PR 24001
* objcopy.c (copy_object): Free dhandle after writing out the
debug information.
* objdump.c (dump_bfd): Free dhandle after printing out the debug
information.
This patch performs a run-time test that a shared libbfd.so has been
compiled with the same size bfd_vma as that of apps using the library.
On a 32-bit host it is easily possible to have one libbfd.so compiled
to support 64-bit targets (or configured with --enable-64-bit-bfd)
while another only supports 32-bit targets. The two libraries will
have differently sized bfd_vma types, and if the wrong one is loaded
all sorts of weird behaviour might be seen.
bfd/
PR 23534
* init.c (BFD_INIT_MAGIC): Define.
(bfd_init): Return BFD_INIT_MAGIC.
bfd-in2.h: Regenerate.
binutils/
PR 23534
* addr2line.c (main): Exit with fatal error if bfd_init
returns an unexpected value.
* ar.c (main): Likewise.
* dlltool.c (identify_dll_for_implib): Likewise.
* nm.c (main): Likewise.
* objcopy.c (main): Likewise.
* objdump.c (main): Likewise.
* size.c (main): Likewise.
* strings.c (main): Likewise.
* windmc.c (main): Likewise.
* windres.c (main): Likewise.
gas/
PR 23534
* as.c (main): Exit with fatal error if bfd_init returns an
unexpected value.
ld/
PR 23534
* ldmain.c (main): Exit with fatal error if bfd_init returns
an unexpected value.
This is the result of an email thread starting here:
https://sourceware.org/ml/binutils/2018-09/msg00031.html
The main point of the thread is this observation:
* Supposing we had an object file with two globals, SomeGlobal and
SomeOtherGlobal, if one were to do "--globalize-symbol SomeGlobal
--keep-global-symbol SomeOtherGlobal", you might expect that both
SomeGlobal and SomeOtherGlobal are global in the output file... but it
isn't. Because --keep-global-symbol is set and doesn't include
SomeGlobal, SomeGlobal will be demoted to a local symbol. And because
the check to see if we should apply the --globalize-symbol flag checks
"flags" (the original flag set), and not "sym->flags", it decides not
to do anything, so SomeGlobal remains a local symbol. Although this is
a weird edge case, should this be changed so that --keep-global-symbol
implicitly keeps anything also specified via --globalize-symbol? (The
code seems technically correct with respect to the documentation, but
IMO the behavior is counter-intuitive).
binutils* objcopy.c (copy_main): Issue a fata error if the
--keep-global-symbol(s) and the --globalize-symbol(s) options are
used together.
* doc/binutils.texi: Document that the two options are
incompatible.
* testsuite/binutils-all/copy-5.d: New test.
* testsuite/binutils-all/objcopy.exp: Run the new test.
add_specific_symbols allocates a buffer to hold symbols. It should be
freed only if it is no longer in use.
PR binutils/23633
* objcopy.c (strip_specific_buffer): New.
(strip_unneeded_buffer): Likewise.
(keep_specific_buffer): Likewise.
(localize_specific_buffer): Likewise.
(globalize_specific_buffer): Likewise.
(keepglobal_specific_buffer): Likewise.
(weaken_specific_buffer): Likewise.
(add_specific_symbols): Add an argument to return pointer to
allocated buffer.
(copy_main): Update add_specific_symbols to update pointers to
allocated buffer. Free pointers to allocated buffer before
return.
* objcopy.c (handle_remove_section_option): Don't require a dot
after .rela and .rel to handle a possible relocation section.
* testsuite/binutils-all/remove-relocs-07.s,
* testsuite/binutils-all/remove-relocs-07.d,
* testsuite/binutils-all/remove-relocs-08.d: New tests.
* testsuite/binutils-all/remove-relocs-01.d,
* testsuite/binutils-all/remove-relocs-04.d,
* testsuite/binutils-all/remove-relocs-05.d,
* testsuite/binutils-all/remove-relocs-06.d: Exclude mips64-openbsd.
BFD handles ELF relocation sections in an executable differently to
relocation sections in a relocatable object. For a relocatable
object, BFD carries the relocations as data associated with the
section to which they apply; The relocation section doesn't appear as
a separate section. For an executable, dynamic relocation sections do
appear as separate sections. This means that objcopy needs to use
different strategies when dealing with relocations.
When --remove-relocations was added to objcopy with commit
d3e5f6c8f1, objcopy lost the ability to remove dynamic relocation
sections such as .rela.plt from executables using the option
"--remove-section=.rela.plt". This patch reinstates that
functionality.
I thought it best to keep --remove-relocations as is, rather than
extending to handle dynamic relocations as per the patch in the PR,
because executables linked with --emit-relocs may have both dynamic
and non-dynamic relocations. In that case --remove-relocataions=* is
useful to remove all the non-dynamic relocations.
PR binutils/23611
* objcopy.c (handle_remove_section_option): Consider .rela and
.rel sections for stripping directly as well as attached to the
associated section they relocate.
* doc/binutils.texi (remove-relocations): Specify that this
option removes non-dynamic relocation sections.
* testsuite/binutils-all/objcopy.exp
(objcopy_remove_relocations_from_executable): New test.
* objcopy.c (add_specific_symbols): Free buffer on exit.
(add_redefine_syms_file): Close file handle on exit.
(copy_object): Close file handle on early exit.
Free buffer on early exit.
Free gaps buffers once they are no longer needed.
* dwarf.c (display_debug_frames): Free allocated memory on exit.
(load_separate_debug_info): Free allocate memory on early exit.
Another fuzzing fix. I think it's reasonable to simply strip out any
group section that is too weird for objcopy to handle.
PR 23141
* objcopy.c (is_strip_section): Strip groups without a valid
signature symbol.
I am checking in the attached patch which updates the binutils
support for version 3 binary annotation notes. (Version 3 adds
an end address to the ranges covered by the notes, so that it
is possible to detect gaps in the coverage).
This patch also stops the note merging feature of objcopy from
executing if the notes have relocations against them. This makes the
code simpler, and prevents the problems with architectures which have
unusual relocation management issues.
* objcopy.c (objcopy_internal_note): New structure.
(gap_exists): New function.
(is_open_note): New function.
(is_func_note): New function.
(is_64bit): New function.
(merge_gnu_build_notes): Handle v3 notes. Do not merge
if there are relocations against the notes.
* readelf.c (get_note_type): Use short names for build attribute
notes.
(print_symbol_for_build_attribute): Rename to
get_symbol_for_build_attribute. Returns the found symbol rather
than printing it.
(print_gnu_build_attribute_description): Maintain address ranges
for function notes as well as global notes. Handle v3 notes.
(print_gnu_build_attribute_name): Use more space for printing the
name in wide mode.
* testsuite/binutils-all/note-2-32.s: Use .dc.l instead of .word.
Eliminate symbol references in order to remove the need for
relocations.
* testsuite/binutils-all/note-2-64.s: Likewise.
* testsuite/binutils-all/note-3-32.s: Add a size to the note_1
symbol.
* testsuite/binutils-all/note-3-64.s: Likewise.
* testsuite/binutils-all/mips/mips-note-2r-n32.d: Update expected
output.
* testsuite/binutils-all/mips/mips-note-2r-n64.d: Likewise.
* testsuite/binutils-all/mips/mips-note-2r.d: Likewise.
* testsuite/binutils-all/note-2-32.d: Likewise.
* testsuite/binutils-all/note-2-64.d: Likewise.
* testsuite/binutils-all/note-3-32.d: Likewise.
* testsuite/binutils-all/note-3-64.d: Likewise.
* testsuite/binutils-all/note-4-64.s: New test. Checks v3 notes.
* testsuite/binutils-all/note-4-32.s: New test.
* testsuite/binutils-all/note-4-64.d: New test result file.
* testsuite/binutils-all/note-4-32.d: New test result file.
A recent Fedora bug (1520805) exposed a problem with objcopy's reloc
copying code, when a binary also contains mergeable notes. The note
merging code would delete some relocs, but then the reloc copying code
would try to put them back again, which did not work.
So I am checking in the patch below to fix the problem. The patch
also tweaks one of the binutils note merging tests so that it is
skipped for the Sparc64 target, since this has funky relocs.
binutils * objcopy.c (copy_relocations_in_section): Use the orelocations
field of the input section, if it has been initialised.
* testsuite/binutils-all/note-2-64.d: Skip test on Sparc64.
bfd * elfcode.h (elf_write_relocs): Check for an empty howto field.
PR 22465
binutils/
* objcopy.c (copy_section): New local extra. If isection->lma not
exactly divisible by interleave, then bias from. Also adjust
osection->lma if necessary.
ld/
* testsuite/ld-elf/interleave-0.d, testsuite/ld-elf/interleave-4.d,
* testsuite/ld-elf/interleave.ld, testsuite/ld-elf/interleave.s: New.
It's nicer than xmalloc followed by bfd_get_section_contents, since
xmalloc exits on failure and needs a check that its size_t arg doesn't
lose high bits when converted from bfd_size_type.
PR binutils/21665
* objdump.c (strtab): Make var a bfd_byte*.
(disassemble_section): Don't limit malloc size. Instead, use
bfd_malloc_and_get_section.
(read_section_stabs): Use bfd_malloc_and_get_section. Return
bfd_byte*.
(find_stabs_section): Remove now unnecessary cast.
* objcopy.c (copy_object): Use bfd_malloc_and_get_section. Free
contents on error return.
* nlmconv.c (copy_sections): Use bfd_malloc_and_get_section.
This patch removes a workaround recently installed in objcopy that
avoided removing duplicated notes in targets for which the number of
internal relocations may be bigger than the number of external
relocations. With the recent fixes in sparc64 and mips64, this
workaround is no longer necessary.
2017-05-19 Jose E. Marchesi <jose.marchesi@oracle.com>
* objcopy.c (merge_gnu_build_notes): Remove workaround that
prevented deleting relocations in duplicated notes in mips64 and
sparc.
The type of relcount is long and the type of sec->reloc_count is
unsigned int. On 32-bit hosts, GCC issues an error:
objcopy.c:2144:20: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]
if (relcount > sec->reloc_count)
Cast relcount to unsigned long to silence GCC.
* objcopy.c (merge_gnu_build_notes): Cast relcount to unsigned
long when comparing with sec->reloc_count.
* objcopy.c: Add --no-merge-notes option to disable note merging.
Add --[no-]merge-notes option to strip, and enable it by default.
(num_bytes): New function.
(merge_gnu_build_notes): Add code to merge stack size notes.
* binutils.texi: Update strip and objcopy documentation.
* readelf.c (print_gnu_build_attribute_name): Use defined
constants for note types.
objcopy/
* objcopy.c (struct redefine_node): Delete the field "next".
(redefine_sym_list): Deleted.
(redefine_specific_htab): New hash table.
(redefine_specific_reverse_htab): Likewise.
(eq_string_redefnode): New function.
(htab_hash_redefnode): Likewise.
(create_symbol2redef_htab): Likewise.
(add_specific_symbol_node): Likewise.
(create_symbol_htabs): Create redefine_specific_htab and
redefine_specific_reverse_htab.
(lookup_sym_redefinition): Use hash table instead of list.
(redefine_list_append): Likewise, and rename to add_redefine_and_check.
(copy_main): Use redefine_specific_htab instead of redefine_sym_list.
Update comments.
* readelf.c (print_gnu_build_attribute_name): Allow stack
protection notes to contain numeric values. Use a colon rather
than a space to separate a string name from its values. Decode
the numeric value of a stack protection note.
* objcopy.c (merge_gnu_build_notes): Allow version notes to
contain extra text after the protocol version number.
The direct references in objcopy kind of look like a hack to me, so
I'm calling these symbols internal too. Certainly they aren't named
and documented as a public BFD symbol today anyway.
So ... give these bfd-internal symbols with external linkage a _bfd_
prefix to avoid collisions in the global symbol namespace.
While at it, give them names that more closely match the corresponding
option name that toggles them.
Also while at it, fix a few related comment typos.
gdb/ChangeLog:
2017-02-17 Pedro Alves <palves@redhat.com>
* srec.c (Chunk): Rename to ...
(_bfd_srec_len): ... this.
(S3Forced): Rename to ...
(_bfd_srec_forceS3): ... this.
* objcopy.c: Adjust all references.
PR binutils/20930
* objcopy.c (mark_symbols_used_in_relocations): Check for a null
symbol pointer pointer before attempting to mark the symbol as
kept.
The objcopy and strip tools make use of the bfd library to manipulate
the state of the input file (to produce an output file). Within the
input file (for ELF at least), relocations are held within a section,
and so, if the user wanted to remove the relocations, but keep the
section to which the relocations would have been applied, it is tempting
to think that specifying the name of a relocation section to objcopy's
--remove-section option might do what you want, for example:
objcopy --remove-section=.rela.text input.elf output.elf
However, this does not work. The reason is that when the input file is
loaded, relocations are not managed as sections, but are, instead,
loaded as data associated with the section to which the relocations
would be applied. In our example above the relocations in '.rela.text'
are held as data on the section '.text' once 'input.elf' is loaded.
One task that objcopy and strip do is copy the relocations from the
input file to the output file if the section is also being copied from
the input file to the output file.
This commit adds a new command line option for objcopy and strip,
--remove-relocations, which can be used to remove the relocations, while
keeping the section that the relocations would have been applied to, for
example:
objcopy --remove-relocations=.text input.elf output.elf
in this case the section '.text' will appear in both 'input.elf' and
'output.elf', but any relocations in 'input.elf' that apply to '.text'
will not be present in 'output.elf'.
I have also added a special case to the handling of --remove-section
that spots if a user tries to remove a relocation section (currently
this is done by spotting the '.rela.' or '.rel.' prefix) and forwards
the request to --remove-relocations.
As with --remove-section and --only-section the --remove-relocations
option supports the '!' prefix on the section-patterns it takes to allow
for sections to be specifically not matched.
There are tests for all the new functionality.
binutils/ChangeLog:
* doc/binutils.texi (objcopy): Document 'remove-relocations'.
(strip): Likewise.
* objcopy.c (SECTION_CONTEXT_REMOVE_RELOCS): Define.
(enum command_line_switch): Add 'OPTION_REMOVE_RELOCS'.
(struct option strip_options): Add 'remove-relocations'.
(struct option copy_options): Likewise.
(copy_usage): Likewise.
(strip_usage): Likewise.
(handle_remove_relocations_option): New function.
(discard_relocations): New function.
(handle_remove_section_option): New function.
(copy_relocations_in_section): Use discard_relocations.
(strip_main): Use handle_remove_section_option for
'remove-section', and handle 'remove-relocations' option.
(copy_main): Likewise.
* testsuite/binutils-all/objcopy.exp: Run new tests.
* testsuite/binutils-all/remove-relocs-01.d: New file.
* testsuite/binutils-all/remove-relocs-01.s: New file.
* testsuite/binutils-all/remove-relocs-02.d: New file.
* testsuite/binutils-all/remove-relocs-03.d: New file.
* testsuite/binutils-all/remove-relocs-04.d: New file.
* testsuite/binutils-all/remove-relocs-05.d: New file.
* testsuite/binutils-all/remove-relocs-06.d: New file.
For symbol matching, prefixing a pattern with '!' will indicate a
non-matching pattern, however, this is not the case for section
patterns. As a result it is not possible to say "apply this action to
all sections except ...".
With this commit the objcopy and strip tools now support '!' prefix for
section patterns, so we can say:
objcopy --remove-section="*" --remove-section="!.text*"
Which will remove all sections, except those matching the pattern
'.text*'.
binutils/ChangeLog:
* objcopy.c (find_section_list): Handle section patterns starting
with '!' being a non-matching pattern.
* doc/binutils.texi (objcopy): Give example of using '!' with
--remove-section and --only-section.
(strip): Give example of using '!' with --remove-section.
* testsuite/binutils-all/data-sections.s: New file.
* testsuite/binutils-all/only-section-01.d: New file.
* testsuite/binutils-all/remove-section-01.d: New file.
* testsuite/binutils-all/objcopy.exp: Run new tests.
* NEWS: Mention new feature.
Some MIPS targets use a named section symbol rather than a symbol with
no name as is used with most ELF targets. When renaming sections, the
named section symbol needs to be renamed too.
Rather than fix this bug, I'd originally intended to just correct the
xfail added recently for update-1.o vs update4.o in update-section.exp,
using the same set of targets for the localize-hidden-1 mips xfail.
I'd extracted that target test into a new function, is_bad_symtab. It
turns out to be useful in readelf.exp too.
bfd/
* config.bfd: Delete mips vxworks patterns matched earlier.
Combine mips*-*-none with mips*-*-elf*.
binutils/
* objcopy.c (find_section_rename): Forward declare. Remove
ibfd and sec_ptr param. Add old_name param. Allow for NULL
returned_flags. Move read of section name and flags to..
(setup_section): ..here. Update find_section_rename call.
(filter_symbols): Rename section symbols for renamed sections.
(copy_object): Call filter_symbols when renamed sections.
* testsuite/lib/binutils-common.exp (is_bad_symtab): New.
* testsuite/binutils-all/update-section.exp: Revert 96037eb0
mips xfail.
* testsuite/binutils-all/objcopy.exp (copy_executable): Use
is_bad_symtab.
(localize-hidden-1): xfail if is_bad_symtab.
* testsuite/binutils-all/readelf.exp: Use is_bad_symtab to select
between mips/tmips.
