_elf_program_header (bfd *ibfd, bfd *obfd)
if (matching_lma == NULL)
matching_lma = suggested_lma;
- map->p_paddr = matching_lma->lma;
+ map->p_paddr = matching_lma->lma * opb;
/* Offset the segment physical address from the lma
to allow for space taken up by elf headers. */
@@ -7278,7 +7278,7 @@ rewrite_elf_program_header (bfd *ibfd, bfd *obfd)
the same alignment. */
if (segment->p_align != 0 && segment->p_align < align)
align = segment->p_align;
- map->p_paddr &= -align;
+ map->p_paddr &= -(align * opb);
}
}
@@ -7322,8 +7322,8 @@ rewrite_elf_program_header (bfd *ibfd, bfd *obfd)
+ (map->includes_phdrs
? iehdr->e_phnum * iehdr->e_phentsize
: 0),
- output_section->alignment_power)
- != output_section->lma)
+ output_section->alignment_power * opb)
+ != output_section->lma * opb)
goto sorry;
}
else
@@ -7389,7 +7389,7 @@ rewrite_elf_program_header (bfd *ibfd, bfd *obfd)
map->p_type = segment->p_type;
map->p_flags = segment->p_flags;
map->p_flags_valid = 1;
- map->p_paddr = suggested_lma->lma;
+ map->p_paddr = suggested_lma->lma * opb;
map->p_paddr_valid = p_paddr_valid;
map->includes_filehdr = 0;
map->includes_phdrs = 0;
diff --git a/bfd/elfcode.h b/bfd/elfcode.h
index 600abfe836e..ec5156a433d 100644
--- a/bfd/elfcode.h
+++ b/bfd/elfcode.h
@@ -1648,10 +1648,11 @@ elf_debug_file (Elf_Internal_Ehdr *ehdrp)
bfd *
NAME(_bfd_elf,bfd_from_remote_memory)
(bfd *templ,
- bfd_vma ehdr_vma,
- bfd_size_type size,
- bfd_vma *loadbasep,
+ bfd_vma ehdr_vma /* Bytes. */,
+ bfd_size_type size /* Octets. */,
+ bfd_vma *loadbasep /* Bytes. */,
int (*target_read_memory) (bfd_vma, bfd_byte *, bfd_size_type))
+ /* (Bytes , , octets ). */
{
Elf_External_Ehdr x_ehdr; /* Elf file header, external form */
Elf_Internal_Ehdr i_ehdr; /* Elf file header, internal form */
@@ -1664,9 +1665,10 @@ NAME(_bfd_elf,bfd_from_remote_memory)
unsigned int i;
bfd_vma high_offset;
bfd_vma shdr_end;
- bfd_vma loadbase;
+ bfd_vma loadbase; /* Bytes. */
char *filename;
size_t amt;
+ unsigned int opb = bfd_octets_per_byte (templ, NULL);
/* Read in the ELF header in external format. */
err = target_read_memory (ehdr_vma, (bfd_byte *) &x_ehdr, sizeof x_ehdr);
@@ -1764,17 +1766,17 @@ NAME(_bfd_elf,bfd_from_remote_memory)
header sits, then we can figure out the loadbase. */
if (first_phdr == NULL)
{
- bfd_vma p_offset = i_phdrs[i].p_offset;
- bfd_vma p_vaddr = i_phdrs[i].p_vaddr;
+ bfd_vma p_offset = i_phdrs[i].p_offset; /* Octets. */
+ bfd_vma p_vaddr = i_phdrs[i].p_vaddr; /* Octets. */
if (i_phdrs[i].p_align > 1)
{
- p_offset &= -i_phdrs[i].p_align;
- p_vaddr &= -i_phdrs[i].p_align;
+ p_offset &= -(i_phdrs[i].p_align * opb);
+ p_vaddr &= -(i_phdrs[i].p_align * opb);
}
if (p_offset == 0)
{
- loadbase = ehdr_vma - p_vaddr;
+ loadbase = ehdr_vma - p_vaddr / opb;
first_phdr = &i_phdrs[i];
}
}
@@ -1830,9 +1832,9 @@ NAME(_bfd_elf,bfd_from_remote_memory)
for (i = 0; i < i_ehdr.e_phnum; ++i)
if (i_phdrs[i].p_type == PT_LOAD)
{
- bfd_vma start = i_phdrs[i].p_offset;
- bfd_vma end = start + i_phdrs[i].p_filesz;
- bfd_vma vaddr = i_phdrs[i].p_vaddr;
+ bfd_vma start = i_phdrs[i].p_offset; /* Octets. */
+ bfd_vma end = start + i_phdrs[i].p_filesz; /* Octets. */
+ bfd_vma vaddr = i_phdrs[i].p_vaddr; /* Octets. */
/* Extend the beginning of the first pt_load to cover file
header and program headers, if we proved earlier that its
@@ -1845,7 +1847,7 @@ NAME(_bfd_elf,bfd_from_remote_memory)
/* Extend the end of the last pt_load to cover section headers. */
if (last_phdr == &i_phdrs[i])
end = high_offset;
- err = target_read_memory (loadbase + vaddr,
+ err = target_read_memory (loadbase + vaddr / opb,
contents + start, end - start);
if (err)
{
diff --git a/bfd/elflink.c b/bfd/elflink.c
index 369f3cb3e7b..5852844498a 100644
--- a/bfd/elflink.c
+++ b/bfd/elflink.c
@@ -11566,8 +11566,8 @@ elf_fixup_link_order (bfd *abfd, asection *o)
struct bfd_link_order *p;
bfd *sub;
struct bfd_link_order **sections;
- asection *s, *other_sec, *linkorder_sec;
- bfd_vma offset;
+ asection *other_sec, *linkorder_sec;
+ bfd_vma offset; /* Octets. */
other_sec = NULL;
linkorder_sec = NULL;
@@ -11577,7 +11577,7 @@ elf_fixup_link_order (bfd *abfd, asection *o)
{
if (p->type == bfd_indirect_link_order)
{
- s = p->u.indirect.section;
+ asection *s = p->u.indirect.section;
sub = s->owner;
if ((s->flags & SEC_LINKER_CREATED) == 0
&& bfd_get_flavour (sub) == bfd_target_elf_flavour
@@ -11632,11 +11632,12 @@ elf_fixup_link_order (bfd *abfd, asection *o)
for (n = 0; n < seen_linkorder; n++)
{
bfd_vma mask;
- s = sections[n]->u.indirect.section;
- mask = ~(bfd_vma) 0 << s->alignment_power;
+ asection *s = sections[n]->u.indirect.section;
+ unsigned int opb = bfd_octets_per_byte (abfd, s);
+
+ mask = ~(bfd_vma) 0 << s->alignment_power * opb;
offset = (offset + ~mask) & mask;
- s->output_offset = offset / bfd_octets_per_byte (abfd, s);
- sections[n]->offset = offset;
+ sections[n]->offset = s->output_offset = offset / opb;
offset += sections[n]->size;
}
@@ -12247,7 +12248,7 @@ bfd_elf_final_link (bfd *abfd, struct bfd_link_info *info)
if (htab->tls_sec)
{
- bfd_vma base, end = 0;
+ bfd_vma base, end = 0; /* Both bytes. */
asection *sec;
for (sec = htab->tls_sec;
@@ -12255,6 +12256,7 @@ bfd_elf_final_link (bfd *abfd, struct bfd_link_info *info)
sec = sec->next)
{
bfd_size_type size = sec->size;
+ unsigned int opb = bfd_octets_per_byte (abfd, sec);
if (size == 0
&& (sec->flags & SEC_HAS_CONTENTS) == 0)
@@ -12262,9 +12264,9 @@ bfd_elf_final_link (bfd *abfd, struct bfd_link_info *info)
struct bfd_link_order *ord = sec->map_tail.link_order;
if (ord != NULL)
- size = ord->offset + ord->size;
+ size = ord->offset * opb + ord->size;
}
- end = sec->vma + size;
+ end = sec->vma + size / opb;
}
base = htab->tls_sec->vma;
/* Only align end of TLS section if static TLS doesn't have special
@@ -12777,6 +12779,8 @@ bfd_elf_final_link (bfd *abfd, struct bfd_link_info *info)
if (bed->dtrel_excludes_plt && htab->srelplt != NULL)
{
+ unsigned int opb = bfd_octets_per_byte (abfd, o);
+
/* Don't count procedure linkage table relocs in the
overall reloc count. */
sh_size -= htab->srelplt->size;
@@ -12796,7 +12800,7 @@ bfd_elf_final_link (bfd *abfd, struct bfd_link_info *info)
/* If .rela.plt is the first .rela section, exclude
it from DT_RELA. */
else if (sh_addr == (htab->srelplt->output_section->vma
- + htab->srelplt->output_offset))
+ + htab->srelplt->output_offset) * opb)
sh_addr += htab->srelplt->size;
}
@@ -14251,7 +14255,7 @@ bfd_elf_discard_info (bfd *output_bfd, struct bfd_link_info *info)
{
asection *i;
int eh_changed = 0;
- unsigned int eh_alignment;
+ unsigned int eh_alignment; /* Octets. */
for (i = o->map_head.s; i != NULL; i = i->map_head.s)
{
@@ -14278,7 +14282,8 @@ bfd_elf_discard_info (bfd *output_bfd, struct bfd_link_info *info)
fini_reloc_cookie_for_section (&cookie, i);
}
- eh_alignment = 1 << o->alignment_power;
+ eh_alignment = ((1 << o->alignment_power)
+ * bfd_octets_per_byte (output_bfd, o));
/* Skip over zero terminator, and prevent empty sections from
adding alignment padding at the end. */
for (i = o->map_tail.s; i != NULL; i = i->map_tail.s)
diff --git a/bfd/merge.c b/bfd/merge.c
index 5a4e709ef38..0c6f7a10d3d 100644
--- a/bfd/merge.c
+++ b/bfd/merge.c
@@ -292,8 +292,9 @@ sec_merge_emit (bfd *abfd, struct sec_merge_hash_entry *entry,
asection *sec = secinfo->sec;
char *pad = NULL;
bfd_size_type off = 0;
- int alignment_power = sec->output_section->alignment_power;
- bfd_size_type pad_len;
+ unsigned int opb = bfd_octets_per_byte (abfd, sec);
+ int alignment_power = sec->output_section->alignment_power * opb;
+ bfd_size_type pad_len; /* Octets. */
/* FIXME: If alignment_power is 0 then really we should scan the
entry list for the largest required alignment and use that. */
@@ -364,9 +365,11 @@ _bfd_add_merge_section (bfd *abfd, void **psinfo, asection *sec,
{
struct sec_merge_info *sinfo;
struct sec_merge_sec_info *secinfo;
- unsigned int align;
+ unsigned int alignment_power; /* Octets. */
+ unsigned int align; /* Octets. */
bfd_size_type amt;
bfd_byte *contents;
+ unsigned int opb = bfd_octets_per_byte (abfd, sec);
if ((abfd->flags & DYNAMIC) != 0
|| (sec->flags & SEC_MERGE) == 0)
@@ -389,10 +392,11 @@ _bfd_add_merge_section (bfd *abfd, void **psinfo, asection *sec,
#ifndef CHAR_BIT
#define CHAR_BIT 8
#endif
- if (sec->alignment_power >= sizeof (align) * CHAR_BIT)
+ alignment_power = sec->alignment_power * opb;
+ if (alignment_power >= sizeof (align) * CHAR_BIT)
return TRUE;
- align = 1u << sec->alignment_power;
+ align = 1u << alignment_power;
if ((sec->entsize < align
&& ((sec->entsize & (sec->entsize - 1))
|| !(sec->flags & SEC_STRINGS)))
@@ -739,7 +743,7 @@ _bfd_merge_sections (bfd *abfd,
for (sinfo = (struct sec_merge_info *) xsinfo; sinfo; sinfo = sinfo->next)
{
struct sec_merge_sec_info *secinfo;
- bfd_size_type align;
+ bfd_size_type align; /* Bytes. */
if (! sinfo->chain)
continue;
@@ -764,8 +768,10 @@ _bfd_merge_sections (bfd *abfd,
return FALSE;
if (align)
{
+ unsigned int opb = bfd_octets_per_byte (abfd, secinfo->sec);
+
align = (bfd_size_type) 1 << secinfo->sec->alignment_power;
- if ((secinfo->sec->size & (align - 1)) != 0)
+ if (((secinfo->sec->size / opb) & (align - 1)) != 0)
align = 0;
}
}
@@ -782,7 +788,7 @@ _bfd_merge_sections (bfd *abfd,
else
{
struct sec_merge_hash_entry *e;
- bfd_size_type size = 0;
+ bfd_size_type size = 0; /* Octets. */
/* Things are much simpler for non-strings.
Just assign them slots in the section. */
diff --git a/include/bfdlink.h b/include/bfdlink.h
index 8d85530e390..40a6d4d40a6 100644
--- a/include/bfdlink.h
+++ b/include/bfdlink.h
@@ -801,9 +801,9 @@ struct bfd_link_order
struct bfd_link_order *next;
/* Type of link_order. */
enum bfd_link_order_type type;
- /* Offset within output section. */
+ /* Offset within output section in bytes. */
bfd_vma offset;
- /* Size within output section. */
+ /* Size within output section in octets. */
bfd_size_type size;
/* Type specific information. */
union
diff --git a/include/elf/internal.h b/include/elf/internal.h
index d626adece5b..9692028eed3 100644
--- a/include/elf/internal.h
+++ b/include/elf/internal.h
@@ -91,7 +91,8 @@ struct elf_internal_phdr {
bfd_vma p_paddr; /* Segment physical address in octets. */
bfd_vma p_filesz; /* Segment size in file in octets. */
bfd_vma p_memsz; /* Segment size in memory in octets. */
- bfd_vma p_align; /* Segment alignment, file & memory. */
+ bfd_vma p_align; /* Segment alignment in bytes, file
+ & memory */
};
typedef struct elf_internal_phdr Elf_Internal_Phdr;
@@ -266,13 +267,13 @@ struct elf_segment_map
unsigned long p_type;
/* Program segment flags. */
unsigned long p_flags;
- /* Program segment physical address. */
+ /* Program segment physical address in octets. */
bfd_vma p_paddr;
/* Program segment virtual address offset from section vma in bytes. */
bfd_vma p_vaddr_offset;
/* Program segment alignment. */
bfd_vma p_align;
- /* Segment size in file and memory */
+ /* Segment size in file and memory in octets. */
bfd_vma p_size;
/* Whether the p_flags field is valid; if not, the flags are based
on the section flags. */
--
2.16.4
next prev parent reply other threads:[~2020-03-02 20:17 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20200215193050.22717-1-ceggers@gmx.de>
2020-02-15 19:31 ` [PATCH " Christian Eggers
2020-03-02 14:18 ` Nick Clifton
2020-03-02 20:17 ` Christian Eggers [this message]
2020-03-12 18:24 ` [PATCH v2 " Christian Eggers
2020-03-13 5:31 ` Alan Modra
2020-02-15 19:31 ` [PATCH 1/2] " Christian Eggers
2020-03-02 14:17 ` Nick Clifton
2020-03-02 20:11 ` [PATCH v4] " Christian Eggers
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200302201524.7398-1-ceggers@gmx.de \
--to=ceggers@gmx.de \
--cc=amodra@gmail.com \
--cc=binutils@sourceware.org \
--cc=nickc@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).