(md_estimate_size_before_relax): Don't set fx_pcrel_adjust any
more.
(md_apply_fix3): Remember addend value for rela relocations.
(tc_gen_reloc): Correctly compute pc-relative relocation addend.
field for pc-relative fixups.
(output_disp): Likewise.
(md_estimate_size_before_relax): Likewise.
(tc_gen_reloc): Subtract fx_pcrel_adjust instead of fx_size for
pc-relative fixups in 64bit mode.
relocs except when BFD64.
* write.c (number_to_chars_bigendian): Don't abort when N is
larger than sizeof (VAL).
(number_to_chars_littleendian): Likewise.
and rearrange for efficiency. For "PIC code" subtraction, use
"rightseg" rather than recalculating. For "symbol OP symbol"
subtract, set "retval" to absolute_section if symbols in same
section.
* symbols.c (resolve_symbol_value): Resolve "sym +/- expr" to an
O_symbol. Simplify a +/- b code. Allow equality and non-equality
comparisons on symbols from any section. Allow other comparison
operators as for subtraction.
(symbol_equated_reloc_p): New predicate function.
* symbols.h (symbol_equated_reloc_p): Declare.
* write.c (adjust_reloc_syms): Use symbol_equated_reloc_p.
(write_relocs): Likewise.
(write_object_file): Likewise.
(relax_segment <rs_machine_dependent>): Ensure segment for
expression syms is set correctly.
* config/tc-mips.c (md_estimate_size_before_relax): Likewise.
* config/tc-i386.c (md_assemble <Output jumps>): Don't lose part
of a complex expression when setting up frag_var.
(md_apply_fix): Use it here. Replace printf with equivalent
as_bad_where.
(tc_gen_reloc): Use as_bad_where instead of as_bad.
(md_apply_fix): Here too.
* config/tc-i386.c (tc_gen_reloc): Use as_bad_where instead of as_bad.
* config/tc-m68k.c (tc_gen_reloc): Likewise.
(md_convert_frag_1): Likewise.
(md_assemble [smallest displacement]): Use correct field of i.op[] union.
(md_assemble [JumpInterSegment output]): Use correct i.disp_reloc[].
(md_assemble [immediate output]): Likewise.
* config/tc-i386.c (tc_gen_reloc): Remove ugly hack which is not needed
anymore since we use bfd_elf_generic_reloc now.
(md_apply_fix3): Only apply hack for partial_inplace if not using RELA.
instruction sequence consisting of a conditional jump of the
opposite sense around an unconditional jump to the target.
Add jumps/nojumps .arch modifier.
* tc-i386.c (md_assemble): Return after the error message;
move testing for 64bit operands to proper place.
* i386.exp: Add tests for presence of 32bit versus 64bit output
format; run both 64bit and 32bit tests when format is available;
add x86_64 test.
* x86_64.s: New file.
* x86_64.d: New file.
(QWORD_MNEM_SUFFIX): New macro.
(CpuK6,CpuAthlon,CpuSledgehammer, Cpu64, CpuNo64, CpuUnknownFlags):
New macros
(CpuMMX,CpuSSE,Cpu3dnow, CpuUnknown): Renumber.
(IgnoreSize, DefaultSize, No_?Suf, FWait, IsString, regKludge, IsPrefix,
ImmExt): Renumber.
(Size64, No_qSuf, NoRex64, Rex64): New macros.
(Reg64, Imm32S, Imm64, Disp32S, Disp64): New macros.
(Imm8, Imm8S, Imm16, Imm32, Imm1, BaseIndex, Disp8, Disp16, Disp32,
InOutPortReg,ShiftCount, Control, Debug, Test, FloatReg, FloatAcc,
SReg2, SReg3, Acc, JumpAbsolute, RegMMX, RegXMM, EsSeg, InvMem): Renumber.
(Reg, WordReg): Add Reg64.
(Imm): Add Imm32S and Imm64.
(EncImm): New.
(Disp): Add Disp64 and Disp32S.
(AnyMem): Add Disp32S.
(RegRex, RegRex64): New macros.
(rex_byte): New type.
* tc-i386.c (set_16bit_code_flag): Kill.
(fits_in_unsigned_long, fits_in_signed_long): New functions.
(reloc): New parameter "signed"; support x86_64.
(set_code_flag): New.
(DEFAULT_ARCH): New macro; default to "i386".
(default_arch): New static variable.
(struct _i386_insn): New fields Operand_PCrel; rex.
(flag_16bit_code): Kill; All tests replaced to "flag_code == CODE_64BIT";
(flag_code): New enum and static variable.
(use_rela_relocations): New static variable.
(flag_code_names): New static variable.
(cpu_arch_flags): Default to CpuUnknownFlags|CpuNo64.
(cpu_arch): Add "sledgehammer"; Add CPUAthlon to Athlon and CpuK6 to
K6 and Athlon.
(i386_align_code): Return plain "nop" for x86_64.
(mode_from_disp_size): Support Disp32S.
(smallest_imm_type): Support Imm32S and Imm64.
(offset_in_range): Support size of 8.
(set_cpu_arch): Do not clobber to Cpu64/CpuNo64.
(md_pseudo_table): Add "code64"; use set_code_flat.
(md_begin): Emit sane error message on hash failure.
(tc_i386_fix_adjustable): Support x86_64 relocations.
(md_assemble): Support QWORD_MNEM_SUFFIX, REX registers,
instructions supported on particular arch just partially,
output of 64bit immediates, handling of Imm32S and Disp32S type.
(i386_immedaite): Support x86_64 relocations; support 64bit constants.
(i386_displacement): Likewise.
(i386_index_check): Cleanup; support 64bit addresses.
(md_apply_fix3): Support x86_64 relocation and rela.
(md_longopts): Add "32" and "64".
(md_parse_option): Add OPTION_32 and OPTION_64.
(i386_target_format): Call even for ELFs; choose between
elf64-x86-64 and elf32-i386.
(i386_validate_fix): Refuse GOTOFF in 64bit mode.
(tc_gen_reloc): Support rela relocations and x86_64.
(intel_e09_1): Support QWORD.
* i386.h (i386_optab): Replace "Imm" with "EncImm".
(i386_regtab): Add flags field.
* config/tc-i386.c (intel_e09_1): Only flag as a memory operand if
it's not an offset expression.
(intel_e10_1): Ditto. Also, if the operand is an offset expression,
keep the braces '[' and ']' in the output string.
(intel_e11): Ditto. Also remove comparison intel_parser.op_modifier
!= FLAT. There is no such op_modifier.
* tc-i386.c (md_assemble): Swap i.disp_relocs when using intel
syntax.
2000-11-30 Diego Novillo <dnovillo@redhat.com>
* intel.s, intel.d: New test for @GOT references.
* tc-i386.c (i386_operand_modifier): Remove.
(build_displacement_string): Remove.
(i386_parse_seg): Remove.
(i386_intel_memory_operand): Remove.
(i386_intel_operand): Re-write using recursive descent parser based
on MASM documentation.
(struct intel_parser_s): New structure.
(intel_parser): New static variable.
(struct intel_token): New structure.
(cur_token, prev_token): New static variables.
(T_NIL): Define.
(T_CONST): Define.
(T_REG): Define.
(T_BYTE): Define.
(T_WORD): Define.
(T_DWORD): Define.
(T_QWORD): Define.
(T_XWORD): Define.
(T_SHORT): Define.
(T_OFFSET): Define.
(T_PTR): Define.
(T_ID): Define.
(intel_match_token): New function.
(intel_get_token): New function.
(intel_putback_token): New function.
(intel_expr): New function.
(intel_e05): New function.
(intel_e05_1): New function.
(intel_e06): New function.
(intel_e06_1): New function.
(intel_e09): New function.
(intel_e09_1): New function.
(intel_e10): New function.
(intel_e10_1): New function.
(intel_e11): New function.
2000-10-24 Diego Novillo <dnovillo@cygnus.com>
* intel.s, intel.d: Add new tests for intel syntax.
2000-10-15 Diego Novillo <dnovillo@cygnus.com>
* config/tc-i386.c (i386_operand_modifier): Only match
modifiers SHORT and FLAT if they are followed by a space.
(parse_register): When `allow_naked_reg' is set, do not confuse
identifiers that start with a register name with a register.
gas/testsuite:
2000-10-15 Diego Novillo <dnovillo@cygnus.com>
* intel.s, intel.d: Add new tests for naked registers using intel
syntax.
the errors exposed by this addition. These were intel mode
"fi... word ptr", "fi... dword ptr", "jmp Imm seg, Imm offset", "out dx,al".
The failure with intel "out dx,al" was also present in att "out al,dx".
Extend testsuite to catch this case too.
assembler. ie. You will be able to do "as --em=i386aout" on an x86
linux-elf assembler to generate aout format object files, rather than
using a separate assembler. The aout emulation is enabled by giving
"--enable-targets=i386-linuxaout" to configure.
Oh yeah, there's a couple of fixes too. Error messages shouldn't be
passed to printf in the format arg just in case someone puts a `%' in
the message.
* config/tc-i386.c (md_shortopts): Check OBJ_MAYBE_ELF as well as
OBJ_ELF. If ELF, add "sq".
(md_parse_option): If ELF, ignore -s and -q.
(md_show_usage): Mention ELF options.
* config/atof-ieee.c (atof_ieee): Change what_kind to int.
* config/atof-vax.c (flonum_gen2vax): Change format_letter to
int.
(md_atof): Return NULL rather than 0.
* config/tc-i386.c (md_atof): Change type to int.
next_set_end even if the end symbol is being discarded.
* gasp.c: Add ATTRIBUTE_UNUSED as needed for non-BFD_ASSEMBLER.
* output-file.c, symbols.c, config/tc-i386.c: Likewise.
* config/obj-coff.c: Likewise.
(seg_info_type): Remove.
(seg_info_off_by_4): Change to array of segT.
(s_get_segment): Adjust accordingly.
(obj_pseudo_table): Fully initialize sentinel entry.
as appropriate. Fill in structure initializations. Add variable
initializations. Add casts.
* dwarf2dbg.c (print_stats): Change i to size_t.
* listing.c (listing_listing): Change list_line to unsigned int.
Move bitfield overflow checks to after the md_apply_fix call in
fixup_segment so that md_apply_fix has a chance to modify value.
Handle 8 and 16-bit pcrel relocs for i386.
Prototypes and other minor code cleanups.
