* config/tc-mips.c (macro_build): Test for currently active
mips16 option.
(mips16_ip): Reject invalid opcodes.
[ opcodes/ChangeLog ]
* mips16-opc.c (I1, I32, I64): New shortcut defines.
(mips16_opcodes): Change membership of instructions to their
lowest baseline ISA.
[ gas/testsuite/ChangeLog ]
* gas/mips/mips.exp: Run new tests.
* gas/mips/mips16e.s, gas/mips/mips16e.d, gas/mips/mips16e-64.s,
gas/mips/mips16e-64.d, gas/mips/mips16e-64.l: New tests.
* config/tc-mips.c (append_insn): Don't check the range of j or
jal addresses.
[ gas/testsuite/ChangeLog ]
* gas/mips/jal-range.l: Don't check the range of j or jal
addresses.
* config/tc-mips.c (append_insn): Only warn about an out-of-range
j or jal address.
[ gas/testsuite/ChangeLog ]
* gas/mips/jal-range.l: Only warn about an out-of-range j or jal
address.
2006-04-30 Thiemo Seufer <ths@mips.com>
David Ung <davidu@mips.com>
* config/tc-mips.c (validate_mips_insn): Handling of udi cases.
(mips_immed): New table that records various handling of udi
instruction patterns.
(mips_ip): Adds udi handling.
[ include/opcode/ChangeLog ]
2006-04-30 Thiemo Seufer <ths@mips.com>
David Ung <davidu@mips.com>
* mips.h: Defines udi bits and masks. Add description of
characters which may appear in the args field of udi
instructions.
[ opcodes/ChangeLog ]
2006-04-30 Thiemo Seufer <ths@mips.com>
David Ung <davidu@mips.com>
* mips-opc.c (mips_builtin_opcodes): Add udi instructions
"udi0" to "udi15".
* mips-dis.c (print_insn_args): Adds udi argument handling.
Daniel Jacobowitz <dan@codesourcery.com>
Phil Edwards <phil@codesourcery.com>
Zack Weinberg <zack@codesourcery.com>
Mark Mitchell <mark@codesourcery.com>
Nathan Sidwell <nathan@codesourcery.com>
bfd/
* bfd-in2.h: Regenerate.
* config.bfd (mips*-*-vxworks*, mips*el-*-vxworks*): New stanzas.
* configure.in (bfd_elf32_bigmips_vxworks_vec): New stanza.
(bfd_elf32_littlemips_vxworks_vec): Likewise.
(bfd_elf32_bigmips_vec): Add elf-vxworks.lo.
(bfd_elf32_littlemips_vec): Likewise.
(bfd_elf32_nbigmips_vec): Likewise.
(bfd_elf32_nlittlemips_vec): Likewise.
(bfd_elf32_ntradbigmips_vec): Likewise.
(bfd_elf32_ntradlittlemips_vec): Likewise.
(bfd_elf32_tradbigmips_vec): Likewise.
(bfd_elf32_tradlittlemips_vec): Likewise.
(bfd_elf64_bigmips_vec): Likewise.
(bfd_elf64_littlemips_vec): Likewise.
(bfd_elf64_tradbigmips_vec): Likewise.
(bfd_elf64_tradlittlemips_vec): Likewise.
* elf32-mips.c: Include elf-vxworks.h.
(mips_info_to_howto_rel): Use elf_backend_mips_rtype_to_howto
instead of calling mips_elf32_rtype_to_howto directly.
(mips_vxworks_copy_howto_rela): New reloc howto.
(mips_vxworks_jump_slot_howto_rela): Likewise.
(mips_vxworks_bfd_reloc_type_lookup): New function.
(mips_vxworks_rtype_to_howto): Likewise.
(mips_vxworks_final_write_processing): Likewise.
(TARGET_LITTLE_SYM, TARGET_LITTLE_NAME): Override for VxWorks.
(TARGET_BIG_SYM, TARGET_BIG_NAME, elf_bed, ELF_MAXPAGESIZE): Likewise.
(elf_backend_want_got_plt): Likewise.
(elf_backend_want_plt_sym): Likewise.
(elf_backend_got_symbol_offset): Likewise.
(elf_backend_want_dynbss): Likewise.
(elf_backend_may_use_rel_p): Likewise.
(elf_backend_may_use_rela_p): Likewise.
(elf_backend_default_use_rela_p): Likewise.
(elf_backend_got_header_size: Likewise.
(elf_backend_plt_readonly): Likewise.
(bfd_elf32_bfd_reloc_type_lookup): Likewise.
(elf_backend_mips_rtype_to_howto): Likewise.
(elf_backend_adjust_dynamic_symbol): Likewise.
(elf_backend_finish_dynamic_symbol): Likewise.
(bfd_elf32_bfd_link_hash_table_create): Likewise.
(elf_backend_add_symbol_hook): Likewise.
(elf_backend_link_output_symbol_hook): Likewise.
(elf_backend_emit_relocs): Likewise.
(elf_backend_final_write_processing: Likewise.
(elf_backend_additional_program_headers): Likewise.
(elf_backend_modify_segment_map): Likewise.
(elf_backend_symbol_processing): Likewise.
* elfxx-mips.c: Include elf-vxworks.h.
(mips_elf_link_hash_entry): Add is_relocation_target and
is_branch_target fields.
(mips_elf_link_hash_table): Add is_vxworks, srelbss, sdynbss, srelplt,
srelplt2, sgotplt, splt, plt_header_size and plt_entry_size fields.
(MIPS_ELF_RELA_SIZE, MIPS_ELF_REL_DYN_NAME): New macros.
(MIPS_RESERVED_GOTNO): Take a mips_elf_link_hash_table argument.
Return 3 for VxWorks.
(ELF_MIPS_GP_OFFSET): Change the argument from a bfd to a
mips_elf_link_hash_table. Return 0 for VxWorks.
(MIPS_ELF_GOT_MAX_SIZE): Change the argument from a bfd to a
mips_elf_link_hash_table. Update the call to ELF_MIPS_GP_OFFSET.
(mips_vxworks_exec_plt0_entry): New variable.
(mips_vxworks_exec_plt_entry): Likewise.
(mips_vxworks_shared_plt0_entry): Likewise.
(mips_vxworks_shared_plt_entry): Likewise.
(mips_elf_link_hash_newfunc): Initialize the new hash_entry fields.
(mips_elf_rel_dyn_section): Change the bfd argument to a
mips_elf_link_hash_table. Use MIPS_ELF_REL_DYN_NAME to get
the name of the section.
(mips_elf_initialize_tls_slots): Update the call to
mips_elf_rel_dyn_section.
(mips_elf_gotplt_index): New function.
(mips_elf_local_got_index): Add an input_section argument.
Update the call to mips_elf_create_local_got_entry.
(mips_elf_got_page): Likewise.
(mips_elf_got16_entry): Likewise.
(mips_elf_create_local_got_entry): Add bfd_link_info and input_section
arguments. Create dynamic relocations for each entry on VxWorks.
(mips_elf_merge_gots): Update the use of MIPS_ELF_GOT_MAX_SIZE.
(mips_elf_multi_got): Update the uses of MIPS_ELF_GOT_MAX_SIZE
and MIPS_RESERVED_GOTNO.
(mips_elf_create_got_section): Update the uses of
MIPS_ELF_GOT_MAX_SIZE. Create .got.plt on VxWorks.
(is_gott_symbol): New function.
(mips_elf_calculate_relocation): Use a dynobj local variable.
Update the calls to mips_elf_local_got_index, mips_elf_got16_entry and
mips_elf_got_page_entry. Set G to the .got.plt entry when calculating
VxWorks R_MIPS_CALL* relocations. Calculate and use G for all GOT
relocations on VxWorks. Add dynamic relocations for references
to the VxWorks __GOTT_BASE__ and __GOTT_INDEX__ symbols. Don't
create dynamic relocations for R_MIPS_32, R_MIPS_REL32 or R_MIPS_64
in VxWorks executables.
(mips_elf_allocate_dynamic_relocations): Add a bfd_link_info argument.
Use MIPS_ELF_RELA_SIZE to calculate the size of a VxWorks entry.
Don't allocate a null entry on VxWorks.
(mips_elf_create_dynamic_relocation): Update the call to
mips_elf_rel_dyn_section. Use absolute rather than relative
relocations for VxWorks, and make them RELA rather than REL.
(_bfd_mips_elf_create_dynamic_sections): Don't make .dynamic
read-only on VxWorks. Update the call to mips_elf_rel_dyn_section.
Create the .plt, .rela.plt, .dynbss and .rela.bss sections on
VxWorks. Likewise create the _PROCEDURE_LINKAGE_TABLE symbol.
Call elf_vxworks_create_dynamic_sections for VxWorks and
initialize the plt_header_size and plt_entry_size fields.
(_bfd_mips_elf_check_relocs): Don't allow GOT relocations to be
used in VxWorks executables. Don't allocate dynamic relocations
for R_MIPS_32, R_MIPS_REL32 or R_MIPS_64 in VxWorks executables.
Set is_relocation_target for each symbol referenced by a relocation.
Allocate .rela.dyn entries for relocations against the special
VxWorks __GOTT_BASE__ and __GOTT_INDEX__ symbols. Create GOT
entries for all VxWorks R_MIPS_GOT16 relocations. Don't allocate
a global GOT entry for symbols mentioned in VxWorks R_MIPS_CALL*,
R_MIPS_32, R_MIPS_REL32 or R_MIPS_64 relocations. Update the calls
to mips_elf_rel_dyn_section and mips_elf_allocate_dynamic_relocations.
Set is_branch_target for symbols mentioned in R_MIPS_PC16 or R_MIPS_26
relocations. Don't set no_fn_stub on VxWorks.
(_bfd_mips_elf_adjust_dynamic_symbol): Update the call to
mips_elf_allocate_dynamic_relocations.
(_bfd_mips_vxworks_adjust_dynamic_symbol): New function.
(_bfd_mips_elf_always_size_sections): Do not allocate GOT page
entries for VxWorks, and do not create multiple GOTs.
(_bfd_mips_elf_size_dynamic_sections): Use MIPS_ELF_REL_DYN_NAME.
Handle .got specially for VxWorks. Update the uses of
MIPS_RESERVED_GOTNO and mips_elf_allocate_dynamic_relocations.
Check for sgotplt and splt. Allocate the .rel(a).dyn contents last,
once its final size is known. Set DF_TEXTREL for VxWorks. Add
DT_RELA, DT_RELASZ, DT_RELAENT, DT_PLTREL, DT_PLTRELSZ and DT_JMPREL
tags on VxWorks. Do not add the MIPS-specific tags for VxWorks.
(_bfd_mips_vxworks_finish_dynamic_symbol): New function.
(mips_vxworks_finish_exec_plt): Likewise.
(mips_vxworks_finish_shared_plt): Likewise.
(_bfd_mips_elf_finish_dynamic_sections): Remove an unncessary call
to mips_elf_rel_dyn_section. Use a VxWorks-specific value of
DT_PLTGOT. Handle DT_RELA, DT_RELASZ, DT_RELAENT, DT_PLTREL,
DT_PLTRELSZ and DT_JMPREL. Update the uses of MIPS_RESERVED_GOTNO
and mips_elf_rel_dyn_section. Use a different GOT header for
VxWorks. Don't sort .rela.dyn on VxWorks. Finish the PLT on VxWorks.
(_bfd_mips_elf_link_hash_table_create): Initialize the new
mips_elf_link_hash_table fields.
(_bfd_mips_vxworks_link_hash_table_create): New function.
(_bfd_mips_elf_final_link): Set the GP value to _GLOBAL_OFFSET_TABLE_
on VxWorks. Update the call to ELF_MIPS_GP_OFFSET.
* elfxx-mips.h (_bfd_mips_vxworks_adjust_dynamic_symbol): Declare.
(_bfd_mips_vxworks_finish_dynamic_symbol): Likewise.
(_bfd_mips_vxworks_link_hash_table_create): Likewise.
* libbfd.h: Regenerate.
* Makefile.am (elfxx-mips.lo): Depend on elf-vxworks.h.
(elf32-mips.lo): Likewise.
* Makefile.in: Regenerate.
* reloc.c (BFD_RELOC_MIPS_COPY, BFD_RELOC_MIPS_JUMP_SLOT): Declare.
* targets.c (bfd_elf32_bigmips_vxworks_vec): Declare.
(bfd_elf32_littlemips_vxworks_vec): Likewise.
(_bfd_target_vector): Add entries for them.
gas/
* config/tc-mips.c (mips_target_format): Handle vxworks targets.
(md_begin): Complain about -G being used for PIC. Don't change
the text, data and bss alignments on VxWorks.
(reloc_needs_lo_p): Don't return true for R_MIPS_GOT16 when
generating VxWorks PIC.
(load_address): Extend SVR4_PIC handling to VXWORKS_PIC.
(macro): Likewise, but do not treat la $25 specially for
VxWorks PIC, and do not handle jal.
(OPTION_MVXWORKS_PIC): New macro.
(md_longopts): Add -mvxworks-pic.
(md_parse_option): Don't complain about using PIC and -G together here.
Handle OPTION_MVXWORKS_PIC.
(md_estimate_size_before_relax): Always use the first relaxation
sequence on VxWorks.
* config/tc-mips.h (VXWORKS_PIC): New.
gas/testsuite/
* gas/mips/vxworks1.s, gas/mips/vxworks1.d,
* gas/mips/vxworks1-xgot.d: New tests.
* gas/mips/mips.exp: Run them. Do not run other tests on VxWorks.
include/elf/
* mips.h (R_MIPS_COPY, R_MIPS_JUMP_SLOT): New relocs.
ld/
* configure.tgt (mips*el-*-vxworks*, mips*-*-vxworks*): Use
separate VxWorks emulations.
* emulparams/elf32ebmipvxworks.sh: New file.
* emulparams/elf32elmipvxworks.sh: New file.
* Makefile.am (ALL_EMULATIONS): Add eelf32ebmipvxworks.o and
eelf32elmipvxworks.o.
(eelf32ebmipvxworks.c, eelf32elmipvxworks.c): New rules.
* Makefile.in: Regenerate.
ld/testsuite/
* ld-mips/vxworks1.dd, ld-mips/vxworks1.ld, ld-mips/vxworks1-lib.dd,
* ld-mips/vxworks1-lib.nd, ld-mips/vxworks1-lib.rd,
* ld-mips/vxworks1-lib.s, ld-mips/vxworks1.rd, ld-mips/vxworks1.s,
* ld-mips/vxworks1-static.d, ld-mips/vxworks2.s, ld-mips/vxworks2.sd,
* ld-mips/vxworks2-static.sd: New tests.
* ld-mips/mips-elf.exp: Run them.
instructions. Define MIPS16_ALL_ARGS and MIPS16_ALL_STATICS for
save/restore encoding of the args field.
* mips16-opc.c: Add MIPS16e save/restore opcodes.
* mips-dis.c (print_mips16_insn_arg): Handle printing of 'm'/'M'
codes for save/restore.
* config/tc-mips.c (mips16_ip): Add handling of 'm' and 'M' codes
for the MIPS16e save/restore instructions.
* gas/mips/mips.exp: Run new save/restore tests.
* gas/testsuite/gas/mips/mips16e-save.s: New test for generating
different styles of save/restore instructions.
* gas/testsuite/gas/mips/mips16e-save.d: New.
into jrc/jalrc versions if ISA_MIPS32+ and not doing the swap,
hence avoiding to emit a nop.
* gas/mips/mips.exp: Run new test.
* gas/testsuite/gas/mips/mips16e-jrc.s: New test for converting
jalr/jr to the compact jalrc/jrc instructions.
* gas/testsuite/gas/mips/mips16e-jrc.d: New.
- allowing true forward references (which will always assume the referenced
symbols have at the point of use) through the new .eqv pseudo-op and the
new == operator
- disallowing changing .equiv-generated equates (so that the protection this
provides is both forward and backward)
- snapshotting equates when their value gets changed so that previous uses
don't get affected by the new value.
- allowing expressions in places where absolute expressions (or register
names) are needed which were not completely resolvable at the point of
their definition but which are fully resolvable at the point of use
In addition it fixes PR/288.
(mips_opts): Add -1 to initialize ase_mt.
(file_ase_mt): New variable for -mmt.
(CPU_HAS_MT): New define.
(validate_mips_insn): Add supports for +t, +T, !, $, *, &, g operand
formats.
(mips_ip): Check ase_mt to enable MT instructions.
Handle !, $, *, &, +T, +t, g operand formats.
For "mftc1", "mfthc1", "cftc1", "mttc1", "mtthc1", "cttc1", we allow
odd float registers.
(OPTION_MT, OPTION_NO_MT): New define.
(OPTION_COMPAT_ARCH_BASE): Change because of inserting MT define.
(md_parse_option): Parse OPTION_MT and OPTION_NO_MT.
(mips_after_parse_args): Set ase_mt based on CPU.
(s_mipsset): Handle ".set mt" and ".set nomt".
(mips_elf_final_processing): Remind of adding new flag for MT ASE.
(md_show_usage): Show usage of -mmt and -mno-mt.
* doc/as.texinfo: Document -mmt and -mno-mt options.
* doc/c-mips.texi: Likewise, and document ".set mt" and ".set nomt"
directives.
when the frags are different for the 2 instructions we want to
swap. If the lengths of the 2 instructions are not the same, we
won't do the swap but emit an nop.
(mips_opts): Add -1 to initialize ase_dsp.
(file_ase_dsp): New variable for -mdsp.
(CPU_HAS_DSP): New define.
(validate_mips_insn): Add supports for 3, 4, 5, 6, 7, 8, 9, 0, ', :, @
operand formats.
(mips_ip): Add min_range and max_range for checking singed numbers.
Check ase_dsp to enable DSP instructions.
Handle 3, 4, 5, 6, 7, 8, 9, 0, ', :, @ operand formats.
(OPTION_DSP, OPTION_NO_DSP): New define.
(OPTION_COMPAT_ARCH_BASE): Change because of inserting DSP define.
(md_parse_option): Parse OPTION_DSP and OPTION_NO_DSP.
(mips_after_parse_args): Set ase_dsp based on CPU.
(s_mipsset): Handle ".set dsp" and ".set nodsp".
(mips_elf_final_processing): Remind of adding new flag for DSP ASE.
(md_show_usage): Show usage of -mdsp and -mno-dsp.
* config/tc-mips.c (load_register): Add leading "0x" to the
output of sprintf_vma().
(macro): Likewise.
gas/testsuite/:
* gas/mips/ldstla-32-1.l: Update to handle leading zeroes.
* gas/mips/ldstla-32-mips3-1.l: Likewise.
* config/tc-mips.c (IS_ZEXT_32BIT_NUM): New macro.
(normalize_address_expr): New function to sign-extend address
offsets that fit into 32 bits in 32-bit mode.
(macro_build_ldst_constoffset): Use normalize_address_expr()
instead of a handcoded sequence.
(load_register): Likewise. Report oversized numbers in a useful
way.
(macro) [ld_st, ldd_std]: Reject all oversized offsets, not only
for constant addresses. Report oversized numbers in a useful way.
(mips_ip): Use normalize_address_expr() for addresses.
gas/testsuite/:
* gas/mips/ldstla-32.s: Exclude offsets that are now meant to fail
and include more instructions/offsets that are meant to succeed.
Use $4 instead $3 to avoid register dependencies.
* gas/mips/ldstla-32.d: Update accordingly.
* gas/mips/ldstla-32-shared.d: Likewise.
* gas/mips/ldstla-32-mips3.d: New test based on the above, except
for mips3.
* gas/mips/ldstla-32-mips3-shared.d: Similarly, for PIC.
* gas/mips/ldstla-32-mips3.s: Source for the new tests.
* gas/mips/ldstla-32-1.s: New test for offsets that are meant to
fail.
* gas/mips/ldstla-32-mips3-1.s: Likewise, for mips3.
* gas/mips/ldstla-32-1.l: Stderr output for the new test.
* gas/mips/ldstla-32-mips3-1.l: Likewise.
* gas/mips/mips.exp: Run the new tests.
(MAX_NOPS): Bump to 4.
(mips_fix_vr4130): New variable.
(nops_for_vr4130): New function.
(nops_for_insn): Use MAX_DELAY_NOPS rather than MAX_NOPS. Use
nops_for_vr4130 if working around VR4130 errata.
(OPTION_FIX_VR4130, OPTION_NO_FIX_VR4130): New macros.
(md_longopts): Add -mfix-vr4130 and -mno-fix-vr4130.
(md_parse_option): Handle them.
(md_show_usage): Print them.
* doc/c-mips.texi: Document -mfix-vr4130 and -mno-fix-vr4130.
(history): Resize to 1 + MAX_NOPS.
(fix_vr4120_class): New enumeration.
(vr4120_conflicts): New variable.
(init_vr4120_conflicts): New function.
(md_begin): Call it.
(insn_uses_reg): Constify first argument.
(classify_vr4120_insn, insns_between, nops_for_insn, nops_for_sequence)
(nops_for_insn_or_target): New functions.
(append_insn): Use the new nops_for_* functions instead of inline
delay checks. Generalize prev_nop_frag handling to handle an
arbitrary history length. Insert nops into the history buffer
once the number of nops in prev_nop_frag is fixed.
(emit_delays): Use nops_for_insn instead of inline delay checks.
(nop_insn, mips16_nop_insn): New variables.
(NOP_INSN): New macro.
(insn_length, create_insn, install_insn, move_insn, add_fixed_insn)
(add_relaxed_insn, insert_into_history, emit_nop): New functions.
(md_begin): Initialize nop_insn and mips16_nop_insn.
(append_insn): Use the new emit_nop function to add nops, recording
them in the history buffer. Use add_fixed_insn or add_relaxed_insn
to reserve room for the instruction and install_insn to install the
final form. Use insert_into_history to record the instruction in
the history buffer. Use move_insn to do delay slot filling.
(mips_emit_delays): Use add_fixed_insn instead of the emit_nop macro.
(macro_build, mips16_macro_build, macro_build_lui, mips_ip)
(mips16_ip): Use create_insn to initialize mips_cl_insns.
(EXTRACT_OPERAND, MIPS16_INSERT_OPERAND, MIPS16_EXTRACT_OPERAND): New.
(insn_uses_reg, reg_needs_delay, append_insn, macro_build)
(mips16_macro_build, macro_build_lui, mips16_macro, mips_ip)
(mips16_ip): Use the new macros instead of explicit masks and shifts.
* elfxx-mips.c (mips_elf_calculate_relocation): Handle special
'__gnu_local_gp' symbol used by gas -mno-shared.
gas/ChangeLog
* config/tc-mips.c (macro_build_lui): Use '__gnu_local_gp'
instead of '_gp' for -mno-shared optimization.
(s_cpload): Ditto.
(s_abicalls): Document it in the comment.
(md_show_usage): Document the -mno-shared option.
gas/testsuite/ChangeLog
* gas/mips/elf-rel23b.d: Use '__gnu_local_gp' instead of '_gp'
for -mno-shared optimization.
* gas/mips/elf-rel25a.d: Ditto.
ld/testsuite/ChangeLog
* ld-mips-elf/multi-got-no-shared-1.s,
ld-mips-elf/multi-got-no-shared-2.s,
ld-mips-elf/multi-got-no-shared.d: New tests.
* ld-mips-elf/mips-elf.exp: Run them.
(mips_opts): Initialize it.
(HAVE_32BIT_ADDRESSES): Set to true if pointers are 32 bits wide.
(HAVE_64BIT_ADDRESSES): Redefine as !HAVE_32BIT_ADDRESSES.
(HAVE_32BIT_SYMBOLS, HAVE_64BIT_SYMBOLS): New macros.
(load_address): Use HAVE_64BIT_SYMBOLS instead of HAVE_64BIT_ADDRESSES
when deciding whether to use a symbolic %highest/%higher expansion.
(macro): Likewise. Remove o64/n32 linux hack. Always use
ADDRESS_ADD*_INSN for address addition in the expansion of "dla"
and "la". Handle constants separately from symbolic expressions in
the "ld_st:" case, using 64-bit arithmetic if HAVE_64BIT_ADDRESSES
and using load_register to load the high part of the address.
(OPTION_MSYM32, OPTION_NO_MSYM32): New macros.
(OPTION_ELF_BASE): Bump by 2.
(md_longopts): Add entries for -msym32 and -mno-sym32.
(md_parse_option): Handle them.
(usage): Document them.
(s_mipsset): Handle ".set sym32" and ".set nosym32".
(s_cpload, s_cpsetup): Use HAVE_64BIT_SYMBOLS instead of
HAVE_64BIT_ADDRESSES to detect 64-bit values of "_gp".
* doc/c-mips.texi: Document ".set sym32", ".set nosym32",
-msym32 and -mno-sym32.
for 64bit address space non-PIC. Fix formatting.
(macro): Likewise. Simplify code.
(md_parse_option): Don't bail out if -G 0 is set for PIC code.
(mips_after_parse_args): Simplify code.
%dtprel_lo, %tprel_hi, %tprel_lo, and %gottprel.
(parse_relocation): Check for a word break after a relocation
operator.
(md_apply_fix3): Handle TLS relocations, and mark thread-local
symbols.