1 Sun Sep 20 19:01:51 1998 Richard Henderson <rth@cygnus.com>
3 * alpha.c (alpha_sr_alias_set): New variable.
4 (override_options): Set it.
5 (alpha_expand_prologue, alpha_expand_epilogue): Use it.
6 (mode_mask_operand): Fix signed-unsigned comparision warning.
7 (alpha_expand_block_move): Likewise.
8 (print_operand): Likewise.
9 (get_aligned_mem): Use change_address.
10 (alpha_set_memflags, alpha_set_memflags_1): Set the alias set.
11 (alphaev4_insn_pipe, alphaev4_next_group): New functions.
12 (alphaev4_next_nop, alphaev5_next_nop): New functions.
13 (alpha_align_insns): Remade from old alphaev5_align_insns
14 to handle multiple processors.
15 (alpha_reorg): Call alpha_align_insns for both ev4 and ev5.
16 * output.h (label_to_alignment): Prototype.
18 * tree.c (new_alias_set): New function.
19 * tree.h (new_alias_set): Declare it.
20 * c-common.c (c_get_alias_set): Use it.
22 Sun Sep 20 12:35:55 1998 Richard Henderson <rth@cygnus.com>
24 * fold-const.c (fold): Yet another COND_EXPR bug: when folding
25 to an ABS expr, convert an unsigned input to signed.
27 Sun Sep 20 12:14:45 1998 Jeffrey A Law (law@cygnus.com)
29 * fold-const.c (fold): Fix another type in COND_EXPR handling code.
31 1998-09-20 Michael Hayes <m.hayes@elec.canterbury.ac.nz>
33 * configure.in: Add support for c4x targets.
36 Sun Sep 20 00:00:51 1998 Richard Henderson <rth@cygnus.com>
38 * combine.c (distribute_notes): If an insn is a cc0 user, only
39 delete it if we can also delete the cc0 setter.
41 Sun Sep 20 00:22:23 1998 Michael Tiemann <michael@impact.tiemann.org>
43 * fold-const.c (fold): Fix typo in COND_EXPR handling code.
44 (invert_truthvalue): Enable truthvalue inversion for
45 floating-point operands if -ffast-math.
47 Sat Sep 19 23:58:07 1998 Melissa O'Neill <oneill@cs.sfu.ca>
49 * configure.in: Disable collect2 for nextstep. Instead use
52 * config/nextstep.h (STARTFILE_SPEC): Add crtbegin.
53 (ENDFILE_SPEC): Define.
54 (OBJECT_FORMAT_MACHO): Define.
55 (EH_FRAME_SECTION_ASM_OP): Define.
56 * crtstuff.c: Handle MACHO.
58 Sun Sep 20 00:24:24 1998 Robert Lipe <robertl@dgii.com>
60 * config/i386/sco5.h (TARGET_MEM_FUNCTIONS): Define.
62 1998-09-19 Torbjorn Granlund <tege@matematik.su.se>
64 * fp-bit.c (pack_d): Do not clear SIGN when fraction is 0.
65 (_fpadd_parts): Get sign right for 0.
67 1998-09-19 Michael Hayes <m.hayes@elec.canterbury.ac.nz>
69 * ginclude/varargs.h: Add support for C4x target.
70 * ginclude/stdargs.h: Likewise.
72 Sat Sep 19 12:05:09 1998 Richard Henderson <rth@cygnus.com>
74 * alpha.c (alpha_return_addr): SET should be VOIDmode.
75 (alpha_emit_set_long_const): Rewrite to be callable from reload
77 (alpha_expand_epilogue): Update for alpha_emit_set_long_const.
78 * alpha.md (movdi): Likewise.
80 Sat Sep 19 07:33:36 1998 Richard Earnshaw (rearnsha@arm.com)
82 * arm.c (add_constant): New parameter address_only, change caller.
83 Set it non-zero if taking the address of an item in the pool.
84 (arm_reorg): Handle cases where we need the address of an item in
87 * arm.c (bad_signed_byte_operand): Check both arms of a sum in
89 * arm.md (splits for *extendqihi_insn and *extendqisi_insn): Handle
90 memory addresses that are not in standard canonical form.
92 Sat Sep 19 01:00:32 1998 Michael Hayes (mph@elec.canterbury.ac.nz)
94 * README.C4X: New file with information about the c4x ports.
95 * ginclude/va-c4x.h: New file for c4x varargs support.
96 * config/c4x: New directory with c4x port files.
98 Fri Sep 18 22:52:05 1998 Jeffrey A Law (law@cygnus.com)
100 * reload.c (find_reloads): Do not replace a pseudo with
101 (MEM (reg_equiv_addr)) in the initializing insn for the
104 Fri Sep 18 23:50:56 1998 David Edelsohn <edelsohn@mhpcc.edu>
106 * toplev.c (rest_of_compilation): Set bct_p on second call to
108 * loop.c (loop_optimize, scan_loop, strength_reduce): New argument
110 (strength_reduce): Only call analyze_loop_iterations and
111 insert_bct if bct_p set.
112 (check_dbra_loop): Fix typo.
113 (insert_bct): Use word_mode instead of SImode.
114 (instrument_loop_bct): Likewise. Do not delete iteration count
115 condition code generation insn. Initialize iteration count before
117 * rtl.h (loop_optimize): Update prototype.
119 * ginclude/va-ppc.h (va_arg): longlong types in overflow area are
120 not doubleword aligned.
122 * rs6000.c (optimization_options): New function.
123 (secondary_reload_class): Only call true_regnum for PSEUDO_REGs.
124 * rs6000.h (OPTIMIZATION_OPTIONS): Define.
125 (REG_ALLOC_ORDER): Allocate highest numbered condition regsiters
126 first; cr1 can be used for FP record condition insns.
128 Fri Sep 18 09:44:55 1998 Nick Clifton <nickc@cygnus.com>
130 * config/m32r/m32r.h (m32r_block_immediate_operand): Add to
133 * config/m32r/m32r.md: Add "movstrsi" and "movstrsi_internal"
136 * config/m32r/m32r.c (m32r_print_operand): Add 's' and 'p'
138 (block_move_call): New function: Call a library routine to copy a
140 (m32r_expand_block_move): New function: Expand a "movstrsi"
141 pattern into a sequence of insns.
142 (m32r_output_block_move): New function: Expand a
143 "movstrsi_internal" pattern into a sequence of assembler opcodes.
144 (m32r_block_immediate_operand): New function: Return true if the
145 RTL is an integer constant, less than or equal to MAX_MOVE_BYTES.
147 Thu Sep 17 16:42:16 EDT 1998 Andrew MacLeod <amacleod@cygnus.com>
149 * except.c (start_catch_handler): Issue 'fatal' instead of 'error' and
151 * libgcc2.c (__eh_rtime_match): fprintf a runtime error. Use <stdio.h>.
153 Thu Sep 17 12:24:33 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
155 * regmove.c (copy_src_to_dest): Check that modes match.
157 Wed Sep 16 22:10:42 1998 Robert Lipe <robertl@dgii.com>
159 * config/i386/sco5.h (SUPPORTS_WEAK): True only if targeting ELF.
161 Wed Sep 16 15:24:54 1998 Richard Henderson <rth@cygnus.com>
163 * i386.h (PREFERRED_RELOAD_CLASS): Respect an existing class
164 narrower than FLOAT_REGS.
166 Wed Sep 16 17:51:00 1998 Alexandre Oliva <oliva@dcc.unicamp.br>
168 * cpplib.c: removed OLD_GPLUSPLUS_INCLUDE_DIR
170 * Makefile.in (old_gxx_include_dir): removed
172 Wed Sep 16 12:29:22 1998 Nick Clifton <nickc@cygnus.com>
174 * config/sh/sh.h: Update definition of HANDLE_PRAGMA to match
177 * config/sh/sh.c (handle_pragma): Rename to sh_handle_pragma().
178 (sh_handle_pragma): Change function arguments to match new
179 specification for HANDLE_PRAGMA.
181 Wed Sep 16 12:43:19 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
183 * gen-protos.c (parse_fn_proto): Cast argument of ISALNUM to
185 (main): Mark parameter `argc' with ATTRIBUTE_UNUSED.
186 When generating output, initialize missing struct member to zero.
188 Wed Sep 16 14:47:43 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
190 * regmove.c (copy_src_to_dest): Don't copy if that requires
193 Wed Sep 16 01:29:12 1998 Bernd Schmidt <crux@pool.informatik.rwth-aachen.de>
195 * global.c (reg_allocno): Now static.
196 * reload1.c (reg_allocno): Delete declaration.
197 (order_regs_for_reload): Take no arguments. Don't treat regs
198 allocated by global differently than those allocated by local-alloc.
200 Wed Sep 16 01:09:01 1998 Kamil Iskra <iskra@student.uci.agh.edu.pl>
202 * m68k/m68k.c (output_function_prologue): Reverse NO_ADDSUB_Q
203 condition, fix format strings.
204 (output_function_epilogue): Likewise.
206 * m68k/m68k.c: Don't include <stdlib.h> directly.
208 Wed Sep 16 00:30:56 1998 Geoff Keating <geoffk@ozemail.com.au>
210 * gcse.c: New definition NEVER_SET for reg_first_set, reg_last_set,
211 mem_first_set, mem_last_set; because 0 can be a CUID.
212 (oprs_unchanged_p): Use new definition.
213 (record_last_reg_set_info): Likewise.
214 (record_last_mem_set_info): Likewise.
215 (compute_hash_table): Likewise.
217 Tue Sep 15 22:59:52 1998 Jeffrey A Law (law@cygnus.com)
219 * rs6000.c (output_epilogue): Handle Chill.
221 * mn10200.h (ASM_OUTPUT_DWARF2_ADDR_CONST): Define.
222 * mn10300.h (ASM_OUTPUT_DWARF2_ADDR_CONST): Define.
224 * combine.c (make_extraction): If no mode is specified for
225 an operand of insv, extv, or extzv, default it to word_mode.
226 (simplify_comparison): Similarly.
227 * expmed.c (store_bit_field): Similarly.
228 (extract_bit_field): Similarly.
229 * function.c (fixup_var_regs_1): Similarly.
230 * recog.c (validate_replace_rtx_1): Similarly.
231 * mips.md (extv, extzv, insv expanders): Default modes for most
232 operands. Handle TARGET_64BIT.
233 (movdi_uld, movdi_usd): New patterns.
235 * pa.c (emit_move_sequence): Do not replace a pseudo with its
236 equivalent memory location unless we have been provided a scratch
237 register. Similarly do not call find_replacement unless a
238 scratch register has been provided.
240 Tue Sep 15 19:23:01 1998 Bernd Schmidt <crux@pool.informatik.rwth-aachen.de>
242 * i386.h (PREFERRED_RELOAD_CLASS): For standard 387 constants,
245 Tue Sep 15 19:09:06 1998 Richard Henderson <rth@cygnus.com>
247 * tree.h (BUILT_IN_CALLER_RETURN_ADDRESS): Unused. Kill.
248 (BUILT_IN_FP, BUILT_IN_SP, BUILT_IN_SET_RETURN_ADDR_REG): Kill.
249 (BUILT_IN_EH_STUB_OLD, BUILT_IN_EH_STUB, BUILT_IN_SET_EH_REGS): Kill.
250 (BUILT_IN_EH_RETURN, BUILT_IN_DWARF_CFA): New.
251 * c-decl.c (init_decl_processing): Update accordingly.
252 * expr.c (expand_builtin): Likewise.
254 * rtl.h (global_rtl): Add cfa entry.
255 (virtual_cfa_rtx, VIRTUAL_CFA_REGNUM): New.
256 (LAST_VIRTUAL_REGISTER): Update.
257 * emit-rtl.c (global_rtl): Add cfa entry.
258 (init_emit): Initialize it.
259 * function.c (cfa_offset): New.
260 (instantiate_virtual_regs): Initialize it.
261 (instantiate_virtual_regs_1): Instantiate virtual_cfa_rtx.
262 (expand_function_end): Call expand_eh_return.
263 * tm.texi (ARG_POINTER_CFA_OFFSET): New.
265 * except.c (current_function_eh_stub_label): Kill.
266 (current_function_eh_old_stub_label): Likwise; update all references.
267 (expand_builtin_set_return_addr_reg): Kill.
268 (expand_builtin_eh_stub_old, expand_builtin_eh_stub): Kill.
269 (expand_builtin_set_eh_regs): Kill.
270 (eh_regs): Produce a third reg for the actual handler address.
271 (eh_return_context, eh_return_stack_adjust): New.
272 (eh_return_handler, eh_return_stub_label): New.
273 (init_eh_for_function): Initialize them.
274 (expand_builtin_eh_return, expand_eh_return): New.
275 * except.h: Update prototypes.
276 * flow.c (find_basic_blocks_1): Update references to the stub label.
277 * function.h (struct function): Kill stub label elements.
279 * libgcc2.c (in_reg_window): For REG_SAVED_REG, check that the
280 register number is one that would be in the previous window.
281 Provide a dummy definition for non-windowed targets.
282 (get_reg_addr): New function.
283 (get_reg, put_reg, copy_reg): Use it.
284 (__throw): Rely on in_reg_window, not INCOMING_REGNO. Kill stub
285 generating code and use __builtin_eh_return. Use __builtin_dwarf_cfa.
287 * alpha.c (alpha_eh_epilogue_sp_ofs): New.
288 (alpha_init_expanders): Initialize it.
289 (alpha_expand_epilogue): Use it.
290 * alpha.h: Declare it.
291 * alpha.md (eh_epilogue): New.
293 * m68h.h (ARG_POINTER_CFA_OFFSET): New.
294 * sparc.h (ARG_POINTER_CFA_OFFSET): New.
296 Tue Sep 15 19:31:58 1998 Michael Meissner <meissner@cygnus.com>
298 * i960.h (CONST_COSTS): Fix thinko. Test flag, not the constant
301 Tue Sep 15 14:10:54 EDT 1998 Andrew MacLeod <amacleod@cygnus.com>
303 * except.h (struct eh_entry): Add false_label field.
304 (end_catch_handler): Add prototype.
305 * except.c (push_eh_entry): Set false_label field to NULL_RTX.
306 (start_catch_handler): When using old style exceptions, issue
307 runtime typematch code before continuing with the handler.
308 (end_catch_handler): New function, generates label after handler
309 if needed by older style exceptions.
310 (expand_start_all_catch): No need to check for new style exceptions.
311 (output_exception_table_entry): Only output the first handler label
312 for old style exceptions.
313 * libgcc2.c (__eh_rtime_match): New routine to lump runtime matching
314 mechanism into one function, if a runtime matcher is provided.
316 Tue Sep 15 13:53:59 EDT 1998 Andrew MacLeod <amacleod@cygnus.com>
318 * config/i960/i960.h (SLOW_BYTE_ACCESS): Change definition to 1.
320 Tue Sep 15 09:59:01 1998 Mark Mitchell <mark@markmitchell.com>
322 * integrate.c (copy_decl_list): Fix typo.
324 Tue Sep 15 04:18:52 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
326 * config/sparc/sparc.md (movdf_const_intreg_sp32): Fix length
329 Mon Sep 14 14:02:53 PDT 1998 Jeff Law (law@cygnus.com)
331 * version.c: Bump for snapshot.
333 Mon Sep 14 10:33:56 PDT 1998 Jeff Law (law@cygnus.com)
335 * version.c: Bump for snapshot.
337 Mon Sep 14 09:51:05 PDT 1998 Jeff Law (law@cygnus.com)
339 * version.c: Bump for snapshot.
341 Sun Sep 13 22:10:18 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
343 * invoke.texi (C Dialect Options): Put back missing @end itemize.
345 Mon Sep 14 02:33:46 1998 Alexandre Oliva <oliva@dcc.unicamp.br>
347 * configure.in: remove usage of `!' to negate the result of a
348 command; some common shells do not support it
350 Sun Sep 13 19:17:35 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
352 * configure.in: in sparc9-sol2 config, use 'if test' not
354 * configure: Rebuilt.
356 * config/sparc/sol2-sld-64.h (SPARC_DEFAULT_CMODEL): Change to
358 (CPP_CPU_SPEC): Do not define _LP64, header files do this.
359 (CPP_CPU_DEFAULT_SPEC): Likewise.
360 * config/sparc/sol2.h (INIT_SUBTARGET_OPTABS): Get the names right
363 * config/sparc/sparc.md (goto_handler_and_restore): Allow any mode
366 Sun Sep 13 09:11:59 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
368 * acconfig.h (NEED_DECLARATION_STRSIGNAL): Provide a stub.
370 * collect2.c: Don't declare `sys_siglist' here.
371 (my_strsignal): Prototype and define new function. Use it in
372 place of `sys_siglist' hacks.
374 * mips_tfile.c: Likewise.
376 * configure.in (AC_CHECK_FUNCS): Check for strsignal.
377 (GCC_NEED_DECLARATIONS): Likewise.
379 * system.h (strsignal): Prototype it, if necessary.
380 (sys_siglist): Declare it, if necessary.
382 Sun Sep 13 04:37:28 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
384 * loop.c (move_movables): While removing insn sequences, preserve
385 the next pointer of the most recently deleted insn when we skip
388 Sun Sep 13 08:13:39 1998 Ben Elliston <bje@cygnus.com>
390 * objc/config-lang.in: Do not output the name of the selected
391 thread file when building the front-end. The Makefile for the
392 runtime library will do this.
394 * objc/Make-lang.in: Do not build the runtime library or install
395 the Objective C header files. The Makefile for the runtime
396 library will do this.
398 * objc/Makefile.in (all.indirect): Only build the front-end.
399 (compiler): Rename to `frontend'.
400 (obj-runtime): Remove target.
401 (copy-headers): Likewise.
402 (clean): No need to remove `libobjc.a' any longer.
404 Sat Sep 12 11:37:19 1998 Michael Meissner <meissner@cygnus.com>
406 * rs6000.h ({ASM,CPP}_CPU_SPEC): Add support for all machines
407 supported with -mcpu=xxx.
409 Fri Sep 11 23:55:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
411 * flow.c (mark_set_1): Recognize multi-register structure return
412 values in CALL insns.
413 (mark_used_regs): Likewise.
414 (count_reg_sets_1): Likewise.
415 (count_reg_references): Likewise.
416 * rtlanal.c (note_stores): Likewise.
417 (reg_overlap_mentioned_p): Likewise.
418 * haifa-sched.c (check_live_1): Likewise.
419 (update_live_1): Likewise.
420 (sched_analyze_1): Likewise.
421 (sched_note_set): Likewise.
422 (birthing_insn_p): Likewise.
423 (attach_deaths): Likewise.
425 * config/sparc/sparc.md (movdf_const_intreg_sp64): Disable.
428 Fri Sep 11 22:57:55 1998 Eric Dumazet <dumazet@cosmosbay.com>
430 * config/i386/sco5.h (ASM_WEAKEN_LABEL): Defined as in svr4.h.
432 Thu Sep 10 22:02:04 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
434 * glimits.h (__LONG_MAX__): Recognize __sparcv9 too.
436 Thu Sep 10 21:19:10 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
438 * configure.in: Add check for GAS subsection -1 support.
439 * acconfig.h (HAVE_GAS_SUBSECTION_ORDERING): Add.
440 * configure config.in: Rebuilt.
441 * config/sparc/sparc.h (CASE_VECTOR_MODE): For V9 flag_pic, use
442 SImode is subsection -1 works, else use DImode.
443 (ASM_OUTPUT_ADDR_VEC_START, ASM_OUTPUT_ADDR_VEC_END): Define if
445 * config/sparc/sparc.c (sparc_output_addr_vec,
446 sparc_output_addr_diff_vec): Use them if defined.
448 Thu Sep 10 10:46:01 1998 Mark Mitchell <mark@markmitchell.com>
450 * tree.h (DECL_ORIGIN): New macro.
451 * integrate.c (copy_and_set_decl_abstract_origin): New function.
452 (copy_decl_list): Use it.
453 (integrate_parm_decls): Likewise.
454 (integrate_decl_tree): Likewise.
455 * dwarf2out.c (decl_ultimate_origin): Simplify.
456 * dwarfout.c (decl_ultimate_origin): Likewise.
457 * c-decl.c (duplicate_decls): Use DECL_ORIGIN.
458 (pushdecl): Likewise.
460 Thu Sep 10 08:01:31 1998 Anthony Green <green@cygnus.com>
462 * config/rs6000/rs6000.c (output_epilog): Add Java support.
464 Thu Sep 10 14:48:59 1998 Martin von Löwis <loewis@informatik.hu-berlin.de>
466 * invoke.texi (C++ Dialect Options): Document -fhonor-std.
468 Thu Sep 10 01:38:05 1998 Jeffrey A Law (law@cygnus.com)
470 * reg-stack.c (straighten_stack): Do nothing if the virtual stack is
471 empty or has a single entry.
473 * toplev.c (rest_of_compilation): Open up the dump file for reg-stack
474 before calling reg_to_stack.
476 Thu Sep 10 00:03:34 1998 Richard Henderson <rth@cygnus.com>
478 * alpha.c (alphaev5_insn_pipe): Abort on default case.
479 (alphaev5_next_group): Swallow CLOBBERs and USEs.
481 * c-tree.h (warn_long_long): Declare it.
483 Wed Sep 9 23:31:36 1998 (Stephen L Moshier) <moshier@world.std.com>
485 * emit-rtl.c (gen_lowpart_common): Disable optimization of
486 initialized float-int union if the value is a NaN.
488 Wed Sep 9 23:00:48 1998 Nathan Sidwell <nathan@acm.org>
490 * c-lex.c (real_yylex): Don't warn about long long constants if
491 we're allowing long long
493 Wed Sep 9 21:58:41 1998 Bernd Schmidt <crux@pool.informatik.rwth-aachen.de>
495 * except.h (current_function_eh_stub_label): Declare.
496 (current_function_eh_old_stub_label): Declare.
497 * function.h (struct function): New members eh_stub_label and
499 * except.c (current_function_eh_stub_label): New variable.
500 (current_function_eh_old_stub_label): New variable.
501 (init_eh_for_function): Clear them.
502 (save_eh_status): Save them.
503 (restore_eh_status): Restore them.
504 (expand_builtin_eh_stub): Set current_function_eh_stub_label.
505 (expand_builtin_eh_stub_old): Set current_function_eh_old_stub_label.
506 * flow.c (find_basic_blocks_1): When handling a REG_LABEL note, don't
507 make an edge from the block that contains it to the block starting
508 with the label if this label is one of the eh stub labels.
509 If eh stub labels exist, show they are reachable from the last block
512 * reload1.c (reload): Break out several subroutines and make some
514 (calculate_needs_all_insns): New function, broken out of reload.
515 (calculate_needs): Likewise.
516 (find_reload_regs): Likewise.
517 (find_group): Likewise.
518 (find_tworeg_group): Likewise.
519 (something_needs_reloads): New global variable, formerly in reload.
520 (something_needs_elimination): Likewise.
521 (caller_save_spill_class): Likewise.
522 (caller_save_group_size): Likewise.
523 (max_needs): Likewise.
524 (group_size): Likewise.
525 (max_groups): Likewise.
526 (max_nongroups): Likewise.
527 (group_mode): Likewise.
528 (max_needs_insn): Likewise.
529 (max_groups_insn): Likewise.
530 (max_nongroups_insn): Likewise.
533 * print-rtl.c (print_rtx): For MEMs, print MEM_ALIAS_SET.
535 Wed Sep 9 13:14:41 1998 Richard Henderson <rth@cygnus.com>
537 * loop.c (load_mems): Copy rtx for output mem.
539 Wed Sep 9 15:16:58 1998 Gavin Romig-Koch <gavin@cygnus.com>
541 * mips/abi64.h (LONG_MAX_SPEC): Don't set LONG_MAX for
542 mips1 or mips2 either.
544 Wed Sep 9 12:31:35 1998 Jeffrey A Law (law@cygnus.com)
546 * pa.c (pa_reorg): New marking scheme for jumps inside switch
548 (pa_adjust_insn_length): Update to work with new marking scheme
549 for jumps inside switch tables.
550 * pa.md (switch_jump): Remove pattern.
551 (jump): Handle jumps inside jump tables.
553 * Makefile.in (profile.o): Depend on insn-config.h
555 Wed Sep 9 09:36:51 1998 Jim Wilson <wilson@cygnus.com>
557 * iris6.h (DWARF2_UNWIND_INFO): Undef.
559 Wed Sep 9 01:32:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
561 Add preliminary native sparcv9 Solaris support.
562 * configure.in: Recognize sparv9-*-solaris2*
564 * config.sub: Recognize sparcv9 just like sparc64.
565 * config/sparc/sol2-c1.asm config/sparc/sol2-ci.asm
566 config/sparc/sol2-cn.asm: Macroize so it can be shared between
567 32-bit and 64-bit Solaris systems.
568 * config/sparc/t-sol2: Assemble those with cpp.
569 * config/sparc/sparc.h (TARGET_CPU_sparcv9): New alias for v9.
570 (*TF*_LIBCALL): If ARCH64 use V9 names.
571 * config/sparc/{xm-sysv4-64,sol2-sld-64}.h: New files.
573 Wed Sep 9 01:07:30 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
575 * config/sparc/sparc.h (TARGET_CM_MEDMID): Fix documentation.
576 (CASE_VECTOR_MODE): Set to SImode even if PTR64, when MEDLOW and
578 (ASM_OUTPUT_ADDR_{VEC,DIFF}_ELT): Check CASE_VECTOR_MODE not
580 * config/sparc/sparc.md (tablejump): Likewise, and sign extend op0
581 to Pmode if CASE_VECTOR_MODE is something else.
583 Wed Sep 9 00:10:31 1998 Jeffrey A Law (law@cygnus.com)
585 * prefix.c (update_path): Correctly handle cases where PATH is
586 a substring of the builtin prefix, but specifies a different
589 Tue Sep 8 23:46:04 1998 Hans-Peter Nilsson <hp@axis.se>
591 * expr.c: Corrected comment about what MOVE_RATIO does.
592 * config/alpha/alpha.h: Ditto.
593 * config/1750a/1750a.h: Ditto.
594 * config/clipper/clipper.h: Ditto.
595 * config/i386/i386.h: Ditto.
597 Tue Sep 8 22:56:12 1998 Jeffrey A Law (law@cygnus.com)
599 * configure.in (m68k-next-nextstep3*): Use collect2.
600 Similarly for x86 NeXT configurations.
601 * configure: Rebuilt.
603 Tue Sep 8 01:38:57 1998 Nathan Sidwell <nathan@acm.org>
605 * configure.in: Don't assume srcdir is .../gcc
606 * configure: Rebuilt.
608 Sat Sep 5 16:34:34 EDT 1998 John Wehle (john@feith.com)
610 * global.c: Update comments.
611 (global_alloc): Assign allocation-numbers
612 even for registers allocated by local_alloc in case
613 they are later spilled and retry_global_alloc is called.
614 (mark_reg_store, mark_reg_clobber,
615 mark_reg_conflicts, mark_reg_death): Always record a
616 conflict with a pseudo register even if it has been
617 assigned to a hard register.
618 (dump_conflicts): Don't list pseudo registers already assigned to
619 a hard register as needing to be allocated, but do list their
621 * local-alloc.c: Update comment.
623 Mon Sep 7 23:38:01 1998 Jeffrey A Law (law@cygnus.com)
625 * configure.in: Check for bogus GCC_EXEC_PREFIX and LIBRARY_PATH.
626 * configure: Rebuilt.
628 Mon Sep 7 22:41:46 1998 Michael Meissner <meissner@cygnus.com>
630 * rs6000.c (rs6000_override_options): Fix name for ec603e, to add
632 * t-ppccomm (MULTILIB_MATCHES_FLOAT): Add support for -mcpu=xxx
633 for all targets that set -msoft-float.
635 Mon Sep 7 23:30:07 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
637 * toplev.c (print_switch_values): Make static to match prototype.
639 Mon Sep 7 19:13:59 1998 Jeffrey A Law (law@cygnus.com)
641 * configure.in: If we are unable to find the "gnatbind" program,
642 then do not configure the ada subdir.
643 * configure: Rebuilt.
645 Sun Sep 6 14:03:58 PDT 1998 Jeff Law (law@cygnus.com)
647 * version.c: Bump for snapshot.
649 Sun Sep 6 13:28:07 PDT 1998 Jeff Law (law@cygnus.com)
651 * version.c: Bump for snapshot.
653 Sun Sep 6 08:54:14 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
655 * Makefile.in (toplev.o): Depend on $(EXPR_H).
656 (insn-extract.o, insn-attrtab.o): Depend on toplev.h.
658 * gansidecl.h: Define ATTRIBUTE_NORETURN.
660 * genattrtab.c: Have insn-attrtab.c include toplev.h.
662 * genextract.c: Have insn-extract.c include toplev.h.
664 * rtl.h: Don't prototype `fatal_insn_not_found' and `fatal_insn'.
666 * toplev.c: Include expr.h.
667 (really_sorry, fancy_abort): Remove prototypes.
668 (set_target_switch): Add argument in prototype.
669 (vfatal): Mark prototype with ATTRIBUTE_NORETURN.
670 (v_really_sorry): Likewise.
671 (print_version, print_single_switch, print_switch_values): Make
672 static and add prototype arguments.
673 (decl_printable_name): Add prototype arguments.
674 (lang_expand_expr_t): New typedef.
675 (lang_expand_expr): Declare as a lang_expand_expr_t.
676 (incomplete_decl_finalize_hook): Add prototype argument.
677 (decl_name): Mark variable `verbosity' with ATTRIBUTE_UNUSED.
678 (botch): Likewise for variable `s'.
679 (rest_of_type_compilation): Mark variables `type' and `toplev'
680 with ATTRIBUTE_UNUSED if none of DBX_DEBUGGING_INFO,
681 XCOFF_DEBUGGING_INFO or SDB_DEBUGGING_INFO are defined.
682 (display_help): Make variable `i' an `unsigned long'.
683 (main): Remove unused parameter `envp'.
684 Cast assignment to `lang_expand_expr' to a `lang_expand_expr_t'.
685 Cast -1 when comparing it with a `size_t'.
687 * toplev.h (fatal, fatal_io_error, pfatal_with_name): Mark
688 prototype with ATTRIBUTE_NORETURN.
689 (fatal_insn_not_found, fatal_insn, really_sorry,
690 push_float_handler, pop_float_handler): Add prototypes.
691 (fancy_abort): Mark prototype with ATTRIBUTE_NORETURN.
692 (do_abort, botch): Add prototypes.
694 Sat Sep 6 12:05:18 1998 John Carr <jfc@mit.edu>
696 * final.c (final): If a label is reached only from a single jump,
697 call NOTICE_UPDATE_CC on the jump and its predecessor before
698 emitting the insn after the label.
700 * i386.h: Add AMD K6 support.
701 Change TARGET_* macros to use table lookup.
702 (INITIALIZE_TRAMPOLINE): Improve trampoline code.
703 (ADJUST_COST): Change definition to call function in i386.c.
704 (ISSUE_RATE): Define as 2 for anything newer than an 80486.
705 * i386.c: Add AMD K6 support.
706 Add constants for feature tests used by TARGET_* macros.
707 (split_di): If before reload, call gen_lowpart and gen_highpart.
708 (x86_adjust_cost): New function.
709 (put_jump_code): New function.
710 (print_operand): New codes 'D' and 'd'.
711 * i386.md: New insn types. New insn attribute "memory".
712 Redefine scheduling parameters to use new types and add AMD K6
713 support. Explicitly set type of most insns.
714 (move insns): K6 prefers movl $0,reg to xorl reg,reg. Pentium
715 Pro and K6 prefer movl $1,reg to incl reg.
716 (adddi3, subdi3): Set cc_status.
717 (DImode shift patterns): Change label counters from HOST_WIDE_INT
718 to int; x86 can't have more than 2^31 DImode shifts per file.
719 (setcc): Combine all setcc patterns. Allow writing memory.
720 Combine all jump patterns using match_operator.
721 (*bzero): Name pattern. Emit mutliple stos instructions when that
722 is faster than rep stos.
723 (xordi3, anddi3, iordi3): Simplify DImode logical patterns and
726 Sun Sep 6 11:17:20 1998 Dave Love <d.love@dl.ac.uk>
728 * config/m68k/x-next (BOOT_LDFLAGS): Define suitably for f771
731 Sat Sep 5 22:05:25 1998 Richard Henderson <rth@cygnus.com>
733 * alpha.c (alpha_ra_ever_killed): Inspect the topmost sequence,
734 not whatever we're generating now.
736 * alpha.c (set_frame_related_p, FRP): New.
737 (alpha_expand_prologue): Mark frame related insns.
738 (alpha_expand_epilogue): Likewise, but with a null FRP.
739 * alpha.h (INCOMING_RETURN_ADDR_RTX): New.
740 * alpha.md (exception_receiver): New.
741 * alpha/crtbegin.asm (.eh_frame): New beginning.
742 (__do_frame_setup, __do_frame_takedown): New.
743 * alpha/crtend.asm (.eh_frame): New ending.
744 * alpha/elf.h (DWARF2_DEBUGGING_INFO): Define.
745 (ASM_SPEC): Don't emit both dwarf2 and mdebug.
746 (ASM_FILE_START): Don't emit .file for dwarf2.
748 * rtl.h (enum reg_note): Add REG_FRAME_RELATED_EXPR.
749 * rtl.c (reg_note_name): Likewise.
750 * rtl.texi (REG_NOTES): Likewise.
751 * dwarf2out.c (dwarf2out_frame_debug): Use it. Recognize a store
754 Sat Sep 5 14:47:17 1998 Richard Henderson <rth@cygnus.com>
756 * i386.h (PREFERRED_RELOAD_CLASS): Standard fp constants load to TOS.
757 * i386.md (movsf, movdf, movxf): Validate memory address returned
758 from force_const_mem. Kill useless REG_EQUAL setting code.
760 Sat Sep 5 14:23:31 1998 Torbjorn Granlund <tege@matematik.su.se>
762 * m68k.md (zero_extendsidi2): Fix typo.
764 Sat Sep 5 13:40:24 1998 Krister Walfridsson <cato@df.lth.se>
766 * configure.in: Removed references to the removed file
767 * config/xm-netbsd.h. Use ${cpu_type}/xm-netbsd.h for
768 * arm*-*-netbsd* and ns32k-*-netbsd*.
769 * config/i386/xm-netbsd.h: Removed unnecessary file.
770 * config/m68k/xm-netbsd.h: Likewise.
771 * config/sparc/xm-netbsd.h: Likewise.
772 * config/mips/xm-netbsd.h: Likewise.
774 Sat Aug 29 13:32:58 1998 Mumit Khan <khan@xraylith.wisc.edu>
776 * i386/cygwin32.h (BIGGEST_ALIGNMENT): Define.
777 (PCC_BITFIELD_TYPE_MATTERS): Define to be 0.
779 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't check for
780 for exact section attributions.
782 * i386/mingw32.h (CPP_PREDEFINES): Add __MSVCRT__ for msvc
784 * i386/crtdll.h (CPP_PREDEFINES): Define.
786 Sat Sep 5 03:23:05 1998 Jeffrey A Law (law@cygnus.com)
788 * m68k.md (5200 movqi): Do not allow byte sized memory references
790 * m68k.c (output_move_qimode): Do not use byte sized operations on
793 * Makefile.in (pexecute.o): Use pexecute.c from libiberty. Provide
794 explicit rules for building. Similarly for alloca, vfprintf,
795 choose-temp and mkstemp, getopt, getopt1, and obstack.
796 (INCLUDES): Add $(srcdir)/../include.
797 * pexecute.c, alloca.c, vfprintf.c, choose-temp.c, mkstemp.c: Delete.
798 * getopt.h, getopt.c getopt1.c, obstack.c, obstack.h: Likewise.
800 Fri Sep 4 11:57:50 1998 Tom Tromey <tromey@cygnus.com>
802 * gcc.c (do_spec_1): [case 'o'] Account for
803 lang_specific_extra_outfiles.
804 (main): Correctly clear all slots in outfiles for
805 lang_specific_extra_outfiles. Set input_file_number before
806 calling lang_specific_pre_link.
808 Fri Sep 4 10:37:07 1998 Jim Wilson <wilson@cygnus.com>
810 * loop.c (load_mems): Fix JUMP_LABEL field after for_each_rtx call.
812 Fri Sep 4 02:01:05 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
814 * config/sparc/sparc.c (output_double_int): In all V9 symbolic
816 (sparc_output_deferred_case_vectors): If no work to do, return.
817 Fix thinko in Sept 1 change.
819 1998-09-03 SL Baur <steve@altair.xemacs.org>
821 * Makefile.in: add semicolon in BISON definition for portability.
823 Thu Sep 3 13:34:41 1998 Toon Moene <toon@moene.indiv.nluug.nl>
825 * config/nextstep.c (handle_pragma): Correct name of third
828 Tue Sep 1 11:30:33 1998 Nick Clifton <nickc@cygnus.com>
830 * config/m32r/m32r.md: Change (reg:CC 17) to (reg:SI 17).
831 * config/m32r/m32r.h: Make register 17 be fixed.
832 * config/m32r/m32r.c: Use SImode for cc operations.
834 Thu Sep 3 18:17:34 1998 Benjamin Kosnik <bkoz@cygnus.com>
836 * invoke.texi (Warning Options): Add -Wnon-template-friend
839 Thu Sep 3 18:16:16 1998 Michael Meissner <meissner@cygnus.com>
841 * rs6000.c (rs6000_override_options): Add -mcpu={401,e603e}.
843 Thu Sep 3 18:05:16 1998 David Edelsohn <edelsohn@mhpcc.edu>
845 * rs6000.md (movsf): Disable explicit secondary-reload-like
846 functionality if TARGET_POWERPC64.
847 (movdf): Remove TARGET_POWERPC64 explicit secondary-reload-like
850 Thu Sep 3 11:41:40 1998 Robert Lipe <robertl@dgii.com>
852 * fixinc.sco: Borrow code to wrap 'bool' typedefs from tinfo.h
853 and term.h from fixinc.wrap.
855 Thu Sep 3 09:47:31 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
857 * aclocal.m4 (GCC_HEADER_STRING): New macro to detect if it is
858 safe to include both string.h and strings.h together.
859 (GCC_NEED_DECLARATION): Test STRING_WITH_STRINGS when deciding
860 which headers to search for function declarations. Continue to
861 prefer string.h over strings.h when both are not acceptable.
863 * acconfig.h (STRING_WITH_STRINGS): Add stub.
865 * configure.in: Call GCC_HEADER_STRING.
867 * system.h: Test STRING_WITH_STRINGS when deciding which headers
868 to include. Continue to prefer string.h over strings.h when both
871 Wed Sep 2 23:56:29 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
873 * config/sparc/sparc.c (output_double_int): If V9 and MEDLOW, do
874 not assume top 32-bits of symbolic addresses are zero if
877 Thu Sep 3 00:23:21 1998 Richard Henderson <rth@cygnus.com>
879 * ginclude/va-alpha.h: Protect entire second portion of the
880 file against double inclusion.
882 Thu Sep 3 00:37:55 1998 Ovidiu Predescu <ovidiu@aracnet.com>
884 Added support for the Boehm's garbage collector.
885 * configure.in: Handle --enable-objc-gc.
886 * configure: Rebuilt.
887 * Makefile.in (CHECK_TARGETS): Add check-objc.
888 (check-objc): New rule.
889 * objc/Make-lang.in: Build a different Objective-C library that
890 runs with the Boehm's collector.
891 * objc/encoding.c (objc_round_acc_size_for_types): New function.
892 * objc/encoding.c: Correctly compute the size of compound types in
893 the presence of bitfields. Skip the variable name of the type if
894 any. Added support for long long.
895 * objc/encoding.h (_C_GCINVISIBLE): New specifier.
896 (_F_GCINVISIBLE): New mask.
897 * objc/gc.c: New file. Compute the type memory mask associated with
898 a class based on the runtime information.
899 * objc/misc.c: Added the hooks that use the Boehm's collector
900 allocation functions.
901 * objc/objc-act.c (build_class_template): Generate a new class
902 member (gc_object_type) to hold the class' type memory mask.
903 (build_shared_structure_initializer): Initialize the new member to
905 (encode_complete_bitfield): New function. Generate the new
907 (encode_field_decl): Generate the new encoding only for the GNU
909 * objc/objc-api.h (_C_LNG_LNG, _C_ULNG_LNG): New specifiers for the
911 (class_get_gc_object_type): New function to mark a pointer instance
912 variable as a weak pointer.
913 * objc/objc-features.texi: New file.
914 * objc/objc.h (gc_object_type): New class member.
915 * objc/objects.c (class_create_instance): Create a typed memory
916 object when compiled with Boehm's collector support.
917 * objc/sendmsg.c (__objc_init_install_dtable): Call
918 __objc_send_initialize instead of setting the initialize flag.
919 (__objc_send_initialize): Call __objc_generate_gc_type_description
920 to generate the class type memory mask. Rewrite the code that
921 sends the +initialize so that it is called only once (bug report
922 and fix from Ronald Pijnacker <Ronald.Pijnacker@best.ms.philips.com>).
923 * testsuite/objc: New testsuite for Objective-C type encoding.
924 * testsuite/lib/objc-torture.exp: New file.
925 * testsuite/lib/objc.exp: New file.
927 Wed Sep 2 14:47:36 1998 Jim Wilson <wilson@cygnus.com>
929 * jump.c (jump_optimize): In if/then/else transformations, add
930 another call to modified_between_p for the jump insn.
932 Wed Sep 2 14:16:49 1998 Jeffrey A Law (law@cygnus.com)
934 * fix-header.c (symlink): Treat like readlink.
936 Wed Sep 2 19:30:06 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
938 * dwarfout.c (fundamental_type_code): Encode 32 bit floats/doubles
941 Wed Sep 2 10:06:07 1998 Nick Clifton <nickc@cygnus.com>
943 * config/nextstep.h: Update HANDLE_PRAGMA macro.
944 * config/h8300/h8300.h: Update HANDLE_PRAGMA macro.
945 * config/i960/i960.h: Update HANDLE_PRAGMA macro.
947 * config/nextstep.c (handle_pragma): Take three arguments, as per
948 the new HANDLE_PRAGMA macro specification.
949 * config/h8300/h8300.c (handle_pragma): Take three arguments, as
950 per the new HANDLE_PRAGMA macro specification.
951 * config/i960/i960.c (process_pragma): Take three arguments, as
952 per the new HANDLE_PRAGMA macro specification.
954 Wed Sep 2 09:25:29 1998 Nick Clifton <nickc@cygnus.com>
956 * c-lex.c (check_newline): Call HANDLE_PRAGMA before
957 HANDLE_SYSV_PRAGMA if both are defined. Generate warning messages
958 if unknown pragmas are encountered.
959 (handle_sysv_pragma): Interpret return code from
960 handle_pragma_token (). Return success/failure indication rather
961 than next unprocessed character.
962 (pragma_getc): New function: retrieves characters from the
963 input stream. Defined when HANDLE_PRAGMA is enabled.
964 (pragma_ungetc): New function: replaces characters back into the
965 input stream. Defined when HANDLE_PRAGMA is enabled.
967 * c-pragma.c (handle_pragma_token): Return success/failure status
970 * c-pragma.h: Change prototype of handle_pragma_token().
972 * varasm.c: (handle_pragma_weak): Only create this function if
973 HANDLE_PRAGMA_WEAK is defined.
975 * c-common,c (decl_attributes): If defined call the expression
976 contained within the INSERT_ATTRIBUTES macro before adding
977 attributes to a decl.
979 * tm.texi (HANDLE_PRAGMA): Document the new verion of
980 HANDLE_PRAGMA, which takes three arguments.
981 (INSERT_ATTRIBUTES): Document this new macro.
983 * LANGUAGES: Document the new version of HANDLE_PRAGMA and the
984 new INSERT_ATTRIBUTES macro.
986 Wed Sep 2 02:03:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
988 * config/sparc/sparc.md (movdf): Only generate special RTL for
990 (move_label_di): Remove
991 (movdi_pic_label_ref, movdi_high_pic_label_ref,
992 movdi_lo_sum_pic_label_ref): New patterns for 64-bit label
994 * config/sparc/sparc.h (ASM_OUTPUT_ADDR_VEC_ELT,
995 ASM_OUTPUT_ADDR_DIFF_ELT): Don't do anything special for MEDLOW,
996 output an .xword for all 64-bit cases.
998 Tue Sep 1 15:55:17 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1000 * config/sparc/sparc.c (finalize_pic): Don't output arbitrary
1001 alignment, use FUNCTION_BOUNDARY instead.
1002 (sparc_output_deferred_case_vectors): Likewise.
1004 Mon Aug 31 17:25:41 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1006 * config/sparc/sparc.md (movsf_const_intreg): Kill warning.
1007 (movtf_insn_sp64, movtf_no_e_insn_sp64): Reorder alternatives.
1009 Mon Aug 31 13:57:55 1998 Richard Henderson <rth@cygnus.com>
1011 * alpha/va_list.h: New file.
1012 * alpha/x-alpha (EXTRA_HEADERS): New. Add va_list.h.
1014 Mon Aug 31 14:55:02 1998 Jeffrey A Law (law@cygnus.com)
1016 * NEWS: Add SCO Openserver and Unixware 7 notes.
1020 Mon Aug 31 15:42:18 1998 Dave Brolley <brolley@cygnus.com>
1022 * varasm.c (compare_constant_1): Handle RANGE_EXPR.
1023 (record_constant_1): Handle RANGE_EXPR.
1025 Mon Aug 31 10:54:03 1998 Richard Henderson <rth@cygnus.com>
1027 * print-rtl.c (print_rtx): NOTE_INSN_LIVE has an rtx not a bitmap.
1028 * haifa-sched.c (sched_analyze): Handle NOTE_INSN_RANGE_START
1029 and NOTE_INSN_RANGE_END specially.
1030 (reemit_notes): Likewise.
1032 Mon Aug 31 10:18:52 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1034 * sparc.c (TMASK, UMASK): Use `(unsigned)1' not `1U'.
1035 (ultrasparc_sched_init): Remove unneeded &.
1037 Mon Aug 31 10:47:16 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
1039 * config/m68k/m68k.h (TARGET_SWITCHES): Don't remove MASK_68040
1040 for m68020-60, to prevent the use of fintrz.
1042 Sun Aug 30 22:17:20 1998 Mark Mitchell <mark@markmitchell.com>
1044 * configure.in: If the native compiler is GCC use $(WARN_CFLAGS)
1046 * Makefile.in: Likewise.
1047 * configure: Regenerated.
1049 Sun Aug 30 22:15:41 1998 H.J. Lu (hjl@gnu.org)
1051 * configure.in (gxx_include_dir): Changed to
1052 '${prefix}/include/g++'-${libstdcxx_interface}.
1053 * configure: Rebuilt.
1055 Sun Aug 30 20:19:43 1998 Hans-Peter Nilsson <hp@axis.se>
1057 * expr.c (expand_expr): Change ">" to ">=" making MOVE_RATIO use
1059 * tm.texi (Costs): Say MOVE_RATIO is number of mem-mem move
1060 *sequences* *below* which scalar moves will be used.
1062 Sun Aug 30 17:18:43 1998 Jeffrey A Law (law@cygnus.com)
1064 * collect2.c (mktemp): Delete unused declaration.
1066 * config/xm-netbsd.h: Remove unnecessary file.
1067 * config/*/xm-netbsd.h: Do not include the generic xm-netbsd.h
1068 file anymore, it is not needed.
1070 Sun Aug 30 16:05:45 1998 Mark Mitchell <mark@markmitchell.com>
1072 * convert.c (convert_to_integer): Issue an error on conversions to
1075 Sun Aug 30 16:47:20 1998 Martin von Lvwis <loewis@informatik.hu-berlin.de>
1077 * Makefile.in: Add lang_tree_files and gencheck.h.
1078 * configure.in: Generate them.
1079 * gencheck.c: Include gencheck.h.
1081 Sat Aug 29 21:38:24 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1083 * config/sparc/sparc.md (pic_lo_sum_di, pic_sethi_di): Rename to
1084 movdi_lo_sum_pic and movdi_high_pic and make visible.
1085 * config/sparc/sparc.c (legitimize_pic_address): For -fPIC,
1086 emit these when Pmode is not SImode.
1087 * config/sparc/linux64.h (SPARC_DEFAULT_CMODEL): Make CM_MEDLOW.
1089 Sat Aug 29 14:59:32 1998 Mumit Khan <khan@xraylith.wisc.edu>
1091 * i386/cygwin32.h (ASM_OUTPUT_SECTION_NAME): Don't emit
1092 .linkonce directive after the first time.
1094 Sat Aug 29 12:39:56 1998 Jeffrey A Law (law@cygnus.com)
1096 * m68k.md (beq0_di): Generate correct (and more efficient) code when
1097 the clobbered operand overlaps with an input.
1098 (bne0_di): Similarly.
1100 * Makefile.in (INSTALL): Remove "--no-header" argument.
1102 * NEWS: Various updates.
1104 Fri Aug 28 19:00:44 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1106 * config/sparc/sparc.c (arith_operand, const64_operand,
1107 const64_high_operand, arith_double_4096_operand): Mark mode as
1109 (create_simple_focus_bits): Remove unused arg highest_bit_set, all
1111 (sparc_emit_set_const64): Remove unused variable i.
1112 (sparc_splitdi_legitimate): Likewise for addr_part.
1113 (ultra_code_from_mask): Likewise for mask.
1114 (ultra_cmove_results_ready_p): Fixup entry modulo calc. and
1115 reverse return values so it matches usage and comments.
1116 (ultra_flush_pipeline): Likewise.
1117 (ultra_fpmode_conflict_exists): Likewise, remove unused variable
1118 this_type, and allow loads and stores of differing FP modes as
1119 they do not create a conflict.
1120 (ultra_find_type): Initialize fpmode to SFmode, fix
1121 parenthesization thinkos in large conditional.
1122 (ultrasparc_sched_init): Mark dump and sched_verbose as unused.
1123 Init free_slot_mask after ultra_cur_hist is reset, not before.
1124 (ultrasparc_rescan_pipeline_state): Remove unused variable ucode.
1125 (ultrasparc_sched_reorder): Don't bzero current pipeline state,
1126 use ultra_flush_pipeline instead, then re-init group pointer.
1127 Fix statement with no effect. If no progress made in, and no
1128 instructions scheduled at all, advance to new pipeline cycle else
1129 we get into an endless loop.
1130 (ultrasparc_adjust_cost): Remove previous arg.
1131 * config/sparc/sparc.h (ADJUST_COST): Update to reflect that.
1133 Fri Aug 28 13:52:35 1998 Jim Wilson <wilson@cygnus.com>
1135 * sparc.md (DImode, DFmode, TFmode splits): Delete self_reference
1136 code. Use reg_overlap_mentioned_p to detect when source and
1137 destination overlap.
1138 (negtf2_notv9+1): Use DFmode instead of SFmode in last two operands.
1140 1998-08-28 Brendan Kehoe <brendan@cygnus.com>
1142 * loop.c (check_dbra_loop): Pass COMPARISON_VALUE, not
1143 COMPARISON_VAL, into invariant_p.
1145 Fri Aug 28 15:13:25 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1147 * regmove.c (regclass_compatible_p): New function.
1148 (regmove_optimize): Use it.
1150 Use NREGS parameter insted of calling max_reg_num.
1152 (fixup_match_1): Don't use code = MINUS when later tieing with
1153 a hard register is likely.
1155 Fri Aug 28 14:54:07 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1157 * loop.c (check_dbra_loop): Fix calculation of FINAL_VALUE when
1158 COMPARISON_VAL was normalized.
1160 Thu Aug 27 20:10:46 1998 Jeffrey A Law (law@cygnus.com)
1162 * loop.c (check_dbra_loop): The loop ending comparison value
1163 must be an invariant or we can not reverse the loop.
1165 * loop.c (scan_loop): Count down from max_reg_num - 1 to
1166 FIRST_PSEUDO_REGISTER to avoid calling max_reg_num each iteration
1168 (load_mems_and_recount_loop_regs_set): Likewise.
1170 * i386.c (print_operand): Remove obsolete 'c' docs.
1172 Wed Aug 26 17:13:37 1998 Tom Tromey <tromey@cygnus.com>
1174 * gthr.h: Document __GTHREAD_MUTEX_INIT_FUNCTION.
1175 * frame.c (init_object_mutex): New function.
1176 (init_object_mutex_once): Likewise.
1177 (find_fde): Call it.
1178 (__register_frame_info): Likewise.
1179 (__register_frame_info_table): Likewise.
1180 (__deregister_frame_info): Likewise.
1182 Thu Aug 27 15:14:18 1998 Jeffrey A Law (law@cygnus.com)
1184 * haifa-sched.c (sched_analyze_insn): Fix thinko in last change.
1186 Thu Aug 27 16:34:51 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1188 * loop.c (check_dbra_loop): Enable code for reversal
1189 of some loops without a known constant loop end.
1191 Wed Aug 26 18:38:15 1998 Richard Henderson <rth@cygnus.com>
1193 * haifa-sched.c (last_clock_var): New.
1194 (schedule_block): Initialize it.
1195 (schedule_insn): Use it to fill insn modes with issue information.
1197 * alpha.c (alpha_handle_trap_shadows): Remove do-nothing exit.
1198 Tag trapb and next insn with TImode.
1199 (alphaev5_insn_pipe, alphaev5_next_group, alphaev5_align_insns): New.
1200 (alpha_reorg): Add conditional for alpha_handle_trap_shadows.
1201 Invoke alphaev5_align_insns as appropriate.
1202 * alpha.h (LABEL_ALIGN_AFTER_BARRIER): Was ALIGN_LABEL_AFTER_BARRIER.
1203 (MD_SCHED_VARIABLE_ISSUE): New.
1204 * alpha.md (attr type): Add multi.
1205 (define_asm_attributes): New.
1206 (prologue_stack_probe_loop, builtin_setjmp_receiver): Set type multi.
1207 (arg_home): Likewise.
1208 (fnop, unop, realign): New.
1210 Wed Aug 26 15:55:41 1998 Jim Wilson <wilson@cygnus.com>
1212 * iris5.h (PREFERRED_DEBUGGING_TYPE): Undef.
1213 * iris5gas.h (PREFERRED_DEBUGGING_TYPE): Define.
1215 * configure.in (powerpc-ibm-aix4.[12]*): Change from 4.[12].*.
1216 (rs6000-ibm-aix4.[12]*): Likewise.
1217 * configure: Regnerate.
1219 Wed Aug 26 09:30:59 1998 Nick Clifton <nickc@cygnus.com>
1221 * config/arm/thumb.c (thumb_exit): Do not move a4 into lr if it
1222 already contains the return address.
1224 Wed Aug 26 12:57:09 1998 Jeffrey A Law (law@cygnus.com)
1226 * calls.c (expand_call): Use bitfield instructions to extract/deposit
1227 word sized hunks when loading unaligned args into registers.
1229 * haifa-sched.c (sched_analyze_insn): Only create scheduling
1230 barriers for LOOP, EH and SETJMP notes on the loop_notes list.
1232 * mn10300.h (RTX_COSTS): Handle UDIV and UMOD too.
1234 Wed Aug 26 16:35:37 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1236 * loop.c (check_dbra_loop): Add some code that would allow reversal
1237 of some loops without a known constant loop end if it were enabled.
1239 Wed Aug 26 11:08:44 1998 Gavin Romig-Koch <gavin@cygnus.com>
1241 * mips.md (lshrsi3_internal2+2): Fix type-o.
1243 Wed Aug 26 10:53:03 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1245 * system.h: Include stdarg.h/varargs.h, make sure they are ordered
1246 correctly with regards to stdio.h.
1248 * calls.c: Remove stdarg.h/varargs.h.
1251 * combine.c: Likewise.
1252 * cpperror.c: Likewise.
1253 * cpplib.c: Likewise.
1254 * cpplib.h: Likewise.
1255 * doprint.c: Likewise.
1256 * emit-rtl.c: Likewise.
1257 * final.c: Likewise.
1258 * fix-header.c: Likewise.
1260 * genattr.c: Likewise.
1261 * genattrtab.c: Likewise.
1262 * gencodes.c: Likewise.
1263 * genconfig.c: Likewise.
1264 * genemit.c: Likewise.
1265 * genextract.c: Likewise.
1266 * genflags.c: Likewise.
1267 * genopinit.c: Likewise.
1268 * genoutput.c: Likewise.
1269 * genpeep.c: Likewise.
1270 * genrecog.c: Likewise.
1271 * mips-tfile.c: Likewise.
1272 * prefix.c: Likewise.
1273 * protoize.c: Likewise.
1274 * regmove.c: Likewise.
1275 * toplev.c: Likewise.
1278 Wed Aug 26 05:09:27 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
1280 * config/sparc/sparc.c (sparc_override_options): If not
1281 TARGET_FPU, turn off TARGET_VIS.
1282 * config/sparc/sparc.h (TARGET_SWITCHES): Add no-vis.
1283 (LEGITIMATE_CONSTANT_P): Allow SF/DF mode zero when TARGET_VIS.
1284 * config/sparc/sparc.md (movsi_insn): Use fzeros not fzero.
1285 (movdi_insn_sp64): Add VIS fzero alternative.
1286 (clear_sf, clear_df): New VIS patterns.
1287 (movsf, movdf expanders): Allow fp_zero_operand flat out when
1289 (one_cmpldi2_sp64): Provide new fnot1 VIS alternative.
1291 Tue Aug 25 10:57:41 1998 Mark Mitchell <mark@markmitchell.com>
1293 * loop.c (n_times_set, n_times_used, may_not_optimize,
1294 reg_single_usage): Convert to varrays. All uses changed.
1295 (insert_loop_mem): Return a value.
1296 (scan_loop): Tweak AVOID_CC_MODE_COPIES code.
1297 (load_mems_and_recount_loop_regs_set): Likewise. Grow the arrays, if
1300 Tue Aug 25 23:57:12 1998 Jeffrey A Law (law@cygnus.com)
1303 * configure.in: Do not set thread_file to "irix" since no such
1306 * sparc.md (float abs/neg splits): Check reload_completed before
1307 calling alter_subreg.
1309 Tue Aug 25 19:17:59 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1311 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): New
1313 * config/sparc/sparc.h: Declare it.
1314 * config/sparc/sparc.md (float abs/neg splits): Use it.
1315 (all other splits): Handle SUBREGs properly where necessary.
1316 (unnamed (1<<x)-1 V8PLUS pattern): Disable for now.
1318 Tue Aug 25 19:48:46 1998 Jeffrey A Law (law@cygnus.com)
1320 * reorg.c (fill_simple_delay_slots): Do not abort if we encounter
1321 an insn on the unfilled_slots_list that has no delay slots.
1322 (fill_eager_delay_slots): Similarly.
1324 Tue Aug 25 13:35:20 1998 Nick Clifton <nickc@cygnus.com>
1326 * config/v850/v850.c (movsi_source_operand): Treat CONSTANT_P_RTX
1327 as an ordinary operand.
1329 Tue Aug 25 12:54:57 1998 Jason Merrill <jason@yorick.cygnus.com>
1331 * tree.c (valid_machine_attribute): Don't apply attributes to both
1334 Tue Aug 25 12:23:20 PDT 1998 Richard Henderson <rth@cygnus.com>
1336 * reload.c (operands_match_p): Handle rtvecs.
1338 * i386.c (legitimate_pic_address_disp_p): New.
1339 (legitimate_address_p): Use it.
1340 (legitimize_pic_address): Use unspecs to represent @GOT and @GOTOFF.
1341 Handle constant pool symbols just like statics.
1342 (emit_pic_move): Use Pmode not SImode for clarity.
1343 (output_pic_addr_const) [SYMBOL_REF]: Remove @GOT and @GOTOFF hacks.
1344 [UNSPEC]: New, handling what we killed above.
1345 [PLUS]: Detect and abort on invalid symbol arithmetic.
1346 * i386.h (CONSTANT_ADDRESS_P): Remove HIGH.
1348 Tue Aug 25 12:02:23 1998 Mark Mitchell <mark@markmitchell.com>
1350 * alias.c: Include output.h.
1351 (DIFFERENT_ALIAS_SETS_P): Don't treat alias sets as
1352 different if we're in a varargs function.
1353 * Makefile.in (alias.o): Depend on output.h
1355 Tue Aug 25 19:20:12 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1357 * sh.h (GIV_SORT_CRITERION): Delete.
1359 Tue Aug 25 13:19:46 1998 Dave Brolley <brolley@cygnus.com>
1361 * regclass.c (regclass): Use xmalloc/free instead of alloca.
1362 * stupid.c (stupid_life_analysis): Ditto.
1363 * reload1.c (reload): Ditto.
1365 Tue Aug 25 05:48:18 1998 Jakub Jelinek <jj@sunsite.ms.mff.cuni.cz>
1367 * config/sparc/sparc.c (arith_4096_operand, arith_add_operand,
1368 arith_double_4096_operand, arith_double_add_operand): New
1370 * config/sparc/sparc.h (PREDICATE_CODES): Add them, declare them.
1371 * config/sparc/sparc.md (adddi3, addsi3, subdi3, subsi3): Use
1372 them to transform add/sub 4096 into add/sub -4096.
1374 Mon Aug 24 23:31:03 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1376 * loop.c (scan_loop): Allocate some slop to handle pseudos
1377 generated by move_movables.
1378 (load_mems_and_recount_loop_regs_set): Honor AVOID_CC_MODE_COPIES
1381 Mon Aug 24 19:45:40 1998 Jim Wilson <wilson@cygnus.com>
1383 * tree.def (DECL_RESULT): Correct documentation.
1385 Tue Aug 25 01:15:27 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1387 * reload1.c (reload_reg_free_before_p): New argument EQUIV; Changed
1388 all callers. Abort for RELOAD_FOR_INSN. RELOAD_FOR_OUTADDR_ADDR:
1389 conflicts will all RELOAD_FOR_OUTPUT reloads.
1391 * reload1.c (reload_cse_regs_1): When deleting a no-op move that
1392 loads the function result, substitute with a USE.
1394 Mon Aug 24 15:20:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
1396 * rs6000.h (GO_IF_LEGITIMATE_ADDRESS): Use TARGET_POWERPC64
1397 when testing LEGITIMATE_INDEXED_ADDRESS_P DFmode and DImode.
1398 (LEGITIMIZE_ADDRESS): Use TARGET_POWERPC64 for INDEXED fixup.
1399 * rs6000.c (print_operand, case 'L'): Add UNITS_PER_WORD, not 4.
1400 (print_operand, cases 'O' and 'T'): Fix typos in lossage strings.
1401 * rs6000.md (fix_truncdfsi2_store): Remove %w from non-CONST_INT
1403 (movdf_softfloat32, movdf_hardfloat64, movdf_softfloat64): Change
1404 'o' to 'm' for GPR variant constraints.
1406 Mon Aug 24 10:25:46 1998 Jeffrey A Law (law@cygnus.com)
1408 * loop.c (scan_loop): Honor AVOID_CC_MODE_COPIES.
1410 * h8300.h (STRIP_NAME_ENCODING): Fix typo.
1412 * sparc.md (TFmode splits): Use reg_overlap_mentioned_p to detect
1413 when the source and destination overlap.
1415 * stmt.c (emit_case_nodes): Change rtx_function to rtx_fn to avoid
1416 clash with global type.
1418 Mon Aug 24 00:53:53 1998 Jason Merrill <jason@yorick.cygnus.com>
1420 * fixinc.irix: Add curses.h handling from fixinc.wrap.
1422 * c-common.c (combine_strings): Also set TREE_READONLY.
1423 Change warn_write_strings to flag_const_strings.
1424 * c-decl.c, c-tree.h: Likewise.
1426 Sun Aug 23 18:39:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1428 * config/sparc/sparc.c (sparc_emit_set_const32): If outputting a
1429 CONST_INT, not a symbolic reference, don't use a HIGH/LO_SUM
1430 sequence, use SET/IOR instead so CSE can see it.
1431 * config/sparc/sparc.md (movhi_const64_special,
1432 movsi_const64_special): New patterns necessitated by that change.
1433 (movhi_high): Remove.
1434 (movhi_lo_sum): Change to match an IOR.
1435 (movdf_insn_sp32): Test TARGET_V9 not TARGET_ARCH64.
1436 (movdf_insn_v9only): New pattern for when V9 but not ARCH64.
1437 (movdf_insn_sp64): Test both TARGET_V9 and TARGET_ARCH64.
1438 (movdf splits): Allow when not V9 or when not ARCH64 and integer
1439 registers are involved.
1440 (snesi_zero_extend split): Remove reload_completed test.
1441 (unnamed plus and minus zero_extend sidi splits): Add it.
1443 Sun Aug 23 11:56:08 1998 Mark Mitchell <mark@markmitchell.com>
1445 * extend.texi: Remove description of extension to explicit
1446 instantiation that is now endorsed by standard C++.
1448 Sun Aug 23 09:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1450 * config/arc/arc.c (arc_initialize_pic): Remove.
1451 * config/arc/arc.h (INITIALIZE_PIC): Similarly, this routine does
1452 nothing on any platform and is invoked by no-one, it does not even
1453 appear in the documentation.
1454 * config/sparc/sparc.h (INITIALIZE_PIC): Likewise.
1455 * config/sparc/sparc.c (initialize_pic): Likewise.
1456 (find_addr_reg): Remove this as well, no longer referenced after
1459 Sun Aug 23 00:17:14 1998 Jeffrey A Law (law@cygnus.com)
1461 * recog.c (validate_replace_rtx_group): New function.
1462 * recog.h (validate_replace_rtx_group): Declare it.
1463 * regmove.c (optimize_reg_copy_3): If any substitution fails, then undo
1464 the entire group of substitutions.
1466 Sat Aug 22 23:31:00 1998 Klaus-Georg Adams (Klaus-Georg.Adams@chemie.uni-karlsruhe.de)
1468 * loop.c (load_mems): Fix initializers.
1470 Fri Aug 21 23:07:46 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1472 * config/sparc/sparc.md (TFmode splits): Handle destination
1473 registers being referenced in the address correctly.
1475 * expmed.c (make_tree) [CONST_INT]: Sign extend even if
1476 TREE_UNSIGNED, when bitsize of type's mode is larger than
1477 HOST_BITS_PER_WIDE_INT.
1479 Fri Aug 21 19:31:31 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
1481 * tree.def (LABELED_BLOCK_EXPR, EXIT_BLOCK_EXPR): New tree nodes.
1482 * tree.h (LABELED_BLOCK_LABEL, LABELED_BLOCK_BODY,
1483 EXIT_BLOCK_LABELED_BLOCK, EXIT_BLOCK_RETURN, LOOP_EXPR_BODY): New
1485 * expr.c (expand_expr): Handle LABELED_BLOCK_EXPR and
1488 Thu Aug 20 19:43:44 1998 Jeffrey A Law (law@cygnus.com)
1490 * h8300.c (h8300_encode_label): Use '&' for tiny data items.
1491 * h8300.h (TINY_DATA_NAME_P): Likewise.
1492 (STRIP_NAME_ENCODING): Handle '&'.
1494 * mn10200.h (REG_OK_FOR_INDEX_P): Do not check the mode of the
1495 register (it could be accessed via an outer SUBREG).
1496 (REG_OK_FOR_BASE_P): Likewwise.
1497 (GO_IF_LEGITIMATE_ADDRESS): Consistently use REGNO_OK_FOR_BASE_P.
1499 * remove.c (optimize_reg_copy_3): Abort instead of silently generating
1502 * jump.c (rtx_renumbered_equal_p): Do not consider PLUS commutative.
1504 Thu Aug 20 17:35:20 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1506 * config/sparc/sparc.md (movtf_insn_sp32): All memory operands
1507 must be offsettable so the splits can be made.
1509 Thu Aug 20 13:56:53 1998 Michael Meissner <meissner@cygnus.com>
1511 * config/i386/winnt.c: Include system.h, not stdio.h to get
1512 sys/param.h pulled in before rtl.h in case the system defines MIN
1515 Thu Aug 20 13:44:20 1998 David Edelsohn <edelsohn@mhpcc.edu>
1517 * rs6000.md (movqi, movhi): Add CONSTANT_P_RTX.
1519 Thu Aug 20 13:15:11 1998 Dave Brolley <brolley@cygnus.com>
1521 * stor-layout.c (layout_type): Compute TYPE_SIZE_UNIT correctly for
1523 * cpplib.c (cpp_define): Handle macros with parameters.
1525 Wed Aug 19 21:33:19 1998 David Edelsohn <edelsohn@mhpcc.edu>
1527 * rs6000.c (rs6000_output_load_toc_table): Use ld for 64-bit.
1528 (output_toc): Use single TOC slot or llong minimal-toc for DFmode
1529 and DImode 64-bit. Use llong for minimal-toc SFmode and
1530 SYMBOL_REF / LABEL_REF 64-bit.
1531 (output_function_profiler): Use llong for profiler label and ld to
1532 load 64-bit label address.
1534 Wed Aug 19 17:52:27 1998 Nick Clifton (nickc@cygnus.com)
1536 * config/arm/thumb.md (extendqisi2_insn): Cope with REG +
1539 Wed Aug 19 14:13:31 PDT 1998 Jeff Law (law@cygnus.com)
1541 * version.c: Bump for snapshot.
1543 Wed Aug 19 13:10:30 PDT 1998 Jeff Law (law@cygnus.com)
1545 * version.c: Bump for snapshot.
1547 Wed Aug 19 13:06:47 1998 Jason Merrill <jason@yorick.cygnus.com>
1549 * collect2.c (extract_init_priority): Use atoi instead of strtoul.
1551 Wed Aug 19 13:51:35 1998 Hans-Peter Nilsson <hp@axis.se>
1553 * tm.texi (Misc): Fix typo "teh".
1555 * tm.texi (PIC): Fix typo "PPIC".
1557 * tm.texi (Caller Saves): Say that DEFAULT_CALLER_SAVES has no
1558 effect when -O2 and higher.
1559 * invoke.texi (Optimize Options): Likewise for -fcaller-saves.
1561 1998-08-19 Michael Hayes <michaelh@ongaonga.chch.cri.nz>
1563 * regclass.c: Changed register set documentation to be consistent
1566 * final.c (final_start_function) Removed redundant test for
1569 Wed Aug 19 13:28:41 1998 Mark Mitchell <mark@markmitchell.com>
1571 * rtl.h (rtx_function): New type.
1572 (for_each_rtx): New function.
1573 * rtlanal.c (for_each_rtx): Define it.
1575 * recog.c (change_t): New type.
1576 (change_objects, change_old_codes, change_locs, change_olds):
1578 (changes): New variable.
1579 (validate_change): Dynamically allocate room for more changes, if
1580 necessary. Uses changes array instead of change_objects, etc.
1581 (apply_change_group): Use changes array instead of
1582 change_objects, etc.
1584 * loop.c (loop_mem_info): New type.
1585 (loop_mems): New variable.
1586 (loop_mems_idx): Likewise.
1587 (looop_mems_allocated): Likewise.
1588 (scan_loop): Remove nregs parameter.
1589 (next_insn_in_loop): New function.
1590 (load_mems_and_recount_loop_regs_set): Likewise.
1591 (load_mems): Likewise.
1592 (insert_loop_mem): Likewise.
1593 (replace_loop_mem): Likewise.
1594 (replace_label): Likewise.
1595 (INSN_IN_RANGE_P): New macro.
1596 (loop_optimize): Don't pass max_reg_num() to scan_loop.
1597 (scan_loop): Remove nregs parameter, compute it after any new
1598 registers are created by load_mems. Use INSN_IN_RANGE_P and
1599 next_insn_in_loop rather than expanding them inline. Call
1600 load_mems to load memory into pseudos, if appropriate.
1601 (prescan_loop): Figure out whether or not there are jumps from the
1602 loop to targets other than the label immediately following the
1603 loop. Call insert_loop_mem to notice all the MEMs used in the
1604 loop, if it could be safe to pull MEMs into REGs for the duration
1606 (strength_reduce): Use next_insn_in_loop. Tweak comments.
1608 Wed Aug 19 08:29:44 1998 Richard Earnshaw (rearnsha@arm.com)
1610 * arm.c (arm_override_options): Remove lie about ignoring PIC flag.
1612 Wed Aug 19 07:08:15 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1614 * config/sparc/sparc.c (finalize_pic): Check for the correct
1615 nonlocal_goto_receiver UNSPEC number.
1616 * config/sparc/sparc.md (nonlocal_goto_receiver): Add comment
1617 making note of this dependency existing in sparc.c
1618 (negtf2_notv9 split): Give NEG SFmode.
1619 (negsf2): Fix insn output string.
1621 Tue Aug 18 12:40:27 1998 Richard Henderson <rth@cygnus.com>
1623 * c-common.c (decl_attributes): Issue an error if the argument
1624 to alias is not a string.
1626 Tue Aug 18 10:33:30 1998 Jeffrey A Law (law@cygnus.com)
1628 * haifa-sched.c (sched_analyze): Put all JUMP_INSNs on the last
1629 pending memory flush list.
1631 * combine.c (can_combine_p): Allow combining insns with REG_RETVAL
1633 (try_combine): Allow combining insns with REG_LIBCALL notes.
1635 * expr.c (emit_block_move): Do not call memcpy as a libcall
1636 instead build up a CALL_EXPR and call it like any other
1638 (clear_storage): Similarly for memset.
1640 * regmove.c (fixup_match_2): Do not call reg_overlap_mentioned_p
1643 * Makefile.in (cplus-dem.o): Provide explicit rules for building
1646 * regmove.c (optimize_reg_copy_1): Update REG_N_CALLS_CROSSED
1647 and REG_LIVE_LENGTH as successful substitutions are made.
1649 Tue Aug 18 07:15:27 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1651 * config/sparc/sparc.c (ultra_find_type): Add empty semicolon
1652 statement after end of loop label.
1654 Tue Aug 18 07:13:27 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1656 * config/sparc/sparc.c (ultra_types_avail): New variable.
1657 (ultra_build_types_avail): New function to record mask of insn
1658 types in ready list at this cycle.
1659 (ultrasparc_sched_reorder): Call it.
1660 (ultra_find_type): Use it to quicken the search. Also simplif
1661 dependency check, don't use rtx_equal_p because we know exactly
1662 what we are looking for.
1664 Tue Aug 18 03:20:53 1998 Richard Earnshaw (rearnsha@arm.com)
1666 * arm.h (SECONDARY_INPUT_RELOAD_CLASS): Return NO_REGS if compiling
1667 for architecture v4.
1669 Mon Aug 17 21:26:38 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1671 * config/sparc/sparc.md (sltu, sgeu): Don't FAIL, call
1673 (movsf_const_intreg, movsf_const_high, movsf_const_lo,
1674 movdf_const_intreg and helper splits): New patterns to move float
1675 constants into integer registers.
1676 (negtf2, negdf2, abstf2, absdf2): Rework using new patterns and
1679 Mon Aug 17 11:46:19 1998 Jeffrey A Law (law@cygnus.com)
1682 * tree.c (build_index_type): Copy TYPE_SIZE_UNIT from sizetype
1684 * c-decl.c (finish_enum): Copy TYPE_SIZ_UNIT from enumtype to tem.
1686 * rs6000.c (secondary_reload_class): For TARGET_ELF, indicate that
1687 a BASE_REGS register is needed as an intermediate when copying
1688 a symbolic value into any register class other than BASE_REGS.
1690 * expr.c (move_by_pieces): No longer static. Remove prototype.
1691 * rtl.h (move_by_pieces): Add extern prototype.
1692 * mips.c (expand_block_move): Handle aligned straight line copy by
1693 calling move_by_pieces.
1695 * expr.c (expand_expr): Allow assignments from TImode PARM_DECLs
1698 Mon Aug 17 10:28:52 1998 Mark Mitchell <mark@markmitchell.com>
1700 * stmt.c (expand_end_loop): Tidy. Allow unconditional
1701 jumps out of the loop to be treated as part of the exit test.
1703 Mon Aug 17 10:06:11 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1704 Jeff Law <law@cygnus.com>
1706 * Makefile.in (cplus-dep.o): Use cplus-dem.c from libiberty.
1707 * cplus-dem.c: Delete.
1709 * Makefile.in (fold-const.o): depend on $(RTL_H).
1711 * fold-const.c: Include rtl.h to get the prototype for
1712 `set_identifier_local_value'.
1714 * loop.c (express_from_1): Remove unused variable `tmp'.
1715 (combine_givs): cast the first argument of bzero to char *.
1717 * toplev.c (display_help): Remove unused variable `looking_for_start'.
1719 * c-decl.c (init_decl_processing): Remove unneeded &.
1721 * alpha.h (alpha_initialize_trampoline): Provide prototype.
1723 * except.c (set_exception_lang_code, set_exception_version_code):
1724 Change parameter from `short' to `int' to avoid using a gcc
1727 * except.h (set_exception_lang_code, set_exception_version_code):
1728 Likewise for prototypes.
1730 * flow.c (count_reg_references): Remove unused variables `regno'
1733 * gcse.c (hash_scan_insn): Declare parameter `in_libcall_block'.
1735 * prefix.c (translate_name): Cast the result of `alloca'.
1737 * varray.h (VARRAY_FREE): Reimplement as a `do-while(0)' statement.
1739 Mon Aug 17 09:23:42 1998 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
1741 * config/m68k/m68k.c: Include "system.h" instead of <stdio.h>.
1743 (valid_dbcc_comparison_p): Mark mode argument as unused.
1744 (symbolic_operand): Likewise.
1745 (legitimize_pic_address): Likewise.
1746 (const_uint32_operand): Likewise.
1747 (const_sint32_operand): Likewise.
1748 * sched.c [!INSN_SCHEDULING]: Define only dummy function
1749 schedule_insns and comment out rest of file.
1751 * m68k.c (output_move_simode_const): Use subl to move a zero into an
1753 (output_move_[hq]imode): Likewise.
1755 Mon Aug 17 09:15:47 1998 Jeffrey A Law (law@cygnus.com)
1757 * toplev.c (main): Enable -fstrict-aliasing for -O2 and above.
1758 * invoke.texi: Corresponding changes.
1760 Mon Aug 17 02:03:55 1998 Richard Henderson <rth@cygnus.com>
1762 * regclass.c (allocate_reg_info): Respect MIN when clearing data.
1764 Sun Aug 16 17:37:06 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1766 * config/sparc/sparc.c (ultra_code_from_mask,
1767 ultra_cmove_results_ready_p, ultra_fpmode_conflict_exists,
1768 ultra_find_type, ultra_schedule_insn, ultra_flush_pipeline,
1769 ultrasparc_sched_init, ultrasparc_variable_issue,
1770 ultra_rescan_pipeline_state, ultrasparc_sched_reorder): New
1771 functions to describe UltraSPARC pipeline exactly to Haifa.
1772 (ultrasparc_adjust_cost): Indicate IMUL type insns have zero cost,
1773 as there is nothing the scheduler can do about it. Indicate that
1774 REG_DEP_OUTPUT's collide. Fixup formatting.
1775 * config/sparc/sparc.h (RTX_COSTS): Fixup integer multiply and
1776 divide costs on Ultra for DImode.
1777 (MD_SCHED_INIT, MD_SCHED_REORDER, MD_SCHED_VARIABLE_ISSUE):
1779 * config/sparc/sparc.md (ieu_unnamed function unit): Rename to
1780 ieuN and add call_no_delay_slot to type list.
1781 (cti function unit): New unit for branches on UltraSPARC.
1782 (subx/addx insns): Set type to misc.
1783 (sidi zero/sign extension insns on arch64): Set type to shift.
1784 (sign_extendhidi2_insn): Set type to sload.
1786 Sun Aug 16 13:52:00 1998 David Edelsohn <edelsohn@mhpcc.edu>
1788 * rs6000.c (rs6000_stack_info): Use if == 0 for sizes.
1789 (output_epilog): Use if != 0 for offset.
1790 (rs6000_fatal_bad_address): Prepare for Intl.
1791 * rs6000.h (rs6000_fatal_bad_address): Declare.
1792 * rs6000.md (movsfcc, movdfcc): Use else if.
1793 (elf_high): Use {liu|lis}.
1794 (elf_low): Use {cal|la}. Remove %a template from old mnemonics.
1795 (movsi): Use rs6000_fatal_bad_address.
1797 Sun Aug 16 01:53:21 1998 Richard Henderson <rth@cygnus.com>
1799 * reload.c (find_equiv_reg): Reject equivalences separated
1800 by a volatile instruction.
1802 Sun Aug 16 00:21:44 1998 Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
1804 * rs6000/linux.h (CPP_OS_DEFAULT_SPEC): Define.
1806 Sat Aug 15 20:51:35 1998 Richard Henderson <rth@cygnus.com>
1808 * alpha.md (movsicc): Fix mode mismatch.
1810 Sat Aug 15 20:22:33 1998 H.J. Lu (hjl@gnu.org)
1812 * config/alpha/alpha.h (ASM_OUTPUT_MI_THUNK): Handle aggregated
1814 * config/alpha/win-nt.h (ASM_OUTPUT_MI_THUNK): Likewise.
1816 Sat Aug 15 08:39:49 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1818 * config/sparc/sparc.md (movsi_lo_sum_pic_label_reg): Remove
1819 write-only modifier from operand 1 constraint.
1821 Sat Aug 15 06:28:19 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1823 * config/sparc/sparc.c (sparc_emit_set_const64_quick1): If
1824 emitting a XOR of -1 at the end, emit a NOT instead for combine's
1826 (sparc_emit_set_const64): Likewise, also when computing trailing
1827 bits do not negate low_bits and make fast_int an int.
1829 Fri Aug 14 21:07:03 1998 Jeffrey A Law (law@cygnus.com)
1831 * loop.c (add_label_notes): Do not ignore references to labels
1832 before dispatch tables. Mirrors Apr 8 change to mark_jump_label.
1833 * gcse.c (add_label_notes): Similarly.
1835 * pa.h (ASM_OUTPUT_MI_THUNK): Strip name encoding.
1837 * m68k.md (adddi_dilshr32): One of the operands must be a register.
1838 (adddi_dishl32): Similarly.
1840 Fri Aug 14 14:12:59 1998 Jason Merrill <jason@yorick.cygnus.com>
1842 * i386.h (MODES_TIEABLE_P): Reorganize to shut up warnings.
1843 * alias.c (memrefs_conflict_p): Add braces to shut up warnings.
1844 * cse.c (cse_basic_block): Add parens to shut up warnings.
1846 Fri Aug 14 12:58:21 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1848 * config/sparc/sparc.c (sparc_emit_set_const64_quick2,
1849 sparc_emit_set_const64_longway, const64_is_2insns,
1850 create_simple_focus_bits, sparc_emit_set_const64): Fix more bugs
1851 in 64-bit constant formation.
1852 * config/sparc/sparc.md (snesi_zero_extend split): Generate
1853 rtl for addx not subx.
1854 (define_insn movdi_const64_special): Make available even when
1855 HOST_BITS_PER_WIDE_INT is not 64.
1856 (movdi_lo_sum_sp64_cint, movdi_high_sp64_cint): Remove.
1857 (losum_di_medlow, sethm, setlo): Make op2 symbolic_operand.
1858 (cmp_siqi_trunc_set, cmp_diqi_trunc_set): Encapsulate both
1859 instances of operand 1 inside a QI subreg.
1860 (xordi3_sp64_dbl): Remove '%' constraint for op1.
1861 (one_cmpldi2_sp64): Fix output string.
1862 (one_cmplsi2_not_liveg0): Rewrite to remove unneeded extra
1864 (unnamed arch64 ashift DI): Truncate shift count if greater than
1867 Fri Aug 14 21:52:53 1998 J"orn Rennecke <amylaar@cygnus.co.uk>
1869 * expr.c (store_expr): Don't optimize away load-store pair
1870 when either source or destination have a side effect.
1872 Fri Aug 14 16:50:10 1998 John Carr <jfc@mit.edu>
1874 * genrecog.c (add_to_sequence): Fatal error if the modes of the
1875 operands of SET are incompatible.
1877 * alpha.md: Fix max and min patterns so modes of SET operands match.
1879 Fri Aug 14 12:22:55 1998 Ian Lance Taylor <ian@cygnus.com>
1881 * configure.in: Avoid [[ by using test and changequote.
1882 * configure: Rebuild.
1884 Fri Aug 14 01:22:31 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1886 * rtl.def (CONSTANT_P_RTX): Fix typo in string name.
1888 * config/sparc/sparc.md (seqdi_special_trunc, snedi_special_trunc,
1889 seqsi_special_extend, snesi_special_extend, snesi_zero_extend and
1890 split, snedi_zero_trunc and split, seqsi_zero_extend and split,
1891 seqdi_zero_trunc and split, pic_lo_sum_di, pic_sethi_di,
1892 movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc, addx_extend_sp32 and
1893 split, addx_extend_sp64, subx_extend_sp64, subx_extend and split):
1894 Fix mismatching modes in SET operands.
1895 (conditional move patterns): Fix formatting.
1896 (unnamed subx arch64 pattern): Remove duplicate insn.
1898 Fri Aug 14 00:34:34 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
1900 * config/sparc/sparc.c (const64_operand, const64_high_operand):
1901 Get it right when HOST_BITS_PER_WIDE_INT is not 64.
1902 (input_operand): Fixup test for what we accept for constant
1904 (sparc_emit_set_const32, sparc_emit_set_symbolic_const64): Give
1906 (safe_constDI): Remove.
1907 (sparc_emit_set_safe_HIGH64, gen_safe_SET64, gen_safe_OR64,
1908 gen_safe_XOR64): New functions.
1909 (sparc_emit_set_const64_quick1, sparc_emit_set_const64_quick2,
1910 sparc_emit_set_const64_longway, sparc_emit_set_const64): Use
1912 * config/sparc/sparc.md (define_insn xordi3_sp64_dbl): Only make
1913 available when HOST_BITS_PER_WIDE_INT is not 64.
1914 (define_insn movdi_sp64_dbl, movdi_const64_special): Likewise and
1915 move before movdi_insn_sp64 pattern.
1916 (define_insn movdi_lo_sum_sp64_dbl, movdi_high_sp64_dbl): Remove.
1917 (define_insn sethi_di_medlow, seth44, setm44, sethh): Use
1918 symbolic_operand as predicate for second operand.
1919 (DImode minus split on arch32, negsi2 expander, one_cmplsi2
1920 expander): Give set VOIDmode.
1922 Fri Aug 14 01:45:06 1998 Mumit Khan <khan@xraylith.wisc.edu>
1924 * i386/cygwin32 (DEFAULT_PCC_STRUCT_RETURN): Define.
1926 Fri Aug 14 01:40:21 1998 Geoffrey Keating <geoffk@ozemail.com.au>
1928 * rs6000/linux.h (LINK_SPEC): Pass -G args to the linker.
1930 Fri Aug 14 01:23:23 1998 Richard Earnshaw (rearnsha@arm.com)
1932 * arm/netbsd.h (TARGET_DEFAULT): Default includes software floating
1934 (CPP_FLOAT_DEFAULT_SPEC): Re-define accordingly.
1936 Fri Aug 14 01:19:08 1998 Robert Lipe <robertl@dgii.com>
1938 * install.texi: Various SCO OpenServer tweaks.
1940 Thu Aug 13 20:14:40 1998 Jim Wilson <wilson@cygnus.com>
1942 * reload1.c (eliminate_regs_in_insn): Handle another case when
1943 eliminating the frame pointer to the hard frame pointer. Add
1944 missing ep->to_rtx check to one existing case.
1946 * mips/mips.md (movhi_internal2+2): Fix typo mem:SI -> mem:HI.
1948 Thu Aug 13 17:08:11 1998 Jason Merrill <jason@yorick.cygnus.com>
1950 * tree.h: De-conditionalize init_priority code.
1952 * mips.h (NM_FLAGS): Change from -Bp to -Bn.
1953 * collect2.c (NM_FLAGS): Change from -p to -n.
1955 * configure.in: Turn on collect2 for mipstx39-elf.
1956 Handle use_collect2=no properly.
1958 * c-common.c: De-conditionalize init_priority code.
1959 * collect2.c (extract_init_priority, sort_ids): New fns.
1960 (main): Call sort_ids.
1961 Move sequence_number to file scope.
1963 * configure.in: Handle --enable-init-priority.
1964 * c-common.c (attrs): Add A_INIT_PRIORITY.
1965 (init_attributes, decl_attributes): Likewise.
1966 * tree.h (DEFAULT_INIT_PRIORITY, MAX_INIT_PRIORITY): New macros.
1967 * tree.c (get_file_function_name_long): Split out...
1968 (get_file_function_name): ...from here.
1970 Thu Aug 13 16:09:53 1998 Martin von Loewis <loewis@informatik.hu-berlin.de>
1972 * expr.c (safe_from_p): Change code to ERROR_MARK only when not
1975 Thu Aug 13 15:24:48 1998 Jason Merrill <jason@yorick.cygnus.com>
1977 * toplev.c (display_help): Add braces to shut up warnings.
1978 * tree.c (simple_cst_equal): Likewise.
1980 * fold-const.c (non_lvalue): Don't deal with null pointer
1982 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
1984 * c-typeck.c (initializer_constant_valid_p): Allow conversion of 0
1985 of any size to a pointer.
1987 Thu Aug 13 12:53:13 1998 Jim Wilson <wilson@cygnus.com>
1989 * i386/winnt.c (i386_pe_asm_file_end): Check TREE_SYMBOL_REFERENCED.
1991 Wed Aug 12 17:25:18 1998 Jeffrey A Law (law@cygnus.com)
1993 * mn10300.c (REG_SAVE_BYTES): Only reserve space for registers
1994 which will be saved.
1995 * mn10300.md (prologue insn): Only save registers which need saving.
1996 (epilogue insn): Similarly.
1998 * mn10300.c, mn10300.h, mn10300.md: Remove "global zero register"
2001 Wed Aug 12 12:39:16 1998 Gavin Romig-Koch <gavin@cygnus.com>
2003 * mips/mips.h (ENCODE_SECTION_INFO): Set SYMBOL_REF_FLAG for
2004 VAR_DECL's in gp addressable sections.
2006 Tue Aug 11 23:02:31 1998 John Carr <jfc@mit.edu>
2008 * sparc.c: Change return <exp> to <exp>; return; in functions
2010 * sparc.md: Add empty semicolon statement after final label in
2013 Tue Aug 11 22:42:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2015 * config/sparc/sparc.md (define_insn addx_extend): Rename to
2016 addx_extend_sp64, only allow when TARGET_ARCH64.
2017 (define_insn addx_extend_sp32 and split): Version that works when
2019 (define_insn subx_extend): Likewise.
2020 (define_split adddi3 and subdi3 with zero extension): Fixup and
2021 correct bugs when not TARGET_ARCH64.
2023 Tue Aug 11 16:04:34 1998 John Carr <jfc@mit.edu>
2025 * except.c (set_exception_lang_code, set_exception_version_code):
2026 Use prototype-style definition if __STDC__, to match declaration
2029 * genemit.c: Change FAIL and DONE macros not to use loops.
2031 Tue Aug 11 12:27:03 1998 Jim Wilson <wilson@cygnus.com>
2033 * dwarf2out.c (ASM_OUTPUT_DWARF_ADDR_CONST): Use
2034 ASM_OUTPUT_DWARF2_ADDR_CONST if defined.
2036 * mips/mips.md (reload_outsi): Use M16_REG_P when TARGET_MIPS16.
2038 Tue Aug 11 18:12:53 1998 Dave Love <d.love@dl.ac.uk>
2040 * README.g77: Update from Craig.
2042 Tue Aug 11 04:46:01 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2044 * config/sparc/sparc.c (sparc_emit_set_const32): INTVAL is of
2046 (safe_constDI sparc_emit_set_const64_quick1,
2047 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
2048 analyze_64bit_constant, const64_is_2insns,
2049 create_simple_focus_bits): Fix some bugs when compiled on real
2051 (function_arg_record_value_3, function_arg_record_value_2,
2052 function_arg_record_value): Add fully prototyped forward decls.
2053 * config/sparc/sparc.md (define_insn cmpsi_insn_sp32): Rename back
2054 to cmpsi_insn and use on both 64 and 32 bit targets.
2055 (define_insn cmpsi_insn_sp64): Remove.
2056 (define_expand zero_extendsidi2): Allow for 32-bit target too.
2057 (define_insn zero_extendsidi2_insn): Rename to
2058 zero_extendsidi2_insn_sp64.
2059 (define_insn zero_extendsidi2_insn_sp32): New pattern and
2060 assosciated forced split for it.
2062 * config/sparc/sparc.c (const64_operand, const64_high_operand):
2064 * config/sparc/sparc.h: Declare them.
2065 (PREDICATE_CODES): Add them.
2066 * config/sparc/sparc.md (movdi_lo_sum_sp64_dbl,
2067 movdi_high_sp64_dbl, xordi3_sp64_dbl): Use them.
2069 Mon Aug 10 22:57:24 1998 John Carr <jfc@mit.edu>
2071 * config/sparc/sparc.md (define_insn jump): Output ba,pt not b,pt
2072 in v9 case as the latter makes the Solaris assembler crash.
2074 Mon Aug 10 22:39:09 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2076 * config/sparc/sparc.c (input_operand): Do not accept a LO_SUM MEM
2077 for TFmode when !v9. We require offsettable memory addresses.
2078 * config/sparc/sparc.h (ALTER_HARD_SUBREG): Handle TFmode to
2079 DFmode register number conversions.
2080 * config/sparc/sparc.md (define_split DFmode moves): If register
2081 is a SUBREG do alter_subreg on it before using.
2082 (define_expand movtf): Fixup comment about alignment on v9.
2083 (define_split TFmode moves): Don't use gen_{high,low}part, create
2084 explicit SUBREGs instead.
2086 Mon Aug 10 19:02:55 1998 John Carr <jfc@mit.edu>
2088 * Makefile.in (mbchar.o): Depend on mbchar.c.
2090 Mon Aug 10 04:28:13 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2091 Richard Henderson <rth@cygnus.com>
2093 Rewrite Sparc backend for better code generation and
2094 improved sparc64 support.
2095 * config/sparc/sp64-elf.h: Set JUMP_TABLES_IN_TEXT_SECTION to
2097 * config/sparc/sysv4.h: Likewise.
2098 * config/sparc/sparc.c (v8plus_regcmp_p, sparc_operand,
2099 move_operand, v8plus_regcmp_op, emit_move_sequence,
2100 singlemove_string, doublemove_string, mem_aligned_8,
2101 output_move_double, output_move_quad, output_fp_move_double,
2102 move_quad_direction, output_fp_move_quad, output_scc_insn):
2104 (small_int_or_double): New predicate.
2105 (gen_compare_reg): Remove TARGET_V8PLUS cmpdi_v8plus emission.
2106 (legitimize_pic_address): Emit movsi_{high,lo_sum}_pic instead of
2107 old pic_{sethi,lo_sum}_si patterns.
2108 (mem_min_alignment): New generic function to replace
2109 mem_aligned_8, which uses REGNO_POINTER_ALIGN information when
2110 available and can test for arbitrary alignments. All callers
2112 (save_regs, restore_regs, build_big_number,
2113 output_function_prologue, output_cbranch, output_return,
2114 sparc_flat_save_restore, sparc_flat_output_function_prologue,
2115 sparc_flat_output_function_epilogue): Prettify
2117 (output_function_epilogue): Likewise and add code to output
2118 deferred case vectors.
2119 (output_v9branch): Likewise, add new arg INSN and use it to tack
2120 on branch prediction settings. All callers changed.
2121 (print_operand): Likewise and output %l44 for LO_SUMs when
2123 (sparc_splitdi_legitimate): New function to make sure DImode
2124 splits can be run properly when !arch64.
2125 (sparc_initialize_trampoline, sparc64_initialize_trampoline):
2126 Reformat example code in comments.
2127 (set_extends): Remove UNSPEC/v8plus_clear_high case.
2128 (sparc_addr_diff_list, sparc_addr_list): New statics to keep track
2129 of deferred case vectors we need to output.
2130 (sparc_defer_case_vector): Record a case vector.
2131 (sparc_output_addr_vec, sparc_output_addr_diff_vec,
2132 sparc_output_deferred_case_vectors): New functions to output them.
2133 (sparc_emit_set_const32): New function to form 32-bit constants in
2134 registers when that requires more than one instruction.
2135 (safe_constDI, sparc_emit_set_const64_quick1,
2136 sparc_emit_set_const64_quick2, sparc_emit_set_const64_longway,
2137 analyze_64bit_constant, const64_is_2insns,
2138 create_simple_focus_bits, sparc_emit_set_const64): New functions
2139 which do the same for 64-bit constants when arch64.
2140 (sparc_emit_set_symbolic_const64): New function to emit address
2141 loading for all code models on v9.
2142 * config/sparc/sparc.h (CONDITIONAL_REGISTER_USAGE): Do not make
2143 %g1 fixed when arch64, unfix %g0 when TARGET_LIVE_G0.
2144 (ALTER_HARD_SUBREG): Fix thinko, return REGNO + 1 not 1.
2145 (SECONDARY_INPUT_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS): Fix
2146 inaccuracies in comments, add symbolic and text_segment operands
2147 when TARGET_CM_MEDANY and TARGET_CM_EMBMEDANY respectively. Use
2148 GENERAL_REGS in these cases as a temp REG is needed to load these
2149 addresses into a register properly.
2150 (EXTRA_CONSTRAINT): Document more accurately, remove Q case as it
2152 (GO_IF_LEGITIMATE_ADDRESS): Allow TFmode for LO_SUM on v9 since fp
2153 quads are guarenteed to have 16-byte alignment.
2154 (LEGITIMIZE_ADDRESS): For SYMBOL_REF, CONST, and LABEL_REF use
2155 copy_to_suggested_reg instead of explicit LO_SUM and HIGH.
2156 (ASM_OUTPUT_ADDR_VEC, ASM_OUTPUT_ADDR_DIFF_VEC): New macros for
2157 deferred case vector implementation.
2158 (ASM_OUTPUT_ADDR_VEC_ELT): Use fputc to output newline.
2159 (ASM_OUTPUT_ADDR_DIFF_ELT): Parenthesize LABEL in macro calls.
2160 Generate "internal label - label" instead of "label - 1b".
2161 (PRINT_OPERAND_ADDRESS): For LO_SUM use %l44 on TARGET_CM_MEDMID.
2162 (PREDICATE_CODES): Remove sparc_operand, move_operand,
2163 v8plus_regcmp_op. Add small_int_or_double, input_operand, and
2165 (doublemove_string, output_block_move, output_fp_move_double,
2166 output_fp_move_quad, output_move_double, output_move_quad,
2167 output_scc_insn, singlemove_string, mem_aligned_8, move_operand,
2168 sparc_operand, v8plus_regcmp_op, v8plus_regcmp_p): Remove externs.
2169 (sparc_emit_set_const32, sparc_emit_set_const64,
2170 sparc_emit_set_symbolic_const64, input_operand, zero_operand,
2171 mem_min_alignment, small_int_or_double): Add externs.
2172 * config/sparc/sparc.md: Document the many uses of UNSPEC and
2173 UNSPEC_VOLATILE in this backend.
2174 (define_function_unit ieu): Rename to ieu_unnamed. Add move and
2175 unary to types which execute in it.
2176 (define_function_unit ieu_shift): Rename to ieu0.
2177 (define_function_unit ieu1): New, executes compare, call, and
2178 uncond_branch type insns.
2179 (define_function_units for type fdivs, fdivd, fsqrt): These
2180 execute in the fpu multiply unit not the adder on UltraSparc.
2181 (define_expand cmpdi): Disallow TARGET_V8PLUS.
2182 (define_insn cmpsi_insn): Rename to cmpsi_insn_sp32.
2183 (define_insn cmpsi_insn_sp64): New, same as sp32 variant except it
2184 allows the arith_double_operand predicate and rHI constraint when
2186 (define_insn cmpdi_sp64, cmpsf_fpe, cmpdf_fpe, cmptf_fpe,
2187 cmpsf_fp, cmpdf_fp, cmptf_fp, sltu_insn, neg_sltu_insn,
2188 neg_sltu_minux_x, neg_sltu_plus_x, sgeu_insn, neg_sgeu_insn,
2189 sltu_plus_x, sltu_plus_x, sltu_plus_x_plus_y, x_minus_sltu,
2190 sgeu_plus_x, x_minus_sgeu, movqi_cc_sp64, movhi_cc_sp64,
2191 movsi_cc_sp64, movdi_cc_sp64, movsf_cc_sp64, movdf_cc_sp64,
2192 movtf_cc_sp64, movqi_cc_reg_sp64, movhi_cc_reg_sp64,
2193 movsi_cc_reg_sp64, movdi_cc_reg_sp64, movsf_cc_reg_sp64,
2194 movdf_cc_reg_sp64, movtf_cc_reg_sp64, zero_extendhisi2_insn,
2195 cmp_siqi_trunc, cmp_siqi_trunc_set, sign_extendhisi2_insn,
2196 sign_extendqihi2_insn, sign_extendqisi2_insn,
2197 sign_extendqidi2_insn, sign_extendhidi2_insn,
2198 extendsfdf2, extendsftf2, extenddftf2, truncdfsf2, trunctfsf2,
2199 trunctfdf2, floatsisf2, floatsidf2, floatsitf2, floatdisf2,
2200 floatdidf2, floatditf2, fix_truncsfsi2, fix_truncdfsi2,
2201 fix_trunctfsi2, fix_truncsfdi2, fix_truncdfdi2, fix_trunctfdi2,
2202 adddi3_sp64, addsi3, cmp_ccx_plus, cmp_cc_plus_set, subdi_sp64,
2203 subsi3, cmp_minus_ccx, cmp_minus_ccx_set, mulsi3, muldi3,
2204 muldi3_v8plus, cmp_mul_set, mulsidi3, mulsidi3_v8plus,
2205 const_mulsidi3_v8plus, mulsidi3_sp32, const_mulsidi3,
2206 smulsi3_highpart_v8plus, unnamed subreg mult,
2207 const_smulsi3_highpart_v8plus, smulsi3_highpart_sp32,
2208 const_smulsi3_highpart, umulsidi3_v8plus, umulsidi3_sp32,
2209 const_umulsidi3, const_umulsidi3_v8plus, umulsi3_highpart_v8plus,
2210 const_umulsi3_highpart_v8plus, umulsi3_highpart_sp32,
2211 const_umulsi3_highpart, divsi3, divdi3, cmp_sdiv_cc_set, udivsi3,
2212 udivdi3, cmp_udiv_cc_set, smacsi, smacdi, umacdi, anddi3_sp64,
2213 andsi3, and_not_di_sp64, and_not_si, iordi3_sp64, iorsi3,
2214 or_not_di_sp64, or_not_si, xordi3_sp64, xorsi3, xor_not_di_sp64,
2215 xor_not_si, cmp_cc_arith_op, cmp_ccx_arith_op,
2216 cmp_cc_arith_op_set, cmp_ccx_arith_op_set, cmp_ccx_xor_not,
2217 cmp_cc_xor_not_set, cmp_ccx_xor_not_set, cmp_cc_arith_op_not,
2218 cmp_ccx_arith_op_not, cmp_cc_arith_op_not_set,
2219 cmp_ccx_arith_op_not_set, negdi2_sp64, cmp_cc_neg, cmp_ccx_neg,
2220 cmp_cc_set_neg, cmp_ccx_set_neg, one_cmpldi2_sp64, cmp_cc_not,
2221 cmp_ccx_not, cmp_cc_set_not, cmp_ccx_set_not, addtf3, adddf3,
2222 addsf3, subtf3, subdf3, subsf3, multf3, muldf3, mulsf3,
2223 muldf3_extend, multf3_extend, divtf3, divdf3, divsf3, negtf2,
2224 negdf2, negsf2, abstf2, absdf2, abssf2, sqrttf2, sqrtdf2, sqrtsf2,
2225 ashlsi3, ashldi3, unnamed DI ashift, cmp_cc_ashift_1,
2226 cmp_cc_set_ashift_1, ashrsi3, ashrdi3, unnamed DI ashiftrt,
2227 ashrdi3_v8plus, lshrsi3, lshrdi3, unnamed DI lshiftrt,
2228 lshrdi3_v8plus, tablejump_sp32, tablejump_sp64, call_address_sp32,
2229 call_symbolic_sp32, call_address_sp64, call_symbolic_sp64,
2230 call_address_struct_value_sp32, call_symbolic_struct_value_sp32,
2231 call_address_untyped_struct_value_sp32,
2232 call_symbolic_untyped_struct_value_sp32, call_value_address_sp32,
2233 call_value_symbolic_sp32, call_value_address_sp64,
2234 call_value_symbolic_sp64, branch_sp32, branch_sp64,
2235 flush_register_windows, goto_handler_and_restore,
2236 goto_handler_and_restore_v9, goto_handler_and_restore_v9_sp64,
2237 flush, all ldd/std peepholes, return_qi, return_hi, return_si,
2238 return_addsi, return_di, return_adddi, return_sf, all call+jump
2239 peepholes, trap, unnamed trap insns): Prettify output strings.
2240 (define_insn anddi3_sp32, and_not_di_sp32, iordi3_sp32,
2241 or_not_di_sp32, xordi3_sp32, xor_not_di_sp32, one_cmpldi2):
2242 Likewise and force + implement splits for integer cases.
2243 (define_insn return_sf_no_fpu): Likewise and allow to match when
2244 no-fpu because of our subreg SFmode splits.
2245 (define_insn zero_extendqihi2, zero_extendqisi2_insn,
2246 zero_extendqidi2_insn, zero_extendhidi2_insn,
2247 zero_extendsidi2_insn, sign_extendsidi2_insn): Likewise and use
2248 input_operand for second operand.
2249 (cmp_minus_cc, cmp_minus_cc_set): Likewise and use
2250 reg_or_0_operand for operand 2 so new splits can use it.
2251 (cmp_zero_extendqisi2, cmp_zero_extendqisi2_set, cmp_cc_plus,
2252 cmp_cc_xor_not): Likewise and don't forget to check TARGET_LIVE_G0
2254 (cmp_zero_extract, cmp_zero_extract_sp64): Likewise and allow
2255 CONST_DOUBLEs for operand 2.
2256 (define_insn move_label_di): Likewise and label distance
2257 optimization because it no longer works with new deferred case
2258 vector scheme. To be revisited.
2259 (define_insn x_minus_y_minus_sltu, x_minus_sltu_plus_y): Likewise
2260 and allow reg_or_0_operand and J constraint for second operand.
2261 (define_insn jump): Set branch predict taken on V9.
2262 (define_insn tablejump): Emit LABEL_REF + PLUS memory address for
2263 new deferred case vector scheme.
2264 (define_insn pic_tablejump_32, pic_tablejump_64): Remove.
2265 (define_insn negdi2_sp32): Force + implement splits.
2266 (define_insn negsi2, one_cmplsi2): Rename to negsi2_not_liveg0 and
2267 one_cmplsi2_not_liveg0 respectively, and create expander of original
2268 names which emit special rtl for TARGET_LIVE_G0.
2269 (define_insn cmpdi_v8plus, scc_si, scc_di): Remove.
2270 (define_insn seq, sne, slt, sge, sle, sltu, sgeu): Don't do
2271 gen_compare_reg, FAIL instead.
2272 (define_insn sgtu, sleu): Likewise and check gen_s*() return
2273 values when trying to reverse condition codes, if they FAIL then
2275 (define_insn snesi_zero, neg_snesi_zero, snesi_zero_extend,
2276 snedi_zero, neg_snedi_zero, snedi_zero_trunc, seqsi_zero,
2277 neg_seqsi_zero, seqsi_zero_extend, seqdi_zero, neg_seqdi_zero,
2278 seqdi_zero_trunc, x_plus_i_ne_0, x_minus_i_ne_0, x_plus_i_eq_0,
2279 x_minus_i_eq_0): Add new splits to perform these multi-insn cases,
2280 set output string to # to indicate they are mandatory splits.
2281 (define_insn pic_lo_sum_si, pic_sethi_si, pic_lo_sum_di,
2282 pic_sethi_di, move_pic_label_si): Remove.
2283 (define_insn movsi_low_sum, movsi_high, movsi_lo_sum_pic,
2284 movsi_high_pic, movsi_pic_label_reg): New patterns to take their
2286 (define_expand movsi_pic_label_ref, define_insn
2287 movsi_high_pic_label_ref, movsi_lo_sum_pic_label_ref): New
2288 expander and insns to handle PIC label references and deferred
2290 (define_insn get_pc_via_rdpc): Comment out as it is no longer
2292 (define_expand movqi, movhi, movsi, movdi, movsf, movdf, movtf):
2293 Rewrite to not use emit_move_sequence, make use of new constant
2294 formation code, and new splits for all multi-insn cases.
2295 (define_insn movqi_insn): Remove sethi case, it can never happen.
2296 Use reg_or_zero_operand instead of const0_rtx explicit test,
2297 use input_operand instead of move_operand for source, and use
2298 general_operand now for dest.
2299 (define_insn movhi_insn): Similar but leave sethi case.
2300 (define_insn lo_sum_qi, store_qi, store_hi): Remove.
2301 (define_insn sethi_hi lo_sum_hi): Rename to movhi_high and
2302 movhi_lo_sum respectively, prettify output string.
2303 (define_insn movsi_zero_liveg0): New pattern to put zero into a
2304 register when needed on TARGET_LIVE_G0.
2305 (define_insn movsi_insn): Use general_operand and input_operand
2306 for dest and src respectively. Simplify applicability test.
2307 Prettify output strings, and add clr alternative for J
2309 (define_insn movdi_sp32_v9, movdi_sp32, define_splits for
2310 deprecated std and reg-reg DI moves): Remove and...
2311 (define_insn movdi_insn_sp32, movdi_insn_sp64): Replace with new
2312 implementation which uses forced splits for all non-single insn
2314 (define_split DI move cases on !arch64): New splits to handle all
2315 situations of 64-bit double register DImode on 32bit, and
2316 unaligned registers and memory addresses for all subtargets.
2317 (define_insn movsf_const_insn, movdf_const_insn, store_sf):
2319 (define_insn movsf_insn, movsf_no_f_insn): Use general_operand and
2320 input_operand for dest and src respectively, prettify output
2322 (define_insn movdf_insn, movdf_no_e_insn, store_df,
2323 movtf_const_insn, movtf_insn, movtf_no_e_insn, store_tf): Remove
2325 (define_insn movdf_insn_sp32, movdf_no_e_insn_sp32,
2326 movdf_insn_sp64, movdf_no_e_insn_sp64, movtf_insn,
2327 movtf_no_e_insn_sp32, movtf_insn_hq_sp64, movtf_insn_sp64,
2328 movtf_no_e_insn_sp64) Replace with new
2329 implementation which uses forced splits for all non-single insn
2331 (define_split DF move cases): New splits in similar vein to DI
2333 (define_insn sethi_di_medlow, sethi_di_medium_pic,
2334 sethi_di_embmedany_data, sethi_di_embmedany_text, sethi_di_sp64,
2335 movdi_sp64_insn): Remove old v9 code model and constant loading
2337 (define_insn pic_lo_sum_di, pic_sethi_di,
2338 sethi_di_medlow_embmedany_pic, sethi_di_medlow, losum_di_medlow,
2339 seth44, setm44, setl44, sethh, setlm, sethm, setlo,
2340 embmedany_sethi, embmedany_losum, embmedany_brsum,
2341 embmedany_textuhi, embmedany_texthi, embmedany_textulo,
2342 embmedany_textlo, movdi_lo_sum_sp64_cint, movdi_lo_sum_sp64_dbl,
2343 movdi_high_sp64_cint, movdi_high_sp64_dbl): Replace with new
2344 scheme, using unspecs, secondary reloads, and one to one sparc
2345 insn to rtl insn mapping for better scheduling and code gen.
2346 (define_expand reload_indi, reload_outdi): Reload helpers for
2347 MEDANY and EMBMEDANY symbol address loading cases which require a
2349 (define_expand movsicc): Remove v8plus_regcmp cases.
2350 (define_insn movdi_cc_sp64_trunc, movdi_cc_reg_sp64_trunc,
2351 cmp_zero_extendqidi2, cmp_zero_extendqidi2_set, cmp_qidi_trunc,
2352 cmp_diqi_trunc_set): New patterns used by some of the new scc
2354 (define_insn xordi3_sp64_dbl): New pattern used for constant
2355 formation when crossing from 32-bit targets.
2356 (define_insn movsi_cc_reg_v8plus, v8plus_clear_high, and helper
2358 (define_insn addx, subx): Make visible and prettify.
2359 (define_insn adddi3_insn_sp32): Likewise and force split.
2360 (define_insn addx_extend, subx_extend, unnamed): New patterns for
2361 64bit scc split usage.
2362 (define_insn unnamed plusDI zero_extend, unnamed minusDI
2363 zero_extend, subdi3): Force and implement splits.
2365 * final.c (final_scan_insn): Don't output labels if target
2366 specifies ASM_OUTPUT_ADDR_{DIFF}_VEC. Do these macro operations
2369 * reorg.c (dbr_schedule): When taking on BR_PRED notes at the end,
2370 don't forget to walk inside SEQUENCESs too as these are what the
2371 delay slot scheduler will create.
2373 Mon Aug 10 01:21:01 1998 Richard Henderson <rth@cygnus.com>
2375 * alpha.md (extxl+1,+2): New patterns to work around
2378 Sat Aug 8 19:20:22 1998 Gary Thomas (gdt@linuxppc.org)
2380 * rs6000.c (rs6000_allocate_stack_space) Fix typo which
2381 caused bad assembly code to be generated.
2383 Sat Aug 8 18:53:28 1998 Jeffrey A Law (law@cygnus.com)
2385 * netbsd.h: Fix typo.
2387 Mon Aug 3 00:06:42 1998 Robert Lipe <robertl@dgii.com>
2389 * config.sub: Fix typo.
2391 Sun Aug 2 22:39:08 1998 Hans-Peter Nilsson <hp@axis.se>
2393 * invoke.texi (Environment Variables): Typo: Change "ascpects"
2395 (Running Protoize): Typo: Change "ther" into "other".
2397 Sun Aug 2 00:42:50 1998 Jeffrey A Law (law@cygnus.com)
2399 * i386/netbsd.h: Undo previous change to DWARF2_UNWIND_INFO.
2400 * m68k/netbsd.h: Likewise.
2401 * ns32k/netbsd.h: Likewise.
2402 * sparc/netbsd.h: Likewise.
2404 Sat Aug 1 17:59:30 1998 Richard Henderson <rth@cygnus.com>
2406 * ginclude/va-alpha.h (va_list): Use a typedef, not a define.
2407 * ginclude/va-clipper.h (va_list): Likewise.
2409 Fri Jul 31 20:22:02 1998 Michael Meissner <meissner@cygnus.com>
2411 * rs6000.c (rs6000_override_options): If big endian and -Os, use
2412 load/store multiple instructions unless user overrides.
2414 Fri Jul 31 17:08:59 1998 Jeffrey A Law (law@cygnus.com)
2416 * ns32k/netbsd.h: Fix typo.
2418 Fri Jul 31 10:23:55 1998 Doug Evans <devans@canuck.cygnus.com>
2420 * m32r/m32r.h (ASM_OUTPUT_SOURCE_LINE): Always output line number
2421 labels with .debugsym if no parallel insns.
2423 Thu Jul 30 19:15:53 1998 Richard Henderson <rth@cygnus.com>
2425 * alpha.md (fp cmp): Replicate patterns for ALPHA_TP_INSN.
2426 (fcmov): Remove ALPHA_TP_INSN patterns -- fcmov doesn't trap.
2428 Thu Jul 30 19:50:15 1998 David Edelsohn <edelsohn@mhpcc.edu>
2430 * rs6000/x-aix43 (AR_FOR_TARGET_FLAGS): Delete.
2431 (AR_FOR_TARGET): Define.
2433 Thu Jul 30 12:29:12 1998 Mark Mitchell <mark@markmitchell.com>
2435 * dyn-string.h: New file.
2436 * dyn-string.c: Likewise.
2437 * Makefile.in (OBJS): Add dyn-string.o.
2438 (dwarf2out.o): Add dyn-string.h dependency.
2439 (dyn-string.o): List dependencies.
2440 * dwarf2out.c: Include dyn-string.h.
2441 (ASM_NAME_TO_STRING): Use dyn_string_append, rather than strcpy.
2442 (addr_const_to_string): Take a dyn_string_t, not a char* as a
2443 prototype. Use dyn_string_append rather than strcat, throughout.
2444 (addr_to_string): Use dyn_string_t.
2446 Thu Jul 30 13:08:07 1998 Ken Raeburn <raeburn@cygnus.com>
2448 Function entry/exit profiling instrumentation:
2449 * expr.h (profile_function_entry_libfunc,
2450 profile_function_exit_libfunc): Declare new variables.
2451 * optabs.c: Define them here.
2452 (init_optabs): Initialize them.
2453 * tree.h (struct tree_decl): New flag
2454 no_instrument_function_entry_exit.
2455 (DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT): New accessor macro.
2456 * c-decl.c (duplicate_decls): Merge it.
2457 * c-common.c (enum attrs): New value A_NO_INSTRUMENT_FUNCTION.
2458 (init_attributes): Use it for "no_instrument_function".
2459 (decl_attributes): Handle it, for functions that have not yet been
2460 compiled. Set decl flag.
2461 * flags.h (flag_instrument_function_entry_exit): Declare new
2463 * toplev.c (flag_instrument_function_entry_exit): Define it here.
2464 (f_options): New option "instrument-functions".
2465 * function.h (struct function): New field instrument_entry_exit.
2466 * function.c (current_function_instrument_entry_exit): New
2468 (push_function_context_to, pop_function_context_from): Save and
2470 (expand_function_start): Set current_ variable, maybe emit return
2471 label and entry profile call.
2472 (expand_function_end): Maybe emit exit profile call.
2474 Thu Jul 30 00:58:34 1998 Jeffrey A Law (law@cygnus.com)
2476 * i386.md (movqi): When optimizing a load of (const_int 1) into a
2477 NON_QI_REG_P, pretend the register is SImode.
2479 Wed Jul 29 23:49:23 1998 Todd Vierling <tv@netbsd.org>
2481 * configure.in: Use xm-netbsd.h as the NetBSD xm file (not xm-siglist).
2482 Accept arm32 as arm, m68k4k as m68k, mipsle as mips-dec, and any
2483 manufacturer id for ns32k.
2484 * configure: Regenerated.
2485 * config/netbsd.h: When using ASM_WEAKEN_LABEL, make it global too.
2486 * config/t-netbsd: Don't compile libgcc1-test as the fns are in libc.
2487 * config/i386/netbsd.h: Undefine DWARF2_UNWIND_INFO, not define as 0.
2488 * config/m68k/netbsd.h: Same.
2489 * config/ns32k/netbsd.h: Same.
2490 * config/sparc/netbsd.h: Same.
2492 Wed Jul 29 22:39:21 1998 Jeffrey A Law (law@cygnus.com)
2494 * unroll.c (unroll_loop): Do not abort for an UNROLL_MODULO
2495 or UNROLL_COMPLETELY loop that starts with a jump to its
2498 Wed Jul 29 22:18:14 1998 David Edelsohn <edelsohn@mhpcc.edu>
2500 * rs6000/rs6000.md (absdi2 define_split): Swap operands of MINUS.
2501 * rs6000/rs6000.c (mask64_operand): Use HOST_BITS_PER_WIDE_INT.
2502 (print_operand, case 'B'): Don't fall through.
2503 (print_operand, case 'S'): Correct mask begin/end computation.
2504 Use HOST_BITS_PER_WIDE_INT.
2505 * rs6000/rs6000.h (CPP_PREDEFINES): Define _LONG_LONG.
2506 (CONDITIONAL_REGISTER_USAGE): GPR13 fixed if TARGET_64BIT.
2507 * rs6000/aix41.h (CPP_PREDEFINES): Same.
2508 * rs6000/aix43.h (CPP_PREDEFINES): Same.
2510 Wed Jul 29 11:47:10 1998 Nick Clifton <nickc@cygnus.com>
2512 * config/arm/thumb.md (extendqisi2_insn): Remove earlyclobber
2513 constraint from second alternative.
2515 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
2517 * configure.in: Fix --without/--disable cases for local-prefix,
2518 gxx-include-dir and checking.
2520 Tue Jul 28 22:01:23 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2522 * configure.in (enable_haifa): Set by default for sparc64 too.
2525 Tue Jul 28 23:29:04 1998 Jason Merrill <jason@yorick.cygnus.com>
2527 * i386/cygwin32.h (VALID_MACHINE_TYPE_ATTRIBUTE): New macro.
2528 * i386/winnt.c (associated_type): New fn.
2529 (i386_pe_valid_type_attribute_p): New fn.
2530 (i386_pe_check_vtable_importexport): Remove.
2531 (i386_pe_dllexport_p): Use associated_type.
2532 (i386_pe_dllimport_p): Likewise.
2534 From Antonio M. O. Neto <anmendes@cruzeironet.com.br>:
2535 * i386.c (i386_valid_type_attribute_p): Also accept
2536 attributes for METHOD_TYPEs.
2538 Tue Jul 28 23:17:39 1998 Peter Gerwinski <peter@gerwinski.de>
2540 * tree.c (build_range_type): Copy TYPE_SIZE_UNIT.
2542 Tue Jul 28 22:31:12 1998 Craig Burley <burley@gnu.org>
2544 * gcc.c: Fix commentary describing %g, %u, %U, and %O.
2546 * gcc.c (do_spec_1): Fix handling of %g%O and %U%O to prevent
2547 them from generating a new base name for each occurence of
2550 1998-07-28 Vladimir N. Makarov <vmakarov@cygnus.com>
2552 * cse.c (cse_insn): Enable subsitution inside libcall only for REG,
2554 * rtlanal.c (replace_rtx): Prohibit replaces in CONST_DOUBLE.
2558 * cplus-dem.c (type_kind_t): New type.
2559 (demangle_template_value_parm): Add type_kind_t parameter. Rely
2560 on this paramter, rather than demangling the type again.
2561 (demangle_integral_value): Pass tk_integral.
2562 (demangle_template_: Pass the value returned from do_type.
2563 (do_type): Return a type_kind_t. Pass tk_integral to
2564 demangle_template_value_parm for array bounds.
2565 (demangle_fund_type): Likewise.
2567 Mon Jul 27 00:54:41 1998 Jason Merrill <jason@yorick.cygnus.com>
2569 * tree.c (simple_cst_equal, case CONSTRUCTOR): OK if the elts are
2572 Mon Jul 27 22:18:36 1998 Jeffrey A Law (law@cygnus.com)
2574 * pa.c (move_operand): Accept CONSTANT_P_RTX.
2576 Mon Jul 27 17:18:52 1998 Dave Brolley <brolley@cygnus.com>
2578 * stor-layout.c (layout_type): Handle arrays of bits, for Chill.
2580 * expr.c (get_inner_reference): Handle zero-based, unsigned, array
2583 Mon Jul 27 14:51:33 1998 Jeffrey A Law (law@cygnus.com)
2585 * mn10300.h (DEBUGGER_AUTO_OFFSET): Define.
2586 (DEBUGGER_ARG_OFFSET): Likewise.
2588 * mn10300.md (movsf): Remove last change. Not needed.
2590 Mon Jul 27 14:22:36 1998 Dave Brolley <brolley@cygnus.com>
2592 * c-lex.c (yylex): Fix boundary conditions in character literal and
2593 string literal loops.
2595 Mon Jul 27 11:43:54 1998 Stan Cox <scox@cygnus.com>
2597 * longlong.h (count_leading_zeros): Sparclite scan instruction was
2598 being invoked incorrectly.
2600 * i386.c (ix86_prologue): Added SUBTARGET_PROLOGUE invocation.
2601 * i386/cygwin32.h (STARTFILE_SPEC, LIB_SPEC, SUBTARGET_PROLOGUE):
2603 * i386/win32.h: New file. Hybrid mingw32.h/cygwin32.h configuration.
2604 * configure.in: Added i[34567]86-*-win32
2605 * config.sub: Ditto.
2606 * configure: Rebuilt.
2608 Sun Jul 26 01:11:12 1998 H.J. Lu (hjl@gnu.org)
2610 * i386.h (CONST_DOUBLE_OK_FOR_LETTER_P): Return 0 when eliminating
2611 the frame pointer and compiling PIC code and reload has not completed.
2613 * i386.c (output_to_reg): Add code to emulate non-popping DImode
2616 Sun Jul 26 01:01:32 1998 Jeffrey A Law (law@cygnus.com)
2618 * regmove.c (regmove_optimize): Fix typo initializing regmove_bb_head.
2620 Sat Jul 25 23:29:23 1998 Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>
2622 * Makefile.in (install-info): Only try to update the info
2623 directory file if it exists in the first place.
2625 Fri Jul 24 18:58:37 1998 Klaus Espenlaub <kespenla@student.informatik.uni-ulm.de>
2627 * rs6000.h (ASM_OUTPUT_CONSTRUCTOR, ASM_OUTPUT_DESTRUCTOR): Delete.
2629 Fri Jul 24 14:20:26 1998 Jeffrey A Law (law@cygnus.com)
2631 * mn10300.md (movqi, movhi, movsi, movsf): Correctly handle
2632 CONST_DOUBLE source.
2634 Fri Jul 24 11:17:04 1998 Nick Clifton <nickc@cygnus.com>
2636 * config/arm/thumb.c (thumb_print_operand): Decode %_ in asm
2637 strings as the insertion of USER_LABEL_PREFIX.
2638 * config/arm/thumb.h (PRINT_OPERAND_PUNCT_VALID_P): Accept _ as a
2640 * config/arm/thumb.md: Use %_ as a prefix to gcc library function
2643 Thu Jul 23 18:53:20 1998 Jim Wilson <wilson@cygnus.com>
2645 * dbxout.c (dbxout_range_type): Only call dbxout_type_index for
2646 already defined type.
2648 Thu Jul 23 13:49:41 1998 Jeffrey A Law (law@cygnus.com)
2650 * expr.c (check_max_integer_computation_mode): Allow conversions
2651 of constant integers to MAX_INTEGER_COMPUTATION_MODE.
2652 (expand_expr): Likewise.
2654 Thu Jul 23 11:12:06 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2656 * expr.c (expand_expr): Expand RETURN_EXPR.
2658 Thu Jul 23 11:00:29 1998 Jim Wilson <wilson@cygnus.com>
2660 * dwarf2out.c (dwarf2out_finish): Call stripattributes on TEXT_SECTION.
2662 Wed Jul 22 19:10:00 1998 Catherine Moore <clm@cygnus.com>
2664 * dwarf2out.c (output_aranges): Call stripattributes
2665 for TEXT_SECTION references.
2666 (output_line_info): Likewise.
2668 Wed Jul 22 14:08:54 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2670 * profile.c (branch_prob): Call allocate_reg_info after outputting
2671 profile rtl in instrument_arcs.
2673 Wed Jul 22 12:47:49 1998 Jim Wilson <wilson@cygnus.com>
2675 * fixinc.irix (math.h): Install wrapper instead of copying.
2677 Wed Jul 22 12:37:14 1998 Alexandre Petit-Bianco <apbianco@cygnus.com>
2679 * tree.def (EXPR_WITH_FILE_LOCATION): Defined as an 'e' expression
2680 so WFL are expanded correctly when contained in a COMPOUND_EXPR.
2681 * tree.h (EXPR_WFL_EMIT_LINE_NOTE): Change macro not to use
2682 lang_flag_0. Added documentation in the flag table.
2684 Tue Jul 21 23:28:35 1998 Klaus Kaempf <kkaempf@rmi.de>
2686 * cccp.c (do_include): Fix vax c style include handling.
2688 Tue Jul 21 13:28:19 1998 Jason Merrill <jason@yorick.cygnus.com>
2690 * cplus-dem.c (do_type): Use demangle_template_value_parm for arrays.
2692 Sun Jul 12 01:27:05 1998 Jason Merrill <jason@yorick.cygnus.com>
2694 * fold-const.c (non_lvalue): Don't deal with null pointer
2696 (fold, case COMPOUND_EXPR): Wrap a constant 0 in a NOP_EXPR.
2698 Tue Jul 21 15:49:31 1998 David Edelsohn <edelsohn@mhpcc.edu>
2700 * rs6000.h (PREDICATE_CODES): Add CONSTANT_P_RTX.
2701 * rs6000.md (movsi, movdi): Add CONSTANT_P_RTX.
2702 * rs6000.c (short_cint_operand): Add CONSTANT_P_RTX.
2703 (u_short_cint_operand): Same.
2704 (reg_or_cint_operand): Same.
2705 (logical_operand): Same.
2706 (input_operand): Same.
2707 (reg_or_short_operand): Use u_short_cint_operand.
2709 Tue Jul 21 08:56:42 1998 Richard Henderson <rth@cygnus.com>
2711 * alpha.md (fix_truncdfsi2, fix_truncsfsi2): Remove the define_expands,
2712 but keep the insns and splits. Adjust so when the ultimate destination
2713 is memory, use cvtql.
2715 Tue Jul 21 08:55:09 1998 Richard Henderson <rth@cygnus.com>
2717 * flow.c (regno_uninitialized): Fixed regs are never uninitialized.
2719 Tue Jul 21 00:31:01 1998 Jeffrey A Law (law@cygnus.com)
2721 * gcc.c (do_spec): Call "error" not "warning".
2723 * configure.in: Fix minor problems with gas feature detection code.
2724 * configure: Rebuilt.
2726 * gcc.c (do_spec): Issue a warning for '%[]' usage.
2729 * gcc.c: Delete %[spec] support.
2730 (do_spec_1, case '('): Likewise.
2731 (do_spec_1, case '['): Call error.
2733 Mon Jul 20 22:34:17 1998 Richard Henderson <rth@cygnus.com>
2735 * alpha.h (CPP_SPEC): Tidy. Hook to cpp_cpu and cpp_subtarget.
2736 (CPP_SUBTARGET_SPEC): Default to empty string.
2737 (CPP_AM_*, CPP_IM_*, CPP_CPU_*, CPP_CPU_SPEC): New.
2738 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS): New.
2739 * alpha/elf.h (LD_SPEC): Use %(elf_dynamic_linker).
2740 * alpha/linux-elf.h (SUBTARGET_EXTRA_SPECS): New.
2742 * alpha/linux.h (CPP_PREDEFINES): Tidy.
2743 * alpha/netbsd-elf.h (SUBTARGET_EXTRA_SPECS): New.
2744 * alpha/netbsd.h (CPP_PREDEFINES): Tidy.
2745 * alpha/osf.h (CPP_PREDEFINES): Remove bits subsumed by CPP_CPU_SPEC.
2746 * alpha/win-nt.h (CPP_PREDEFINES): Likewise.
2747 * alpha/vsf.h (CPP_PREDEFINES): Likewise.
2748 (CPP_SUBTARGET_SPEC): New. Do this instead of overriding CPP_SPEC.
2749 * alpha/vxworks.h: Likewise.
2751 Mon Jul 20 22:51:57 1998 Ken Raeburn <raeburn@cygnus.com>
2753 * mips.md (reload_outsi): Added missing REGNO call.
2754 (smulsi3_highpart, umulsi3_highpart): Provide prototype for
2756 (mul_acc_di, mul_acc_64bit_di): Don't use match_op_dup, use
2757 another match_operator and compare the codes.
2759 * mips.h (MASK_DEBUG_E, MASK_DEBUG_I): Set to zero.
2761 * MIPS multiply pattern fixes:
2762 * mips.h (enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
2763 Add union classes for HI, LO, or HILO plus general registers.
2764 (GENERATE_MADD): Deleted.
2765 * mips.md (mulsi3_mult3): Don't disparage output-LO alternative.
2766 Add TARGET_MAD to condition.
2767 (mulsi3): Test HAVE_mulsi3_mult3, not specific flags.
2768 (mul_acc_si): Expand GENERATE_MADD here; it's the only use. Use
2769 "*d" for accumulator, to give preference to LO initially but not
2772 Mon Jul 20 16:16:38 1998 Dave Brolley <brolley@cygnus.com>
2774 * configure.in (enable_c_mbchar): New configure option.
2775 (extra_cpp_objs): Always available now.
2777 * cexp.y (mbchar.h): #include it.
2778 (yylex): Handle Multibyte characters in character literals.
2780 * cccp.c (mbchar.h): #include it.
2781 (main): Set character set based on LANG environment variable.
2782 (rescan): Handle multibyte characters in comments.
2783 (skip_if_group): See above.
2784 (validate_else): See above.
2785 (skip_to_end_of_comment): See above.
2786 (macarg1): See above.
2787 (discard_comments): See above.
2788 (rescan): Handle multibyte characters in string and character literals.
2789 (collect_expansion): See above.
2790 (skip_quoted_string): See above.
2791 (macroexpand): See above.
2792 (macarg1): See above.
2793 (discard_comments): See above.
2794 (change_newlines): See above.
2796 * c-lex.c (mbchar.h): #include it.
2797 (GET_ENVIRONMENT): New macro.
2798 (init_lex): Set character set based on LANG environment variable.
2799 (yylex): Handle multibyte characters in character literals.
2800 (yylex): Handle multibyte characters in string literals.
2802 * Makefile.in (mbchar.o): New target.
2803 (cccp$(exeext)): @extra_cpp_objs@ is always available.
2804 (cppmain$(exeext)): @extra_cpp_objs@ is always available.
2806 * mbchar.[ch]: New files for multibyte character handling.
2808 Mon Jul 20 01:11:11 1998 David S. Miller <davem@pierdol.cobaltmicro.com>
2810 * jump.c (jump_optimize): When simplifying noop moves and
2811 PUSH_ROUNDING, fix thinko so we use same criterion for identifying
2812 the PUSHes to rewrite in second loop as we did in the first.
2814 Sun Jul 19 08:23:53 1998 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
2816 * cplus-dem.c (demangle_nested_args): Make function definition
2817 static to match the prototype.
2819 Fri Jul 17 14:58:44 1998 Richard Henderson <rth@cygnus.com>
2821 * alloca.c: Respect USE_C_ALLOCA.
2822 * gencheck.c (xmalloc): Ignore __GNUC__ for definition.
2823 * gengenrtl.c (xmalloc): Likewise.
2825 Fri Jul 17 14:18:14 1998 Richard Henderson <rth@cygnus.com>
2827 * loop.h (struct induction): Add no_const_addval.
2828 * loop.c (the_movables, reg_address_cost): New variables.
2829 (init_loop): Init reg_address_cost.
2830 (loop_optimize): Call end_alias_analysis.
2831 (scan_loop): Init the_movables.
2832 (record_giv): Init induction->no_const_addval.
2833 (basic_induction_var) [PLUS]: Use rtx_equal_p instead of ==.
2834 [REG]: Rearrange loop search test to catch more cases.
2835 (general_induction_var): Return success not benefit; take an extra
2836 argument for that. Change all callers.
2837 (simplify_giv_expr) [PLUS]: Always combine invariants. Use sge_plus.
2838 [MULT]: Use rtx_equal_p instead of ==. Combine simple invariants.
2839 [default]: Search the_movables for additional combinations.
2840 (sge_plus_constant, sge_plus): New functions.
2841 (express_from_1): New function.
2842 (express_from): Always define. Rewrite using express_from_1.
2843 (combine_givs_p): Handle more cases. Ignore address cost.
2844 (cmp_combine_givs_stats): New function.
2845 (combine_givs_used_once, combine_givs_benefit_from): New functions.
2846 (combine_givs): Rewrite to do best-fit combination.
2848 * fold-const.c (operand_equal_p): Handle RTL_EXPR.
2849 (fold): Do a complete (A*C)+(B*C) association check.
2851 Fri Jul 17 11:21:55 1998 Jim Wilson <wilson@cygnus.com>
2853 * function.c (fixup_var_refs_insns): Handle CLOBBER of a CONCAT.
2855 Fri Jul 17 11:48:55 1998 Jeffrey A Law (law@cygnus.com)
2857 * mn10300.c (MODES_TIEABLE_P): Fix typo.
2859 Fri Jul 17 03:26:12 1998 Rihcard Earnshaw (rearnsha@arm.com)
2861 * tree.c (valid_machine_attribute): Only create a new type variant if
2862 there is a decl to use it.
2864 Thu Jul 16 14:48:04 1998 Nick Clifton <nickc@cygnus.com>
2866 * gcc.c (do_spec_1): Cope with %g/%u/%U options which do not have
2869 Fri Jul 17 03:24:40 1998 Hans-Peter Nilsson <hp@axis.se>
2871 * extend.texi (Explicit Reg Vars): Typo: change "may deleted" into "may
2874 Thu Jul 16 14:48:47 1998 Jeffrey A Law (law@cygnus.com)
2876 * mn10300.c (count_tst_insns): New arg oreg_countp. Callers changed.
2877 Simplify tests for clearing an address register.
2878 (expand_prologue): Corresponding changes.
2880 * mn10300.md (movXX patterns): Make sure the destination is an
2881 ADDRESS_REG when substituting "zero_areg" for (const_int 0).
2882 (logical patterns): Split into expanders + patterns
2883 (zero and sign extension patterns): Similarly.
2884 (shift patterns): Similarly.
2886 Thu Jul 16 01:17:44 1998 Richard Henderson <rth@cygnus.com>
2888 * loop.c (emit_iv_add_mult): Scan the entire insn list generated
2889 for the sequence, recording base values.
2891 Wed Jul 15 10:49:55 1998 Richard Henderson <rth@cygnus.com>
2893 * i386.h (CPP_CPU_SPEC): Remove -Asystem(unix).
2895 Tue Jul 14 14:15:30 1998 Nick Clifton <nickc@cygnus.com>
2897 * gcc.c: Remove ANSI-C ism from --help code.
2899 * toplev.c: Support --help with USE_CPPLIB.
2901 Tue Jul 14 14:46:08 1998 Jeffrey A Law (law@cygnus.com)
2903 * configure.in: Rework gas feature code to work with symlink based
2906 * extend.texi: Clarify some issues related to local variables
2907 assigned to explicit registers.
2909 * mn10300.md (mulsi): Turn into expander + pattern.
2911 * mn10300.md (movsi, movsf, movdi, movdf): Remove "x" from I -> a
2914 Tue Jul 14 07:41:59 1998 Richard Earnshaw (rearnsha@arm.com)
2916 * arm/tcoff.h (USER_LABEL_PREFIX): Make it empty to match coff.h.
2918 Tue Jul 14 03:02:44 1998 Jeffrey A Law (law@cygnus.com)
2920 * version.c: Bump again to distinguish mainline tree from the
2923 See ChangeLog.0 for earlier changes.
2926 add-log-time-format: current-time-string
OSDN Git Service
