3 * config/tc-i386.c (i386_align_code): Add comments explaining the
8 * write.c (relax_segment): Give an error if a .space symbol is
11 * read.c (read_a_source_file): Don't handle mri_pending_align if
12 the handler is s_globl or s_ignore.
16 * config/tc-d10v.c (find_opcode): Remove unused variable "numops".
18 * config/tc-tic80.c: Many additions to previous placeholder file.
19 * config/tc-tic80.h: Ditto.
24 * config/tc-i386.c (i386_align_code): Improve the nop patterns.
28 * config/tc-mips.c (mips_fix_adjustable): New function.
29 * config/tc-mips.h (tc_fix_adjustable): Call mips_fix_adjustable.
30 (mips_fix_adjustable): Declare.
33 * read.c (read_a_source_file): After doing an mri_pending_align,
34 adjust the line_label if there is one.
35 (s_space): Set mri_pending_align if an odd number of bytes were
41 * config/tc-d10v.h (md_do_align): Add this hook to call
42 d10v_cleanup() when a ".align" is detected. Fixes PR11487.
44 * config/tc-d10v.c (find_opcode): Correctly calculate
45 branch displacement when .aligns are present.
50 * config/tc-mn10200.c (md_relax_table): Define.
51 (md_convert_frag): Implement.
52 (md_assemble): Handle relaxable operands/instructions correctly.
53 (md_estimate_size_before_relax): Implement.
54 * config/tc-mn10200.h (TC_GENERIC_RELAX_TABLE): Define.
58 * config/tc-mips.c (append_insn): Give an error for jumps to a
60 (md_apply_fix): Make a branch to an odd address an error rather
63 * config/tc-mips.c (md_convert_frag): If the user explicitly
64 requested an extended opcode, pass warn as true to mips16_immed.
66 * config/tc-mips.c (mips16_ip): Handle a missing expression like
67 an explicit 0, so that explicitly extended instructions work
72 * ecoff.c (ecoff_build_symbols): Don't generate a local ECOFF
73 symbol for a common symbol.
77 Patch presumed to have been checked in awhile ago but wasn't.
79 * write.c: Delete "ifndef md_relax_frag" around is_dnrange.
80 (relax_segment, case rs_org): Move code inside braces. Move locals
81 target,after inside too.
82 (relax_segment, case rs_machine_dependent): Guts moved to ...
83 (relax_frag): New function.
84 Call md_prepare_relax_scan if defined.
88 * config/tc-m68k.c (m68k_ip): Reject pc-relative addresses for the
89 'p' operand specifier.
93 * config/tc-m68k.c (HAVE_LONG_BRANCH): New macro, returns true for
94 m68k family cpus which support long branch addressing modes.
95 (m68k_ip, md_convert_frag_1, md_estimate_size_before_relax,
96 md_create_long_jump): Use it.
100 * config/tc-mips.c (md_begin): Don't set SEC_ALLOC or SEC_LOAD for
101 the .reginfo or .MIPS.options section if configured for an
104 * config/tc-mips.c (md_begin): Don't set interlocks for
109 * read.c (read_a_source_file): Make sure the symbol ends with
110 whitespace before checking whether the next character is '='.
114 * config/tc-i386.c (sco_id): Moved from here...
115 * config/obj-elf.c (sco_id): ...to here. Adding the identifier
116 really is an SCO ELF specific thing, not just a SCO x86 specific
122 * configure.in (tic80-*-*): Don't require 'coff'.
123 * configure: Regenerate.
128 * read.c (emit_expr): Check for overflow of a negative value
130 * write.c (fixup_segment): Likewise.
131 * config/obj-coff.c (fixup_segment): Likewise.
133 * config/tc-m68k.c (struct label_line): Define.
134 (labels, current_label): New static variables.
135 (md_assemble): Mark current_label as text, and clear it.
136 (m68k_frob_label): New function.
137 (m68k_flush_pending_output): New function.
138 (m68k_frob_symbol): New function.
139 * config/tc-m68k.h (tc_frob_label): Define.
140 (md_flush_pending_output): Define.
141 (tc_frob_symbol): Don't warn, just call m68k_frob_symbol.
142 (tc_frob_coff_symbol): Likewise.
144 * read.c (read_a_source_file): When defining a macro in MRI mode,
145 don't add the symbol to the symbol table.
149 * config/tc-mn10300.c (tc_gen_reloc): Handle sym1-sym2 fixups
150 here since fixup_segment doesn't (linkrelax is set).
151 * config/tc-mn10200.c (tc_gen_reloc): Likewise.
155 * config/tc-mn10200.c (md_assemble): Tweak fx_offset for pc-relative
160 * config/tc-hppa.c (struct hppa_fix_struct): Tweak fx_r_field's type
161 to avoid warnings with the native HP compiler.
162 (fix_new_hppa): Similarly for the r_type argument.
163 (pa_build_unwind_subspace, hppa_elf_mark_end_of_function): Enclose
164 in an #if OBJ_ELF to keep gcc -Wall quiet.
165 (md_apply_fix): Always initialize "result".
167 * config/tc-mn10200.c (md_assemble): Generate relocations.
171 * config/tc-m68k.c (s_even): Adjust the alignment of the current
176 * config/obj-elf.c (elf_file_symbol): When using ECOFF debugging,
177 pass on the new file hook.
179 * config/tc-alpha.c (alpha_fix_adjustable): Not quite the same as
180 !alpha_force_relocation, as local LITERALs can be adjusted to be
181 relative to the section.
185 * config/obj-coff.c (yank_symbols): If tc_frob_coff_symbol is
187 * config/tc-m68k.h (tc_frob_symbol): Check whether text label is
188 aligned to odd boundary.
189 (tc_frob_coff_symbol): Define.
191 * doc/as.texinfo (Set): Change parenthesized @xref to @pxref.
193 * macro.c (macro_expand_body): In MRI mode, just copy a single &.
195 * config/tc-m68k.c (m68k_ip): Call frag_grow before adding a
198 * config/tc-m68k.c (m68k_ip): Accept 'B' as a size for an
200 (md_assemble): If the size is 'B', set fx_signed.
201 (md_apply_fix_2): Use fx_signed when checking for overflow.
203 * write.h (struct fix): Add fx_signed field.
204 * write.c (fix_new_internal): Initialize fx_no_overflow and
206 (fixup_segment): Use fx_signed when checking for overflow.
207 * config/obj-coff.c (fixup_segment): Check fx_no_overflow and
208 fx_signed when checking for overflow.
212 * NOTES, NOTES.config: Removed. These are rarely, if ever,
213 updated, and all the useful information is in doc/internals.texi.
216 * read.c (read_a_source_file): Check for conditional operators
217 before doing an MRI pending alignment.
218 * config/tc-m68k.h (m68k_conditional_pseudoop): Declare.
219 (tc_conditional_pseudop): Define.
220 * config/tc-m68k.c (m68k_conditional_pseudop): New function.
221 * doc/internals.texi (CPU backend): Describe
222 tc_conditional_pseudoop.
225 * config/tc-m68k.c (m68k_rel32_from_cmdline): New static
227 (md_begin): Check m68k_rel32_from_cmdline before setting
229 (m68k_mri_mode_change): Likewise.
230 (md_longopts): Add --disp-size-default-16 and
231 --disp-size-default-32.
232 (md_parse_option): Handle new options.
233 (md_show_usage): Mention new options.
234 * doc/c-m68k.texi (M68K-Opts): Document new options.
237 * config/tc-m68k.c (m68k_index_width_default): New static
239 (m68k_ip): Use m68k_index_width_default to set the size of a base
240 register whose size was not given.
241 (md_longopts): Add --base-size-default-16 and
242 --base-size-default-32.
243 (md_parse_option): Handle new options.
244 (md_show_usage): Mention new options.
245 * doc/c-m68k.texi (M68K-Opts): Document new options.
247 * doc/c-mips.texi: Mention ISA level 4, and the -mips16 option.
249 * configure.in: Recognize mips-*-linux* target.
250 * configure: Rebuild.
252 * config/tc-mips.c (load_register): Rewrite 64 bit handling to
253 work if valueT is only 32 bits.
255 * config/tc-mips.c: Throughout, check target_big_endian rather
257 (byte_order): Remove.
258 (mips_init_after_args): Remove.
259 * config/tc-mips.h (LITTLE_ENDIAN, BIG_ENDIAN): Don't define.
260 (mips_init_after_args): Don't declare.
261 (tc_init_after_args): Don't define.
263 * config/tc-mips.h (tc_frob_after_relocs): Define if
265 (mips_elf_final_processing): Likewise.
266 (ELF_TC_SPECIAL_SECTIONS): Likewise.
271 * config/tc-v850.c (md_assemble): If this is sst.{h,w} or
272 sld.{h,w} and the operand is relocatable, adjust the adend by
273 shifting it right one bit.
278 * read.c (read_a_source_file): Check mri_pending_align after
279 checking for a macro. From Ronald F. Guilmette
282 * Makefile.in (ALL_CFLAGS): Add -D_GNU_SOURCE.
284 * config/tc-sparc.c (md_apply_fix3): Rename from md_apply_fix, and
285 add segment argument. If OBJ_ELF, treat a relocation against a
286 symbol in a linkonce section like a relocation against an external
288 * config/tc-sparc.h (MD_APPLY_FIX3): Define.
292 * config/tc-mips.c (mips16_macro): Add case for M_ABS.
296 * NOTES.config (Implementation): as.h #define's "GAS" not "gas",
297 includes config.h instead of host.h, tc.h instead of tp.h, and
298 targ-env.h instead of target-environment.h.
299 Also, obj-format.h includes targ-cpu.h instead of
302 * configure.in (case ${generic_target}): Add tic80-*-coff entry.
303 * configure: Rebuild with autoconf.
304 * config/obj-coff.h (coff/tic80.h): Include if TC_TIC80 defined.
305 (TARGET_FORMAT): Define to "coff-tic80".
306 * config/tc-tic80.c: New file for TIc80 support.
307 * config/tc-tic80.h: New file for TIc80 support.
312 * doc/as.texinfo (M): Mention explicitly that -M changes macro
317 * write.c (adjust_reloc_syms): If the fixup symbol has been
318 equated to an undefined symbol, convert the fixup to being against
319 the target symbol. Remove obsolete code handling a special case
324 * mpw-make.sed: Use NewFolderRecursive for installation.
329 * config/tc-d10v.c (do_assemble): Correct previous bug fix.
333 * config/tc-d10v.c (md_assemble): Fix bug which caused
334 second instruction in a line to be case sensitize. PR11312
339 * config/tc-mn10200.c (mn10200_insert_operand): Don't
340 range check operands with MN10200_OPERAND_NOCHECK set.
341 (check_operand): Likewise.
345 * config/tc-mips.c: Undo part of last Friday's alignment changes.
346 (md_begin): Always align the text section to a four byte
348 (append_insn): Remove call to record_align.
350 * config/tc-mips.c (insn_label): Remove.
351 (struct insn_label_list): Define.
352 (insn_labels, free_insn_labels): New static variables.
353 (mips_clear_insn_labels): New static function.
354 (append_insn): Mark all mips16 text labels, and make them odd.
355 Handle all labels after emitting a nop, not just one. Call
356 mips_clear_insn_labels rather than just clearing insn_label.
357 (mips_emit_delays): Add insns parameter, and use it to decide
358 whether to mark mips16 labels. Handle all labels, not just one.
359 Force mips16 labels to be odd. Change all callers.
360 (mips16_immed): Don't check for an odd branch target.
361 (md_apply_fix): Don't check mips16 mode for a branch reloc.
362 (mips16_extended_frag): Ignore the low bit in a branch target.
363 (md_convert_frag): Likewise.
364 (mips_no_prev_insn): Call mips_clear_insn_labels rather than just
366 (mips_align, mips_flush_pending_output, s_cons): Likewise.
367 (s_float_cons, s_gpword): Likewise.
368 (s_align): Use insn_labels rather than insn_label.
369 (s_cons, s_float_cons, s_gpword): Likewise.
370 (mips_frob_file_after_relocs): New function.
371 (mips_define_label): Rewrite to add to insn_labels list.
372 * config/tc-mips.h (tc_frob_file_after_relocs): Define.
373 * ecoff.c (ecoff_build_symbols): If the size of a function comes
374 out odd, increment it.
376 * config/tc-mips.c (append_insn): Only update prev_insn when not
377 reordering if place is NULL.
379 * config/tc-mips.c (mips16_ip): Check for a missing expression
380 when using the register indirect addressing mode.
384 * config/tc-mn10200.c (mn10200_insert_operand): Don't
385 check 24bit operands for overflow.
386 (check_operand): Likewise.
390 * doc/as.texinfo (Section): Document how to use the .section
391 pseudo-op for COFF and ELF.
395 * write.c (adjust_reloc_syms): Fix linkonce check for ELF.
399 * config/tc-mips.c (prev_insn_reloc_type): New static variable.
400 (RELAX_MIPS16_ENCODE): Add dslot and jal_dslot arguments, and
401 store them. Adjust other RELAX_MIPS16 macros.
402 (RELAX_MIPS16_DSLOT): Define.
403 (RELAX_MIPS16_JAL_DSLOT): Define.
404 (append_insn): Pass new arguments to RELAX_MIPS16_ENCODE. Correct
405 handling of whether previous instruction has a fixup. Set
406 prev_insn_reloc_type.
407 (mips_no_prev_insn): Clear prev_insn_reloc_type.
408 (mips16_extended_frag): Use the right base address for a PC
409 relative add or load.
410 (md_convert_frag): Likewise. If a PC relative add or load is
411 used, record the alignment for the section.
415 * write.c (adjust_reloc_syms): Don't reduce a reloc against a
416 linkonce section into a reloc against the section symbol.
418 * config/tc-mips.c (mips16_macro): Remove nop instructions after
421 * config/tc-mips.c (md_begin): If configured for an embedded ELF
422 system, don't set the section alignment to 2**4.
423 (s_change_sec): Likewise.
424 (append_insn): Call record_alignment for the section.
425 (md_section_align): Don't align the section size for an embedded
430 * write.c (adjust_reloc_syms): Make sure that symbols are
431 resolved; expression symbols may have been skipped.
432 * config/obj-coff.c (fixup_segment): Likewise.
436 * config/tc-ppc.c (ppc_elf_suffix): Move @plt to
437 BFD_RELOC_24_PLT_PCREL relocation.
438 (md_apply_fix3): Support BFD_RELOC_24_PLT_PCREL.
443 * config/tc-d10v.c (write_2_short): Remove code that called
444 parallel_ok() when the programmer specified parallel instructions.
449 * config/tc-mn10300.c (md_assemble): Update to handle endianness
452 * config/tc-mn10200.c (md_assemble): Opcode 0x0 is valid!
453 * config/tc-mn10300.c (md_assemble): Likewise.
457 * config/tc-mips.c (append_insn): Make sure there is enough room
458 in a frag after a mips16 instruction to switch it with a jump
461 * config/tc-mips.c (mips16_extended_frag): Give an error for an
462 attempt to use a non absolute symbol in an extending frag.
466 * config/tc-mn10200.c: Flesh out assembler support for MN10200.
467 * config/tc-mn10200.h: Likewise.
471 * app.c (do_scrub_chars): At the end of a C comment, pass space to
472 UNGET rather than PUT. Set old_state before setting state to -2.
474 * config/tc-mips.c (mips16_extended_frag): Avoid an infinite loop
475 when extending because the value is exactly maxtiny + 1.
477 * config/tc-mips.c (RELAX_MIPS16_ENCODE): Add small and ext
478 arguments, and store them. Adjust other RELAX_MIPS16 macros.
479 (RELAX_MIPS16_USER_SMALL): Define.
480 (RELAX_MIPS16_USER_EXT): Define.
481 (mips16_small, mips16_ext): New static variables.
482 (append_insn): Pass mips16_small and mips16_ext to
484 (mips16_ip): Set mips16_small and mips16_ext.
485 (mips16_immed): Don't check mips16_autoextend.
486 (mips16_extended_frag): Check USER_SMALL and USER_EXT.
488 * write.c (write_relocs): Print an error for an out of range
489 fixup, rather than calling abort.
491 * as.c (main): Unlink the output file if there are errors while
492 generating the fixups.
496 * config/tc-mips.c (mips16_extended_frag): Don't call
498 (md_convert_frag): Call resolve_symbol_value before calling
499 S_GET_VALUE, and don't add in the frag address.
501 * config/tc-mips.c (mips16_immed): Add file and line parameters,
502 and use them when reporting errors. Change all callers.
506 * config/tc-mn10300.c: Fix various gcc -Wall warnings.
507 Remove '$' prefixing for registers.
512 * config/tc-d10v.c (md_assemble): Check to see if prev_seg
513 is initialized before using it.
514 (d10v_cleanup): No longer uses its argument, so make it void.
516 * config/tc-d10v.h (d10v_cleanup): Change prototype.
521 * write.c (fixup_segment): Don't discard the symbol for a PC
522 relative fixup to an absolute symbol.
527 * config/tc-d10v.c (md_assemble, d10v_cleanup): Fix bug
528 with multiple sections.
533 * config/tc-mips.c (md_longopts): Rename mips-16 to mips16, and
534 no-mips-16 to no-mips16.
535 (s_mipsset): Accept .set mips16 and .set nomips16.
539 * config/tc-ppc.c (ppc_elf_suffix): Take expressionS pointer
540 argument, and check for +/- constant following the suffix, folding
541 it into the expression.
542 (ppc_elf_cons): Change ppc_elf_suffix calls.
543 (md_assemble): Ditto.
544 (shlib): Replace boolean mrelocatable with enumeration shlib.
545 (md_parse_option): Discriminate between PIC style shared libraries
547 (ppc_elf_validate_fix): Don't report warnings for PIC style shared
552 * config/tc-ppc.h ({tc,ppc}_comment_chars): Define, so that we can
553 change the comment characters.
555 * config/tc-ppc.c (comment_chars): Delete in favor of
557 (ppc_{eabi,solaris}_comment_chars): Eabi and Solaris versions of
559 (ppc_comment_chars): Select appropriate comment chars by default.
560 (msolaris): New flag for -m{,no-}solaris.
561 (md_parse_option): Recognize -K pic. Add support for
563 (md_show_usage): Update.
564 (md_begin): Do not set ELF flags if Solaris.
565 (ppc_elf_suffix): @local sets R_PPC_LOCAL24PC relocation.
566 (md_apply_fix3): Add support for R_PPC_LOCAL24PC.
570 * as.c (main): Correct handling of flag_always_generate_output.
574 * config/tc-mn10300.c (tc_gen_reloc): Get the addend from
575 fx_offset, not fx_addnumber.
577 * config/tc-mn10300.h (tc_fix_adjustable): Don't do any
582 * config/tc-i386.c: Correct misspelling: balenced to balanced.
586 * config/tc-mips.c (md_section_align): Check for an alignment of
587 4, not an alignment of 16. Corrects August 7 patch.
591 * configure, conf.in: Rebuild with autoconf 2.12.
593 * config/tc-ppc.c (ppc_elf_lcomm): Don't give an error if no
594 alignment is specified.
596 Add support for mips16 (16 bit MIPS implementation):
597 * config/tc-mips.c: Extensive additions for mips16 support, not
599 (RELAX_OLD, RELAX_NEW): Use only 7 bits each.
600 (insn_uses_reg): Change last parameter to an enum.
601 * config/tc-mips.h (LOCAL_LABELS_DOLLAR): Define as 0.
602 (md_relax_frag): Define as mips_relax_frag.
603 (mips_relax_frag): Declare.
604 (struct mips_cl_insn): Add use_extend and extend fields.
605 (tc_fix_adjustable): Define.
606 * config/obj-elf.h (S_GET_OTHER): Define.
607 (S_SET_OTHER): Define.
611 * config/tc-m68k.c (m68k_ip): Implement cases for new <, >, m, n,
612 o and p operand specifiers.
616 * write.c: Delete "ifndef md_relax_frag" around is_dnrange.
617 (relax_segment, case rs_org): Move code inside braces. Move locals
618 target,after inside too.
619 (relax_segment, case rs_machine_dependent): Guts moved to ...
620 (relax_frag): New function.
621 Call md_prepare_relax_scan if defined.
622 * config/tc-m68k.h (md_prepare_relax_scan): Renamed from
627 * config/tc-mn10300.c (address_registers): Use '$' as register
628 prefix instead of '%'.
629 (data_registers, other_registers, md_assemble): Likewise.
631 * config/tc-mn10300.c (address_registers): Use '%' prefix for regs.
632 (data_registers, other_registers, md_assemble): Likewise.
634 * config/tc-mn10300.c (md_assemble): Correctly determine the
635 correct location and type for each relocation.
636 (md_pcrel_from): Simplify.
640 * config/tc-sh.c (md_convert_frag): Improve warning when branch is
641 converted into branch around branch.
645 * config/tc-mn10300.h (DIFF_EXPR_OK): Don't define this.
646 (tc_fix_adjustable): Don't adjust relocs against weak symbols or
648 * config/tc-mn10300.c (md_begin): Set linkrelax.
649 (md_assemble): Create fixups as needed.
650 (md_apply_fix3): Gut. It shouldn't ever get called anymore.
655 * config/tc-d10v.c (parallel_ok): When automatically converting
656 serial ops to parallel, do not consider a branch as the first
662 * config/tc-mn10300.c (md_assemble): Handle MN10300_OPERAND_REG_LIST.
666 * config/tc-mn10300.c (mn10300_insert_operand): Provide prototype
668 (check_operand): Likewise.
673 * config/tc-d10v.c (parallel_ok): Branch and link instructions
675 (write_2_short): Call parallel_ok to check whether two short
676 instructions the user requested execute in parallel, can be
681 * config/tc-d10v.c (write_2_short): Fix bug that wouldn't
682 allow a branch and link in parallel with an exe instruction.
686 * doc/c-d10v.texi: Add info on @word modifier.
691 * config/tc-mn10300.c (mn10300_insert_operand): MN10300_OPERAND_SPLIT
692 operands are assumed to be 32bits. Use "bits" field to hold the
693 number of bits in the main instruction word for MN10300_OPERAND_SPLIT.
694 (mn10300_check_operand): MN10300_OPERAND_SPLIT operands are assumed
697 * config/tc-mn10300.c (mn10300_insert_operand): Shift low part
698 of a MN10300_OPERAND_SPLIT operand by operand->shift.
700 * config/tc-mn10300.c (mn10300_insert_operand): Handle
701 MN10300_OPERAND_SPLIT.
705 * config/tc-mn10300.c (md_assemble): Insert operands into
706 the extension part of the instruction if necessary.
707 (mn10300_insert_operand): Accept pointer to extension word
708 argument. Make insn a pointer argument too. Return type
709 is now void. All callers changed.
713 * config/tc-mn10300.c (mn10300_insert_operand): Handle
714 repeated register operands.
718 * doc/as.texinfo: Added section on reporting bugs.
720 * config/tc-alpha.c: Change uses of void * to PTR. Change the
721 alpha_macro emit field to expect a const argument, and change the
722 arg field to be const. Fix some spacing to follow the GNU
727 * config/tc-alpha.c (md_parse_option): Add knowledge of 21164pc
728 (pca56) and 21264 (ev6) cpus.
729 (md_apply_fix): Private relocation types are now negative.
730 (alpha_force_relocation): Likewise.
731 (tc_gen_reloc): Likewise.
732 (emit_insn): Likewise.
733 (emit_ldXu): Do the right thing when the hardware can do byte insns.
734 (emit_stX): Likewise.
735 (emit_sextX): Likewise.
739 * config/obj-coff.c (do_relocs_for): Call resolve_symbol_value on
740 a symbol found in a reloc.
742 * symbols.c (resolve_symbol_value): Improve the error message if
743 an undefined symbol is used in an expression.
747 * doc/internals.texi: Rewrite, and add a lot of documentation.
748 * doc/Makefile.in (internals.info): New target.
753 * config/tc-v850.h (tc_fix_adjustable): Don't adjust relocs
754 against weak symbols.
758 * config/tc-v850.c (md_assemble): Don't lose for relaxable
761 * config/tc-v850.c (md_convert_frag): Make sure we insert the
762 fixup at the right address within the frag.
764 * config/tc-v850.c (md_convert_frag): Don't set fragP->fr_fix
765 to an absolute value, instead increment it as needed.
767 * config/tc-v850.h (TC_GENERIC_RELAX_TABLE): Define.
768 * config/tc-v850.c: Fix some indention problems.
769 (md_relax_table): Define for D9->D99 branch displacement
771 (md_convert_frag): Do something useful instead of aborting.
772 (md_estimate_size_before_relax): Likewise.
773 (md_assemble): Note if the matching instruction has a relaxable
774 operand. If it does, allocate frag with frag_var and don't
781 * config/tc-d10v.h (md_cleanup): New function. This is needed to
782 write out any buffered instructions when a ".end" is found.
787 * read.c (read_a_source_file): New hook md_cleanup().
791 * write.c (fix_new_exp): Use make_expr_symbol to build an
792 expression symbol for a complex fixup.
797 * config/tc-v850.c (v850_reloc_prefix): Several disgusting
798 hacks to improve parsing of complex hi, lo, zda, etc
800 (md_assemble): Don't demand and eat a trailing ')' after finding
801 a v850 relocation prefix. Sign extend the constant in a
802 BFD_RELOC_LO16 expression. Do eat a trailing ')' after a complete
804 (parse_cons_expression_v850): Don't eat a trailing ')' after
805 finding a v850 relocation prefix.
807 * config/tc-v850.h (TC_PARSE_CONS_EXPRESSION): Define.
808 (TC_CONS_FIX_NEW): Likewise.
809 * config/tc-v850.c (parse_cons_expression_v850): New function.
810 (cons_fix_new_v850): Likewise.
812 * config/tc-v850.h (tc_fix_adjustable): Don't adjust TDA relocs.
817 * config/tc-ppc.c (md_apply_fix3): Give a better warning message
818 for an unknown relocation type.
823 * config/tc-v850.c (md_pseudo_table): Add .word; allocates
828 * config/tc-v850.c (md_assemble): Handle TDAOFF relocs
829 differently for movea & sst/sld insns.
835 * config/tc-d10v.c (parallel_ok): Don't allow illegal combinations
841 * obj.h (struct format_ops): Add frob_file_after_relocs field.
842 * config/obj-multi.h (obj_frob_file_after_relocs): Define.
843 * config/obj-ecoff.c (ecoff_format_ops): Initialize new
844 frob_file_after_relocs field.
845 * config/obj-elf.c (elf_format_ops): Likewise.
846 * config/tc-mips.c: Undefine obj_frob_file_after_relocs before
851 * config/tc-mips.c (cons_fix_new_mips): Only treat 8 byte reloc
852 specially if not ELF.
853 (md_apply_fix): Handle BFD_RELOC_64.
854 (tc_gen_reloc): Handle BFD_RELOC_64.
856 * config/tc-i386.c (md_apply_fix3): Don't increment value for a PC
857 relative reloc when BFD_ASSEMBLER and OBJ_AOUT (more ugly gas
860 * config/obj-aout.h (S_IS_DEFINE): non BFD_ASSEMBLER version:
861 Don't check S_GET_OTHER.
865 * config/tc-mips.c (mips_ip): Accept an odd floating point
866 register with l.s or s.s.
868 * config/obj-aout.c (obj_pseudo_table): Use obj_aout_type for
870 (obj_aout_type): New static function.
874 * Makefile.in ($(OBJS)): Depend upon libiberty.h.
879 * config/tc-v850.c (v850_reloc_prefix): Recognize zdaoff, tdaoff
880 and sdaoff expressions.
882 * write.c (fixup_segment): Don't add symbol value to addend if
884 * config/tc-v850.h (tc_fix_adjustable): Don't adjust any
887 * config/tc-v850.c (md_pcrel_from): Undo yesterday's changes.
888 (md_pcrel_from_section): Likewise.
889 * config/tc-v850.h (MD_PCREL_FROM_SECTION): Likewise.
893 * config/tc-v850.c (md_pcrel_from): Delete unused function.
894 (md_pcrel_from_section): New function.
895 * config/tc-v850.h (MD_PCREL_FROM_SECTION): Define.
900 * config/tc-mips.c (load_register): Add cast to offsetT when using
905 * config/obj-elf.c (elf_frob_file): Move ECOFF debug processing to ...
906 (elf_frob_file_after_relocs): ... here. New function.
907 * config/obj-elf.h (obj_from_file_after_relocs): New macro.
908 * write.c (write_object_file): Call *frob_after_relocs after the
909 call to write_relocs.
911 * config/tc-alpha.c: Use new BFD_RELOC_ALPHA_ELF_LITERAL reloc.
913 * config/tc-alpha.c (load_expression): Don't SET_VALUE on the section
914 symbol, as this messes up linking. Instead, expand the recursive call
915 inline and change up the appropriate bits to get the 0x8000 offset
920 * config/tc-sparc.h (tc_fix_adjustable): Permit the difference of
921 two symbols in the same segment to be adjusted.
923 * configure.in: Don't get confused by CPU-VENDOR-linux-gnu.
924 * configure: Rebuild.
928 * config/tc-ppc.c (ppc_insert_operand): Change most warnings into
930 (ppc_elf_validate_fix): Ditto.
931 (md_assemble): Ditto.
933 (ppc_pe_section): Ditto.
934 (ppc_frob_symbol): Ditto.
938 * config/mn10300.c (md_assemble): Pass an extra shift count
939 to mn10300_insert_operand based on the opcode format.
940 (mn10300_insert_operand): Accept and use extra shift count
943 * config/tc-mn10300.c (md_assemble): Use FMT_* macros for
944 formats rather than hard-coded constants.
946 * config/tc-mn10300.c (md_assemble): Format D5 instructions
947 are 7 bytes long. Write out instructions in big-endian format.
951 * config/tc-mn10300.c (md_assemble): Tweak further so
952 that all instructions are parsed correctly.
956 * as.h: Include libiberty.h.
957 (xmalloc, xrealloc): Don't declare.
958 * as.c: Don't include libiberty.h.
959 * expr.c, read.c, stabs.c, config/obj-coff.c: Likewise.
960 * config/tc-mips.c: Likewise.
961 * messages.c: Likewise.
962 (xstrerror): Don't declare.
967 * config/tc-mn10300.h (pre_defined_registers) Remove.
968 (system_registers, cc_names): Likewise.
969 (address_registers, data_registers, other_registers): New register
971 (register_name, system_register_name, cc_name): Remove.
972 (mn10300_reloc_prefix): Likewise.
973 (data_register_name): New function.
974 (address_register_name, other_register_name): Likewise.
975 (md_assemble): Rough cut at parsing operands. Remove lots of
977 (md_apply_fix3): Disable for now.
981 * config/tc-m68k.c (select_control_regs): New function, extracted
982 out of m68k_init_after_args.
983 (m68k_init_after_args): Use it.
984 (mri_chip): Use it here as well to update set of allowed control
989 * config/obj-elf.c (elf_begin): New function.
990 (obj_elf_section): Add the section symbol to the symbol table.
991 * config/obj-elf.h (obj_begin): Define.
992 (elf_begin): Declare.
993 * as.c (perform_an_assembly_pass): Call obj_begin if it is
998 * config/obj-coff.c (fixup_segment): Subtract the section address
999 from a PC relative reloc if TC_M68K.
1003 * config/tc-sparc.c (md_pseudo_table): Make .uahalf, .uaword, and
1004 .uaxword available even if not OBJ_ELF.
1005 (md_atof): Remove unused local variable wordP.
1009 * config/tc-mn10x00.c, config/tc-mn10x00.h: New files
1010 for Matsushita MN10x00 support.
1011 * configure.in: Recognize mn10x00-*-*
1012 * configure: Rebuilt.
1016 * obj-evax.h: move openvms definitions from here to tc-alpha.c.
1017 * tc-alpha.c: add support for vms_case_hack like in vax/vms.
1018 (load_expression): track clobbering of base reg before jmp/jsr.
1019 (s_alpha_file): pass case_hack flags and source filename via
1020 symbol table to bfd.
1021 * tc-alpha.h (TC_CONS_FIX_NEW): define
1025 * configure.in (mips-*-rtems*): New target, like mips-*-elf*.
1026 * configure: Rebuild.
1030 * read.c (s_macro): Warn if a macro has the same name as a
1032 (s_space): In m68k MRI mode, align to a word boundary.
1033 * macro.c (define_macro): Add namep parameter. Change all
1035 * macro.h (define_macro): Update declaration.
1037 * as.c (show_usage): Print bug report address.
1038 (parse_args): Change version printing to match current GNU
1040 * gasp.c (show_usage): Print bug report address.
1041 (main): Change version printing to match current GNU standards.
1043 * config/tc-m68k.c (init_table): Correct access control unit
1046 * config/tc-alpha.c: Add some static function prototypes.
1047 (alpha_macros): Move to top of file. Make static.
1048 (alpha_num_macros): Move to top of file.
1053 * tc-v850.h: Define LOCAL_LABEL to recognise _.L_* symbols
1059 * listing.c (list_symbol_table): Remove bogus code in BFD64 case,
1060 and just call sprintf_vma.
1064 * expr.c (expr): Change >>= to >> (fix typo). (From meissner).
1068 * read.c (float_cons): Call md_flush_pending_output if it is
1074 * config/tc-d10v.c (md_operand): Created. Allows operands to
1076 * config/tc-d10v.h (md_operand): Undefined.
1081 * config/tc-m68k.c (add_fix): Treat a width of '3' like 'B'.
1082 (md_assemble): A fixup width of '3' means a 1 byte reloc.
1086 * config/obj-coff.c (fixup_segment): Don't adjust PC relative
1087 reloc for the i960 for a reloc in the same section. This undoes
1088 one of the two changes made Aug 19.
1092 * config/obj-coff.c (obj_coff_endef): Both versions: Move C_STAT
1093 symbols to the position of the debugging information.
1097 * expr.c (expr): Always use unsigned right shifts for >>.
1101 * config/tc-arm.c (md_apply_fix3): Update two thumb instruction
1102 slots when processing BL fixups.
1104 * config/tc-arm.c (output_inst): Ensure Thumb BL fixup is marked
1105 on the first half of the instruction.
1109 * ecoff.c (ecoff_stab): Create an expression symbol for a complex
1110 stabs expression, rather than giving an error.
1112 * ecoff.c (ecoff_new_file): Don't do anything if we are still in
1117 * config/tc-mips.c (append_insn): Fill in the value for a constant
1118 jump, rather than creating a reloc.
1122 * config/tc-mips.c (append_insn): Don't swap an instruction which
1123 sets a condition code with an instruction which uses a condition
1125 (mips_ip): In cases 'N' and 'M', look for $fccN rather than an
1128 * config/tc-mips.c (md_begin): Recognize r5000 for cpu.
1129 (mips_ip): Give a better error message if the ISA level is wrong.
1130 (md_parse_option): Recognize -mcpu=[v][r]5000.
1134 * config/tc-mips.c (COUNT_TOP_ZEROES): Added macro to count
1136 (load_register): Ensure hi32 bits are not lost during lo32bit
1137 processing. Fix shift offset that was overflowing into the next
1138 instruction field. Add code to generate shorter sequences for
1139 constants with a single contiguous seqeuence of ones.
1144 * config/tc-d10v.c (d10v_dot_word): New function to support
1145 "@word" with the word pseudo-op.
1146 (md_apply_fix3): Cleanup and changes to support correct sizes
1147 for 16 and 18-bit relocs.
1152 * configure.in (sparc-*-aout): Set `em'.
1153 * configure: Regenerated.
1154 * config/te-sparcaout.h: New file.
1155 * config/tc-sparc.h (TARGET_BYTES_BIG_ENDIAN): Define.
1156 Ifdef TE_SPARCOUT define TARGET_FORMAT and SPARC_BIENDIAN.
1157 * config/tc-sparc.c (INSN_BIG_ENDIAN): New macro.
1158 (SPECIAL_CASE_{SETSW,SETX}): Define.
1159 ({NOP,OR,FMOVS,SETHI,SLLX,SRA}_INSN): Define.
1160 (md_begin): Delete setting of `target_big_endian'.
1161 (output_insn): New function.
1162 (md_assemble): Rewrite. Add `setx' support.
1163 (sparc_ip): Handle `0' operand char. Recognize setuw, setsw, setx
1165 (md_atof): Add little endian support.
1166 (md_number_to_chars): Likewise.
1167 (md_apply_fix): Likewise.
1168 (md_longopts): Recognize -EL,-EB ifdef SPARC_BIENDIAN.
1169 (md_parse_option): Likewise.
1170 (md_show_usage): Print -EL, -EB ifdef SPARC_BIENDIAN.
1174 * ecoff.c (ecoff_new_file): New function.
1175 * ecoff.h (ecoff_new_file): Declare.
1176 * config/obj-ecoff.h (obj_app_file): Define.
1180 * config/tc-alpha.c (load_expression): Bias the .lit8 section
1181 symbol by 32k so that our 16-bit signed offset can address the
1186 * config/tc-mips.c (load_register): Remove unused variable tmp.
1190 * config/tc-mips.c (load_register): Remove unnecessary code that
1191 was causing the high 32bits of 64bit constants to be lost.
1196 * config/tc-d10v.c: Added changes to support function
1197 pointers and "@word" syntax.
1203 * config/tc-v850.c: Remove commented out and #if 0'd code.
1204 (v850_reloc_prefix): Provide prototype.
1205 (postfix, get_reloc, build_insn): Remove prototypes for nonexistant
1207 (md_begin, md_assemble, md_apply_fix3): Remove unused variables.
1208 (md_assemble): Add default to case statement.
1212 * config/tc-v850.c (md_assemble): Compute size of the instrction
1215 * config/tc-v850.c (md_apply_fix3): Do simple byte, short and
1220 * config/tc-v850.c (md_apply_fix3): Use little endian get/put
1221 routines to fetch/store the updated instruction from/to memory.
1222 (v850_insert_operand): If the operand has a specialized insert
1227 * config/tc-v850.c (reg_name_search): Align calling convention to
1228 be like identical function found in tc-ppc.c.
1229 (get_reloc): Removed.
1230 (v850_reloc_prefix): New function, parse lo(), hi() and hi0().
1231 (md_assemble): emit fixups.
1232 (md_pcrel_from): renamed from md_pcrel_from_section, emit proper
1234 (md_apply_fix3): handle fixups/relocs.
1235 * config/tc-v850.h (MD_PCREL_FROM_SECTION): Removed definition.
1241 * configure.in (sh-*-elf*): New target.
1242 * config/tc-sh.h (TARGET_ARCH): Define.
1243 (WORKING_DOT_WORD): Define.
1244 (TC_COFF_FIX2RTYPE): Only define if OBJ_COFF.
1245 (BFD_ARCH, COFF_MAGIC, TC_COUNT_RELOC): Likewise.
1246 (TC_RELOC_MANGLE, tc_coff_symbol_emit_hook): Likewise.
1247 (DO_NOT_STRIP, NEED_FX_R_TYPE, TC_KEEP_FX_OFFSET): Likewise.
1248 (TC_COFF_SIZEMACHDEP, tc_frob_file): Likewise.
1249 (SUB_SEGMENT_ALIGN): Likewise.
1250 (RELOC_32): Don't define.
1251 (tc_frob_file_before_adjust): Define if BFD_ASSEMBLER.
1252 (target_big_endian): Declare if OBJ_ELF.
1253 (TARGET_FORMAT): Define if OBJ_ELF.
1254 * config/tc-sh.c: Use BFD reloc codes instead of SH COFF reloc
1256 (tc_crawl_symbol_chain): Only define if OBJ_COFF.
1257 (tc_headers_hook, tc_coff_sizemachdep): Likewise.
1258 (struct sh_count_relocs): Define.
1259 (sh_count_relocs): New static function, broken out of
1260 sh_frob_file. Add BFD_ASSEMBLER code.
1261 (sh_frob_section): Likewise.
1262 (sh_frob_file): Call sh_frob_section.
1263 (md_convert_frag): If BFD_ASSEMBLER, change type of headers, and
1264 call section_symbol rather than seg_info (seg)->dot.
1265 (md_section_align): Add OBJ_ELF version.
1266 (SWITCH_TABLE_CONS): Define.
1267 (SWITCH_TABLE): Use SWITCH_TABLE_CONS.
1268 (md_apply_fix): Change parameter types if BFD_ASSEMBLER. Only
1269 handle fx_r_type == 0 if not BFD_ASSEMBLER. Return 0 if
1271 (struct reloc_map): Define if not BFD_ASSEMBLER.
1272 (coff_reloc_map): Likewise.
1273 (sh_coff_reloc_mangle): Use coff_reloc_map to convert fx_r_type.
1274 (tc_gen_reloc): New function if BFD_ASSEMBLER.
1275 * write.c (write_relocs): Ifdef out fx_where test which triggers
1276 inappropriately for SH ELF.
1277 (write_object_file): Call tc_frob_file_before_adjust and
1278 obj_frob_file_before_adjust if they are defined.
1280 * write.c (write_object_file): Use BFD_RELOC_16, not
1281 BFD_RELOC_NONE, when calling fix_new_exp for a broken word.
1283 * read.c (emit_expr): Fix conversion of byte count to BFD reloc
1289 * config/tc-d10v.c (find_opcode): Fix problem with calculating
1290 branch sizes in across sections.
1296 * config/tc-850.c (md_assemble): Handle hi() correctly. Handle
1301 * config/tc-v850.c (md_begin): Deal with end of opcode
1308 * config/tc-d10v.c (find_opcode): Fix a bug which could generate
1309 the wrong opcode for cases like st2w where there are many forms
1310 of the same instruction.
1315 * expr.c (operand): If md_parse_name is defined, call it before
1316 calling symbol_find_or_make.
1317 * config/tc-ppc.h (md_parse_name): Define.
1318 (ppc_parse_name): Declare.
1319 * config/tc-ppc.c (reg_name_search): Add regs and regcount
1321 (register_name): Update call to reg_name_search.
1322 (cr_operand): New static variable.
1323 (cr_names): New static const array.
1324 (ppc_parse_name): New function.
1325 (md_assemble): If PPC_OPERAND_CR is set in the operand flags, set
1326 cr_operand before calling expression.
1330 * config/tc-hppa.c (tc_gen_reloc): Add new argument to
1331 hppa_gen_reloc_type call.
1336 * config/tc-d10v.c: Fixed ".word". Fixed problem with range checking
1337 on addresses. Improved error messages.
1338 * doc/c-d10v.texi: Added docs for register pairs.
1342 * config/tc-d10v.c (parallel_ok): Fix bug in parallel
1348 * ecoff.c (init_file): Initialize fMerge to 1.
1349 (add_file): Restore old file merging code, but only merge files if
1351 (ecoff_directive_loc): Clear fMerge field of current file.
1352 (ecoff_generate_asm_lineno): Likewise.
1357 * doc/c-d10v.texi: Fix typo.
1363 * config/tc-v850.c (md_assemble): Correct bit masking for
1364 hi and lo expressions.
1366 * config/tc-v850.c (md_assemble): Rough cut at demanding
1367 "ep" or "r30" in sst and sld instructions.
1368 (md_apply_fix3): Don't abort. Just warn that we don't
1371 * config/tc-v850.c (CC_NAME_CNT): Define.
1372 (cc_name): New function.
1373 (md_assemble): Handle V850_OPERAND_CC correctly.
1375 * config/tc-v850.c (md_assemble): Don't forget to initialize
1378 * config/tc-v850.c (reg_name_search): Generalize to search
1379 any given register table.
1380 (register_name): Pass appropriate table and size to reg_name_search.
1381 (system_register_name): New function.
1382 (SYSREG_NAME_CNT): Define.
1383 (md_assemble): Handle operands which are system registers.
1385 * config/tc-v850.c (md_assemble): If we find a register, but the
1386 opcode doesn't want a register, then we don't have a match.
1387 (md_assemble): Get size of the instruction from the opcode table.
1392 * configure.in: Set and substitute HLDENV.
1393 * configure: Rebuild.
1394 * Makefile.in (HLDENV): New variable.
1395 (as.new): Use $(HLDENV).
1397 * ecoff.c (ecoff_directive_endef): Avoid a division by zero error
1398 if an array dimension is not known.
1403 * config/tc-d10v.c: Fix a reloc bug caused by my last change.
1404 * doc/c-d10v.texi: Cleanup.
1410 * config/tc-v850.c: New file.
1411 * config/tc-v850.h: New file.
1412 * configure (v850-*-elf): New target.
1413 * configure.in (v850-*-elf): New target.
1419 * doc/c-d10v.texi: New file.
1420 * doc/all.texi: Added D10V stuff.
1421 * doc/as.texinfo: Added D10V stuff.
1425 * config/tc-d10v.c: All references to defined symbols should
1426 now use the optimal instruction. .float and .double now work.
1431 * config/obj-coff.c (fixup_segment): Adjust PC relative reloc by
1432 section address for the i960 as is done for the i386.
1436 * mpw-config.in: Add wildcards for config matching, add mips-*-*
1437 case, forward-include bfd/elf-bfd.h.
1442 * config/tc-d10v.c: Add additional information to the opcode
1443 table to help determinine which instructions can be done
1449 * config/tc-arm.c: Major changes to add Thumb support, with lots
1451 Reverted to INSN_SIZE macro, rather than insn_size variable.
1452 (insns): Added ARM "bx" instruction support.
1453 (tinsns): Added Thumb instruction definition structure.
1454 (arm_tops_hsh): Added hash structure for Thumb opcodes.
1455 (md_pseudo_table): Added ".arm", ".thumb" and ".code" pseudo-ops.
1456 (opcode_select,s_arm,s_thumb,s_code): Added.
1457 (decode_shift): Allow upper-case RRX.
1458 (do_ldst): Simpler halfword support.
1459 (do_ldmstm): Improved.
1460 (reg_list, do_bx, thumb_reg, thumb_add_sub, thumb_shift,
1461 thumb_mov_compare, thumb_load_store, do_t_arit, do_t_add,
1462 do_t_asr, do_t_branch, do_t_bx, do_t_compare, do_t_ldmstm,
1463 do_t_ldrb, do_t_ldrh, do_t_lds, do_t_lsl, do_t_lsr, do_t_mov,
1464 do_t_push_pop, do_t_str, do_t_strb, do_t_strh, do_t_sub, do_t_swi,
1466 (md_apply_fix3): Add support for BFD_RELOC_ARM_THUMB_* relocations.
1467 (md_parse_option): Add support for -mthumb.
1468 (md_show_usage): Updated to reflect new command line option.
1469 (arm_data_in_code, arm_canonicalize_symbol_name): Added.
1470 * config/tc-arm.h: Provide TC_FIX_TYPE to allow private ARM
1471 fragment information to be held.
1475 * tc-arm.c (md_apply_fix3): Also set fixP->fx_done if fx_addsy is
1476 non-null, but is a constant.
1477 (fix_new_arm): Call make_expr_symbol to make the expression symbol
1478 so that error reporting will work correctly.
1482 * config/tc-i386.c (tc_i386_fix_adjustable): Don't adjust relocs
1483 against weak symbols.
1487 * config/tc-ppc.h (TC_FORCE_RELOCTION): Define if OBJ_XCOFF.
1488 (ppc_force_relocation): Declare if OBJ_XCOFF.
1489 * config/tc-ppc.c (ppc_force_relocation): New function if
1494 * config/tc-mips.h (BYTE_ORDER): Don't define. No longer used.
1499 * config/tc-d10v.c: Fix problem with relocs.
1504 * config/tc-sh.c (sh_do_align): If not BFD_ASSEMBLER, always align
1505 with nops if not in data_section or bss_section.
1509 Add support for openVMS/Alpha.
1510 * as.h (PRINTF_LIKE): Don't define if VMS, for now.
1511 * config/obj-evax.c: New file.
1512 * config/obj-evax.h: New file.
1513 * config/tc-alpha.c: Add support for EVAX format if OBJ_EVAX is
1515 * config/tc-alpha.h: Add support for EVAX format if OBJ_EVAX is
1516 defined. Add case for bfd_target_evax_flavour.
1517 * config/vms-a-conf.h: New file.
1518 * conf-a-gas.com: New file.
1519 * configure.in: Add target alpha-*-*vms*.
1520 * configure: Rebuild.
1521 * makefile.vms: New file.
1522 * read.c (s_lcomm): Align bss_seg on 8 byte boundary if OBJ_EVAX.
1523 Don't call ffs on openVMS/Alpha.
1527 * configure.in: Make GAS_CHECK_DECL_NEEDED include <string.h> or
1528 <strings.h> if they exist. Call GAS_CHECK_DECL_NEEDED on strstr
1530 * acconfig.h (NEED_DECLARATION_STRSTR): New macro.
1531 (NEED_DECLARATION_SBRK): New macro.
1532 * configure, conf.in: Rebuild.
1533 * as.h: Only include <strings.h> if HAVE_STRINGS_H.
1534 (strstr): Declare if NEED_DECLARATION_STRSTR.
1535 * as.c: If HAVE_SBRK and NEED_DECLARATION_SBRK, declare sbrk.
1539 * symbols.c (resolve_symbol_value): Handle addition or subtraction
1540 by a constant before entering the main switch. Reject attempts to
1541 apply an arithmetic function to non-absolute symbols, except for
1542 the special case of subtraction of two symbols in the same
1545 * config/tc-mips.c (md_section_align): Do align if OBJ_ELF, but
1546 not to more than a 16 byte boundary.
1548 * config/tc-i386.c (tc_gen_reloc): Accept all relocs; remove
1550 (tc_gen_reloc): If out of memory call as_fatal rather than
1551 assert. If no howto found, call as_bad_where rather than
1552 as_fatal. Change the error message slightly. Set howto to a
1553 non-NULL value in order to keep going.
1558 * config/tc-d10v.c: Added code to support 32-bit fixups for stabs.
1563 * config/tc-h8300.c (get_specific): New operand "size" derived
1564 from ".b", ".w" and ".l" extensions. All callers changed. If
1565 the base instruction has no operands, then use the size to
1566 determine which specific instruction to use.
1570 * config/tc-i960.c (mem_fmt): Call parse_expr before emit.
1574 * config/tc-mips.c (md_section_align): Don't change addr if
1579 * config/tc-hppa.c: Revert yesterday's changes.
1584 * config/tc-d10v.c: Disable range checking on 16-bit values.
1589 * config/tc-m68k.c (m68k_ip): Set ok_arch for every instruction,
1590 not just the ones that don't match.
1595 * config/tc-d10v.c: Fixed bugs in short relocs and range checking.
1600 * config/tc-arm.c: Changed INSN_SIZE to variable insn_size, as
1601 pre-cursor to adding Thumb support. Also added cpu_variant flag
1602 information to each of the asm_flg structures.
1603 (md_parse_option): Updated ARM7 parsing to allow 't' for
1604 thumb/halfword support, aswell as 'm' for long multiply.
1605 (md_show_usage): Updated help message.
1606 (md_assemble): Check that instruction flags are applicated to the
1607 current cpu variant.
1608 (md_apply_fix3, tc_gen_reloc): Add BFD_RELOC_ARM_OFFSET_IMM8 and
1609 BFD_RELOC_ARM_HWLITERAL relocation support for new halfword and
1610 signextension instructions.
1611 (do_ldst): Generate halfword and signextension variants if
1612 mnemonic flags match.
1613 (ldst_extend): Do not allow shifts in the offset field of halfword
1614 or signextension instructions.
1615 (validate_offset_imm): Provide check on halfword and signextension
1617 (add_to_lit_pool): Merge identical literal pool values.
1621 * config/tc-hppa.c (selector_table): Add 'E' selector.
1622 (cons_fix_new_hppa): Don't coke on e_esel.
1623 (tc_gen_reloc, SOM version): Handle R_COMP2 when used
1624 to help generate exception handling tables.
1625 (md_apply_fix): Don't try to apply fixups with an e_esel
1627 (hppa_fix_adjustable): Fixups with e_esel selectors
1632 * config/tc-sparc.c (md_pseudo_table): Add 2byte, 4byte, and 8byte
1638 * config/tc-d10v.c: Added lots of error checking. Added hacks
1639 to support accumulator shifts.
1644 * symbols.c (S_SET_EXTERNAL): Let .weak override.
1645 (S_CLEAR_EXTERNAL): Likewise.
1646 (S_SET_WEAK): Remove error; just let .weak override.
1651 * config/tc-d10v.c (md_assemble): Now handles multiline
1656 * config/tc-d10v.c: Fix packaging bug. Added range checking.
1657 Added kludge for divs instruction. Fixed minor problem with
1658 multiple text sections.
1659 * config/tc-d10v.h (d10v_cleanup): Change prototype.
1663 * config/tc-d10v.c (md_apply_fix3): Fix all instruction
1664 addresses to be right-shifted by 2.
1668 * config/tc-d10v.c: Many changes to get relocs working.
1669 (register_name): No longer creates a symbol for register names.
1670 (pre_defined_registers): moved to opcodes/d10v-opc.c.
1671 (d10v_insert_operand): Now works correctly for either container.
1672 * config/tc-d10v.h (d10v_cleanup): Declare.
1677 * config/tc-mips.c (tc_gen_reloc): BFD_RELOC_PCREL_HI16_S and
1678 BFD_RELOC_PCREL_LO16 are expected to be PC relative.
1682 * tc-alpha.c: Patches to track current minimum alignment to reduce
1683 the number of fragments created with frag_align.
1684 (alpha_current_align): New static variable.
1685 (s_alpha_text): Reset alignment to 0.
1686 (s_alpha_data, s_alpha_rdata, s_alpha_sdata): Likewise.
1687 (s_alpha_stringer, s_alpha_space): New functions.
1688 (s_alpha_cons, alpha_flush_pending_output): Remove functions.
1689 (alpha_cons_align): New function to replace both of them.
1690 (emit_insn): Only align if alpha_current_align is less than 2;
1691 reset alpha_current_align to 2.
1692 (s_alpha_gprel32): Likewise.
1693 (s_alpha_section): New function. Basically duplicate the other
1694 alpha section change hooks. Only define for ELF.
1695 (s_alpha_float_cons): Simplify alignment handling.
1696 (md_pseudo_table): Only define "rdata" and "sdata" if OBJ_ECOFF.
1697 If OBJ_ELF, define "section", "section.s", "sect", and "sect.s".
1698 Don't define the s_alpha_cons pseudo-ops. Do define
1699 s_alpha_stringer and s_alpha_space pseudo-ops.
1700 (alpha_align): Skip if less than current default alignment. Set
1702 * tc-alpha.h (md_flush_pending_output): Remove.
1703 (md_cons_align): Add.
1705 * tc-alpha.c: Add oodles of function description comments.
1706 (md_bignum_to_chars): Remove; there are no callers.
1707 (md_show_usage): Mention some more variants.
1712 * configure.in (sparc-*-sysv4*): New target.
1713 * configure: Rebuild.
1715 * config/tc-sparc.c (md_pseudo_table): Change uahalf, uaword, and
1716 uaxword to use s_uacons.
1717 (sparc_no_align_cons): New static variable.
1718 (s_uacons): New static function.
1719 (sparc_cons_align): If sparc_no_align_cons is set, just clear it
1722 * config/tc-sparc.c (s_common): Remove unused label allocate_bss.
1724 * configure.in: Add mips-*-irix6* target. Handle Irix 6 like Irix
1725 5 with regard to shared libraries.
1726 * configure: Rebuild.
1728 * config/tc-m68k.c (m68k_ip): Use the correct length when
1729 allocating space for the unsupported architecture error message.
1734 * configure.in (d10v-*-*): Allow d10v-*-*, don't require d10v-*-elf*.
1738 * config/tc-d10v.c: New file.
1739 * config/tc-d10v.h: New file.
1740 * configure (d10v-*-elf): New target.
1741 * configure.in (d10v-*-elf): New target.
1746 * config/tc-ppc.c (md_parse_option): Recognize -K PIC.
1750 * config/tc-alpha.c (alpha_align): Change fill parameter
1751 to a pointer. Take NULL as 0 or nop depending on section. Change
1753 (s_alpha_align): Rename local variables.
1755 * doc/as.texinfo (.align): Document action of omitted
1760 * config/tc-ppc.c (md_apply_fix3): Give a useful error message
1761 when an unsupported PC relative reloc is seen, rather than calling
1764 * app.c (do_scrub_chars): Remove not_cpp_line local variable.
1765 Instead, check state when '#' comment is seen.
1769 * config/tc-mips.c (mips_regmask_frag): Only define if OBJ_ELF or
1771 (tc_gen_reloc): If fixup was changed to be PC relative, change
1772 reloc type accordingly. Use name of reloc in error message.
1774 * as.h: Don't define const or volatile.
1775 * flonum.h: Don't define const.
1777 * config/tc-m68k.c (tc_gen_reloc): Change the code appropriately
1778 if fx_pcrel is set. Correct setting the addend case in the
1779 OBJ_ELF case (from Andreas Schwab
1781 (md_show_usage): Correct -mfc5200 to -m5200.
1785 * doc/c-m68k.texi: Document -m5200 flag.
1786 * doc/as.texinfo: Likewise.
1788 * config/tc-m68k.c (m68k_ip): The coldfire does not support 8x
1793 * symbols.c (S_SET_EXTERNAL): Change as_warn to as_bad.
1794 (S_CLEAR_EXTERNAL, S_SET_WEAK): Likewise.
1798 * Makefile.in (VERSION): Set to cygnus-2.7.1.
1800 * Released binutils 2.7.
1804 * config/tc-mips.c (mips_ip): Only perform range check when
1805 dealing with O_constant expressions.
1809 * m68k-parse.h (m68k_register): Add new coldfile control
1812 * config/tc-m68k.c (mcf5200_control_regs): New variable,
1813 array of control registers for the coldfire.
1814 (cpu_of_arch): Added mcf5200.
1815 (archs): Added mcf5200.
1816 (init_table): Add new control registers.
1817 (m68k_ip): Added support for new control registers.
1818 (m68k_init_after_args): Likewise.
1820 * config/tc-m68k.c (md_show_usage): Add -m5200 to usage text.
1824 * read.h (is_it_end_of_statement): Declare.
1825 * read.c (is_it_end_of_statement): Remove declaration.
1827 * config/tc-ppc.c (ppc_elf_suffix): Correct parenthesization of ||
1829 (md_assemble): Fix handling of @l with an unsigned constant. Add
1830 default case to reloc switch.
1832 * config/tc-i386.h (AOUT_MACHTYPE): Define as 0 if TE_386BSD.
1835 * ecoff.c (last_lineno): New static variable.
1836 (add_procedure): Set last_lineno.
1837 (ecoff_directive_loc): Likewise.
1838 (ecoff_generate_asm_lineno): Likewise.
1839 (ecoff_fix_loc): New function.
1840 * ecoff.h (ecoff_fix_loc): Declare.
1841 * config/tc-mips.c (append_insn): When inserting nops, and using
1842 ECOFF debugging, call ecoff_fix_loc.
1846 * config/tc-h8300.c (build_bytes): If an operand type is
1847 marked as SRC_IN_DST retrieve it from the "destination" op.
1851 * configure.in (arm-*-riscix*): Set emulation to riscix.
1852 * configure: Rebuild.
1853 * config/te-riscix.h: New file to define TE_RISCIX.
1855 * config/tc-sh.h (SUB_SEGMENT_ALIGN): Define.
1859 * Makefile.in (config.status): Just run config.status as other
1864 * mpw-config.in (TARGET_OS): Add definition to conf.
1868 * config/tc-mips.c (append_insn): Parenthesize
1869 cop_interlocks expressions.
1873 * listing.c (listing_print): Close the listing file if it is not
1874 stdout. Close the other files opened for the listing.
1876 * config/tc-sparc.h (md_cons_align): Define.
1877 (sparc_cons_align): Declare.
1878 (HANDLE_ALIGN): Define.
1879 (sparc_handle_align): Declare.
1880 * config/tc-sparc.c (sparc_cons_align): New function.
1881 (sparc_handle_align): New function.
1882 * read.c (cons_worker): Call md_cons_align if it is defined.
1884 * as.h (struct frag): Add fr_file and fr_line fields.
1885 * frags.c (frag_new): Set fr_file and fr_line.
1886 (frag_var): Likewise.
1887 (frag_variant): Likewise.
1889 * as.h (struct frag): Remove unused align_mask and align_offset
1892 * listing.c (calc_hex): Offset by fr_fix when examining fr_var.
1897 * configure.in (mips-*-osf*): New target.
1898 * configure: Rebuild.
1900 * config/tc-m68k.c: Add 68ec060 as a synonym for 68060.
1904 * config/tc-mips.c: Added cop_interlocks, to avoid NOP insertion
1905 between co-processor comparisons and branches for the VR4300.
1909 * Makefile.in (bindir, libdir, datadir, mandir, infodir, includedir,
1910 INSTALL_PROGRAM, INSTALL_DATA): Use autoconf-set values.
1912 * configure.in (AC_PREREQ): autoconf 2.5 or higher.
1913 * doc/Makefile.in (bindir, libdir, datadir, mandir, infodir,
1914 includedir): Use autoconf set values.
1919 * listing.c (listing_eject): Don't do anything if listing is 0.
1920 (listing_list): Likewise.
1921 (listing_source_line): Likewise.
1922 (listing_title): Don't save title if listing is 0.
1923 (listing_source_file): Check listing rather than listing_tail.
1925 * configure.in: On alpha*-*-osf*, link against libbfd.a if not
1926 using shared libraries.
1927 * configure: Rebuild.
1931 * config/tc-mips.c (mips_ip): In case 'i'/'j', don't require an
1932 absolute expression if a relocation type was specified.
1936 * configure.in: Add support for *-*-rtems* configurations.
1937 * configure: Rebuild.
1941 * configure.in: Add alpha-*-linuxecoff* target. Use elf for
1942 alpha-*-linux* target. Force bfd_gas for alpha-*. Require
1943 opcodes library for alpha.
1944 * configure: Rebuild with autoconf 2.10.
1945 * config/tc-alpha.c: Substantial rewrite to add ELF support and
1946 use new opcode table.
1947 * config/tc-alpha.h (md_undefined_symbol): Don't define.
1948 (LOCAL_LABEL): Define differently if OBJ_ELF.
1949 (FAKE_LABEL_NAME): Define if OBJ_ELF.
1950 * config/alpha-opcode.h: Remove.
1951 * config/obj-elf.h: If TC_ALPHA, define ECOFF_DEBUGGING.
1952 * Makefile.in (TARG_CPU_DEP_alpha): Depend upon
1953 include/opcode/alpha.h rather than config/alpha-opcode.h.
1957 * config/obj-aout.c (obj_emit_relocations): Give an error if the
1958 relocation symbol was not resolved.
1959 * config/obj-coff.c (do_relocs_for): Likewise.
1961 * write.c (adjust_reloc_syms): Refetch the symbol section after
1962 calling S_GET_VALUE, since it may have changed.
1964 * expr.c (struct expr_symbol_line): Define.
1965 (expr_symbol_lines): New static variable.
1966 (make_expr_symbol): Add entry to expr_symbol_lines.
1967 (expr_symbol_where): New function.
1968 * expr.h: Use extern on function declarations.
1969 (expr_symbol_where): Declare.
1970 * symbols.c (resolve_symbol_value): Try to use expr_symbol_where
1971 rather than printing the meaningless name of an expression
1976 * config/tc-i386.c (md_number_to_chars): Deleted.
1977 * config/tc-i386.h (md_number_to_chars): New macro.
1979 * config/tc-alpha.c (build_operate_n, build_mem): Moved earlier in
1981 (load_symbol_address, load_expression): Use build_mem.
1982 (build_operate): New function.
1983 (emit_addq_r): Use it.
1987 * symbols.c (colon): #if VMS, use S_SET_OTHER to store `const_flag'.
1991 * config/tc-vax.h (NOP_OPCODE): Define.
1995 * config/obj-vms.h (S_IS_COMMON): Define.
1996 (S_IS_LOCAL): Check for \002 as well as \001.
1997 (LONGWORD_ALIGNMENT): New macro.
1998 (SUB_SEGMENT_ALIGN): Use it.
2002 * config/vms-conf.h: Reconcile with conf.in.
2006 * write.c (is_dnrange): Only define if TC_GENERIC_RELAX_TABLE is
2009 * doc/as.texinfo: Document that any number of hex digits can
2012 * as.c (struct defsym_list): Define.
2013 (defsyms): New static variable.
2014 (parse_args): Just put --defsym arguments on defsyms list, rather
2016 (main): Define defsyms after output file is created.
2018 * config/tc-m68k.c (m68k_ip): Reject PRE and POST indexing mode on
2021 * config/tc-mips.c (mips_ip): In cases 'I', 'i', and 'j', set
2022 insn_error rather than calling check_absolute_expr.
2024 * as.c (emulation_name): Remove unused static variable.
2025 (default_emul_bfd_name): Add return NULL to avoid warning.
2026 * ecoff.c (ecoff_stab): Remove unused variables name and
2028 * frags.c (frag_new): Remove unused variable tmp.
2029 * hash.c (hash_grow): Parenthesize + within <<.
2030 (hash_print_statistics): Use %lu, not %d, to print unsigned
2032 * messages.c: Include "libiberty.h".
2033 (fprint_value): Add cast to avoid printf warning.
2034 (sprint_value): Likewise.
2035 * read.c: Include "ecoff.h".
2036 (emit_expr): Add casts to avoid printf warnings.
2037 * read.h: Use extern for function declarations.
2038 (pop_insert): Declare.
2039 * stabs.c: Include "ecoff.h".
2040 * subsegs.c (subseg_set_rest): Remove unused variables tmp,
2041 former_last_fragP, and new_fragP.
2042 * subsegs.h (subsegs_print_statistics): Declare.
2043 * symbols.c (debug_verify_symchain): Change macro to discard
2045 * write.c (dump_section_relocs): Likewise.
2046 * write.h: Use extern for function declarations.
2047 (write_print_statistics): Declare.
2048 * config/e-mipsecoff.c (mipsecoff_bfd_name): Return NULL to avoid
2050 * config/e-mipself.c (mipself_bfd_name): Likewise.
2051 * config/obj-elf.h (elf_ecoff_set_ext): Declare.
2053 * config/tc-sparc.h (TC_RELOC_RTSYM_LOC_FIXUP): If OBJ_ELF, always
2054 emit relocations against external symbols.
2056 * config/tc-alpha.c (tc_gen_reloc): Output a sensible error
2057 message if bfd_reloc_type_lookup fails, rather than calling
2060 * config/tc-alpha.c (alpha_force_relocation): Add
2061 BFD_RELOC_12_PCREL to switch.
2065 * config/tc-i386.h (LOCAL_LABEL,FAKE_LABEL_NAME): Use defaults for
2066 TE_PE (Lfoo, not .Lfoo).
2070 * read.c (s_fill): Don't warn about a zero repeat count.
2072 * config/tc-mips.c (mips_ip): Don't warn about using AT as a
2073 coprocessor register.
2075 * config/tc-i386.c (md_assemble): When checking the size of a
2076 register to set the size of an instruction, do a bitwise and with
2077 Reg8 and Reg16 rather than requiring the type to be exactly Reg8
2082 * config/tc-h8300.c (parse_reg): Tweak error messages.
2083 (build_bytes): Likewise.
2084 (skip_colonthing): Handle :32 suffix.
2085 (get_specific): Promote L_24 to L_32 if it makes a match.
2086 Don't always promote L_8 to L_16.
2087 (do_a_fix_imm): Clean up L_32 and L_24 handling.
2089 * config/tc-h8300.c (Smode): New variable.
2090 (h8300hmode): Turn off Hmode.
2091 (h8300smode): New function. Turn on Smode and Hmode.
2092 (md_pseudo_table): New ".h8300s" pseudo-op.
2093 (parse_reg): Handle "exr" register.
2094 (get_operand): Handle bizarre syntax for "stm.l" and "ldm.l".
2095 Handle "mach" and "machl" operands for ldmac.
2096 (get_specific): Handle "stm.l" and "ldm.l".
2097 (build_bytes): Handle "stm.l" and "ldm.l"; handle MACREG operands.
2098 * config/tc-h8300.h (COFF_MAGIC): Handle H8/S magic number.
2103 * doc/as.texinfo: Reorder chapter of machine dependent options so
2104 that it is sorted by chip name.
2106 * doc/as.texinfo: Use consistant spelling of Vax.
2107 * doc/c-vax.texi: Likewise.
2111 * config/tc-hppa.c (md_pseudo_table): Add ".begin_try" and ".end_try"
2113 (tc_gen_reloc, SOM version): Handle R_BEGIN_TRY and R_END_TRY.
2114 (md_apply_fix): Likewise.
2115 (pa_try): New function.
2116 (hppa_force_relocation): Force relocs for BEGIN_TRY and END_TRY.
2120 * config/tc-hppa.c (md_pseudo_table): Add ".level" pseudo op.
2121 (pa_level): New function.
2125 * listing.c (listing_newline): Don't do anything if listing is 0.
2129 * subsegs.c (section_symbol): If symbol_table_frozen is set, call
2130 symbol_create, not symbol_new.
2134 * write.c (adjust_reloc_syms): Don't set sy_used_in_reloc for an
2135 absolute symbol unless TC_FORCE_RELOCATION returns true.
2137 * config/obj-coff.c (previous_file_symbol): Remove BFD_ASSEMBLER
2139 (c_dot_file_symbol): BFD_ASSEMBLER version: Don't set the value of
2140 the symbol to a pointer. Don't set previous_file_symbol.
2141 Simplify symbol list rearrangement.
2142 (coff_frob_symbol): Don't do anything with C_FILE symbols.
2143 (coff_adjust_symtab): Don't check previous_file_symbol.
2147 * config/tc-ppc.c (ppc_elf_lcomm): New function for .lcomm
2149 (md_pseudo_table): Add ppc_elf_lcomm.
2153 * config/tc-m68k.c (m68k_ip): Accept ABSL for 'O', so that `bfextu
2154 d0{24:1},d0' works without an immediate prefix on the bit numbers.
2155 (md_begin): Add digits to alt_notend_table.
2156 (md_parse_option): Make s a const pointer.
2158 * config/tc-sparc.c (md_pseudo_table): Add "empty".
2159 (s_empty): New static function.
2161 * config/obj-coff.c (struct filename_list): Only define if not
2163 (filename_list_head, filename_list_tail): Likewise.
2164 (c_section_symbol): Remove unused BFD_ASSEMBLER version.
2165 (obj_coff_endef, BFD_ASSEMBLER version): Don't set the debugging
2166 flag for C_MOS, C_MOE, C_MOU, or C_EOS symbols, since they should
2167 have a section of N_ABS rather than N_DEBUG. If we do a merge,
2168 remove the new symbol from the list.
2169 (obj_coff_endef, both versions): Call tag_insert even if there is
2170 an old symbol with the same name, if the old symbol does not
2172 (coff_frob_symbol): Check SF_GET_TAG, C_EOF, and C_FILE outside of
2173 the SF_GET_DEBUG condition. Don't call SA_SET_SYM_ENDNDX with a
2174 symbol that will be moved to the end of the symbol list.
2175 (coff_adjust_section_syms): Always call section_symbol for .text,
2177 (coff_frob_section): Likewise. Also, remove unused variable
2180 * config/tc-ns32k.c (convert_iif): Call frag_grow rather than
2181 manipulating frags directly.
2182 (md_number_to_field): Adjust mem_ptr correctly if ENDIAN is
2185 * app.c (do_scrub_chars): If '/' is LINE_COMMENT_START, check
2186 whether the next character is '*' before checking whether we are
2187 at the start of a line. Permit LINE_COMMENT_START to start a
2188 comment in state 1 (seen some whitespace) as well, to match the
2191 * gasp.c (do_align): Permit a fill value for .align.
2195 * read.c (next_char_of_string): Warn if a newline is seen in the
2196 middle of a string. Call bump_line_counters when appropriate.
2200 * symbols.c (colon): Use LOCAL_LABEL.
2204 * Makefile.in (install): Don't check to see if tooldir exists.
2205 Make $(tooldir) and $(tooldir)/bin.
2209 * config/ppc-sol.mt (TDEFINES): Don't turn on -mregnames by
2214 * config/tc-mips.c (mips_ip): Don't call as_warn if we are setting
2215 insn_error. Don't put the string "ERROR" in insn_error. Set
2216 insn_error rather than calling as_warn for an unsupported opcode.
2220 * config/tc-mips.c (md_parse_option): Check for a 64 bit format
2221 before permitting -64.
2222 * output-file.c (output_file_create): Remove duplicate
2227 * config/tc-mips.c (md_begin): If -64, create a .MIPS.options
2228 section rather than a .reginfo section.
2229 (mips_elf_final_processing): If -64, write out 64 bit RegInfo
2232 * config/tc-mips.c (load_register): If mips_isa < 3, permit a 32
2233 bit value with the high bit set.
2237 * read.c (s_lcomm): Set section flags for .sbss section.
2239 * config/tc-mips.c (mips_64): New static variable.
2240 (mips_target_format): If mips_64, return elf64 targets rather than
2242 (md_longopts): Add "32" and "64".
2243 (md_parse_option): Handle -32 and -64.
2244 (md_show_usage): Mention -32 and -64.
2245 (cons_fix_new_mips): If mips_64, don't convert an 8 byte reloc to
2250 * config/tc-ppc.c (comment_chars): Make '!' a comment character
2251 for Solaris compatibility.
2253 * stabs.c (s_stab_generic): Under PowerPC Solaris, convert a
2254 .stabd with 4 arguments into a .stabn.
2258 * config/tc-mips.c (macro): When passing X_add_number to
2259 macro_build, cast it to int first.
2263 * config/tc-z8k.c (md_apply_fix): Handle fx_r_type of 0, as
2264 created by emit_expr.
2266 * symbols.c (symbol_create): If bfd_make_empty_symbol fails, call
2267 as_perror rather than assert.
2271 * config/tc-mips.c (mips_ip): Mark sections created to hold
2272 floating point information as read only.
2276 * config/tc-ppc.c (ppc_set_cpu): Change defaults to match AIX.
2280 * read.c (potable): Add .skip as a synonym for .space.
2282 * stabs.c (s_stab_generic): For PowerPC ELF, allow .stabd to take
2283 4 arguments, providing the 4th argument is 0, to allow
2284 compatibility with the Solaris assembler.
2288 * config/tc-sh.h (struct sh_segment_info_type): Define.
2289 (TC_SEGMENT_INFO_TYPE): Define.
2290 (sh_frob_label): Declare.
2291 (tc_frob_label): Define.
2292 (sh_flush_pending_output): Declare.
2293 (md_flush_pending_output): Define.
2294 * config/tc-sh.c (md_assemble): If relaxing, emit a R_SH_CODE
2295 reloc before the instruction if necessary.
2296 (sh_frob_label): New function.
2297 (sh_flush_pending_output): New function.
2298 (sh_coff_frob_file): Ignore ALIGN, CODE, DATA, and LABEL relocs
2299 when looking for the reloc for the target of .uses.
2300 (md_convert_frag): Fix printf format (%0xlx to 0x%lx).
2301 (sh_force_relocation): Force CODE, DATA, and LABEL relocs to be
2303 (md_apply_fix): Ignore CODE, DATA, and LABEL relocs.
2304 (sh_coff_reloc_mangle): Force CODE, DATA, and LABEL relocs to use
2305 the absolute symbol.
2307 * subsegs.h (segment_info_type): Add tc_segment_info_data field if
2308 TC_SEGMENT_INFO_TYPE is defined.
2312 * config/tc-i386.c (md_assemble): Make sure the opcode suffix
2313 matches the register size.
2317 * config/obj-coff.c (count_entries_in_chain): Ignore Fixups with
2319 (do_relocs_for): Likewise.
2320 (fixup_segment): Don't just quit if linkrelax is set. Try to
2321 apply non pc-relative sym1-sym2 fixups, even if linkrelax is
2326 * config/tc-ppc.c (ppc_elf_validate_fix): Allow GOT and section
2327 relative relocations with -mrelocatable. Also allow unfixed
2328 relocs in .ex_shared.
2332 * config/obj-coff.c (yank_symbols): Check that FNAME_OFFSET is
2333 non-zero before assuming this is a long file name.
2334 (w_strings): Likewise.
2335 (c_dot_file_symbol): Set FNAME_OFFSET to 1 for a long file name.
2337 * config/obj-coff.c (w_strings): Move declaration of i inside
2338 #ifdef block which uses it.
2342 * config/tc-h8300.c (do_a_fix_imm): Rename last argument to
2343 "relaxmode". Output relocs which identify various relaxing
2344 possibilities for mov.[bwl] instructions.
2345 (build_bytes): Pass in a relaxing mode to do_a_fix_imm.
2349 * config/tc-arm.h (TC_HANDLES_FX_DONE): Define.
2350 (MD_APPLY_FIX3): Define.
2351 * config/tc-arm.c (my_get_expression): Only watch for bad segments
2353 (md_apply_fix3): Renamed from md_apply_fix.
2354 If pcrel reloc and symbol is in different section, undo effects
2359 * config/tc-hppa.c (hppa_fix_adjustable): Don't adjust
2360 any reloc with an LR% or RR% field selector for SOM.
2364 * Makefile.in: Add subsegs.h to appropriate TARG_CPU_DEP_*
2369 * config/obj-coff.c (coff_frob_symbol): Don't merge a symbol with
2371 (yank_symbols): Likewise.
2375 * subsegs.h (segment_info_type): If MANY_SEGMENTS and not
2376 BFD_ASSEMBLER, add name field.
2377 * config/obj-coff.c: Include "libiberty.h".
2378 (coff_header_append): Handle long section names.
2379 (crawl_symbols): Just use the name field for the symbol name,
2380 without worrying about null byte termination.
2381 (w_strings): Handle long section names.
2382 (write_object_file): Likewise. Also, use the name field, rather
2384 (obj_coff_add_segment): Permit long section names.
2385 (obj_coff_init_stab_section): Use the name field, rather than
2387 (adjust_stab_section): Likewise.
2388 * config/te-pe.h (COFF_LONG_SECTION_NAMES): Define.
2390 * config/tc-i960.c (brtab_emit): Don't set fx_im_disp field.
2391 (mem_fmt): Likewise.
2392 (md_apply_fix): Don't check fx_im_disp field.
2396 * configure.in: Add * after sparc*-*-vxworks.
2397 * configure: Rebuild.
2399 * app.c (do_scrub_begin): If tc_comment_chars is not defined,
2400 define it to comment_chars. Use tc_comment_chars rather than
2402 (do_scrub_chars): Use tc_comment_chars rather than comment_chars.
2403 * config/tc-m68k.h (tc_comment_chars): Define.
2404 (m68k_comment_chars): Declare.
2405 * config/tc-m68k.c (m68k_comment_chars): Rename from
2406 comment_chars. Change into a pointer rather than an array.
2407 (md_longopts): Add "bitwise-or".
2408 (md_parse_option): Handle OPTION_BITWISE_OR.
2409 (md_show_usage): Mention --bitwise-or.
2410 * doc/c-m68k.texi: Document --bitwise-or.
2414 * config/tc-m68k.c (m68k_ip): Prevent attempts to use long offsets
2417 * config/obj-coff.c (obj_coff_section): BFD_ASSEMBLER version:
2418 call demand_empty_rest_of_line. Non BFD_ASSEMBLER version:
2419 correct handling of input line pointer, and call
2420 demand_empty_rest_of_line.
2424 * config/tc-sparc.c (in_bitfield_range): New static function.
2425 (sparc_ip): New cases X,Y. Use SPARC_OPCODE_ARCH_V9_P.
2426 (md_apply_fix, cases BFD_RELOC_32_PCREL_S2,
2427 BFD_RELOC_SPARC_{WDISP16,WDISP19}): Fix undefined code.
2428 (md_apply_fix): New cases BFD_RELOC_SPARC_[56].
2429 (tc_gen_reloc): New cases BFD_RELOC_SPARC_[56].
2433 * config/obj-coff.c: BFD_ASSEMBLER:
2434 (coff_last_bf): New static variable.
2435 (coff_frob_symbol): Set endndx of a .bf symbol.
2437 (obj_coff_endef): Call SF_SET_PROCESS on a .bf symbol.
2438 (last_bfP): New static variable.
2439 (yank_symbols): Set endndx of a .bf symbol.
2443 * config/tc-ppc.c (md_parse_option): Add support for Solaris's -le
2444 and -s options. Add -be for good measure.
2448 * read.c (s_space): Support non-constant fill value. Handle fill
2449 value correctly for a size other than 1.
2453 * config/tc-arm.c (my_get_float_expression): Update call to
2454 gen_to_words, X_PRECISION changed from 6 to 5.
2458 * config/tc-ppc.c (register_name,reg_name_search): Move register
2459 name lookup from PE specific code to all targets. Add support for
2460 -mregnames/-mno-regnames to control whether register names are
2462 (md_assemble): Call register_name for all platforms.
2463 (md_parse_option): Add support for -mregnames/-mno-regnames.
2465 * configure.in (powerpcle*-*-solaris): Add support.
2466 (powerpc*-*-linux): Ditto.
2467 * configure: Regenerate.
2469 * config/ppc-sol.mt: New config file for PowerPC Solaris.
2473 * config/tc-mips.c (mips_frob_file): Permit multiple %hi relocs to
2474 be associated with a single %lo reloc.
2476 * config/tc-mips.c (load_address): Cast X_add_number to valueT
2477 before comparing against MAX_GPREL_OFFSET, so that negative
2478 numbers are handled correctly.
2483 * config/tc-sparc.c (last_insn): New static variable.
2484 (md_assemble): Warn about putting floating point branches in a
2485 delay slot. If architecture is less than v9, insert NOP
2486 instructions between floating point instructions and floating
2487 point branches. (The SunOS assembler does both these operations.)
2488 Save the last instruction opcode.
2489 (sparc_ip): Add pinsn parameter. Change caller.
2491 * config/tc-m68k.c (md_estimate_size_before_relax): Correct check
2492 for byte jump to next instruction to skip empty frags.
2496 * config/tc-alpha.c (alpha_ip): If we are going to call emit_add64
2497 for addq with a 16 bit signed value, just emit a lda instruction
2502 * config/tc-h8300.c (do_a_fix_imm): Don't cut off high bits
2507 * configure.in: Permit --enable-shared to specify a list of
2509 * configure: Rebuild.
2513 * config/tc-h8300.c (get_specific): Remove some #if 0 code.
2514 (build_bytes): Remove all ABSMOV related code; it's unnecessary.
2518 * config/atof-ieee.c: Fix handling of denormalized extended
2519 precision numbers and overflow/underflow detection.
2520 (MAX_PRECISION, X_PRECISION, P_PRECISION): Changed from 6 to 5, to
2521 not include the 16 bit gap in the m68k extended precision format.
2525 * configure.in: Add i386-*-freebsdelf* target; from John Polstra
2527 * configure: Rebuild.
2531 * config/tc-mips.c: Allow non-zero offsets from .sdata symbols to
2532 be accessed using the $gp register.
2533 * config/tc-mips.h (MAX_GPREL_OFFSET): Added.
2537 * config/tc-sparc.c (sparc_md_end): Set bfd machine number to
2538 bfd_mach_sparc_sparclet if current_architecture is sparclet.
2542 * read.c (get_line_sb): Bump line counters based on
2543 input_line_pointer[-1], not *input_line_pointer. Don't bother to
2544 call LISTING_NEWLINE.
2545 (s_macro): Don't call demand_empty_rest_of_line.
2546 * app.c (do_scrub_chars): When handling C style comments, unget
2551 * read.h (enum linkonce_type): Define.
2552 (s_linkonce): Declare.
2553 * read.c (potable): Add "linkonce".
2554 (s_linkonce): New function.
2555 * subsegs.h (segment_info_type): Add linkonce field to
2556 MANY_SEGMENTS && ! BFD_ASSEMBLER section.
2557 * config/obj-coff.h (obj_handle_link_once): Define if TE_PE.
2558 (obj_coff_pe_handle_link_once): Declare if TE_PE.
2559 * config/obj-coff.c: If TE_PE and not BFD_ASSEMBLER, #include
2561 (obj_coff_pe_handle_link_once): New function, defined if TE_PE.
2562 (c_section_symbol): If TE_PE, set the x_comdat field in the aux
2563 entry based on the linkonce field in segment_info.
2564 * doc/as.texinfo: Document .linkonce.
2568 * doc/as.1: Changed to be recognized by catman -w on Solaris.
2572 * stabs.c (s_stab_generic): Call the listing functions before
2573 doing the rest of the processing, which may involve freeing the
2574 string. Pass string, not string + stroff, to OBJ_PROCESS_STAB in
2575 SEPARATE_STAB_SECTIONS case.
2577 * config/tc-hppa.c: Remove nested comment.
2578 (tc_gen_reloc): Move label done inside the ifdef in which it is
2580 (md_apply_fix): Pass pointers to correct types to libhppa.h
2581 functions. Always return a value.
2583 * config/tc-mips.h (tc_frob_file): Define.
2584 (mips_frob_file): Declare.
2585 * config/tc-mips.c (struct mips_hi_fixup): Define.
2586 (mips_hi_fixup_list): New static variable.
2587 (imm_unmatched_hi): New static variable.
2588 (md_assemble): Clear imm_reloc, imm_unmatched_hi, and
2589 offset_reloc. Pass imm_unmatched_hi to append_insn.
2590 (append_insn): Add unmatched_hi parameter. If it is set, add the
2591 new fixup to mips_hi_fixup_list. Change all callers.
2592 (mips_ip): Set imm_unmatched_hi when appropriate.
2593 (mips_frob_file): New function.
2597 * configure.in (sparc-*-solaris2*): Renamed from sparc*-*-solaris2*.
2598 * configure: Regenerated.
2602 * as.c (main): Call bfd_set_error_program_name.
2606 * as.h (strdup): Don't declare.
2607 * stabs.c: Include libiberty.h
2608 (get_stab_string_offset): Use xstrdup rather than strdup.
2609 (s_stab_generic): Likewise.
2610 * as.c (parse_args): Likewise.
2611 * read.c (s_mri_sect): Likewise.
2613 * gasp.c (change_base): Recognize \(...) construct documented to
2614 pass through enclosed characters literally through to the output.
2615 (process_assigns): Likewise. Also, be more careful to avoid
2616 looking past the end of the buffer.
2620 * config/tc-i386.c (md_parse_option): If OBJ_ELF, ignore -k for
2625 * doc/as.texinfo, doc/c-i960.texi: Fix typos.
2629 * config/alpha-opcode.h: Added cvtst instruction.
2633 * ecoff.c (ecoff_stab): Don't try to make a symbol out of the stab
2634 string. Extract the addend from the result of expression.
2638 * app.c (do_scrub_chars): If whitespace is seen in state 11, and
2639 LABELS_WITHOUT_COLONS is not defined, and we are not in m68k MRI
2640 mode, change the state to 3 rather than 1.
2644 * config/obj-coff.h (C_REGISTER_SECTION): Change from 20 to 50, to
2645 correspond to 11 March change.
2649 * config/tc-h8300.c (build_bytes, MEMIND case): Generate
2650 an R_MEM_INDIRECT reloc rather than R_RELBYTE.
2654 * configure: Rebuild with autoconf 2.8.
2658 * config/atof-ieee.c (gen_to_words): Improve handling of
2659 X_PRECISION numbers. Based on patches from Andreas Schwab
2664 * as.h (SEG_NORMAL, SEG_LIST): Bump segment limit from 10 to 40.
2666 * subsegs.c (MANY_SEGMENTS): Increase segment limit.
2667 * obj-coff.c (seg_N_TYPE, seg_info_off_by_4): Likewise.
2668 (do_relocs_for, w_symbols, obj_coff_add_segment, do_linenos_for,
2669 crawl_symbols, coff_header_append): Loop to SEG_LAST rather than
2674 * config/tc-sparc.c (sparc_ip): Handle operand char 'O' (neg reg).
2678 * config/tc-hppa.c (SUBSEG_MILLI): Define.
2679 (pa_def_subspaces): Add $MILLICODE$.
2680 (pa_spaces_begin): Set section flags for $MILLICODE$.
2684 * config/obj-elf.c (obj_elf_section): Only SEC_LOAD if the type is
2685 not SHT_NOBITS. Don't tamper with flags based on type if a
2686 special section was found (revert Feb 29 change).
2688 * config/tc-sh.c (sh_do_align): Only align using the nop pattern
2689 if aligning to a longword boundary or greater.
2693 * config/tc-sh.c (sh_do_align): Pass 1 not 2 to frag_align.
2697 * configure.in (i386-*-cygwin32): Don't use bfd_gas.
2698 * configure: Regenerated.
2702 * config/tc-hppa.c: Add default definitions for R_N0SEL and
2703 R_N1SEL since they're not defined for old versions of hpux.
2705 * config/tc-hppa.c (tc_gen_reloc): Fix typo in R_COMP2 code.
2706 Set "sym_ptr_ptr" and "addend" fields to dummy values for
2707 R_N0SEL and R_N1SEL.
2711 * flonum-konst.c: Add two more constants for 1e+-2048 and
2712 1e+-4096, and correct the other constants.
2714 * symbols.c (resolve_symbol_value): Handle O_logical_not.
2718 * config/obj-elf.c (obj_elf_section): Allow predefined section
2719 types to set the nobits type. Avoid a shadowed declaration.
2723 * config/tc-hppa.c (hppa_fix_adjustable): For SOM, don't
2724 reduce relocs using e_nlrsel field selectors.
2726 * write.c (fix_new_exp): Don't use #elif. Some compilers
2729 * config/tc-hppa.c (selector_table): Add "n", "nl", and "nlr" to
2731 (pa_chk_field_selector): Handle new field selectors for SOM.
2735 * configure.in (m68k-*-linux*aout*, m68k-*-linux*): New targets.
2736 * configure: Rebuild.
2737 * config/te-linux.h (LOCAL_LABELS_FB): Define.
2738 * config/tc-m68k.h (TARGET_FORMAT) [TE_LINUX]: Define to
2740 * config/tc-m68k.c (comment_chars): Don't include '#' if TE_LINUX
2745 * mpw-make.sed: Update to handle shared library support.
2749 * config/tc-sparc.c (sparc_ip): Print all architectures that support
2750 the insn on mismatch.
2754 * configure.in: Add support for a29-coff.
2755 * configure: Rebuild.
2759 * config/tc-sh.c (sh_coff_frob_file): Don't consider the address
2760 of the section when looking for the R_SH_USES fixup, because the
2761 frag addresses have not yet been adjusted.
2763 * gdbinit.in: Set a breakpoint on as_warn_where.
2765 * config/tc-mips.c (macro): Add missing arguments to macro_build
2770 * config/tc-h8300.c (tc_reloc_mangle): Change reloc based on size
2771 if it is TC_CONS_RELOC. Set a size of 4 to R_RELLONG.
2775 * config/tc-sparc.c (sparc_ip): Recognize %asr0 for v8.
2779 * config/tc-sparc.c (parse_keyword_arg): Accept leading '%'.
2780 (sparc_ip): Accept %asr[1..31] for v8 and %asr[%16..31] for v9.
2781 Recognize [uU] format args as sparclet cpregs.
2785 * config/tc-sh.c (sh_handle_align): Don't emit R_SH_ALIGN relocs
2790 * config/tc-sparc.h (TC_RELOC_RTSYM_LOC_FIXUP): Check S_IS_WEAK as
2791 well as S_IS_EXTERNAL.
2792 (tc_fix_adjustable): Likewise.
2793 * config/tc-sparc.c (md_apply_fix): In OBJ_ELF case, check for
2794 S_IS_WEAK as well as S_IS_EXTERNAL when deciding whether to return
2796 (tc_gen_reloc): Check S_IS_WEAK as wel as S_IS_EXTERNAL when
2797 deciding whether to convert BFD_RELOC_32_PCREL_S2 if PIC.
2801 * config/tc-sparc.c (max_architecture): Change to sparclite for
2803 (default_compatible): Delete.
2804 (sparc_ffs): New function.
2805 (md_begin): Only call SPARC_OPCODE_CONFLICT_P once.
2806 (sparc_ip): Rewrite architecture match and bump logic.
2810 * configure.in: Check for 'do not mix' from native linker before
2811 trying to use -rpath.
2812 * configure: Rebuild.
2816 * config/obj-coff.h (SF_ADJ_LNNOPTR): Define (non BFD_ASSEMBLER).
2817 (SF_GET_ADJ_LNNOPTR): Define (non BFD_ASSEMBLER).
2818 (SF_SET_ADJ_LNNOPTR): Define (non BFD_ASSEMBLER).
2819 * config/obj-coff.c (obj_coff_endef): Set ADJ_LNNOPTR when LNNOPTR
2821 (w_symbols): If ADJ_LNNOPTR is set, add the section lnnoptr field
2822 to the symbol lnnoptr field, to get the correct file offset.
2826 * config/obj-elf.c (elf_frob_symbol): On the PowerPC, force all
2827 symbols that are not function, file, or section symbols to be
2832 * configure: Set and substitute RPATH_ENVVAR.
2833 * configure: Rebuild.
2834 * Makefile.in (RPATH_ENVVAR): New variable.
2835 (check): Use $(RPATH_ENVVAR) rather than LD_LIBRARY_PATH.
2838 doesn't need opcodes. If configuring shared, opcodes needs bfd.
2839 * configure: Rebuild.
2843 * read.c (s_mri_sect): Don't return '\0' in type. Set all
2844 appropriate flags in BFD section.
2846 * configure.in (m68k-*-psos*): New target.
2847 * configure: Rebuild.
2848 * config/te-psos.h: New file.
2849 * config/tc-m68k.c (comment_chars): Don't include '#' if TE_PSOS
2855 * configure.in: Remove duplicate setting of cpu_type. Check
2856 whether opcodes library is required for on all targets, not just
2858 * configure: Rebuild.
2860 * config/tc-mips.c (mips_big_got): New static variable.
2861 (s_extern): Don't declare.
2862 (reg_needs_delay): New static function.
2863 (macro_build): Permit GOT/CALL_HI/LO relocs.
2864 (macro_build_lui): If place is not NULL, use the number in the
2866 (load_address): Handle mips_big_got case.
2867 (macro): Handle mips_big_got for M_LA_AB, M_JAL_A, and load and
2869 (OPTION_XGOT): Define.
2870 (md_longopts): Add "xgot" if OBJ_ELF.
2871 (md_parse_option): Handle -xgot.
2872 (md_show_usage): Mention -xgot.
2873 (md_apply_fix): Permit GOT/CALL_HI/LO relocs.
2874 (tc_gen_reloc): Handle GOT/CALL_HI/LO relocs.
2878 * config/tc-m68k.c (m68k_ip) [operand kind '#']: When fixing
2879 the byte relocation, point it to the low byte of the word.
2883 * configure.in: Set HDLFLAGS for *-*-hpux with --enable-shared.
2884 * configure: Rebuild.
2888 * configure.in: Recognize any sparc* cpu.
2889 * configure: Regenerated.
2893 * read.c (potable): Add "mri" and ".mri".
2894 (s_mri): New function.
2895 * read.h (s_mri): Declare.
2896 * app.c (scrub_m68k_mri): New static variable.
2897 (mri_pseudo): New static variable.
2898 (do_scrub_begin): Add m68k_mri parameter. Use it rather than
2899 flag_m68k_mri. Initialize scrub_m68k_mri.
2900 (mri_state, mri_last_ch): New static variables.
2901 (struct app_save): Add scrub_m68k_mri, mri_state, and mri_last_ch
2903 (app_push): Save new fields.
2904 (app_pop): Restore new fields.
2905 (do_scrub_chars): Check scrub_m68k_mri rather than flag_mri_mri.
2906 If TC_M68K, use a trivial state machine to look for occurrences of
2907 the .mri pseudo-op, and change the mode appropriately.
2908 * as.h (do_scrub_begin): Update prototype.
2909 * input-scrub.c (input_scrub_begin): Pass flag_m68k_mri to
2911 * config/tc-m68k.c (reg_prefix_optional_seen): New static
2913 (m68k_mri_mode_change): New function.
2914 (md_parse_option): Set reg_prefix_optional_seen.
2915 * config/tc-m68k.h (m68k_mri_mode_change): Declare.
2916 (MRI_MODE_CHANGE): Define.
2917 * doc/as.texinfo: Document .mri pseudo-op.
2919 * app.c (do_scrub_chars): In MRI mode, don't treat '#' as a
2924 Support for OBJ_ELF on m68k, mostly inside #ifdef OBJ_ELF:
2925 * config/m68k-parse.h (enum pic_relocation): Define.
2926 (struct m68k_exp): Add pic_reloc field.
2927 * config/tc-m68k.h (TC_RELOC_RTSYM_LOC_FIXUP): Define.
2928 (tc_fix_adjustable): Define to call tc_m68k_fix_adjustable.
2929 (NO_RELOC): Define to BFD_RELOC_NONE if BFD_ASSEMBLER, to zero
2931 * config/tc-m68k.c: Delete definition of NO_RELOC.
2932 (struct m68k_it): Add pic_reloc field.
2933 (add_fix): Copy over pic_reloc field.
2934 (md_pseudo_table): Interpret .align parameter as byte count.
2935 (mote_pseudo_table): Likewise.
2936 (tc_m68k_fix_adjustable): New function.
2937 (get_reloc_code): New function.
2938 (md_assemble): Use it as last argument to fix_new_exp.
2939 (md_apply_fix_2): For a relocation against a symbol don't put the
2940 addend into the data.
2941 (tc_gen_reloc): Different addend computation for OBJ_ELF.
2942 (m68k_ip): Don't relax an operand that requires pic relocation.
2943 (md_begin): Align .text, .data and .bss on 4 byte boundary by
2945 * write.c (fixup_segment): Don't add symbol value to addend if
2946 TC_M68K and OBJ_ELF.
2947 * config/m68k-parse.y (yylex): Handle @PLTPC, etc.
2948 (motorola_operand): Add rule for `(zapc, EXPR)'.
2952 * ecoff.c (ecoff_directive_weakext): Fixed so that whitespace
2953 *really* is permissible before the comma.
2957 * config/tc-sh.c (sh_do_align): Align to a 2 byte boundary before
2958 inserting nop instructions.
2962 * config/te-aux.h: Change include of aux.h to aux-coff.h.
2966 * config/tc-i960.c (tc_coff_symbol_emit_hook): Correct storage
2967 class setting for a CALLNAME symbol in COFF.
2969 * read.c (potable): Pass negative numbers for new .balign[wl] and
2970 .p2align[wl] pseudo-ops.
2971 (s_align_bytes): Treat a negative argument as specifying the fill
2973 (s_align_ptwo): Likewise.
2977 * read.c (potable): Add balignw, balignl, p2alignw, and p2alignl.
2978 (do_align): Take new len parameter. Change all callers. Pass it
2980 (s_align_bytes): Arg now indicates the length of the fill pattern.
2981 (s_align_ptwo): Likewise.
2982 * config/obj-coff.c (write_object_file): Pass length to
2984 * config/tc-i386.h (md_do_align): Take new len parameter.
2985 * config/tc-m88k.h (md_do_align): Likewise.
2986 * config/tc-m88k.c (m88k_do_align): Likewise.
2987 * config/tc-sh.h (md_do_align): Likewise.
2988 * config/tc-sh.c (sh_do_align): Likewise.
2989 * doc/as.texinfo: Document new pseudo-ops.
2991 * config/obj-coff.c (fixup_mdeps): Divide offset by fr_var, as is
2992 done in cvt_frag_to_fill.
2994 * config/tc-sh.h (sh_do_align): Declare.
2995 (md_do_align): Define.
2996 * config/tc-sh.c (sh_do_align): New function.
2998 * ecoff.c (ecoff_build_lineno): Don't try to store the address
2999 difference if the next address is before the current one.
3001 * config/tc-m68k.c (struct m68k_cpu): Add alias field.
3002 (archs): Initialize new field.
3003 (m68k_ip): Don't list alias names when listing CPUs which support
3006 * as.c (main): Call parse_args before read_begin.
3007 * app.c (do_scrub_chars): If flag_m68k_mri, don't put a dot in
3008 front of generated pseudo-ops.
3009 * read.c (potable): Ignore "name".
3010 (s_app_file): Permit a single quote after the string, since one
3011 may appear in m68k MRI mode.
3013 * configure.in: Check for --enable-shared. If linking against
3014 shared BFD and opcodes, fix library name on SunOS, and try to set
3016 * configure: Rebuild.
3020 * as.h (flag_m68k_mri): Declare.
3021 * as.c (parse_args): If TC_M68K, set flag_m68k_mri for -M.
3022 * Many files: For MRI syntax that is specific to the m68k MRI
3023 assembler, check flag_m68k_mri rather than flag_mri or
3024 MRI_MODE_NEEDS_PSEUDO_DOT.
3028 * config/tc-i960.c (ARCH_HX): Define.
3030 (targ_has_sfr): Handle ARCH_HX.
3031 (targ_has_iclass): Handle ARCH_HX.
3032 (tc_coff_fix2rtype): Add return 0 to avoid warning.
3033 (tc_headers_hook): If the architecture was specified explicitly,
3034 use it when setting the flags. Set the extern variable coff_flags
3035 rather than headers->filehdr.f_flags, since the latter is set
3036 unconditionally in obj-coff.c.
3037 (i960_handle_align): Remove unused variable fixp.
3039 Support for building bfd and opcodes as shared libraries, based on
3041 * configure.in: Set OPCODES and BFD to search directories.
3042 Substitute OPCODES_DEP and BFDDEP. On SunOS, set HLDFLAGS.
3043 * configure: Rebuild.
3044 * Makefile.in (LDFLAGS, HLDFLAGS): New variables.
3045 (LIBDEPS): New variable.
3046 (as.new0: Depend upon $(LIBDEPS) rather than $(LIBS). Use
3047 $(HLDFLAGS) in link.
3048 (check): Set LD_LIBRARY_PATH in the environment.
3052 * config/tc-ppc.h (ELF_TC_SPECIAL_SECTIONS): Make .sdata2, .sbss2,
3053 .PPC.EMB.sdata0, and .PPC.EMB.sbss0 sections all default to
3054 read-only, not read/write.
3058 * Makefile.in (INSTALL_XFORM): Remove -e.
3062 * write.c (write_relocs): Use S_IS_DEFINED and S_IS_COMMON rather
3063 than comparing S_GET_SEGMENT to undefined_section.
3064 (write_object_file): Skip symbols which were equated to an
3065 undefined or common symbol.
3066 * symbols.c (resolve_symbol_value): Use S_IS_DEFINED and
3067 S_IS_COMMON rather than comparing S_GET_SEGMENT to
3069 (S_GET_VALUE): Likewise. Avoid recursion problems if S_IS_DEFINED
3070 or S_IS_COMMON call S_GET_VALUE.
3071 * config/obj-aout.h (S_IS_COMMON): Define if not BFD_ASSEMBLER.
3072 * config/obj-aout.c (obj_emit_relocations): If a reloc is equated
3073 to an undefined or common symbol, convert the reloc to be against
3075 (obj_crawl_symbol_chain): Skip symbols which were equated to an
3076 undefined or common symbol.
3077 * config/obj-bout.h (S_IS_COMMON): Define if not BFD_ASSEMBLER.
3078 * config/obj-bout.c (obj_emit_relocations): If a reloc is equated
3079 to an undefined or common symbol, convert the reloc to be against
3081 (obj_crawl_symbol_chain): Skip symbols which were equated to an
3082 undefined or common symbol.
3083 * config/obj-coff.c (do_relocs_for): Use S_IS_DEFINED and
3084 S_IS_COMMON rather than comparing S_GET_SEGMENT to
3086 (yank_symbols): Skip symbols which were equated to an undefined or
3091 * config/obj-aout.h (S_IS_LOCAL): Check for \002 as well as \001.
3092 * config/obj-bout.h (S_IS_LOCAL): Likewise.
3094 * configure.in: Make sure we only add m68k-parse.o to
3095 ${extra_objects} once, no matter how many m68k targets have been
3097 * configure: Rebuild.
3101 * configure.in (i386-*-cygwin32, ppc-*-cygwin32): New.
3102 * configure: Rebuild.
3106 * config/tc-m68k.c (md_pseudo_table): Add "extend" and "ldouble".
3107 * doc/c-m68k.texi: Document .extend and .ldouble.
3109 * configure.in (m68*-apple-aux*): New target.
3110 * config/te-aux.h: New file.
3111 * config/obj-coff.c (compare_external_relocs): New static function
3113 (do_relocs_for): Sort relocs if TE_AUX.
3114 (fixup_segment): If TE_AUX, store common symbol value in segment.
3115 * config/tc-m68k.h (TARGET_FORMAT): Define if TE_AUX.
3119 * config/obj-coff.h (S_IS_LOCAL): Check for \002 as well as \001.
3121 * config/tc-mips.c (s_mips_globl): Set BSF_OBJECT if it is not
3123 (s_cpload): Set BSF_OBJECT for _gp_disp symbol.
3124 * read.c (s_lcomm): If S_SET_SIZE is defined, set the size of the
3126 * ecoff.c (add_procedure): Set the BSF_FUNCTION flag.
3127 (ecoff_build_symbols): If S_SET_SIZE is defined, set the size of
3128 an undefined symbol and the size of a function symbol.
3129 * config/obj-elf.c (elf_frob_symbol): If TC_MIPS, set BSF_OBJECT
3130 for all common symbols.
3134 * config/tc-i960.c (parse_memop): In MRI mode, don't use implicit
3137 * expr.c (operand): Accept 0x hex constants in MRI mode if not on
3142 * config/obj-elf.c (obj_elf_type): Set BSF_OBJECT flag for a type
3145 * ecoff.c (localsym_t): Add addend field.
3146 (add_ecoff_symbol): Add addend argument. Change all callers.
3147 (coff_sym_value): Make static.
3148 (coff_sym_addend): New static variable.
3149 (ecoff_directive_def): Initialize coff_sym_addend.
3150 (ecoff_directive_val): Accept symbol + constant.
3151 (ecoff_directive_endef): Pass coff_sym_addend to add_ecoff_symbol.
3152 (ecoff_build_symbols): Include the addend in the symbol value.
3156 * config/tc-ppc.c (md_assemble): Ignore overflow on
3157 BFD_RELOC_16_GOTOFF and BFD_RELOC_PPC_TOC16.
3161 * config/tc-ppc.c (md_apply_fix3): SDA21 relocations are now 4
3162 bytes in size, so offset appropriately in big endian mode when
3163 writing the bottom 2 bytes.
3167 * config/tc-sparc.c (default_compatible): New static local.
3168 (md_begin): Initialize it. Rewrite warn_on_bump handling.
3169 (sparc_ip): If no architecture or -bump specified, don't mark as
3170 mismatched those in default_compatible.
3175 * configure.in (i386-*-sco*elf*): Use fmt elf, targ sco5.
3176 * configure: Rebuild.
3177 * config/sco5.mt: New file; set TDEFINES to -DSCO_ELF.
3178 * config/tc-i386.c (sco_id): New function, if SCO_ELF.
3179 * config/tc-i386.h (tc_init_after_args): Define if SCO_ELF.
3180 (sco_id): Declare if SCO_ELF.
3184 * config/tc-sparc.c (initial_architecture,can_bump_v9_p): Deleted.
3185 ({max,warn_after}_architecture): New static locals.
3186 (md_begin): Replace NUMOPCODES with sparc_num_opcodes.
3187 If both architecture and -bump requested, set max_architecture to max.
3188 (sparc_md_end): Simplify.
3189 (sparc_ip): Replace references to can_bump_v9_p with max_architecture.
3190 Rewrite code to bump architecture and check for conflicts.
3191 (md_longopts): Recognize -xarch={v8plus,v8plusa} for compatibility
3192 with Solaris assembler.
3193 (md_parse_option): Likewise. Call sparc_opcode_lookup_arch.
3194 (md_show_usage): Update.
3198 * Makefile.in (RUNTEST): Fix reference to $${srcdir}.
3202 * config/tc-sparc.h (TARGET_FORMAT): Use #ifdef SPARC_ARCH64 instead of
3203 #ifdef sparcv9 when choosing value.
3206 (sparc_md_end): Declare.
3207 * config/tc-sparc.c (SPARC_V9): Renamed from sparcv9.
3208 (initial_architecture): New static local.
3209 (can_bump_v9_p): Likewise.
3210 (NO_V9): Delete all occurrences.
3211 (sparc_md_end): New function.
3212 (sparc_ip): New local v9_arg_p. Rework fp reg number test.
3213 Don't bump architecture to v9 unless can_bump_v9_p set.
3214 (md_parse_option): -A<arch> passed, set can_bump_v9_p accordingly.
3215 * configure.in (sparc64 target cpu): Don't set obj_format here.
3216 (SPARC_V9): Renamed from sparcv9.
3217 (sparc64-*-elf*): Define SPARC_ARCH64.
3218 * configure: Regenerated.
3219 * acconfig.h (SPARC_V9): Renamed from sparcv9.
3220 (SPARC_ARCH64): Add.
3221 * conf.in: Regenerated.
3222 * config/vmsconf.h: Update.
3226 * config/tc-mips.c (load_register): Optimise "dli" loads.
3227 (md_show_usage): add "-mcpu=vr4100" to help text.
3231 * symbols.c (resolve_symbol_value): If a symbol is equated to an
3232 undefined symbol, preserve the X_op of O_symbol.
3233 (S_GET_VALUE): Fix check to permit this case.
3234 * write.c (write_relocs): If a reloc is against an undefined
3235 symbol equated to another symbol, change the reloc to be against
3237 * config/obj-coff.c (do_relocs_for): Likewise.
3239 * config/tc-ppc.c (ppc_csect): An unnamed csect is storage class
3244 * config/obj-elf.c (elf/ppc.h): Include elf/ppc.h if target
3245 computer is PowerPC.
3247 * config/tc-ppc.c (md_apply_fix3): Add more embedded relocations.
3249 * config/tc-ppc.h (ELF_TC_SPECIAL_SECTIONS): Add sections
3250 mentioned in the eabi.
3254 * config/tc-ppc.c (ppc_reldata): Changed alignement on reldata_section
3255 * config/tc-ppc.c (ppc_pdata): Changed the alignment on pdata_section
3259 * config/tc-ppc.c (mapping): Add more relocation suffixes.
3263 * config/tc-ppc.c (ppc_elf_validate_fix): Allow .gcc_except_table
3264 as a section it is ok to have unadorned -mrelocatable pointers in.
3268 * config/tc-ppc.c (ppc_section*): Wrap these functions inside
3273 * config/obj-elf.c (obj_elf_section): Add hooks so machine
3274 dependent section attributes can be handled.
3276 * config/tc-ppc.h: (md_elf_section_{letter,type,word,flags}): New
3277 macros to add support for exclude section flag and ordered section
3280 * config/tc-ppc.c (ppc_elf_section_{letter,type,word,flags}): New
3281 functions to add support for exclude section flag and ordered
3286 * subsegs.c (section_symbol): Don't try to look up the section
3287 symbol in the hash table. It should be possible to have a symbol
3288 with the same name as a section, but no connection to it.
3290 * read.c (cons_worker): Only call mri_comment_end from flag_mri.
3293 * expr.c (operand): Skip whitespace after a close parenthesis.
3298 * config/tc-sh.c (md_apply_fix): Call as_bad_where instead of
3299 as_warn_where for relocation overflow.
3300 (parse_reg): Accept register name only if next character is
3305 * ecoff.c (ecoff_stab): Simplify. Correctly handle sym + offset
3306 addresses for static variables.
3310 * config/tc-ppc.c (mapping): Make @got give a real GOT relocation,
3311 and xgot give the old toc16 relocation.
3312 (md_apply_fix3): Support all GOT relocations.
3316 * config/tc-mips.c (load_address): Correctly handle a constant in
3321 * config/tc-sh.c (parse_reg): Recognize SH3 registers.
3322 (get_specific): Handle A_SSR, A_SPC and A_REG_B.
3323 (build_Mbytes): Handle REG_B.
3327 * ecoff.c (ecoff_build_aux): Use new bfd_big_endian macro.
3331 * mpw-make.sed: If linking, edit ALL_CFLAGS to CFLAGS.
3335 * config/obj-coff.c (write_object_file): Set the s_align field to
3336 the number of bytes, rather than to the power of 2.
3340 * Makefile.in (DISTCLEAN_HERE): New variable.
3341 (distclean): Use it.
3342 (maintainer-clean): Depend upon clean-here rather than clean,
3343 distclean, and clean-info. Run make maintainer-clean in doc.
3344 Remove files listed in DISTCLEAN_HERE.
3345 * doc/Makefile.in (maintainer-clean realclean): Split out from
3346 distclean. Depend upon clean-info and distclean.
3350 * mac-as.r: Fix copyright and version strings.
3351 (cfrg): Use PROG_NAME instead of literal name.
3355 * read.c (read_a_source_file): If tc_unrecognized_line is defined,
3357 * config/tc-a29k.h (tc_unrecognized_line): Define.
3358 * config/tc-a29k.c (a29k_unrecognized_line): New function.
3359 (md_operand): Handle a29k style local dollar labels.
3363 * config/obj-multi.h: If OBJ_MAYBE_ELF, define OBJ_SYMFIELD_TYPE.
3367 * read.c (s_fill): If md_flush_pending_output is defined, call
3372 * config/obj-coff.c (size_section, fill_section, fixup_mdeps):
3373 Treat rs_align_code like rs_align.
3377 * config/tc-arm.c (cp_address_required_here): Set pre_inc when
3378 converting an absolute address into a PC-relative one.
3382 * mpw-config.in: Don't always use te-generic.h for emulation.
3383 (powerpc-apple-macos): Use emulation te-macos.h.
3385 (mips-gm-magic): New configuration.
3387 * mpw-make.sed (install, install-only): Edit in Mac-specific
3392 * configure.in: Improve message about unsupported ELF targets.
3393 * configure: Rebuild.
3395 * config/tc-m88k.c (m88k_do_align): Correct check for whether fill
3396 pattern is zero. From Manfred Hollstein.
3400 * config/tc-ppc.c (ppc_pe_section): To get the alignment right for
3401 the various idata sections, we check the name on the .section pseudo.
3405 * config/obj-coff.c (fixup_segment): If TC_M88K is defined, do not
3406 add section's paddr to add_number; compatibility to native as and
3411 * configure.in: Treat m68k-sysv4 like m68k-elf, not m68k-sysv3.
3413 * hash.c (struct hash_entry): Moved here...
3414 * hash.h (struct hash_entry): ...from here.
3416 * config/obj-elf.c (elf_frob_symbol): Don't free and clear sy_obj
3417 if it's already known to be null.
3421 * config/tc-ppc.c (md_apply_fix3): Don't adjust the value for 32
3422 bit relocs converted to PC relative relocs. This turned out to
3423 add the offset from the beginning of .text twice.
3427 * stabs.c (s_stab_generic): In 's' case, free string from
3430 * config/obj-elf.h (ELF_TARGET_SYMBOL_FIELDS): Remove unused field
3432 * config/obj-multi.h (ELF_TARGET_SYMBOL_FIELDS) [OBJ_MAYBE_ELF]:
3435 * subsegs.h (segment_info_type): Make bitfields unsigned.
3437 * expr.h (struct expressionS): Make X_op and X_unsigned bitfields,
3438 and move them together. On most systems this will reduce the
3439 structure size by one word.
3440 (operatorT): Define O_max.
3441 * expr.c (expr_begin): Verify that X_op is wide enough to hold
3444 * read.c (pop_insert): Print error returned by hash table
3447 * as.c (dump_statistics): Split out from main; dump some hash
3448 table stats and target-specific stats.
3449 (start_time): No longer automatic to main.
3450 (main): Set file-level start_time and call dump_statistics at
3451 exit. Exit by calling xexit.
3452 (show_usage): Make --statistics description less specific.
3453 * subsegs.c (subsegs_print_statistics): New function.
3454 * write.c (write_print_statistics): New function.
3455 (n_fixups): New static variable.
3456 (fix_new_internal): Increment it.
3457 * read.c (read_print_statistics): New function.
3458 * read.h (read_print_statistics): Declare.
3459 * symbol.c (symbol_print_statistics): New function.
3460 * symbol.h (symbol_print_statistics): Declare.
3461 * hash.c (hash_print_statistics): New function.
3462 * hash.h (hash_print_statistics): Declare.
3463 * config/tc-i386.c (i386_print_statistics): New function.
3464 * config/tc-i386.h (i386_print_statistics): Declare.
3465 (tc_print_statistics): New macro.
3466 * messages.c (as_fatal, as_assert, as_abort): Use xexit, not
3469 * hash.c (DELETED): Rewrite to use a valid but unique address.
3470 (START_POWER): Reduce to 10.
3471 (enum stat_enum): New enumerator, replacing STAT_* index macros.
3472 Add new values for counting strcmp calls.
3473 (GROW_FACTOR): New macro.
3474 (hash_grow): Use GROW_FACTOR. Rewrite for quick returns instead
3476 (FULL_VALUE): New macro. Use 1/4 of table size instead of 1/2.
3477 (hash_new): Use FULL_VALUE.
3478 (struct hash_control): Definition moved here.
3479 (hash_code): Don't mask to low bits.
3480 (hash_ask): Mask returned hash code here. Check hash value before
3481 calling strcmp; count strcmp calls.
3482 * hash.h (struct hash_control): Declare, don't define, here.
3483 (HASH_STATLENGTH): Deleted.
3484 (struct hash_entry): Add field for hash code.
3485 (hash_say, hash_apply): Don't declare.
3487 * hash.c (destroy): Return void.
3488 (applicatee): Ditto.
3489 (main): Fix declarations.
3490 (hash_apply): Return void. Argument `function' returns void. Put
3491 inside "#ifdef TEST".
3492 (hash_say): Define only if TEST is defined.
3493 * hash.h (hash_apply, hash_say): Declarations deleted.
3497 * configure: Rebuild with autoconf 2.7.
3501 * aclocal.m4 (AC_PROG_CC): Remove local definition.
3502 * configure: Rebuild with autoconf 2.6.
3506 * config/tc-ppc.c (ppc_debug_name_section_size): Remove.
3507 (ppc_stabx): Don't increment ppc_debug_name_section_size.
3509 (ppc_frob_file): Remove.
3510 * config/tc-ppc.h (tc_frob_file): Don't define.
3511 (ppc_frob_file): Don't declare.
3515 * Makefile.in (TARG_CPU_DEP_alpha): Mention alpha-opcode.h.
3516 * config/alpha-opcode.h: Include one-operand variants of jmp and
3519 * config/te-delt88.h: Renamed from te-delta88.h, to avoid conflict
3520 with te-delta.h in 8.3 file systems.
3521 * configure.in: Adjusted.
3525 * config/tc-mips.c (s_err): Remove; just use the one in read.c.
3527 * config/m68k-parse.y (yylex): In MRI mode, '@' can start an octal
3529 * expr.c (operand): Handle MRI suffixes after unadorned 0.
3533 Version 2.6 released.
3534 * Makefile.in (VERSION): Updated to 2.6.
3536 * config/obj-coff.c (write_object_file): Change use of md_do_align
3537 to pass a pointer rather than a fill value, to match other uses.
3541 * config/tc-ns32k.h (TC_FIX_TYPE): Add missing semicolon.
3543 * as.c (main): Move md_end call to just after call to
3544 perform_an_assembly_pass. Delete cpu-specific code here.
3545 * config/tc-i960.h (md_end): New macro, calls brtab_emit.
3546 * config/tc-arm.c (md_end): Unused function deleted.
3547 * config/tc-ns32k.c (md_end): Ditto.
3549 * config/tc-i386.c (i386_align_code): New function, moved here
3550 from HANDLE_ALIGN macro.
3551 * config/tc-i386.h (HANDLE_ALIGN): Call it.
3555 * config/tc-i386.h (md_do_align): cast fill and 0x90 to char
3560 * config/tc-i386.h (md_do_align): Make ".align n,0x90" generate
3561 multi-byte nops to avoid changing gcc. The necessary gcc change
3562 might break old assemblers.
3566 * config/tc-i386.h (md_do_align, HANDLE_ALIGN): Add macros to
3567 generate optimal multi-byte nop instructions for ".align n"
3568 ".align n,0x90", and aligns requiring more than 15 bytes of
3569 padding still generate multiple 0x90's as before.
3573 * config/tc-m68k.c (s_mri_until): Call pop_mri_control.
3577 * configure.in (ppc-*-macos*, ppc-*-mpw*): New configurations.
3578 * configure: Update.
3579 * mpw-make.sed: Reorder commands to make sed happier.
3580 * config/te-macos.h: New file.
3581 * config/tc-ppc.h (TARGET_FORMAT): Set correctly for PowerMac.
3585 * config/tc-hppa.c (pa_ip): Fix off-by-2 bug in length check for
3586 conditional branches.
3587 (md_apply_fix): Likewise.
3591 * config/tc-a29k.c (md_apply_fix): Warn if an attempt is made to
3592 generate a reloc which the linker will not handle correctly. Fix
3593 overflow checking--R_IREL is 18 bits, not 17.
3597 * config/obj-coff.c (fixup_segment): Don't subtract md_pcrel_from
3598 from a PC relative reloc if TC_A29K.
3600 * config/tc-a29k.c (line_separator_chars): Restore '@'. Existing
3601 code depends upon it.
3603 * config/tc-a29k.c (md_operand): Handle $float, $double, and
3604 $extend. Based on code from Eric Freudenthal
3606 * config/tc-a29k.h (LEX_DOLLAR): Define.
3607 * read.c (LEX_DOLLAR): Define if not defined.
3608 (lex_type): Use LEX_DOLLAR.
3612 * configure.in (a29k-nyu-sym1): New target, just like other a29k
3617 * config/obj-coff.c (c_dot_file_symbol): Cast xmalloc return.
3621 * config/tc-ppc.c (md_apply_fix3): Added BFD_RELOC_RVA. Currently
3622 used only by "dlltool.c".
3626 * config/tc-alpha.c: Undefine inline if not __GNUC__.
3627 (md_pseudo_table): Don't define "extern".
3631 * config/tc-ppc.c (ppc_biei): Force symbol into text_section.
3633 * config/tc-ppc.c (md_show_usage): Put backslash at end of line.
3637 * macro.c (macro_expand_body): Don't warn about == with a
3638 nonexistent parameter, in case it is in a comment field.
3640 * as.c (main): On TC_A29K, call macro_init with macro_alternate
3642 * macro.c (get_any_string): Don't keep quotes if macro_strip_at is
3643 set, even if macro_alternate is set.
3644 (get_apost_token): If macro_strip_at, only skip kind if it is '@'.
3645 (sub_actual): If macro_strip_at, and kind is '@', don't look up
3646 the token unless it ended in '@'.
3647 * config/tc-a29k.c (line_separator_chars): Remove '@'.
3648 * doc/c-a29k.texi: Document macro usage on A29K.
3652 * config/tc-m68k.c (m68k_ip): Handle new 'W' place, meaning a
3654 (install_operand): Likewise.
3656 * config/obj-elf.c (ecoff_debug_pseudo_table): Add "extern".
3660 * configure.in (m88k-motorola-sysv*): New target.
3661 * configure: Rebuild.
3662 * config/te-delta88.h: New file.
3663 * config/obj-coff.c (write_object_file): Use md_do_align if it is
3665 * config/tc-m88k.h (SUB_SEGMENT_ALIGN): Define.
3666 (md_do_align): Define.
3667 * config/tc-m88k.c: Include "subsegs.h".
3668 (m88k_do_align): New function.
3670 * config/te-delta.h (STRIP_UNDERSCORE): Don't define.
3671 (COFF_NOLOAD_PROBLEM): Define.
3672 (LOCAL_LABELS_DOLLAR, LOCAL_LABELS_FB): Define.
3676 * config/tc-i386.c (md_assemble): For a jump instruction with
3677 non-constant target, require 7 available bytes in the current
3682 * config/obj-elf.h: Include bfd/elf-bfd.h rather than
3687 * configure.in (alpha-*-linux*): Use ecoff.
3688 * configure: Rebuild.
3689 * ecoff.c (ecoff_directive_extern): New function.
3690 (ecoff_directive_weakext): New function.
3691 (ecoff_build_symbols): Handle weak symbols.
3692 (ecoff_setup_ext): Likewise.
3693 (ecoff_frob_symbol): Warn about weak common symbols.
3694 * ecoff.h (ecoff_directive_extern): Declare.
3695 (ecoff_directive_weakext): Declare.
3696 * symbols.c (S_IS_WEAK): New function.
3697 * symbols.h (S_IS_WEAK): Declare.
3698 * config/obj-ecoff.c (obj_pseudo_table): Add "extern" and
3700 * config/tc-mips.c (mips_pseudo_table): Remove "extern".
3705 * config/tc-ppc.c (ppc_lglobl): Do the right thing.
3707 * config/tc-ppc.c (ppc_bb): Call SF_SET_PROCESS.
3708 (ppc_eb): Likewise. Set the storage class to C_BLOCK, not C_FCN.
3709 (ppc_frob_symbol): Don't change C_BLOCK symbols to C_HIDEXT.
3710 * config/obj-coff.c (coff_frob_symbol): Don't call
3711 SA_SET_SYM_ENDNDX with the current symbol; call it with the next
3712 one. If OBJ_XCOFF, try to figure out whether the symbol is going
3715 * config/tc-ppc.c (md_pseudo_table): Add "bc" and "ec".
3716 (ppc_stab_symbol): New static variable.
3717 (ppc_change_csect): Check that ppc_toc_csect is not NULL.
3718 (ppc_stabx): Set ppc_stab_symbol around call to symbol_make. Set
3719 sy_tc.real_name to the stab string.
3720 (ppc_bc, ppc_ec): New static functions.
3721 (ppc_canonicalize_symbol_name): If ppc_stab_symbol is set, don't
3723 (ppc_symbol_new_hook): If ppc_stab_symbol is set, don't look for a
3725 (ppc_frob_symbol): Set BSF_NOT_AT_END for symbols with csect aux
3728 * input-scrub.c (input_scrub_push): Reset sb_index.
3732 * listing.c (listing_newline): Don't create a frag in the absolute
3737 * config/tc-ppc.c (md_pseudo_table): Add "data" and "text".
3738 (ppc_csect): Move most of the code to ppc_change_csect, and call
3740 (ppc_change_csect): New static function, taken from ppc_csect.
3741 (ppc_section): New static function.
3742 (ppc_saw_abs): New static varable.
3743 (ppc_frob_symbol): Create aux entry for absolute symbols. Warn if
3744 a symbol has no csect.
3745 (ppc_adjust_symtab): New function.
3746 * config/tc-ppc.h (tc_adjust_symtab): Define if OBJ_XCOFF.
3747 (ppc_adjust_symtab): Declare if OBJ_XCOFF.
3749 * write.c (write_object_file): If tc_adjust_symtab is defined,
3750 call it just before the call to obj_adjust_symtab.
3752 * symbols.c (symbol_find_or_make): Change name to be const.
3753 * symbols.h (symbol_find_or_make): Update declaration.
3757 * doc/as.texinfo (Align): Mention SH.
3758 * doc/c-m68k.texi (M68K-Directives, .even): Describe behavior, not
3760 * doc/c-z8k.texi (Z8000 Directives, global): Fix minor typo.
3761 (Z8000 Directives, even): Don't give numeric align value, instead
3766 * tc-arm.c (do_ldst): Assemble ldr/str r0, [r1] as a pre-increment
3771 * Makefile.in (diststuff): Also make info.
3772 (maintainer-clean realclean): Also make clean-info.
3776 * config/tc-hppa.c (md_pseudo_table): Add new ".nsubspa" opcode.
3777 (pa_subspace): For ".nsubspa", always create a new subspace
3778 with the given attributes, even if one already exists with the
3783 * config/tc-ppc.h (TC_FORCE_RELOCATION_SECTION): Rename from
3784 TC_FORCE_RELOCATION, taking an additional section argument. If
3785 the section of the target symbol is not the same as the current
3786 section, always force the relocation to be used.
3787 (MD_PCREL_FROM_SECTION): New macro to call md_pcrel_from_section.
3789 * config/tc-ppc.c (md_pcrel_from_section): Rename from the
3790 md_pcrel_from function, taking an additional section argument.
3791 Invoke TC_FORCE_RELOCATION_SECTION instead of TC_FORCE_RELOCATION.
3793 * write.c (TC_FORCE_RELOCATION_SECTION): Define in terms of the
3794 older TC_FORCE_RELOCATION if not defined.
3795 (MD_PCREL_FROM_SECTION): If not defined, invoke md_pcrel_from.
3796 (fixup_segment): Use MD_PCREL_FROM_SECTION instead of
3797 md_pcrel_from, and TC_FORCE_RELOCATION_SECTION instead of
3798 TC_FORCE_RELOCATION.
3802 * input-scrub.c (as_where): Set name to null pointer if we don't
3804 * messages.c (identify): Only print filename if non-null.
3805 (as_show_where): Ditto, for line number too.
3806 (as_warn_internal, as_bad_internal): Ditto.
3808 * input-file.c (input_file_open): If the input file can't be
3809 opened, consider it an error.
3813 * config/tc-mips.c: Added mips_4100 control, and support for
3814 accepting the 4100 as a MIPS architecture variant (md_begin,
3815 macro_build, mips_ip, md_parse_option). Adding suitable
3816 command-line OPTIONs, and updating the help text (md_show_usage).
3820 * subsegs.c (subseg_begin): Only set absolute_frchain.fix_* when
3821 BFD_ASSEMBLER is defined.
3823 * Use one active frag and one obstack per frag chain:
3824 * frags.c (frags): Variable deleted.
3825 (frag_alloc): New function.
3826 (frag_grow, frag_more, frag_variant, frag_now_fix,
3827 frag_append_1_char): Refer to frchain_now->frch_obstack instead of
3829 (frag_new): Ditto. Verify that frch_last and frag_now match on
3830 entry and exit, and that old frag_now has non-zero type. Replace
3831 "know" uses with "assert". Use frag_alloc instead of mucking with
3833 * frags.h (frags): Declaration deleted.
3834 * subsegs.h (struct frchain): Add new field frch_frag_now.
3835 * subsegs.c (frchains, dummy_frag, absolute_frchain): New static
3837 (subsegs_begin): Initialize frchains obstack. Under gcc, don't
3838 give it any stricter alignment than frchainS structures need. Do
3839 not initialize frags obstack. Set frag_now to point to
3840 dummy_obstack. Initialize absolute_frchain.
3841 (subseg_set_rest): Save and restore frag_now in frch_frag_now
3842 field of frchainS. Don't create new frags on section switch, and
3843 use frag_alloc when creating a new frag chain. For absolute
3844 section, set frchain_now to absolute_frchain. Verify that
3845 frch_last and frag_now match on entry and exit. Initialize
3846 per-chain obstack, and under gcc, set required alignment to that
3847 needed by fragS structure.
3849 * write.c (chain_frchains_together_1): Verify fr_type is nonzero.
3851 * stabs.c (get_stab_string_offset): Only copy input string if a
3852 fresh copy is needed, not if the section already exists.
3853 (s_stab_generic): Cache stab section name to bypass lookups, since
3854 usually it will match. Could be made faster still by changing the
3855 memory allocation rules.
3856 (s_xstab): Cache section name to bypass repeated string
3859 * frags.c (frag_new): Deleted register declarations.
3861 * listing.c (frag_now): Don't declare.
3863 * as.c (chunksize): New variable.
3864 (debug_memory): New variable.
3865 (main): If debug_memory is set, reduce chunksize and
3867 * as.h (chunksize): Declare it.
3868 * read.c (read_begin): Use it.
3870 * config/tc-alpha.c (md_shortopts): Include 'g'.
3871 (md_parse_option): Ignore it.
3873 * Makefile.in (distclean): Remove Makefile and config.status from
3874 testsuite directory.
3875 (clean-here): Don't delete testsuite. Instead, delete only the
3876 files within it that would be generated by running tests.
3878 * config/tc-hppa.c (hppa_elf_mark_end_of_function): Call
3879 frag_now_fix instead of accessing obstack info directly.
3880 * config/tc-arm.c (s_ltorg): Ditto.
3881 (md_assemble): Ditto.
3883 * config/tc-i386.c (md_assemble): Call frag_grow instead of
3888 * stabs.c (aout_process_stab): Insert debug symbol into symbol
3889 chain after parsing value expression, if any, to avoid separating
3890 continued .stabs lines.
3894 * config/tc-mips.c (mips_elf_pseudo_table): Remove.
3895 (mips_pop_insert): Don't call pop_insert on mips_elf_pseudo_table.
3899 * config/tc-ppc.c (md_begin): Use new flags PPC_OPCODE_COMMON for
3900 -mcom support and PPC_OPCODE_ANY for -many.
3901 (md_parse_option): Ditto.
3903 (md_begin): For duplicate instructions, print all duplicates
3908 * config/tc-ppc.c (md_parse_option): Support for -mcom to turn on
3909 common mode operation.
3910 (md_show_usage): Add -mcom to usage message.
3914 * expr.c (op_rank): Add O_symbol_rva.
3915 * expr.h (operatorT): Add O_symbol_rva.
3916 * read.c (cons_worker): Set O_symbol_rva when necessary.
3917 * write.c (fix_new_exp): Understand O_symbol_rva.
3921 * config/tc-mips.c: Correct s_cons arguments. From Michael
3926 * config/tc-ppc.c (ppc_macro): Make count unsigned.
3927 (ppc_biei): Set segment to now_seg and value to coff_n_line_nos.
3928 (ppc_frob_symbol): Handle C_BINCL and C_EINCL symbols by setting
3930 * config/obj-coff.c (coff_n_line_nos): Rename from n_line_nos, and
3931 make non-static. Change all users.
3932 * config/obj-coff.h (coff_n_line_nos): Declare.
3938 * configure.in (AC_ARG_WITH(bfd-assembler)): Fix help message.
3940 * config/obj-elf.c (obj_elf_common): Convert specified byte
3941 alignment to power of two. Set size of local bss symbol.
3943 * config/tc-m68k.c (tc_gen_reloc): Fix typo in variable name.
3947 * sb.c, macro.c: Decide whether to include <string.h> or
3948 <strings.h> just as as.h does.
3952 * Makefile.in (site.exp): Fix setting of $srcdir.
3954 * config/tc-arm.c (md_atof): Fix little-endian output.
3955 * config/tc-arm.h (ARM_BI_ENDIAN): Move definition so defined for
3960 * doc/as.texinfo: Split out the NS32k family documentation,
3961 despite its being commented out for now.
3962 * doc/c-ns32k.texi: New file.
3964 * sb.c, macro.c: Include string.h.
3966 * Makefile.in (comparison): Only check *.o; we don't care if
3967 timestamps inserted by the native linker differ.
3969 * config/tc-alpha.c (alpha_align): Only fill with a no-op pattern
3970 if alignment stricter than 4 bytes is requested; in that case,
3971 align to a 4-byte boundary first.
3975 * config/obj-vms.c (VMS_RSYM_Parse): eliminate "might be used
3976 uninitialized" warning for `Max_Source_Offset'.
3980 * config/tc-ppc.c (parse_toc_entry): New function to parse [toc]
3981 qualifiers and detect errors if present.
3982 (md_assemble): Add call to parse_toc_entry. Also added some support
3983 for the [tocv] qualifier.
3984 (ppc_pe_tocd): New function to support data in the toc section.
3988 * config/tc-ppc.c (ppc_frob_symbol): Don't create an aux entry for
3993 * config/tc-m68k.c (isword): Accept all values from -65536 to
3994 +65535, so ~VAL will not be rejected.
3996 * cond.c (s_endif): Call demand_empty_rest_of_line. In MRI mode,
3997 skip characters after the pseudo-op.
3999 * read.c (get_line_sb): Don't look past buffer_limit.
4000 (s_include): In MRI mode, skip characters after the file name.
4004 * config/m68k-parse.y (m68k_reg_parse): In MRI mode, permit
4005 periods in register names.
4007 For older changes see ChangeLog.1.