Paolo Bonzini [Wed, 16 Jun 2004 07:35:00 +0000 (07:35 +0000)]
configure.ac: New name of configure.in.
boehm-gc/ChangeLog:
2005-06-16 Paolo Bonzini <bonzini@gnu.org>
* boehm-gc/configure.ac: New name of configure.in.
Remove useless multilib stuff. Change uses of
INCLUDES and CXXINCLUDES to AM_CPPFLAGS.
* boehm-gc/configure.in: New name of configure.in.
* boehm-gc/Makefile.am: Update for automake 1.8.5.
* boehm-gc/acinclude.m4: Include ../config/acx.m4.
* boehm-gc/aclocal.m4: Regenerate.
* boehm-gc/configure: Regenerate.
contrib/ChangeLog:
2004-06-16 Paolo Bonzini <bonzini@gnu.org>
* gcc_update (boehm-gc/configure): Depend on
boehm-gc/configure.ac instead of boehm-gc/configure.in
gcc/ChangeLog:
2004-06-16 Paolo Bonzini <bonzini@gnu.org>
* doc/install.texi: boehm-gc now uses automake 1.8.5.
Paolo Bonzini [Wed, 16 Jun 2004 07:25:53 +0000 (07:25 +0000)]
coverage.c: Remove argument to rest_of_compilation.
gcc/ChangeLog:
2004-06-16 Paolo Bonzini <bonzini@gnu.org>
* coverage.c: Remove argument to rest_of_compilation.
* expr.c (execute_expand, set_save_expr_context, pass_expand): New.
* passes.c (rest_of_compilation): Remove argument.
(pass_rest_of_compilation): New.
(rest_of_handle_final, rest_of_handle_delay_slots,
rest_of_handle_stack_regs, rest_of_handle_variable_tracking
rest_of_handle_machine_reorg, rest_of_handle_regrename
rest_of_handle_sched, rest_of_handle_sched2, rest_of_handle_gcse2
rest_of_handle_regmove, rest_of_handle_tracer
rest_of_handle_if_conversion, rest_of_handle_if_after_combine
rest_of_handle_web, rest_of_handle_branch_prob
rest_of_handle_value_profile_transformations, rest_of_handle_cfg
rest_of_handle_addressof, rest_of_handle_jump_bypass
rest_of_handle_life, rest_of_handle_cse, rest_of_handle_cse2):
Check that the two arguments are actually superfluous.
* toplev.h (rest_of_compilation): Adjust prototype.
* tree-optimize.c (register_dump_files): Add properties argument.
Track validity of passes. Only initialize dump files for
tree-based passes. Store the full set of provided passes in
the pass.
(init_tree_optimization_passes): Register pass_expand and
pass_rest_of_compilation.
(execute_one_pass): Do not track the presence of required properties
here. Set in_gimple_form. Do not update current_properties.
(current_properties): Remove.
(set_save_expr_context): Remove.
(tree_rest_of_compilation): Do not set in_gimple_form. Do not
expand to RTL here, and do not call rest_of_compilation. Push
GGC context even before gimplification.
* tree-pass.h (PROP_rtl, PROP_trees): New flags.
(pass_expand, pass_rest_of_compilation): Declare.
gcc/java/ChangeLog:
2004-06-16 Paolo Bonzini <bonzini@gnu.org>
Jeff Law [Wed, 16 Jun 2004 05:09:41 +0000 (23:09 -0600)]
fold-const.c (swap_tree_comparison): No longer static.
* fold-const.c (swap_tree_comparison): No longer static.
(tree_swap_operands_p): Similarly. Return true if both operands
are SSA_NAMEs and the first operand has a higher version number than
the second operand.
* tree.h (swap_tree_comparison): Prototype.
(tree_swap_operands_p): Prototype.
* tree-ssa-operands.c (get_expr_operands): For commutative
operators and relational comparisons, canonicalize the
order of the operands.
* gcc.dg/tree-ssa/20040615-1.c: New test.
* gcc.dg/tree-ssa/20030824-1.c: Update expected output to
be less sensitive to operand ordering.
* gcc.dg/tree-ssa/20030824-2.c: Likewise.
Zdenek Dvorak [Tue, 15 Jun 2004 22:15:21 +0000 (00:15 +0200)]
df.c (df_reg_clobber_gen): Removed.
* df.c (df_reg_clobber_gen): Removed.
(df_bb_rd_local_compute, df_insn_refs_record, df_rd_local_compute):
Make more effective for hard regs.
* ra-build.c (livethrough_conflicts_bb): Check contains_call.
Tobias Schlüter [Tue, 15 Jun 2004 21:50:50 +0000 (23:50 +0200)]
re PR fortran/15962 (constant expression not recognized as such)
fortran/
PR fortran/15962
* match.c (match_case_selector): Call gfc_match_init_expr
instead of gfc_match_expr.
* resolve.c (validate_case_label_expr): No need to check for
constant, since it wouldn't have been matched with the fix to
match.c.
testsuite/
PR fortran/15962
* gfortran.fortran-torture/execute/select_1.f90: New test.
Alexandre Oliva [Tue, 15 Jun 2004 21:43:21 +0000 (21:43 +0000)]
c-pragma.h (c_lex_string_translate): Change type to int.
gcc/ChangeLog:
* c-pragma.h (c_lex_string_translate): Change type to int.
* c-parse.in: Change all assignments of c_lex_string_translate
to true and false to 1 and 0.
* c-lex.c (c_lex_string_translate): Likewise.
(lex_string): Convert string without translation in the -1
case.
gcc/cp/ChangeLog:
* parser.c: Change all assignments of c_lex_string_translate
to true and false to 1 and 0.
(cp_lexer_read_token): Convert type of the translated string.
(cp_parser_skip_to_closing_parentheses): Preserve original
value of c_lex_string_translate, and set it to -1 while
running.
(cp_parser_cache_group): Likewise.
(cp_parser_cache_group_1): Renamed.
(cp_parser_asm_operand_list): Remove redundant setting of
c_lex_string_translate.
(cp_parser_primary_expression) [CPP_STRING, CPP_WSTRING]:
Handle chained strings.
Roger Sayle [Tue, 15 Jun 2004 15:06:14 +0000 (15:06 +0000)]
expmed.c (synth_mult): Mask bits of the multiplier to the machine mode of the multiplication.
* expmed.c (synth_mult): Mask bits of the multiplier to the
machine mode of the multiplication. Don't consider shifts
by more than (or equal to) the width of the operation's mode.
Paolo Bonzini [Tue, 15 Jun 2004 14:47:55 +0000 (14:47 +0000)]
configure.ac: New name of configure.in.
libmudflap/ChangeLog:
2005-06-15 Paolo Bonzini <bonzini@gnu.org>
* configure.ac: New name of configure.in. Update
AC_INIT, AC_CONFIG_SRCDIR, AC_CONFIG_HEADERS, AC_CONFIG_FILES,
AC_OUTPUT, AM_INIT_AUTOMAKE to the preferred style for
Autoconf 2.5x and Automake 1.7 or later.
* configure.in: Remove.
* configure: Regenerate.
zlib/ChangeLog.gcj:
2005-06-15 Paolo Bonzini <bonzini@gnu.org>
* Makefile.am: Remove useless multilib rules.
* configure.ac: Rewrite multilib support to use
features of recent automakes.
* aclocal.m4: Regenerate with Automake 1.8.5.
* Makefile.in: Regenerate with Automake 1.8.5.
* configure: Regenerate.
libffi/ChangeLog:
2005-06-15 Paolo Bonzini <bonzini@gnu.org>
gcc/ChangeLog:
2004-06-15 Paolo Bonzini <bonzini@gnu.org>
* function.h (struct function): Remove cannot_inline field.
(current_function_cannot_inline): Remove.
* passes.c (rest_of_compilation): Reset DECL_DEFER_OUTPUT.
Simplify conditionals to ignore warn_return_type.
* tree-optimize.c (tree_rest_of_compilation): Do not reset
DECL_DEFER_OUTPUT.
* objc/objc-act.c (build_module_descriptor, finish_method_def):
Do not set current_function_cannot_inline.
gcc/java/ChangeLog:
2004-06-15 Paolo Bonzini <bonzini@gnu.org>
* class.c (emit_register_classes): Make the function uninlinable,
do not set current_function_cannot_inline.
* resource.c (write_resource_constructor): Do not reset
flag_inline_functions around rest_of_compilation.
gcc/cp/ChangeLog:
2004-06-15 Paolo Bonzini <bonzini@gnu.org>
* cp-tree.h (struct language_function): Remove cannot_inline.
* decl.c (save_function_data): cannot_inline is no more.
(cxx_push_function_context): Likewise.
* decl2.c (start_objects, start_static_storage_duration_function):
Reset DECL_INLINE, set DECL_UNINLINABLE.
Jeff Law [Mon, 14 Jun 2004 20:41:41 +0000 (14:41 -0600)]
tree-ssa.c (kill_redundant_phi_nodes): More correctly handle PHIs where the destination or an argument is marked with...
* tree-ssa.c (kill_redundant_phi_nodes): More correctly handle
PHIs where the destination or an argument is marked with
SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
Tobias Schlüter [Mon, 14 Jun 2004 15:56:50 +0000 (17:56 +0200)]
re PR fortran/14928 (minloc intrinsic does not understand mask)
fortran/
2004-06-05 Tobias Schlueter <tobias.schlueter@physik.uni-muenchen.de>
Andrew Vaught <andyv@firstinter.net>
PR fortran/14928
* gfortran.h (gfc_check_f): Add new field f3ml.
* check.c (gfc_check_minloc_maxloc): Take argument list instead
of individual arguments, reorder if necessary.
* intrinsic.h (gfc_check_minloc_maxloc): ... adapt prototype.
* intrinsic.c (add_sym_3ml): New function.
(add_functions): Change to add_sym_3ml for MINLOC, MAXLOC.
(check_specific): Catch special case MINLOC, MAXLOC.
testsuite/
PR fortran/14928
* gfortran.fortran-torture/compile/mloc.f90: New test.
Co-Authored-By: Andrew Vaught <andyv@firstinter.net>
From-SVN: r83111
Zdenek Dvorak [Mon, 14 Jun 2004 14:18:01 +0000 (16:18 +0200)]
Makefile.in (FLAGS_H): New.
* Makefile.in (FLAGS_H): New.
(flags.h): Replace by FLAGS_H.
* c.opt: Document Var, VarExists, Init and Report attributes.
* common.opt: Fill the values of the attributes.
* diagnostic.c (flag_fatal_errors): Do not define.
* except.c (flag_non_call_exceptions): Do not define.
* flags.h: Include options.h. Remove declarations conflicting with
the automatically defined ones.
* opts.c: Remove automatically defined variables.
(handle_option): Perform default initialization.
(common_handle_option): Do not handle options covered by the
default initialization.
* opts.h (struct cl_option): Add flag_var, has_set_value and set_value
fields.
(CL_REPORT): New.
* opts.sh: Generate variable declarations, handle CL_REPORT.
* toplev.c: Remove automatically defined variables.
(f_options): Removed.
(print_switch_values): Use cl_options instead of f_options.
* toplev.h (version_flag): Declaration removed.
* ada-tree.def (LOOP_STMT, EXIT_STMT): Update documentation.
* ada-tree.h (EXIT_STMT_LABEL): Renamed from EXIT_STMT_LOOP.
* decl.c (gnat_to_gnu_entity): Also set force_global for imported
subprograms.
* trans.c (gnu_loop_label_stack): Renamed from gnu_loop_stmt_stack;
all callers changed.
(gnat_to_gnu, case N_Loop_Statement, case N_Exit_Statement): Change
the way that EXIT_STMT finds the loop label.
(gnat_gimplify_stmt, case LOOP_STMT, EXIT_STMT): Likewise.
(gnat_gimplify_stmt, case DECL_STMT): Handle variable-sized decls here.
(add_stmt): Use annotate_with_locus insted of setting directly.
(pos_to_construct): Set TREE_PURPOSE of each entry to index.
(gnat_stabilize_reference, case ARRAY_RANGE_REF): Merge with ARRAY_REF.
* utils.c (gnat_install_builtins): Install __builtin_memcmp.
(build_vms_descriptor): Add extra args to ARRAY_REF.
(convert): Use VIEW_CONVERT_EXPR between aggregate types.
* utils2.c (gnat_truthvalue_conversion, case INTEGER_CST, REAL_CST):
New cases.
(build_binary_op): Don't make explicit CONVERT_EXPR.
Add extra rgs to ARRAY_REF.
Arnaud Charlet [Mon, 14 Jun 2004 13:19:14 +0000 (15:19 +0200)]
[multiple changes]
2004-06-14 Pascal Obry <obry@gnat.com>
* gnat_ugn.texi: Document relocatable vs. dynamic Library_Kind on
Windows. Fix minor typo.
* mlib-tgt-mingw.adb: New implementation using the GCC -shared option
which is now supported on Windows. With this implementation using the
Library Project feature is no different on Windows than on UNIX.
2004-06-14 Vincent Celier <celier@gnat.com>
* makegpr.adb (Compile_Sources): Nothing to do when there are no
non-Ada sources.
* prj-part.adb (Parse_Single_Project): When a duplicate project name is
found, show the project name and the path of the previously parsed
project file.
2004-06-14 Ed Schonberg <schonberg@gnat.com>
* exp_ch6.adb (Add_Call_By_Copy_Code): For an out-parameter that is an
array, avoid copying the actual before the call.
2004-06-14 Thomas Quinot <quinot@act-europe.fr>
* g-debpoo.adb: Remove alignment assumptions from GNAT.Debug_Pools.
Instead, allocate memory on worst-case alignment assumptions, and then
return an aligned address within the allocated zone.
2004-06-14 Robert Dewar <dewar@gnat.com>
* bindgen.adb (Gen_Adainit_Ada): Do not generate external references to
elab entities in predefined units in No_Run_Time_Mode.
(Gen_Adainit_C): Same fix
(Gen_Elab_Calls_Ada): Do not generate calls to elaborate predefined
units in No_Run_Time_Mode
(Gen_Elab_Calls_C): Same fix
* symbols-vms-alpha.adb: Minor reformatting
* g-debpoo.ads: Minor reformatting
* lib.adb (In_Same_Extended_Unit): Version working on node id's
* lib.ads (In_Same_Extended_Unit): Version working on node id's
* lib-xref.adb: Minor cleanup, use new version of In_Same_Extended_Unit
working on nodes.
* make.adb: Minor reformatting
* par-ch12.adb: Minor reformatting
* par-prag.adb: Add dummy entry for pragma Profile_Warnings
* prj-strt.adb: Minor reformatting
* restrict.ads, restrict.adb: Redo handling of profile restrictions to
be more general.
* sem_attr.adb: Minor reformatting
* sem_ch7.adb: Minor reformatting
* sem_elab.adb (Check_A_Call): Deal with problem of calling init proc
for type in the same unit as the object declaration.
* sem_prag.adb (Check_Arg_Is_External_Name): New procedure, allows
static string expressions and not just string literals.
Minor reformatting
(Set_Warning): Reset restriction warning flag for restriction pragma
Implement pragma Profile_Warnings
Implement pragma Profile (Restricted)
Give obolescent messages for old restrictions and pragmas
* snames.h, snames.ads, snames.adb: Add new entry for pragma
Profile_Warnings.
* s-rident.ads: Add declarations for restrictions required by profile
Restricted and profile Ravenscar.
* targparm.ads, targparm.adb: Allow pragma Profile in system.ads
* gnat_ugn.texi: Correct some missing entries in the list of GNAT
configuration pragmas.
J"orn Rennecke [Mon, 14 Jun 2004 12:09:08 +0000 (12:09 +0000)]
basic-block.h (could_fall_through): Declare.
* basic-block.h (could_fall_through): Declare.
* cfganal.c (can_fallthru): Succeed if the target is EXIT_BLOCK_PTR.
Fail if the source already has a fallthrough edge to the exit
block pointer.
(could_fall_through): New function.
* cfgbuild.c (make_edges): Check if we already have a fallthrough
edge to the exit block pointer.
* cfglayout.c (fixup_fallthru_exit_predecessor): Check that it is
not called before reload has completed.
Handle special case of first block having a fall-through exit edge.
(cfg_layout_finalize): Don't call it before reload or if we have
rtl epilogues.
(fixup_reorder_chain): A fall through to the exit block does not
require the block to come last. Add sanity checks.
* cfgrtl.c (rtl_split_edge): Add special handling of fall through
edges to the exit block.
* function.c (cfglayout.h): #include.
(thread_prologue_and_epilogue_insns): If we have neither return nor
epilogue, but a fall through to the exit block from mid-function,
force a non-fall-through exit.
* Makefile.in (function.o): Depend on CFGLAYOUT_H.
Paolo Carlini [Mon, 14 Jun 2004 10:49:36 +0000 (10:49 +0000)]
locale_facets.tcc (time_get<>::_M_extract_name): Don't use the 'magic number' 10 in the computation of __minlen...
2004-06-14 Paolo Carlini <pcarlini@suse.de>
* include/bits/locale_facets.tcc (time_get<>::_M_extract_name):
Don't use the 'magic number' 10 in the computation of __minlen;
never access __name past __minlen; in the loop over __i3, don't
decrease __nmatches and increase __i3 at once, only either of
the two; tidy.
Fix PR tree-optimization/15979
Fix PR tree-optimization/15981
* tree-ssa-pre.c (insert_aux): Fix faulty logic so that we don't
try to insert values undefined along some path.
Andrew Pinski [Sun, 13 Jun 2004 22:15:51 +0000 (22:15 +0000)]
fold-const.c (fold_checksum_tree <case 't'>): Only look at TREE_VALUES if the EXPR is an ENUMERAL_TYPE.
2004-06-13 Andrew Pinski <pinskia@physics.uc.edu>
* fold-const.c (fold_checksum_tree <case 't'>): Only
look at TREE_VALUES if the EXPR is an ENUMERAL_TYPE.
Only look at TYPE_MIN_VALUE and TYPE_MAX_VALUE if
EXPR is an INTEGERAL_TYPE or a scalar float type.