This patch creates dynamic sections in i386/x86-64 create_dynamic_sections
instead of creating them on demend. Linker will strip them if they are
empty. It changes order in x86-64 .eh_frame section. The extra DW_CFA_nop
paddings is due to
https://sourceware.org/bugzilla/show_bug.cgi?id=21441
bfd/
* elf32-i386.c (elf_i386_create_dynamic_sections): Create the
.plt.got section here.
(elf_i386_check_relocs): Don't create the .plt.got section.
* elf64-x86-64.c (elf_x86_64_create_dynamic_sections): Create
the .plt.got and .plt.bnd sections here.
(elf_x86_64_check_relocs): Don't create the .plt.got nor
.plt.bnd sections.
ld/
* testsuite/ld-x86-64/pr21038a.d: Update DW_CFA_nop paddings
in .eh_frame section.
* testsuite/ld-x86-64/pr21038c.d: Update .eh_frame order.
x86, PLT relocation may contain R_386_TLS_DESC or R_X86_64_TLSDESC
even though there is no real PLT. We need to add DT_PLTRELSZ, DT_PLTREL
and DT_JMPREL if there is a .rel.plt/.rela.plt section.
bfd/
* elf32-i386.c (elf_i386_size_dynamic_sections): Alwasys add
DT_PLTRELSZ, DT_PLTREL and DT_JMPREL for .rel.plt section.
* elf64-x86-64.c (elf_x86_64_size_dynamic_sections): Alwasys
add DT_PLTRELSZ, DT_PLTREL and DT_JMPREL for .rela.plt section.
ld/
* testsuite/ld-i386/tlsdesc2.d: New test.
* testsuite/ld-x86-64/tlsdesc2.d: Likewise.
Complement commit e17b0c351f ("MIPS/BFD: Respect the ELF gABI dynamic
symbol table sort requirement") and correct an inconsistency in dynamic
symbol accounting data causing an assertion failure in the MIPS backend:
ld: BFD (GNU Binutils) 2.28.51.20170330 assertion fail
../../binutils-gdb/bfd/elfxx-mips.c:3860
in the course of making a GOT entry in a static binary to satisfy a GOT
relocation present in input, due to the local dynamic symbol count not
having been established.
To do so let backends request `_bfd_elf_link_renumber_dynsyms' to be
always called, rather than where a dynamic binary is linked only, and
then make this request in the MIPS backend.
bfd/
PR ld/21334
* elf-bfd.h (elf_backend_data): Add `always_renumber_dynsyms'
member.
* elfxx-target.h [!elf_backend_always_renumber_dynsyms]
(elf_backend_always_renumber_dynsyms): Define.
(elfNN_bed): Initialize `always_renumber_dynsyms' member.
* elfxx-mips.h (elf_backend_always_renumber_dynsyms): Define.
* elflink.c (bfd_elf_size_dynamic_sections): Also call
`_bfd_elf_link_renumber_dynsyms' if the backend has requested
it.
(bfd_elf_size_dynsym_hash_dynstr): Likewise.
ld/
PR ld/21334
* testsuite/ld-mips-elf/pr21334.dd: New test.
* testsuite/ld-mips-elf/pr21334.gd: New test.
* testsuite/ld-mips-elf/pr21334.ld: New test linker script.
* testsuite/ld-mips-elf/pr21334.s: New test source.
* testsuite/ld-mips-elf/mips-elf.exp: Run the new tests.
Force symbol dynamic if it isn't undefined weak. Generate relative
relocation for GOT reference against non-dynamic symbol in PIC to
avoid unnecessary dynamic symbols.
bfd/
* elf64-x86-64.c (elf_x86_64_link_hash_entry): Add
no_finish_dynamic_symbol.
(elf_x86_64_link_hash_newfunc): Set no_finish_dynamic_symbol to
0.
(elf_x86_64_allocate_dynrelocs): If a symbol isn't undefined
weak symbol, don't make it dynamic.
(elf_x86_64_relocate_section): If a symbol isn't dynamic in PIC,
set no_finish_dynamic_symbol and generate R_X86_64_RELATIVE
relocation for GOT reference.
(elf_x86_64_finish_dynamic_symbol): Abort if
no_finish_dynamic_symbol isn't 0.
ld/
* testsuite/ld-x86-64/no-plt.exp: Also check no-plt-1e.nd.
* testsuite/ld-x86-64/no-plt-1e.nd: New file.
commit f129e49f4d
Author: H.J. Lu <hjl.tools@gmail.com>
Date: Tue Jan 10 11:30:25 2017 -0800
Don't use elf_i386_eh_frame_plt directly
fixed i386 PLT eh_frame generation. Skip pr12570 tests since they are
for non-nacl targets.
* testsuite/ld-i386/pr12570a.d: Skip for nacl targets.
* testsuite/ld-i386/pr12570b.d: Likewise.
commit a27e437177
Author: Roland McGrath <roland@gnu.org>
Date: Thu Jul 28 22:35:15 2011 +0000
BFD vector for elf32-i386-nacl:
changed ELF_MAXPAGESIZE to 0x10000 for VxWorks. This patch fixes it
and updated testsuite/ld-i386/vxworks2.sd to add space for program
headers.
bfd/
PR ld/21425
* elf32-i386.c (ELF_MAXPAGESIZE): Set to 0x1000 for VxWorks.
ld/
PR ld/20815
* testsuite/ld-i386/vxworks2.sd: Add space for program headers.
A few tests in the ld testsuite were expecting the disassembler to
emit `rett' instructions in V9. This patch updates the tests to
expect `return' instead.
ld/ChangeLog:
2017-04-24 Jose E. Marchesi <jose.marchesi@oracle.com>
* testsuite/ld-sparc/tlssunbin64.dd: Expect `return' instructions
instead of `rett' in V9.
* testsuite/ld-sparc/tlssunnopic64.dd: Likewise.
* testsuite/ld-sparc/tlssunpic64.dd: Likewise.
GOT reference to global symbol in PIE will lead to dynamic symbol. It
becomes a problem when "time" or "times" is defined as a variable in
an executable, clashing with functions of the same name in libc. If
a symbol isn't undefined weak symbol, don't make it dynamic in PIE and
generate R_386_RELATIVE relocation.
bfd/
PR ld/21402
* elf32-i386.c (elf_i386_link_hash_entry): Add
no_finish_dynamic_symbol.
(elf_i386_link_hash_newfunc): Set no_finish_dynamic_symbol to 0.
(elf_i386_allocate_dynrelocs): If a symbol isn't undefined weak
symbol, don't make it dynamic in PIE.
(elf_i386_relocate_section): If a symbol isn't dynamic in PIE,
set no_finish_dynamic_symbol and generate R_386_RELATIVE
relocation for R_386_GOT32
(elf_i386_finish_dynamic_symbol): Abort if no_finish_dynamic_symbol
isn't 0.
ld/
PR ld/21402
* testsuite/ld-elf/indirect.exp: Don't skip PIE indirect5 and
indirect6 tests on i386.
When -static -E/--dynamic-list are passed to linker, linker may create
executable with dynamic sections which aren't supported by run-time.
We require --no-dynamic-linker together with -static -E/--dynamic-list
before adding dynamic symbol table to static executable.
bfd/
PR ld/19617
PR ld/21086
* elflink.c (elf_link_add_object_symbols): Require
--no-dynamic-linker with -E/--dynamic-list when creating
dynamic sections.
ld/
PR ld/19617
PR ld/21086
* testsuite/ld-elf/pr19617a.d: Pass --no-dynamic-linker to ld.
* testsuite/ld-elf/pr19617b.d: Likewise.
* testsuite/ld-elf/pr19617c.d: Likewise.
*testsuite/ld-i386/pr19636-4d.d: Likewise.
* testsuite/ld-elf/readelf.exp: Pass --no-dynamic-linker to ld
with --export-dynamic.
* testsuite/ld-elf/shared.exp: Pass --no-dynamic-linker to ld
with -E.
On i386, since GOT reference is needed to access global symbols in PIE,
those symbols are made dynamic. Crash happens when there is a reference
to the same global symbol with a different symbol type in a shared
object. Since mixing different types of the same symbol doesn't work in
general, this patch skips those tests on i386 as well as compiles non-PIE
indirect5 and indirect6 tests with $NOPIE_LDFLAGS and $NOPIE_CFLAGS.
PR ld/21402
* testsuite/ld-elf/indirect.exp: Pass $NOPIE_LDFLAGS and
$NOPIE_CFLAGS to non-PIE indirect5 and indirect6 tests. Skip
PIE indirect5 and indirect6 tests on i386.
Complement commit 3807734dbe ("PR ld/15428: MIPS/LD/testsuite:
Un-KFAIL `__ehdr_start' test 2") and join tests that do not need to be
split anymore.
ld/
* testsuite/ld-mips-elf/mips-elf.exp: Join `__ehdr_start' tests.
We need to make an IR symbol visible if it is defined in an IR object
and referenced in a dynamic object. When --as-needed is used, since
linker removes the IR symbol reference of the dynamic object if the
dynamic object isn't needed in the first pass, the IR definition isn't
visible to the dynamic object even if the dynamic object becomes needed
in the second pass. Add dynamic_ref_after_ir_def to bfd_link_hash_entry
to track IR symbol which is defined in an IR object and later referenced
in a dynamic object. dynamic_ref_after_ir_def is preserved when restoring
the symbol table for unneeded dynamic object.
bfd/
PR ld/21382
* elflink.c (elf_link_add_object_symbols): Preserve
dynamic_ref_after_ir_def when restoring the symbol table for
unneeded dynamic object.
include/
PR ld/21382
* bfdlink.h (bfd_link_hash_entry): Add dynamic_ref_after_ir_def.
ld/
PR ld/21382
* plugin.c (is_visible_from_outside): Symbol may be visible
from outside if dynamic_ref_after_ir_def is set.
(plugin_notice): Set dynamic_ref_after_ir_def if the symbol is
defined in an IR object and referenced in a dynamic object.
* testsuite/ld-plugin/lto.exp: Run PR ld/21382 tests.
* testsuite/ld-plugin/pr21382a.c: New file.
* testsuite/ld-plugin/pr21382b.c: Likewise.
Since undefined IFUNC symbol is treated as normal FUNC symbol, don't
abort on undefined IFUNC symbol in the second PLT.
bfd/
PR ld/21401
* elf64-x86-64.c (elf_x86_64_finish_dynamic_symbol): Don't abort
on on undefined IFUNC symbol in the second PLT.
ld/
PR ld/21401
* testsuite/ld-ifunc/ifunc.exp: Add a libtest-2-now.so test with
-z now.
-z nodynamic-undefined-weak is only implemented for x86. (The sparc
backend has some support code but doesn't enable the option by
including ld/emulparams/dynamic_undefined_weak.sh, and since the
support looks like it may be broken I haven't enabled it.) This patch
adds the complementary -z dynamic-undefined-weak, extends both options
to affect building of shared libraries as well as executables, and
adds support for the option on powerpc.
include/
* bfdlink.h (struct bfd_link_info <dynamic_undefined_weak>):
Revise comment.
bfd/
* elflink.c (_bfd_elf_adjust_dynamic_symbol): Hide undefweak
or make dynamic for info->dynamic_undefined_weak 0 and 1.
* elf32-ppc.c:Formatting.
(ensure_undefweak_dynamic): Don't make dynamic when
info->dynamic_undefined_weak is zero.
(allocate_dynrelocs): Discard undefweak dyn_relocs for
info->dynamic_undefined_weak. Discard undef dyn_relocs when
not default visibility. Discard undef and undefweak
dyn_relocs earlier.
(ppc_elf_relocate_section): Adjust to suit.
* elf64-ppc.c: Formatting.
(ensure_undefweak_dynamic): Don't make dynamic when
info->dynamic_undefined_weak is zero.
(allocate_dynrelocs): Discard undefweak dyn_relocs for
info->dynamic_undefined_weak. Discard them earlier.
ld/
* ld.texinfo (dynamic-undefined-weak): Document.
(nodynamic-undefined-weak): Document that this option now can
be used with shared libs.
* emulparams/dynamic_undefined_weak.sh: Support -z
dynamic-undefined-weak.
* emulparams/elf32ppccommon.sh: Include dynamic_undefined_weak.sh.
* testsuite/ld-undefined/weak-undef.exp (undef_weak_so),
(undef_weak_exe): New. Use them. Add -z dynamic-undefined-weak
and -z nodynamic-undefined-weak tests.
* Makefile.am: Update powerpc dependencies.
* Makefile.in: Regenerate.
commit 902e9fc76a
Author: Maciej W. Rozycki <macro@imgtec.com>
Date: Tue Feb 21 01:46:42 2017 +0000
PR ld/20828: Move symbol version processing ahead of GC symbol sweep
breaks version definition with --version-script --soname. This patch
fixes it by getting soname index before generating the version definition
section.
bfd/
PR ld/21389
* elflink.c (bfd_elf_size_dynamic_sections): Get soname index
before generating the version definition section.
ld/
PR ld/21389
* testsuite/ld-elf/pr21389.map: New file.
* testsuite/ld-elf/pr21389.s: Likewise.
* testsuite/ld-elf/pr21389a.d: Likewise.
* testsuite/ld-elf/pr21389b.d: Likewise.
* testsuite/ld-elf/pr21389c.d: Likewise.
We already handle the case of an object file first defining a symbol
that a later shared library also defines where the symbol types are
incompatible. This patch handles the reverse, when a later object
file defines an incompatible symbol defined by an earlier shared
library.
bfd/
* elflink.c (_bfd_elf_merge_symbol): Undo dynamic linking
state when a regular object file defines a symbol with
incompatible type to that defined by an earlier shared lib.
ld/
* testsuite/ld-elf/indirect5a.c,
* testsuite/ld-elf/indirect5b.c,
* testsuite/ld-elf/indirect5.map,
* testsuite/ld-elf/indirect5.out: New test.
* testsuite/ld-elf/indirect6a.c: Likewise.
* testsuite/ld-elf/indirect.exp (check_dynamic_syms): New proc.
Run new tests and check dynsyms.
PR21287 notes that .init_array is correctly given a type of
SHT_INIT_ARRAY while .init_array.nnn gets SHT_PROGBITS. This patch
fixes that problem, and properly drops warnings from the compiler that
would cause the testsuite to fail. My a44d0bd78 change to check
ld_compile status, necessary to pick up compile errors, also meant
warnings were not ignored.
bfd/
PR 21287
* elf.c (special_sections_f): Match .fini_array and .fini_array.*.
(special_sections_i): Likewise for .init_array.
(special_sections_p): Likewise for .preinit_array.
ld/
PR 21287
* testsuite/ld-elf/init-fini-arrays.d: Match INIT_ARRAY and FINI_ARRAY.
* testsuite/ld-elf/init-fini-arrays.s: Use %init_array and %fini_array
section types.
* testsuite/lib/ld-lib.exp (default_ld_compile): Trim assembler
warnings about "ignoring incorrect section type".
(run_ld_link_exec_tests, run_cc_link_tests): Delete old comment.
My mips-linux and mips64-linux testsuite runs have been failing a
bunch of visibility tests, seemingly randomly. It turns out the
problem occurs when object files are left over in ld/tmpdir from a
previous run.
* testsuite/ld-elfvsb/elfvsb.exp (visibility_run): Delete
sh1p.o, sh2p.o, sh1np.o and sh2np.o before compiling. Use
remote_file host exists rather than file exists.
Since not all compilers support -no-pie, NOPIE_CFLAGS and NOPIE_LDFLAGS
should be used to disable PIE.
PR ld/21090
* testsuite/ld-x86-64/x86-64.exp (undefined_weak): Use
NOPIE_CFLAGS and NOPIE_LDFLAGS to disable PIE for the non-pie
version of the test.
Since common symbols that are turned into definitions don't have the
DEF_REGULAR flag set, we need to check ELF_COMMON_DEF_P for common
symbols.
bfd/
PR ld/19579
PR ld/21306
* elf32-s390.c (elf_s390_finish_dynamic_symbol): Check
ELF_COMMON_DEF_P for common symbols.
* elf64-s390.c (elf_s390_finish_dynamic_symbol): Likewise.
* elf64-x86-64.c (elf_x86_64_relocate_section): Likewise.
* elflink.c (_bfd_elf_merge_symbol): Revert commits
202ac193bb and
07492f668d.
ld/
PR ld/19579
PR ld/21306
* testsuite/ld-elf/pr19579a.c (main): Updated.
bfd/
* coffgen.c (_bfd_coff_gc_mark_hook): Handle PE weak
external symbols with a definition.
(_bfd_coff_gc_mark_extra_sections): Fix typo.
ld/
* testsuite/ld-pe/pe.exp: New test.
* testsuite/ld-pe/weakdef-1.s: New test source.
* testsuite/ld-pe/weakdef-1.d: New test.
On powerpc-linux, the second PT_LOAD header is RWE. On hppa64-linux,
only one PT_LOAD header is present. Since the tests are really about
the new GNU_MBIND headers, ignore PT_LOAD and its section mapping.
* testsuite/ld-elf/mbind1a.d: Remove matches for PT_LOAD segments.
* testsuite/ld-elf/mbind1b.d: Likewise.
* testsuite/ld-elf/mbind1c.d: Likewise.
Complement commit 902e9fc76a ("PR ld/20828: Move symbol version
processing ahead of GC symbol sweep"), commit b531344c34 ("PR
ld/20828: Reorder the symbol sweep stage of section GC") and commit
81ff47b3a5 ("PR ld/20828: Fix linker script symbols wrongly forced
local with section GC"), and prevent symbols forcibly entered in the
output file with the use of the `--undefined=' or `--require-defined='
linker command line options or the EXTERN linker script command from
being swept in section garbage collection and consequently recorded in
the dynamic symbol table as local entries. This happens in certain
circumstances, where a symbol reference also exists in one of the static
input files, however only in a section which is garbage-collected and
does not make it to the output file, and the symbol is defined in a
dynamic object present in the link.
For example with the `i386-linux' target and the `pr21233.s' and
`pr21233-l.s' sources, and the `pr21233.ld' linker script included with
this change we get:
$ as -o pr21233-l.o pr21233-l.s
$ ld -shared -T pr21233.ld -o libpr21233.so pr21233-l.o
$ as -o pr21233.o pr21233.s
$ ld --gc-sections -e foo --require-defined=bar -T pr21233.ld -o pr21233 pr21233.o libpr21233.so
$ readelf --dyn-syms pr21233
Symbol table '.dynsym' contains 2 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 00000000 0 NOTYPE LOCAL DEFAULT UND
1: 00000000 0 OBJECT LOCAL DEFAULT UND bar
$
which makes the run-time `bar' dependency of the `pr21233' executable
different from its corresponding link-time dependency, i.e. the presence
of `libpr21233.so' and its `bar' symbol is required at the link time,
however at the run time a copy of `libpr21233.so' without `bar' will do.
Similarly with `--undefined=' and EXTERN which do not actually require
the reference to the symbol requested to be satisfied with a definition
at the link time, however once the definition has been pulled at the
link time, so it should at the dynamic load time.
Additionally with the `mips-linux' target we get:
$ ld --gc-sections -e foo --require-defined=bar -T pr21233.ld -o pr21233 pr21233.o libpr21233.so
ld: BFD (GNU Binutils) 2.28.51.20170324 assertion fail .../bfd/elfxx-mips.c:3861
$
as the target is not prepared to handle such a local dynamic symbol.
With this change in effect we get:
$ readelf --dyn-syms pr21233
Symbol table '.dynsym' contains 2 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 00000000 0 NOTYPE LOCAL DEFAULT UND
1: 00000000 0 OBJECT GLOBAL DEFAULT UND bar
$
instead, for both targets.
ld/
PR ld/21233
* ldlang.c (insert_undefined): Set `mark' for ELF symbols.
* testsuite/ld-elf/pr21233.sd: New test.
* testsuite/ld-elf/pr21233-l.sd: New test.
* testsuite/ld-elf/pr21233.ld: New test linker script.
* testsuite/ld-elf/pr21233-e.ld: New test linker script.
* testsuite/ld-elf/pr21233.s: New test source.
* testsuite/ld-elf/pr21233-l.s: New test source.
* testsuite/ld-elf/shared.exp: Run the new tests.
Mark an ALLOC section, which should be placed in special memory area,
with SHF_GNU_MBIND. Its sh_info field indicates the special memory
type. GNU_MBIND section names start with ".mbind" so that they are
placed as orphan sections by linker. All input GNU_MBIND sections
with the same sh_type, sh_flags and sh_info are placed in one output
GNU_MBIND section. In executable and shared object, create a
GNU_MBIND segment for each GNU_MBIND section and its segment type is
PT_GNU_MBIND_LO plus the sh_info value. Each GNU_MBIND segment is
aligned at page boundary.
The assembler syntax:
.section .mbind.foo,"adx",%progbits
^ 0: Special memory type.
|
'd' for SHF_GNU_MBIND.
.section .mbind.foo,"adx",%progbits,0x1
^ 1: Special memory type.
|
'd' for SHF_GNU_MBIND.
.section .mbind.bar,"adG",%progbits,.foo_group,comdat,0x2
^ 2: Special memory type.
|
'd' for SHF_GNU_MBIND.
bfd/
* elf.c (get_program_header_size): Add a GNU_MBIND segment for
each GNU_MBIND section and align GNU_MBIND section to page size.
(_bfd_elf_map_sections_to_segments): Create a GNU_MBIND
segment for each GNU_MBIND section.
(_bfd_elf_init_private_section_data): Copy sh_info from input
for GNU_MBIND section.
binutils/
* NEWS: Mention support for ELF SHF_GNU_MBIND and
PT_GNU_MBIND_XXX.
* readelf.c (get_segment_type): Handle PT_GNU_MBIND_XXX.
(get_elf_section_flags): Handle SHF_GNU_MBIND.
(process_section_headers): Likewise.
* testsuite/binutils-all/mbind1.s: New file.
* testsuite/binutils-all/objcopy.exp: Run readelf test on
mbind1.s.
gas/
* NEWS: Mention support for ELF SHF_GNU_MBIND.
* config/obj-elf.c (section_match): New.
(get_section): Match both sh_info and group name.
(obj_elf_change_section): Add argument for sh_info. Pass both
sh_info and group name to get_section. Issue an error for
SHF_GNU_MBIND section without SHF_ALLOC. Set sh_info.
(obj_elf_parse_section_letters): Set SHF_GNU_MBIND for 'd'.
(obj_elf_section): Support SHF_GNU_MBIND section info.
* config/obj-elf.h (obj_elf_change_section): Add argument for
sh_info.
* config/tc-arm.c (start_unwind_section): Pass 0 as sh_info to
obj_elf_change_section.
* config/tc-ia64.c (obj_elf_vms_common): Likewise.
* config/tc-microblaze.c (microblaze_s_data): Likewise.
(microblaze_s_sdata): Likewise.
(microblaze_s_rdata): Likewise.
(microblaze_s_bss): Likewise.
* config/tc-mips.c (s_change_section): Likewise.
* config/tc-msp430.c (msp430_profiler): Likewise.
* config/tc-rx.c (parse_rx_section): Likewise.
* config/tc-tic6x.c (tic6x_start_unwind_section): Likewise.
* doc/as.texinfo: Document 'd' for SHF_GNU_MBIND.
* testsuite/gas/elf/elf.exp: Run section12a, section12b and
section13.
* testsuite/gas/elf/section10.d: Updated.
* testsuite/gas/elf/section10.s: Likewise.
* testsuite/gas/elf/section12.s: New file.
* testsuite/gas/elf/section12a.d: Likewise.
* testsuite/gas/elf/section12b.d: Likewise.
* testsuite/gas/elf/section13.l: Likewise.
* testsuite/gas/elf/section13.d: Likewise.
* testsuite/gas/elf/section13.s: Likewise.
include/
* elf/common.h (PT_GNU_MBIND_NUM): New.
(PT_GNU_MBIND_LO): Likewise.
(PT_GNU_MBIND_HI): Likewise.
(SHF_GNU_MBIND): Likewise.
ld/
* NEWS: Mention support for ELF SHF_GNU_MBIND and
PT_GNU_MBIND_XXX.
* emultempl/elf32.em (gld${EMULATION_NAME}_place_orphan): Place
input GNU_MBIND sections with the same type, attributes and
sh_info field into a single output GNU_MBIND section.
* testsuite/ld-elf/elf.exp: Run mbind2a and mbind2b.
* testsuite/ld-elf/mbind1.s: New file.
* testsuite/ld-elf/mbind1a.d: Likewise.
* testsuite/ld-elf/mbind1b.d: Likewise.
* testsuite/ld-elf/mbind1c.d: Likewise.
* testsuite/ld-elf/mbind2a.s: Likewise.
* testsuite/ld-elf/mbind2b.c: Likewise.
The Nios II processor documentation defines relocations for PIC and
shared libraries as part of the GNU/Linux ABI only; GCC rejects -fpic
on bare-metal.
2017-03-21 Sandra Loosemore <sandra@codesourcery.com>
ld/
* testsuite/lib/ld-lib.exp (check_shared_lib_support): Return
false for nios2-*-elf.
ld/
* ldlang.c (lang_check_section_addresses): Check for address space
overflow.
* testsuite/ld-checks/checks.exp (overflow_check): New procedure
* testsuite/ld-checks/over.s: New test source.
* testsuite/ld-checks/over.d: New test.
* testsuite/ld-checks/over2.s: New test source.
* testsuite/ld-checks/over2.d: New test.
This gcc option isn't well supported, so use the actual linker option
we want to test.
* testsuite/ld-elf/shared.exp: Use -Wl,-export-dynamic rather
than -rdynamic.
These targets use the generic ELF support, so don't handle orphans
well. The patch also updates the orphan doco to reflect this fact,
and deletes some ELF details that don't really add anything.
* ld.texinfo (Orphan Sections): Mention that not all targets
handle orphans well. Delete ELF details.
* testsuite/ld-elf/orphan-9.d: Don't run for i860 and i960.
* testsuite/ld-elf/orphan-10.d: Likewise.
Given a linker script fragment like this:
SECTIONS {
. = 0x1000;
.text : AT(0x100) { *(.text) }
.data : AT(0x200) { *(.data) }
.rodata : AT(0x300) { *(.rodata) }
}
and an input file containing sections, '.text', '.data.1', and
'.rodata', then we'd expect the linker to place '.text' and '.rodata' in
the obvious way, and the '.data.1' orphan section would be located after
the '.data' section (assuming similar section properties).
Further, I believe that the expectation would be that the LMA for the
orphan '.data.1' section would start from 0x200 (as there is no '.data'
content).
However, right now, the LMA for '.data.1' would be 0x101, following on
from the '.text' section, this is because the change in LMA for the
'.data' section is not noticed by the linker, if there's no content in
the '.data' section.
What can be even more confusing to a user (though the cause is obvious
once you understand what's going on) is that adding some content to
'.data' will cause the orphan '.data.1' to switch to an LMA based off of
0x200.
This commit changes the behaviour so that an empty section that is in
the default lma region, and sets its lma, will adjust the lma of the
default region, this change will then be reflected in following sections
within the default lma memory region.
There's a new test to cover this issue that passes on a range of
targets, however, some targets generate additional sections, or have
stricter memory region size requirements that make it harder to come
up with a generic pass pattern, that still tests the required
features. For now I've set the test to ignore these targets.
ld/ChangeLog:
* ldlang.c (lang_size_sections_1): Shortcut loop only after
tracking changes to the default regions LMA.
* testsuite/ld-elf/orphan-9.ld: Extend header comment.
* testsuite/ld-elf/orphan-10.d: New file.
* testsuite/ld-elf/orphan-10.s: New file.
* NEWS: Mention change in behaviour.
When picking an lma_region for an orphan section we currently create a
new lang_output_section_statement_type and then populate this with the
orphan section.
The problem is that the lang_output_section_statement_type has a prev
pointer that links back to the previous output section. For non-orphan
output sections, that are created in linker script order, the prev
pointer will point to the output section that appears previous in linker
script order, as you'd probably expect.
The problem is that orphan sections are placed after processing the
linker script, and so, in the case of an output section created for an
orphan input section, the prev pointer actually points to the last
output section created.
This causes some unexpected behaviour when the orphan section is not
placed after the last non-orphan section that was created.
For example, consider this linker script:
MEMORY {
TEXT : ORIGIN = 0x200, LENGTH = 0x10
RODATA : ORIGIN = 0x400, LENGTH = 0x10
}
SECTIONS {
.text : {*(.text) } AT>TEXT
.data : AT(0x300) { *(.data) }
.rodata : { *(.rodata) } AT>RODATA
}
If we are processing an orphan section '.data.1' and decide to place
this after '.data', then the output section created will have a prev
pointer that references the '.rodata' output section. The result of
this is that '.data.1' will actually be assigned to the RODATA lma
region, which is probably not the expected behaviour.
The reason why '.data.1' is placed into the lma region of the '.rodata'
section is that lma region propagation is done at the time we create the
output section, based on the previous output section pointer, which is
really just a last-output-section-created pointer at that point in time,
though the prev point is fixed up later to reflect the true order of the
output sections.
The solution I propose in this commit is to move the propagation of lma
regions into a separate pass of the linker, rather than performing this
as part of the enter/exit of output sections during linker script
parsing.
During this later phase we have all of the output sections to hand, and
the prev/next points have been fixed up by this point to reflect the
actual placement ordering.
There's a new test to cover this issue that passes on a range of
targets, however, some targets generate additional sections, or have
stricter memory region size requirements that make it harder to come
up with a generic pass pattern, that still tests the required
features. For now I've set the test to ignore these targets.
ld/ChangeLog:
* ldlang.c (lang_leave_output_section_statement): Move lma_region
logic to...
(lang_propagate_lma_regions): ...this new function.
(lang_process): Call new function.
* testsuite/ld-elf/orphan-9.d: New file.
* testsuite/ld-elf/orphan-9.ld: New file.
* testsuite/ld-elf/orphan-9.s: New file.
* NEWS: Mention change in behaviour.
A number of places in elf32-nios.c created dynamic sections but didn't
set the hash table dynobj. That meant we could have duplicate dynamic
sections connected to a number of bfds, so size_dynamic_sections
didn't properly discard or allocate contents.
Also, the entire set of dynamic sections was created in check_relocs
on seeing GOT relocs, when only .got related sections are needed,
probably done to hide segfaults later in finish_dynamic_sections.
The patch fixes these issues and makes the assembler emit errors when
nios2 lacks the necessary pc-relative relocs for subtraction
expressions, rather than silently generating bad code.
eg. ld-elf/merge. I've also tidied uses of elf32_nios2_hash_table and
elf_hash_table.
bfd/
PR 20995
* elf32-nios2.c (nios2_elf32_relocate_section): Use htab
rather than elf32_nios2_hash_table or elf_hash_table.
(create_got_section): Likewise.
(nios2_elf32_finish_dynamic_symbol): Likewise.
(nios2_elf32_adjust_dynamic_symbol): Likewise.
(nios2_elf32_size_dynamic_sections): Likewise.
(nios2_elf32_check_relocs): Delete dynobj, sgot, and srelgot
vars. Use htab equivalents directly instead. Don't create
all dynamic sections on needing just the GOT. Use a goto
rather than a fall-through with reloc test. Ensure
htab->dynobj is set when making dynamic sreloc section.
(nios2_elf32_finish_dynamic_sections): Delete dynobj, use htab
equivalent directly instead. Don't segfault on looking for
.dynamic when dynamic sections have not been created. Don't
segfault on .got.plt being discarded.
(nios2_elf32_size_dynamic_sections): Delete plt and got vars.
Don't set "relocs" on .rela.plt. Do handle .sbss. Delete
fixme and another not so relevant comment.
(nios2_elf_add_symbol_hook): Delete dynobj var. If not
already set, set hash table dynobj on creating .sbss.
gas/
* config/tc-nios2.h (TC_FORCE_RELOCATION_SUB_LOCAL): Define.
ld/
* testsuite/ld-elf/merge.d: xfail for nios.
This came up because I was looking at ld/tmpdir/addpcis.o and noticed
the odd addends on REL16DX_HA. They ought to both be -4. The error
crept in due REL16DX_HA howto being pc-relative (as indeed it should
be), and code at gas/write.c:1001 after this comment
/* Make it pc-relative. If the back-end code has not
selected a pc-relative reloc, cancel the adjustment
we do later on all pc-relative relocs. */
*not* cancelling the pc-relative adjustment. So I've made a dummy
non-relative split reloc so that the generic code handles this, rather
than attempting to add hacks later in md_apply_fix which would not be
very robust. Having the new internal reloc also makes it easy to
support
addpcis rx,sym@ha
as an equivalent to
addpcis rx,(sym-0f)@ha
0:
The patch also fixes overflow checking, which must test whether the
addi will overflow too since @l relocs don't have any overflow check.
Lastly, since I was poking at md_apply_fix, I arranged to have the
generic gas/write.c code emit errors for subtraction expressions where
we lack reloc support.
include/
* elf/ppc64.h (R_PPC64_16DX_HA): New. Expand fake reloc comment.
* elf/ppc.h (R_PPC_16DX_HA): Likewise.
bfd/
* reloc.c (BFD_RELOC_PPC_16DX_HA): New.
* elf64-ppc.c (ppc64_elf_howto_raw <R_PPC64_16DX_HA>): New howto.
(ppc64_elf_reloc_type_lookup): Translate new bfd reloc.
(ppc64_elf_ha_reloc): Correct overflow test on REL16DX_HA.
(ppc64_elf_relocate_section): Likewise.
* elf32-ppc.c (ppc_elf_howto_raw <R_PPC_16DX_HA>): New howto.
(ppc_elf_reloc_type_lookup): Translate new bfd reloc.
(ppc_elf_check_relocs): Handle R_PPC_16DX_HA to pacify gcc.
* libbfd.h: Regenerate.
* bfd-in2.h: Regenerate.
gas/
* config/tc-ppc.c (md_assemble): Use BFD_RELOC_PPC_16DX_HA for addpcis.
(md_apply_fix): Remove fx_subsy check. Move code converting to
pcrel reloc earlier and handle BFD_RELOC_PPC_16DX_HA. Remove code
emiiting errors on seeing fx_pcrel set on unexpected relocs, as
that is done now by the generic code via..
* config/tc-ppc.h (TC_FORCE_RELOCATION_SUB_LOCAL): ..this. Define.
(TC_VALIDATE_FIX_SUB): Define.
ld/
* testsuite/ld-powerpc/addpcis.d: Define ext1 and ext2 at
limits of addpcis range.
Interpret the `jalr $0, $25' instruction encoding with an R_MIPS_JALR
relocation attached as an alias to `jr $25' and convert the jump to an
equivalent branch where possible, consequently covering the MIPSr6
architecture for the purpose of this optimization too.
bfd/
* elfxx-mips.c (mips_elf_perform_relocation): Also handle the
`jalr $0, $25' instruction encoding.
gas/
* testsuite/gas/mips/jalr4.s: Add `jalr $0, $25' instructions.
* testsuite/gas/mips/jalr4.d: Adjust accordingly. Remove MIPSr6
encoding patterns.
* testsuite/gas/mips/jalr4-n64.d: Likewise.
* testsuite/gas/mips/mipsr6@jalr4.d: New test.
* testsuite/gas/mips/mipsr6@jalr4-n32.d: New test.
* testsuite/gas/mips/mipsr6@jalr4-n64.d: New test.
ld/
* testsuite/ld-mips-elf/jalr4.dd: Adjust for `jalr $0, $25'
instructions.
* testsuite/ld-mips-elf/jalr4-r6.dd: New test.
* testsuite/ld-mips-elf/mips-elf.exp: Run the new test.
HPUX has a different .comm syntax, and anything in the first column
is a label.
gas/
* testsuite/gas/elf/strtab.s: Don't put directives on first
column or continuation with labels not in first column.
ld/
* testsuite/ld-elf/elf.exp: Xfail pr20995 tests on hppa64-hpux.
Set up HPUX defsym. Run pr14170 tests and build symbol3 objects,
defining HPUX where necessary. Define HPUX for implib tests.
* testsuite/ld-elf/comm-data4.d: Run for hpux.
* testsuite/ld-elf/endsym.d: Likewise.
* testsuite/ld-elf/linkoncerdiff.d: Likewise.
* testsuite/ld-elf/comm-data4.s: Add alternate .comm when HPUX.
* testsuite/ld-elf/comm-data5.s: Likewise.
* testsuite/ld-elf/endsym.s: Likewise.
* testsuite/ld-elf/pr14170c.s: Likewise.
* testsuite/ld-elf/symbol3.s: Likewise.
* testsuite/ld-elf/implib.s: Likewise. Don't start directives
in first column.
* testsuite/ld-elf/linkoncerdiff2.s: Don't use numeric labels.
* testsuite/ld-elf/warn3.d: Run for hpux.
* testsuite/ld-scripts/rgn-at10.d: Xfail for hpux.
* testsuite/ld-scripts/rgn-at11.d: Likewise.
* testsuite/ld-scripts/size-2.d: Remove xfail for hpux.
Fix a commit 74e1a04b97 ("More fixes for reading corrupt ELF files.")
`readelf --version-info' regression that caused "Version definition past
end of section" to be always printed at the end, even with good section
data.
For example with the `mips-linux' target we get:
$ cat ver_def.s
.data
.globl new_foo
.type new_foo, %object
new_foo:
.symver new_foo, foo@@ver_foo
$ cat ver_def.ver
{ global: *foo*; local: *; };
$ as -o ver_def.o ver_def.s
$ ld -e 0 --export-dynamic --version-script=ver_def.ver -o ver_def ver_def.o
$ readelf -V ver_def
Version symbols section '.gnu.version' contains 4 entries:
Addr: 000000000000007e Offset: 0x01007e Link: 2 (.dynsym)
000: 0 (*local*) 2 (ver_foo) 1 (*global*) 2 (ver_foo)
Version definition section '.gnu.version_d' contains 2 entries:
Addr: 0x0000000000000088 Offset: 0x010088 Link: 3 (.dynstr)
000000: Rev: 1 Flags: BASE Index: 1 Cnt: 1 Name: ver_def
0x001c: Rev: 1 Flags: none Index: 2 Cnt: 1 Name: ver_foo
Version definition past end of section
$
The cause is the `if (idx + ent.vd_next <= idx)' condition introduced to
ensure forward progress, which however always triggers for good version
definition section data as the last entry will have its `vd_next' value
set to 0.
Adjust the condition then, to say `if (idx + ent.vd_next < idx)' instead
and to ensure forward progress limit the number of entries processed to
the size of the version definition section, removing the problematic
message from output quoted above, while ensuring the original PR 17531
test case is still handled gracefully.
Add a suitable test case so that we have `readelf --version-info'
coverage; due to the lack of infrastructure needed to run the linker in
the `binutils' test suite and limited justification to implement it add
a new `readelf.exp' script to the `ld' test suite instead, intended to
gather any `readelf' test cases that require the linker to be run. If
ever we decide to have linker infrastructure added to the `binutils'
test suite, then the script can be moved between the test suites.
binutils/
* readelf.c (process_version_sections) <SHT_GNU_verdef>: Limit
the number of entries processed by the section size. Don't
break out of the loop if `ent.vd_next' is 0.
ld/
* testsuite/ld-elf/ver_def.d: New test.
* testsuite/ld-elf/ver_def.ld: New test linker script.
* testsuite/ld-elf/ver_def.ver: New test version script.
* testsuite/ld-elf/ver_def.s: New test source.
* testsuite/ld-elf/readelf.exp: New test script.
Remove separate original NewABI JALR relocation handling, introduced
with commit d06471104a ("relax jalr $t9 [R_MIPS_JALR symbol] to bal
symbol"), <https://sourceware.org/ml/binutils/2003-03/msg00394.html>,
and only used by LD with the `--relax' option specified, and rely solely
on `mips_elf_perform_relocation' code, which has been introduced with
commit 1367d393bb ("On the RM9000 convert jal to bal if in range"),
<https://www.sourceware.org/ml/binutils/2004-12/msg00088.html> and since
made more complete, across all the three ABIs.
Also remove the `--relax' option, now irrelevant, from the tests added
with the former commit.
bfd/
* elfxx-mips.h (_bfd_mips_relax_section): Remove prototype.
* elfxx-mips.c (_bfd_mips_relax_section): Remove function.
* elf64-mips.c (bfd_elf64_bfd_relax_section): Remove macro.
* elfn32-mips.c (bfd_elf32_bfd_relax_section): Likewise.
ld/
* testsuite/ld-mips-elf/relax-jalr-n32.d: Remove `--relax'
option.
* testsuite/ld-mips-elf/relax-jalr-n32-shared.d: Likewise.
* testsuite/ld-mips-elf/relax-jalr-n64.d: Likewise.
* testsuite/ld-mips-elf/relax-jalr-n64-shared.d: Likewise.
Discard R_MIPS_JALR and R_MICROMIPS_JALR relocations associated with
jumps that cannot be converted to an equivalent branch right away in
`mips_elf_calculate_relocation' rather than letting them through to
`mips_elf_perform_relocation'. This includes cross-mode jumps which
need to flip the ISA bit or jumps to a misaligned location that cannot
be encoded with a branch, in addition to preemptible symbol references
already handled.
Cross-mode jumps are actually already rejected as the conversion is made
in `mips_elf_perform_relocation', so in this case this change only saves
some processing. Jumps to a misaligned location are however converted,
with bits causing misalignment lost, making resulting code functionally
different even if the lone effect is avoiding an address error exception
with an instruction fetch at the jump destination requested.
Add test cases suitable, also including GAS verification to confirm that
the JALR relocations explicitly requested have indeed been output in the
intermediate objects used.
bfd/
* elfxx-mips.c (mips_elf_calculate_relocation) <R_MIPS_JALR>
<R_MICROMIPS_JALR>: Discard relocation if `cross_mode_jump_p'
or misaligned.
gas/
* testsuite/gas/mips/jalr4.d: New test.
* testsuite/gas/mips/jalr4-n32.d: New test.
* testsuite/gas/mips/jalr4-n64.d: New test.
* testsuite/gas/mips/jalr4.s: New test source.
* testsuite/gas/mips/mips.exp: Run the new tests.
ld/
* testsuite/ld-mips-elf/jalr4.dd: New test.
* testsuite/ld-mips-elf/mips-elf.exp: Run the new test.
PR 20744
bfd/
* elf32-ppc.c (ppc_elf_howto_raw): Correct dst_mask on all VLE
16D relocations.
(ppc_elf_vle_split16): Correct field mask and shift for 16D relocs.
(ppc_elf_relocate_section): Correct calculation for VLE SDAREL
relocs.
ld/
* testsuite/ld-powerpc/vle-reloc-2.s: Use r6 for last insn of
each group.
* testsuite/ld-powerpc/vle-reloc-2.d: Update for above change
and sdarel reloc fix.
Complement commit b531344c34 ("PR ld/20828: Reorder the symbol sweep
stage of section GC") and commit 81ff47b3a5 ("PR ld/20828: Fix linker
script symbols wrongly forced local with section GC") and move symbol
version processing ahead of the symbol sweep stage of section GC, all in
`bfd_elf_size_dynamic_sections', so that version symbols created stay in
the global scope and are not output as local symbols to the dynamic
symbol table in the presence of corresponding symbol definitions pulled
from a DSO involved in a link.
Consolidate the whole of symbol version processing into a single block
from all parts scattered across the function and rearranging the local
variables used as necessary, however leaving the setting of dynamic
entries associated with the DT_VERDEF, DT_VERDEFNUM, DT_VERNEED and
DT_VERNEEDNUM tags and the SEC_EXCLUDE flag for unused `.gnu.version'
section in the original places.
With the rearrangement of code blocks `Elf_Internal_Verneed *t' would
shadow the previous definition of `struct bfd_elf_version_tree *t', so
rename the former variable to `vn'.
bfd/
PR ld/20828
* elflink.c (bfd_elf_size_dynamic_sections): Move symbol version
processing ahead of the call to `elf_gc_sweep_symbol'.
ld/
PR ld/20828
* testsuite/ld-elf/pr20828-d.sd: New test.
* testsuite/ld-elf/pr20828-e.sd: New test.
* testsuite/ld-elf/pr20828-v.od: New test.
* testsuite/ld-elf/pr20828-v.ver: New test version script.
* testsuite/ld-elf/pr20828-v.ld: New test linker script.
* testsuite/ld-elf/pr20828.ld: Add `.gnu.version' and
`.gnu.version_d'.
* testsuite/ld-elf/shared.exp: Run the new tests.