* alpha-tdep.c (alpha_push_arguments): Fix typo (TYPE_VALUE ->
VALUE_TYPE). Do the cast for TYPE_CODE_BOOL, TYPE_CODE_CHAR,
TYPE_CODE_ENUM, and TYPE_CODE_RANGE as well as TYPE_CODE_INT.
Sat Apr 15 14:04:32 1995 Per Bothner <bothner@cygnus.com>
* alpha-tdep.c (alpha_push_arguments): Only cast to long for
TYPE_CODE_INT.
* config/tc-sh.c (md_convert_frag): Instead of aborting on large displacements,
print an error message. Don't invoke the code that followed the abort call
until it's been verified.
* test-build.mk: Enable building of shared libraries on IRIX 5 and
OSF/1. Fix compiler flags.
* build-all.mk: Support Linux and OSF/1 3.0. Fix compiler flags.
(md_assemble): Use hashed list of insns.
Print warning if 8 byte insn appears in delay slot.
(md_operand): Handle %st(sym1-sym2).
(get_arc_exp_reloc_type): Likewise.
(md_apply_fix, case BFD_RELOC_ARC_B26): Now using implicit addends
so must store them here.
some AIX versions.
* configure: Regenerated.
* configure.in: Add m68k-*-elf.
* config/tc-m68k.c (comment_chars) [OBJ_ELF]: Include '#'.
(md_pseudo_table) [OBJ_ELF]: Ignore "swbeg".
(md_begin): Set alt_notend_table['&'], so svr4-style immediate operands are
accepted.
(md_apply_fix): Argument VALP should point to valueT.
(md_convert_frag): Argument SEC should be type segT.
(md_shortopts) [OBJ_ELF]: Accept 'Q' with an argument.
(md_parse_option): Ignore it.
(md_convert_frag_1): Add in frag address for the symbol in the displacement
calculation.
* config/tc-m68k.h (TARGET_FORMAT) [OBJ_ELF]: Use "elf32-m68k".
(TARGET_ARCH): Define.
(REGISTER_PREFIX_OPTIONAL) [OBJ_ELF]: Default to 0.
(LOCAL_LABEL, FAKE_LABEL_NAME, REGISTER_PREFIX_OPTIONAL): Handle these the same
way for OBJ_ELF as for M68KCOFF.
consistency with error_begin. Also print warning_pre_print.
Document it better.
* utils.c (warning): Use it.
* utils.c (error_begin): Doc fix.
* rs6000-nat.c (vmap_ldinfo): If symfile_objfile is not part of
any vmap, nuke it.
Do generic rs_machine_dependent relaxation only if TC_GENERIC_RELAX_TABLE is
defined. Use its value as the base of the table, so the actual declaration
can be cpu-specific (and const or not as appropriate). For other cpus where
this isn't done (the majority), delete md_relax_table altogether.
For machines with WORKING_DOT_WORD, delete md_{short,long}_jump_size.
(md_relax_table): No longer const. Add PCINDEX entries.
(m68k_ip): For AINDX with simple symbol operand, generate a PCINDEX frag if PC
is used, or do normal non-AINDX processing for address register.
(m68k_init_after_args): If cpu is 68000 or 68010, fix md_relax_table to prevent
relaxation of PCINDEX/BYTE mode to SHORT, since they don't support that mode.
(md_convert_frag_1, case PCLEA/LONG): Add 4 to offset, not 2. Add support for
new PCINDEX modes.
(md_estimate_size_before_relax): Process PCINDEX/SZ_UNDEF mode.
(md_convert_frag_1, case PCLEA/SHORT): Add 2 to offset.
(m68k_ip, case most punctuation/AOFF): If using PC, call add_frag using PCLEA.
(nopic_need_relax): New static function, split out from
md_estimate_size_before_relax.
(md_estimate_size_before_relax): Call it.
(load_address, macro): In NO_PIC branches, if nopic_need_relax returns nonzero,
don't attempt GP optimization.
comments.
(struct m68k_cpu): New type.
(archs, n_archs): New variables, with single list of name/enum mapping and
aliases.
(m68k_ip): Delete the table here.
(m68k_init_after_args): Use the new table here instead of open-coding it.
(md_parse_option, case 'm'): Ditto.
