* elf64-ppc.c: Define more insns used in plt call stubs.
(ppc64_elf_brtaken_reloc): Assume isa version 2 or above.
(ppc64_elf_relocate_section): Likewise.
(enum ppc_stub_type): Add ppc_stub_plt_call_r2save.
(struct ppc_link_hash_table): Increase size of stub_count array.
Add plt_stub_align and plt_thread_safe.
(ALWAYS_USE_FAKE_DEP, ALWAYS_EMIT_R2SAVE): Define.
(plt_stub_size, plt_stub_pad): New functions.
(build_plt_stub): Emit barriers for power7 thread safety. Don't
emit needless save of r2.
(build_tls_get_addr_stub): Adjust params.
(ppc_build_one_stub): Handle ppc_stub_plt_call_r2save and aligning
plt stubs. Adjust build_*plt_stub calls.
(ppc_size_one_stub): Similarly.
(ppc64_elf_size_stubs): Accept plt_thread_safe and plt_stub_align
params. Choose default for plt_thread_safe based on existence of
calls to thread creation functions. Modify plt_call to
plt_call_r2save when no tocsave reloc found. Align tail of stub
sections.
(ppc64_elf_build_stubs): Align tail of stub sections. Adjust
output of stub statistics.
(ppc64_elf_relocate_section): Handle ppc_stub_plt_call_r2save.
* elf64-ppc.h (ppc64_elf_size_stubs): Update prototype.
ld/
* emultempl/ppc64elf.em (PARSE_AND_LIST_PROLOGUE,
PARSE_AND_LIST_LONGOPTS, PARSE_AND_LIST_OPTIONS,
PARSE_AND_LIST_ARGS_CASES): Handle --{no-,}plt-thread-safe and
--{no-,}plt-align.
(plt_thread_safe, plt_stub_align): New vars.
(gld${EMULATION_NAME}_after_allocation): Pass them to
ppc64_elf_size_stubs. Align stub sections according to plt_stub_align.
* ld.texinfo: Document new command line options, and an old
undocumented option.
Fix zero registers core files when built by gcc-4.7.
* elf64-x86-64.c (elf_x86_64_write_core_note): Remove variables p and
size. Call elfcore_write_note for the local variables. Remove the
final elfcore_write_note call. Add NOTREACHED comments.
* dwarf2.c (_bfd_dwarf2_slurp_debug_info): Factor out the part
which reads DWARF2 and stores in stash from find_line.
(find_line) Call _bfd_dwarf2_slurp_debug_info.
* libbfd-in.h (_bfd_dwarf2_slurp_debug_info): Add declaration.
* libbfd.h (_bfd_dwarf2_slurp_debug_info): Regenerate.
* mach-o.c (dsym_subdir): The name of subdir where debug
information may be stored.
(bfd_mach_o_lookup_uuid_command): New. Lookup a load command whose
type is UUID.
(bfd_mach_o_dsym_for_uuid_p): New. Check if the specified BFD is
corresponding to the executable.
(bfd_mach_o_find_dsym): New. Find a debug information BFD in the
specified binary file.
(bfd_mach_o_follow_dsym): New. Find a debug information BFD for
the original BFD.
(bfd_mach_o_find_nearest_line): Check dSYM files for Mach-O
executables, dylibs, and bundles.
(bfd_mach_o_close_and_cleanup): Clean up BFDs for the dSYM file.
* mach-o.h (dsym_bfd): The BFD of the dSYM file.
bfd:
* mach-o.c (bfd_mach_o_mangle_symbols): Put in the section index
for stabd symbols.
(bfd_mach_o_primary_symbol_sort_key): Adjust for stabs.
(bfd_mach_o_cf_symbols): Likewise.
gas:
* config/obj-macho.c (obj_macho_process_stab): New.
* config/obj-macho.h (OBJ_PROCESS_STAB): Define.
(obj_macho_process_stab): Declare.
bfd:
* mach-o.c (bfd_mach_o_write_symtab): Fill in the string table index
as the value of an indirect symbol. Keep the string table index in
non-indirect syms for reference.
(bfd_mach_o_write_dysymtab): New.
(bfd_mach_o_primary_symbol_sort_key): New.
(bfd_mach_o_cf_symbols): New.
(bfd_mach_o_sort_symbol_table): New.
(bfd_mach_o_mangle_symbols): Return early if no symbols. Sort symbols.
If we are emitting a dysymtab, process indirect symbols and count the
number of each other kind.
(bfd_mach_o_mangle_sections): New.
(bfd_mach_o_write_contents): Split out some pre-requisite code into
the command builder. Write dysymtab if the command is present.
(bfd_mach_o_count_sections_for_seg): New.
(bfd_mach_o_build_seg_command): New.
(bfd_mach_o_build_dysymtab_command): New.
(bfd_mach_o_build_commands): Reorganize to support the fact that some
commands are optional and should not be emitted if there are no
sections or symbols.
(bfd_mach_o_set_section_contents): Amend comment.
* mach-o.h: Amend and add to comments.
(mach_o_data_struct): Add fields for dysymtab symbols counts and a
pointer to the indirects, when present.
(bfd_mach_o_should_emit_dysymtab): New macro.
(IS_MACHO_INDIRECT): Likewise.
gas/testsuite:
* gas/mach-o/dysymtab-1-64.d: New.
* gas/mach-o/dysymtab-1.d: New.
* gas/mach-o/symbols-1-64.d: New.
* gas/mach-o/symbols-1.d: New.
* gas/mach-o/symbols-base-64.s: New.
* gas/mach-o/symbols-base.s: New.
* elf32-rl78.c (rl78_elf_howto_table): Add R_RL78_RH_RELAX.
(rl78_reloc_map): Add BFD_RELOC_RL78_RELAX.
(rl78_elf_relocate_section): Add R_RL78_RH_RELAX, R_RL78_RH_SFR,
and R_RL78_RH_SADDR.
(rl78_elf_finish_dynamic_sections): Only validate PLT section if
we didn't relax anything, as relaxing might remove a PLT reference
after we've set up the table.
(elf32_rl78_relax_delete_bytes): New.
(reloc_bubblesort): New.
(rl78_offset_for_reloc): New.
(relax_addr16): New.
(rl78_elf_relax_section): Add support for relaxing long
instructions into short ones.
[gas]
* config/rl78-defs.h (rl78_linkrelax_addr16): Add.
(rl78_linkrelax_dsp, rl78_linkrelax_imm): Remove.
* config/rl78-parse.y: Tag all addr16 and branch patterns with
relaxation markers.
* config/tc-rl78.c (rl78_linkrelax_addr16): New.
(rl78_linkrelax_branch): New.
(OPTION_RELAX): New.
(md_longopts): Add relax option.
(md_parse_option): Add OPTION_RELAX.
(rl78_frag_init): Support relaxation.
(rl78_handle_align): New.
(md_assemble): Support relaxation.
(md_apply_fix): Likewise.
(md_convert_frag): Likewise.
* config/tc-rl78.h (MAX_MEM_FOR_RS_ALIGN_CODE): New.
(HANDLE_ALIGN): New.
(rl78_handle_align): Declare.
* config/rl78-parse.y (rl78_bit_insn): New. Set it for all bit
insn patterns.
(find_bit_index): New. Strip .BIT suffix off relevent
expressions for bit insns.
(rl78_lex): Exclude bit suffixes from expression parsing.
[include/elf]
* rl78.h (R_RL78_RH_RELAX, R_RL78_RH_SFR, R_RL78_RH_SADDR): New.
(RL78_RELAXA_BRA, RL78_RELAXA_ADDR16: New.
* elf32-s390.c (elf_s390_relocate_section): Add check for debugging
section in LD to LE linker relaxation for R_390_TLS_LDO32.
* elf64-s390.c (elf_s390_relocate_section): Likewise for
R_390_TLS_LDO64.
* mach-o-i386.c (bfd_mach_o_section_type_valid_for_tgt): Define NULL.
* mach-o-target.c (bfd_mach_o_backend_data): Initialize bfd_mach_o_section_type_valid_for_tgt
* mach-o-x86-64.c (bfd_mach_o_section_type_valid_for_x86_64): New.
(bfd_mach_o_section_type_valid_for_tgt): Set to bfd_mach_o_section_type_valid_for_x86_64.
* mach-o.c (bfd_mach_o_section_type_name): Reorder and eliminate dup.
(bfd_mach_o_section_attribute_name): Reorder.
(bfd_mach_o_get_section_type_from_name): If the target has defined a validator for section
types, then use it.
* mach-o.h (bfd_mach_o_get_section_type_from_name): Alter declaration to include the bfd.
gas:
* config/obj-macho.c (obj_mach_o_section): Account for target-dependent section
types. Improve error handling when wrong section types/attributes are specified.
gas/testsuite:
* gas/mach-o/err-sections-1.s: New.
* gas/mach-o/err-sections-2.s: New.
* gas/mach-o/sections-3.d: New.
* gas/mach-o/sections-3.s: New.
Catherine Moore <clm@codesourcery.com>
Sandra Loosemore <sandra@codesourcery.com>
Richard Sandiford <rdsandiford@googlemail.com>
* elfxx-mips.c (mips_elf_local_pic_function_p): Return true when
H is a MIPS16 function with a kept 32-bit stub. Update comments.
(mips_elf_get_la25_target): New function.
(mips_elf_add_la25_intro): Change to use mips_elf_get_la25_target().
(mips_elf_add_la25_stub): Move compute of use_trampoline_p down,
change to use mips_elf_get_la25_target().
(mips_elf_relocation_needs_la25_stub): Add target_is_16_bit_code_p
parameter, add switch case for R_MIPS16_26.
(mips_elf_calculate_relocation): Redirect relocation to point to the
LA25 stub if it exists, instead of the MIPS16 stub. Update arguments
of call to mips_elf_relocation_needs_la25_stub(), don't use la25 stub
for mips16->mips16 calls.
(_bfd_mips_elf_check_relocs): Update arguments of call to
mips_elf_relocation_needs_la25_stub().
(mips_elf_create_la25_stub): Change to use mips_elf_get_la25_target().
* mach-o-i386.c (TARGET_PRIORITY): Define as 0 (top priority)
* mach-o-target.c (TARGET_NAME): Use TARGET_PRIORITY
* mach-o-x86-64.c (TARGET_PRIORITY): Define as 0 (top priority)
* mach-o.c (bfd_mach_o_header_p): Remove special handling for
mach-o-i386.
(TARGET_PRIORITY) Set 1 for mach-o-be and mach-o-le, and set 0 for
mach-o-fat.
* mach-o.c (bfd_mach_o_canonicalize_reloc): Update relocation
table only when there isn't the cahce.
(bfd_mach_o_get_dynamic_reloc_upper_bound): Need one more space
for a pointer for the watchdog.
(bfd_mach_o_canonicalize_dynamic_reloc): Utilize cache like
bfd_mach_o_canonicalize_reloc.
(bfd_mach_o_close_and_cleanup): Call bfd_mach_o_free_cached_info.
(bfd_mach_o_free_cached_info): Free up cache data.
* mach-o.h (reloc_cache): A place to store cache of dynamic relocs.
(bfd_mach_o_free_cached_info): Add declaration.
* elfcode.h (elf_checksum_contents): Read in the section's
contents if they are not already available.
* compress.c (bfd_get_full_section_contents): Use zmalloc to
allocate the buffers so that excess bytes are guaranteed to be
zero.
* dwarf2.c (bfd_dwarf2_cleanup_debug_info): Accept stash as an
argument like other functions to support formats other than ELF.
* elf-bfd.h (bfd_dwarf2_cleanup_debug_info): Move to bfd-in.h.
* elf.c (_bfd_elf_close_and_cleanup): Pass dwarf2_find_line_info
in tdata as a parameter.
* libbfd-in.h (bfd_dwarf2_cleanup_debug_info): Move from
elf-bfd.h.
* libbfd.h (bfd_dwarf2_cleanup_debug_info): Regenerate.
* mach-o-target.c (bfd_mach_o_close_and_cleanup): Remove the
fallback macro.
(bfd_mach_o_find_nearest_line): Likewise.
* mach-o.c (bfd_mach_o_find_nearest_line): Add the definition
which calls _bfd_dwarf2_find_nearest_line.
(bfd_mach_o_close_and_cleanup): Likewise.
* mach-o.h (mach_o_data_struct): Add dwarf2_find_line_info.
(bfd_mach_o_find_nearest_line): Add declaration.
(bfd_mach_o_close_and_cleanup): Add declaration.
* elf32-hppa.c (final_link_relocate): Convert R_PARISC_TLS_GD21L,
R_PARISC_TLS_LDM21L and R_PARISC_TLS_IE21L relocations to
R_PARISC_DPREL21L when not doing a shared link. Likewise convert
R_PARISC_TLS_GD14R, R_PARISC_TLS_LDM14R and R_PARISC_TLS_IE14R to
R_PARISC_DPREL14R. Handle R_PARISC_TLS_GD21L, R_PARISC_TLS_LDM21L
and R_PARISC_TLS_IE21L with R_PARISC_DLTIND21L.
* elfxx-mips.c (mips_elf_link_hash_table.rld_value): Remove.
(mips_elf_link_hash_table.rld_symbol): New field;
(MIPS_ELF_RLD_MAP_SIZE): New macro.
(_bfd_mips_elf_add_symbol_hook): Remember __rld_obj_head symbol
in rld_symbol.
(_bfd_mips_elf_create_dynamic_sections): Remember __rld_map symbol
in rld_symbol.
(_bfd_mips_elf_size_dynamic_sections): Set correct size for .rld_map.
(_bfd_mips_elf_finish_dynamic_symbol): Remove .rld_map handling.
(_bfd_mips_elf_finish_dynamic_sections): Use rld_symbol to
calculate DT_MIPS_RLD_MAP value.
(_bfd_mips_elf_link_hash_table_create): Initialize rld_symbol,
quit initializing rld_value.
2011-12-08 Andrew Pinski <apinski@cavium.com>
Adam Nemet <anemet@caviumnetworks.com>
* archures.c (bfd_mach_mips_octeon2): New macro
* bfd-in2.h: Regenerate.
* cpu-mips.c (I_mipsocteon2): New enum value.
(arch_info_struct): Add bfd_mach_mips_octeon2.
* elfxx-mips.c (_bfd_elf_mips_mach): Support E_MIPS_MACH_OCTEON2.
(mips_set_isa_flags): Add bfd_mach_mips_octeon2.
(mips_mach_extensions): Add bfd_mach_mips_octeon2.
gas:
2011-12-08 Andrew Pinski <apinski@cavium.com>
Adam Nemet <anemet@caviumnetworks.com>
* tc-mips.c (CPU_IS_OCTEON): Add Octeon2.
(mips_cpu_info_table): Add Octeon2.
* doc/c-mips.texi: Document octeon2 as an acceptable value for -march=.
gas/testsuite:
2011-12-08 Andrew Pinski <apinski@cavium.com>
Adam Nemet <anemet@caviumnetworks.com>
* gas/mips/mips.exp: Add Octeon2 for an architecture.
Run octeon2 test.
* gas/mips/octeon2.d: New file.
* gas/mips/octeon2.s: New file.
include/opcode:
2011-12-08 Andrew Pinski <apinski@cavium.com>
Adam Nemet <anemet@caviumnetworks.com>
* mips.h (INSN_CHIP_MASK): Update according to INSN_OCTEON2.
(INSN_OCTEON2): New macro.
(CPU_OCTEON2): New macro.
(OPCODE_IS_MEMBER): Add Octeon2.
opcodes:
2011-12-08 Andrew Pinski <apinski@cavium.com>
Adam Nemet <anemet@caviumnetworks.com>
* mips-dis.c (mips_arch_choices): Add Octeon2.
For "octeon+", just include OcteonP for the insn.
* mips-opc.c (IOCT): Include Octeon2.
(IOCTP): Include Octeon2.
(IOCT2): New macro.
(mips_builtin_opcodes): Add "laa", "laad", "lac", "lacd", "lad",
"ladd", "lai", "laid", "las", "lasd", "law", "lawd".
Move "lbux", "ldx", "lhx", "lwx", and "lwux" up to where the standard
loads are, and add IOCT2 to them.
Add "lbx" and "lhux".
Add "qmac.00", "qmac.01", "qmac.02", "qmac.03", "qmacs.00",
"qmacs.01", "qmacs.01", "qmacs.02" and "qmacs.03".
Add "zcb" and "zcbt".
* mach-o.c (bfd_mach_o_read_command): Add the bfd in the error
message.
(bfd_mach_o_openr_next_archived_file): Use arch name as member name.
(bfd_mach_o_fat_extract): Ditto.
2011-11-29 Andrew Pinski <apinski@cavium.com>
* mips-dis.c (mips_arch_choices): Add Octeon+.
* mips-opc.c (IOCT): Include Octeon+.
(IOCTP): New macro.
(mips_builtin_opcodes): Add "saa" and "saad".
bfd/
2011-11-29 Andrew Pinski <apinski@cavium.com>
* archures.c (bfd_mach_mips_octeonp): New macro.
* bfd-in2.h: Regenerate.
* bfd/cpu-mips.c (I_mipsocteonp): New enum value.
(arch_info_struct): Add bfd_mach_mips_octeonp.
* elfxx-mips.c (mips_set_isa_flags): Add bfd_mach_mips_octeonp.
(mips_mach_extensions): Add bfd_mach_mips_octeonp.
include/opcodes/
2011-11-29 Andrew Pinski <apinski@cavium.com>
* mips.h (INSN_CHIP_MASK): Update according to INSN_OCTEONP.
(INSN_OCTEONP): New macro.
(CPU_OCTEONP): New macro.
(OPCODE_IS_MEMBER): Add Octeon+.
(M_SAA_AB, M_SAAD_AB, M_SAA_OB, M_SAAD_OB): New enum values.
gas/
2011-11-29 Andrew Pinski <apinski@cavium.com>
* config/tc-mips.c (CPU_IS_OCTEON): New macro function.
(CPU_HAS_SEQ): Change to use CPU_IS_OCTEON.
(NO_ISA_COP): Likewise.
(macro) <ld_st>: Add support when off0 is true.
Add support for M_SAA_AB, M_SAA_OB, M_SAAD_OB and M_SAAD_AB.
(mips_cpu_info_table): Add octeon+.
* doc/c-mips.texi: Document octeon+ as an acceptable value for -march=.
gas/testsuite/
2011-11-29 Andrew Pinski <apinski@cavium.com>
* gas/mips/mips.exp: Add octeon+ for an architecture.
Run octeon-saa-saad test.
(run_dump_test_arch): For Octeon architectures, also try octeon@.
* gas/mips/octeon-pref.d: Remove -march=octeon from command line.
* gas/mips/octeon.d: Likewise.
* gas/mips/octeon-saa-saad.d: New file.
* gas/mips/octeon-saa-saad.s: New file
* elf64-ppc.c (struct ppc64_elf_obj_tdata): Rename
ha_relocs_not_using_r2 to unexpected_toc_insn.
(ok_lo_toc_insn): New function.
(ppc64_elf_edit_toc): Check insn on lo toc reloc. Emit warning.
(ppc64_elf_relocate_section): Don't check insn on lo toc reloc here.
Handle addic on lo toc reloc.
ld/testsuite/
* ld-powerpc/powerpc.exp: Expect ld warnings for tocopt test.
* ld-powerpc/tocopt.out: New file.
* elf32-hppa.c (elf32_hppa_hide_symbol): Make STT_GNU_IFUNC symbol
go through PLT. Reset plt field with init_plt_offset.
(elf32_hppa_adjust_dynamic_symbol): Ensure that a PLT slot is
allocated for symbols referenced by a plabel.
(where relocatable objects are in ELF format) so far led to all local (aka
static) symbols to be discarded, making debugging quite a bit more difficult
(like Linux, Xen builds an internal symbol lookup table from nm output
generated on the binary produced by an earlier linking pass). Therefore, this
patch arranges to insert all (relevant) local symbols from non-COFF objects
into the final executable's symbol table between those coming from COFF input
files and the global ones.
bfd/
2011-11-02 Jan Beulich <jbeulich@suse.com>
* coffgen.c (coff_write_alien_symbol): Make public. Add 'struct
internal_syment *' parameter. Extend 'dummy' to an array with two
elements. Set n_numaux early. Handle BSF_FILE.
(coff_write_symbols): Pass NULL as new third argument to
coff_write_alien_symbol().
* cofflink.c (_bfd_coff_final_link): Don't use COFF-specific
obj_raw_syment_count() on non-COFF input BFD. Insert local symbols
from non-COFF input BFDs.
* libcoff-in.h (coff_write_alien_symbol): Declare.
* libcoff.h (coff_write_alien_symbol): Re-generate.
