cplus-dem.c (demangle_function_name): Changed to return value indicating if a name was correctly demangled.

2008-01-23  Thiago Jung Bauermann  <bauerman@br.ibm.com>

* cplus-dem.c (demangle_function_name): Changed to return value
indicating if a name was correctly demangled.
(iterate_demangle_function): Use demangle_function_name return
value.

From-SVN: r131754

* config.sub, config.guess: Update from upstream sources.

From-SVN: r131753

bfin-protos.h (WA_RETS, [...]): New macros.

* config/bfin/bfin-protos.h (WA_RETS, ENABLE_WA_RETS): New macros.
* config/bfin/bfin.c (bfin_cpus): Add WA_RETS everywhere.
(cputype_selected): New static variable.
(bfin_handle_option): Set it if -mcpu is used.
(override_option): Select default set of workarounds if no cpu type
selected on the command line.
(workaround_rts_anomaly): Only run if ENABLE_WA_RETS.

From-SVN: r131752

re PR c++/33984 (bit-fields, references and overloads)

gcc/cp/ChangeLog:
PR c++/33984
* call.c (reference_binding): For bitfields use the declared bitfield
type.
(add_builtin_candidates): Likewise.
* class.c (layout_class_type): For bitfields copy over the
original type quals.
gcc/testsuite/ChangeLog:
PR c++/33984
* g++.dg/conversion/bitfield7.C: New test.
* g++.dg/cpp0x/decltype4.C: Fixed xfail.

From-SVN: r131751

Daily bump.

From-SVN: r131749

gcc/
From  Michael Frysinger  <michael.frysinger@analog.com>
* config/bfin/bfin-protos.h (enum bfin_cpu_type): Add
BFIN_CPU_BF547, BFIN_CPU_BF523, BFIN_CPU_BF524, and BFIN_CPU_BF526.

* config/bfin/elf.h (LIB_SPEC): Use proper linker script
for bf547, bf523, bf524, and bf526.
* config/bfin/bfin.c (bfin_cpus[]): Add bf547,  bf523, bf524, and
bf526.
* config/bfin/bfin.h (TARGET_CPU_CPP_BUILTINS): Define
__ADSPBF523__ for bf523, __ADSPBF524__ for bf524,
__ADSPBF526__ for bf526, __ADSPBF52x__ for all three, as well as
__ADSPBF547__ and __ADSPBF54x__ for bf547.
* doc/invoke.texi (Blackfin Options): Document that
-mcpu now accept bf547, bf523, bf524, and bf526.

gcc/testsuite/
From Michael Frysinger  <michael.frysinger@analog.com>
* gcc.target/bfin/mcpu-bf547.c: New test.
* gcc.target/bfin/mcpu-bf523.c: New test.
* gcc.target/bfin/mcpu-bf524.c: New test.
* gcc.target/bfin/mcpu-bf526.c: New test.

From-SVN: r131746

re PR rtl-optimization/34628 (problems with inlining on ARM)

PR rtl-optimization/34628
* combine.c (try_combine): Stop and undo after the first combination
if an autoincrement side-effect on the first insn has effectively
been lost.

From-SVN: r131744

re PR c++/34859 (g++ -D__STDC_LIMIT_MACROS -D__STDC_LIMIT_MACROS causes error)

libcpp
PR c++/34859:
* macro.c (_cpp_create_definition): Handle __STDC_LIMIT_MACROS and
__STDC_CONSTANT_MACROS.
gcc/testsuite
PR c++/34859:
* gcc.dg/cpp/pr34859.c: New file.

From-SVN: r131743

re PR fortran/34875 (read into vector-valued section doesn't transfer any values)

2008-01-22  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34875
* trans-io.c (gfc_trans_transfer): If the array reference in a
read has a vector subscript, use gfc_conv_subref_array_arg to
copy back the temporary.

2008-01-22  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34875
* gfortran.dg/vector_subscript_3.f90: New test.

From-SVN: r131742

re PR target/34529 (Wrong code with altivec stores and offsets)

        PR target/34529
        * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
        Offset addresses are not valid for Altivec or paired float modes.

From-SVN: r131741

re PR c++/34912 (ICE with friend in local class)

        PR c++/34912
        * friend.c (do_friend): Check for prior declaration of a friend
        function of a local class.
        * name-lookup.c (lookup_name_innermost_nonclass_level):
        No longer static.
        * name-lookup.h: Declare it.

From-SVN: r131740

re PR fortran/34848 (internal compiler error with optional argument of character type and array return type)

2008-01-22  Tobias Burnus  <burnus@net-b.de>

       PR fortran/34848
       * interface.c (compare_actual_formal): Fix adding type
       to missing_arg_type for absent optional arguments.

2008-01-22  Tobias Burnus  <burnus@net-b.de>

       PR fortran/34848
       * gfortran.dg/missing_optional_dummy_4.f90

From-SVN: r131738

vect-ifcvt-9.c: Use inline.

* gcc.dg/vect/vect-ifcvt-9.c: Use inline.

From-SVN: r131735

Changelog: Insert forgotten entry of commit 131705.

2008-01-22  Tobias Grosser  <grosser@fim.uni-passau.de>

* Changelog: Insert forgotten entry of commit 131705.
Commit 131705 was:
* MAINTAINERS (Write After Approval): Add myself.

From-SVN: r131734

decl.c (components_to_record): Do not reuse the empty union type if there is a representation clause on the record.

* decl.c (components_to_record): Do not reuse the empty union type
if there is a representation clause on the record.
* trans.c (addressable_p): Return true for INTEGER_CST.

From-SVN: r131733

re PR c++/34829 (placement new with primitive Java types rejected)

gcc/testsuite
PR c++/34829:
* g++.dg/ext/pr34829.C: New file.
gcc/cp
PR c++/34829:
* init.c (build_new_1): Only disallow Java aggregates.

From-SVN: r131732

user.cfg.in: Set SHORT_NAMES to YES.

2008-01-21  Matthias Klose  <doko@ubuntu.com>

* doc/doxygen/user.cfg.in: Set SHORT_NAMES to YES.

From-SVN: r131731

re PR c++/34607 (ICE with invalid for loop after #pragma omp for)

PR c++/34607
* c-parser.c (c_parser_omp_for_loop): Don't call c_finish_omp_for
if DECL_INITIAL (decl) is error_mark_node.

* semantics.c (finish_omp_for): Don't call c_finish_omp_for
if decl or init is error_mark_node.

* gcc.dg/gomp/pr34607.c: New test.
* g++.dg/gomp/pr34607.C: New test.

From-SVN: r131730

re PR c++/34914 (Member pointer to vector rejected)

PR c++/34914
* c-common.c (handle_vector_size_attribute): Only allow
integral, scalar float and fixed point types.  Handle OFFSET_TYPE
the same way as pointer, array etc. types.
* tree.c (reconstruct_complex_type): Handle OFFSET_TYPE.

* g++.dg/ext/vector10.C: New test.

From-SVN: r131729

re PR fortran/34907 (valgrind error indication from testsuite trans-types.c: gfc_typenode_for_spec)

2008-01-22  Tobias Burnus  <burnus@net-b.de>

       PR fortran/34907
       * parse.c (parse_spec): Change = into ==.

From-SVN: r131728

re PR c++/34918 (Broken diagnostic: 'vector_cst' not supported by dump_expr)

PR c++/34918
* error.c (dump_expr): Handle VECTOR_CST.

* g++.dg/other/error23.C: New test.

From-SVN: r131727

re PR c++/34917 (ICE with const vector)

PR c++/34917
* tree.c (build_type_attribute_qual_variant): Call
build_qualified_type if attributes are equal, but quals are not.

* g++.dg/ext/vector12.C: New test.

From-SVN: r131726

re PR c++/33959 (ICE in instantiate_class_template, at cp/pt.c:6649)

        PR c++/33959
        * pt.c (tsubst_aggr_type): Make sure our context is complete.

        PR c++/34573
        * pt.c (retrieve_local_specialization): Robustify.
        (tsubst_pack_expansion, tsubst_decl): Remove redundant checks.

        PR c++/34846
        * pt.c (tsubst): Only call retrieve_local_specialization if the
        original typedef was in a function template.

From-SVN: r131724

Missed testcases in earlier commit.

2008-01-22  Manuel Lopez-Ibanez  <manu@gcc.gnu.org>

PR 32102
testsuite/
* gcc.dg/Wstrict-overflow-21.c: New.
* g++.dg/warn/Wstrict-aliasing-8.C: New.

From-SVN: r131722

Missed testcase in earlier commit.

2008-01-22  Manuel Lopez-Ibanez  <manu@gcc.gnu.org>

PR middle-end/33092
testsuite/
* gcc.dg/pr33092.c: New.

From-SVN: r131721

re PR c/32102 (-Wall stomps on -Wstrict-overflow)

2008-01-22  Manuel Lopez-Ibanez  <manu@gcc.gnu.org>

        PR 32102
        * doc/invoke.texi (-Wall): -Wall enables -Wstrict-overflow=1.
        * flags.h (warn_strict_aliasing): Remove.
        (warn_strict_overflow): Remove.
        * opts.c (warn_strict_aliasing): Remove.
        (warn_strict_overflow): Remove.
        * c-opts.c (c_common_handle_option): -Wall only sets
        -Wstrict-aliasing or -Wstrict-overflow if they are uninitialized.
        (c_common_post_options): Give default values to -Wstrict-aliasing
        and -Wstrict-overflow if they are uninitialized.
        * common.opt (Wstrict-aliasing): Specify Var and Init.
        (Wstrict-overflow): Likewise.

testsuite/
        * gcc.dg/Wstrict-overflow-21.c: New.
        * g++.dg/warn/Wstrict-aliasing-8.C: New.

From-SVN: r131720

re PR tree-optimization/26854 (Inordinate compile times on large routines)

2008-01-22  Kenneth Zadeck <zadeck@naturalbridge.com>

PR rtl-optimization/26854
PR rtl-optimization/34400
PR rtl-optimization/34884
* ddg.c (create_ddg_dep_from_intra_loop_link): Use
DF_RD->gen.
* df.h (df_changeable_flags.DF_RD_NO_TRIM): Deleted
(df_rd_bb_info.expanded_lr_out): Deleted
* loop_invariant.c (find_defs): Deleted DF_RD_NO_TRIM flag.
* loop_iv.c (iv_analysis_loop_init): Ditto.  * df-problems.c
(df_rd_free_bb_info, df_rd_alloc, df_rd_confluence_n,
df_rd_bb_local_compute, df_rd_transfer_function, df_rd_free):
Removed code to allocate, initialize or free expanded_lr_out.
(df_rd_bb_local_compute_process_def): Restructured to make more
understandable.
(df_rd_confluence_n): Removed code to no apply invalidate_by_call
sets if the sets are being trimmed.

From-SVN: r131719

re PR bootstrap/32287 (gas version style changed causes warnings with configure)

2008-01-22  H.J. Lu  <hongjiu.lu@intel.com>

PR bootstrap/32287
* configure.ac (ld_vers): Support GNU linker version xx.xx.*
(as_vers): Likewise.
* configure: Regenerated.

From-SVN: r131718

re PR tree-optimization/33092 (Using -O1 -fno-tree-salias results in ICE)

2008-01-22  Manuel Lopez-Ibanez  <manu@gcc.gnu.org>

PR middle-end/33092
* tree-pass.h (pass_build_alias): New pass.
* tree-ssa-alias.c (gate_build_alias): New.
(pass_build_alias): New.
* passes.c (init_optimization_passes): Add pass_build_alias after
pass_create_structure_vars.

testsuite/
* gcc.dg/pr33092.c: New.

From-SVN: r131717

20020926-1.c: Add -m31 option.

2008-01-22  Andreas Krebbel  <krebbel1@de.ibm.com>

* gcc.target/s390/20020926-1.c: Add -m31 option.

From-SVN: r131716

re PR fortran/34915 (-std=f95 rejects len_trim() and index() in init expressions)

gcc/fortran:
2008-01-22  Daniel Franke  <franke.daniel@gmail.com>

PR fortran/34915
* expr.c (check_elemental): Fix check for valid data types.

gcc/testsuite:
2008-01-22  Daniel Franke  <franke.daniel@gmail.com>

PR fortran/34915
* gfortran.dg/initialization_18.f90: New test.

From-SVN: r131715

s390.h (S390_TDC_POSITIVE_NORMALIZED_NUMBER): Renamed to S390_TDC_POSITIVE_NORMALIZED_BFP_NUMBER.

2008-01-22  Wolfgang Gellerich  <gellerich@de.ibm.com>

* config/s390/s390.h (S390_TDC_POSITIVE_NORMALIZED_NUMBER):
Renamed to S390_TDC_POSITIVE_NORMALIZED_BFP_NUMBER.
(S390_TDC_NEGATIVE_NORMALIZED_NUMBER): Renamed to
S390_TDC_NEGATIVE_NORMALIZED_BFP_NUMBER.
(S390_TDC_POSITIVE_DENORMALIZED_NUMBER): Renamed to
S390_TDC_POSITIVE_DENORMALIZED_BFP_NUMBER.
(S390_TDC_NEGATIVE_DENORMALIZED_NUMBER): Renamed to
S390_TDC_NEGATIVE_DENORMALIZED_BFP_NUMBER.
(S390_TDC_POSITIVE_NORMALIZED_BFP_NUMBER): New constant.
(S390_TDC_NEGATIVE_NORMALIZED_BFP_NUMBER): New constant.
(S390_TDC_POSITIVE_DENORMALIZED_BFP_NUMBER): New constant.
(S390_TDC_NEGATIVE_DENORMALIZED_BFP_NUMBER): New constant.
* config/s390/s390.md (FP_ALL): New mode iterator.
(_d): New mode attribute.
("*signbit<mode>2>"): Changed mode of first operand.
("isinf<mode>2"): Changed mode of first operand.
("*TDC_insn"): Adaptation for DFP modes.

From-SVN: r131714

re PR fortran/34899 (Continuation lines with <tab><number> not recognized)

2008-01-22  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34899
        * scanner.c (load_line): Support <tab><digit> continuation
        * lines.
        * invoke.texi (-Wtabs): Document this.

2008-01-22  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34899
        * gfortran.dg/tab_continuation.f: New.

From-SVN: r131713

re PR fortran/34896 (libgomp.fortran/reduction5.f90)

2008-01-22  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34896
* module.c (read_module): Set use_rename attribute.

2008-01-22  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34896
* gfortran.dg/use_rename_2.f90: New test.

From-SVN: r131712

* tree.c (check_qualified_type): Improve function description.

From-SVN: r131711

re PR c++/34196 (uninitialized variable warning in dead exception region)

        PR c++/34196
        * tree.h (TRY_CATCH_IS_CLEANUP): New macro.
        * cp/decl.c (wrap_cleanups_r): Set TRY_CATCH_IS_CLEANUP.
        * tree-eh.c (honor_protect_cleanup_actions): Strip TRY_CATCH_EXPR
        if it is set.

From-SVN: r131710

Daily bump.

From-SVN: r131708

MAINTAINERS (Write After Approval): Add myself.

2008-01-22  Tobias Grosser  <grosser@fim.uni-passau.de>

        * MAINTAINERS (Write After Approval): Add myself.

From-SVN: r131705

* sv.po: Update.

From-SVN: r131704

tm.texi (HARD_REGNO_NREGS): Note that this macro must not return zero.

* doc/tm.texi (HARD_REGNO_NREGS): Note that this macro must not
return zero.

From-SVN: r131703

update_web_docs_libstdcxx_svn: "docs" dir moved to "doc".

2008-01-21  Jonathan Wakely  <jwakely.gcc@gmail.com>

* update_web_docs_libstdcxx_svn: "docs" dir moved to "doc".

From-SVN: r131702

re PR libgcj/34369 (java.net.URI.relativize(URI) method returns incorrect results)

2008-01-21  Luciano Chavez  <lnx1138@us.ibm.com>

PR libgcj/34369:
* java/net/URI.java (relativize): Check initial segment for
trailing "/".

From-SVN: r131701

re PR fortran/34901 (add kind information to mismatched-types error message)

2007-01-21  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34901
        * interface.c (compare_parameter): Improved error message
        for arguments of same type and mismatched kinds.

From-SVN: r131700

re PR target/34856 (ICE with some constant vectors)

2008-01-21  Richard Guenther  <rguenther@suse.de>

PR middle-end/34856
* tree-cfg.c (verify_expr): Allow all invariant expressions
instead of just constant class ones as reference argument.
* tree-ssa-loop-im.c (for_each_index): Handle CONSTRUCTOR
like any other constant.
* tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.

* gcc.c-torture/compile/pr34856.c: New testcase.

From-SVN: r131696

regmove.c (fixup_match_1): Update call crossed frequencies.

2008-01-21  H.J. Lu  <hongjiu.lu@intel.com>

* regmove.c (fixup_match_1): Update call crossed frequencies.

From-SVN: r131695

re PR tree-optimization/34885 (ICE in copy_reference_ops_from_ref, at tree-ssa-sccvn.c:574)

2008-01-21  Richard Guenther  <rguenther@suse.de>

PR c/34885
* tree-inline.c (setup_one_parameter): Deal with mismatched
types using a VIEW_CONVERT_EXPR.

* gcc.c-torture/compile/pr34885.c: New testcase.

From-SVN: r131694

Update copyright year.

From-SVN: r131693

trans.c (gnat_to_gnu): Use POINTER_PLUS_EXPR in pointer arithmetics.

* trans.c (gnat_to_gnu) <N_Free_Statement>: Use POINTER_PLUS_EXPR
in pointer arithmetics.
* utils2.c (build_allocator): Likewise.

From-SVN: r131692

* de.po: Update.

From-SVN: r131691

re PR c++/34850 (Recursive BLOCK tree causes compilation to hang during diagnostics)

2008-01-21  Richard Guenther  <rguenther@suse.de>

PR c++/34850
cp/
* error.c (cp_print_error_function): Deal with recursive
BLOCK trees.

* g++.dg/torture/pr34850.C: New testcase.

From-SVN: r131690

re PR middle-end/34701 (ICE in tree-ssa-ccp.c with -fipa-struct-reorg)

2008-01-21  Alon Dayan  <alond@il.ibm.com>
    Olga Golovanevsky  <olga@il.ibm.com>

PR tree-optimization/34701
* gcc.dg/struct/wo_prof_array_field.c: New test.

2008-01-21  Alon Dayan  <alond@il.ibm.com>
    Olga Golovanevsky  <olga@il.ibm.com>

PR tree-optimization/34701
* ipa-struct-reorg.c (gen_size): Fix the malloc parameter calculation
when the structure size is not a power of 2.

Co-Authored-By: Olga Golovanevsky <olga@il.ibm.com>
From-SVN: r131689

re PR c++/34891 (Broken diagnostic: 'view_convert_expr' not supported by dump_expr)

/cp
2008-01-20  Paolo Carlini  <pcarlini@suse.de>

        PR c++/34891
        * error.c (dump_expr): Deal with VIEW_CONVERT_EXPR.

/testsuite
2008-01-20  Paolo Carlini  <pcarlini@suse.de>

        PR c++/34891
        * g++.dg/ext/vector9.C: New.

From-SVN: r131687

re PR c++/34776 (ICE with invalid member declaration in template class)

/cp
2008-01-20  Paolo Carlini  <pcarlini@suse.de>

        PR c++/34776
PR c++/34486
        * name-lookup.c (do_class_using_decl): Do not call constructor_name_p
on non-IS_AGGR_TYPE type scope.
(constructor_name_p): Assert IS_AGGR_TYPE.

/testsuite
2008-01-20  Paolo Carlini  <pcarlini@suse.de>

        PR c++/34776
PR c++/34486
        * g++.dg/template/crash75.C: New.
        * g++.dg/template/crash76.C: Likewise.

From-SVN: r131686

install.texi: Add doc for --enable-checking=df.

2008-01-20  Kenneth Zadeck <zadeck@naturalbridge.com>

* doc/install.texi: Add doc for --enable-checking=df.

From-SVN: r131685

Daily bump.

From-SVN: r131683

re PR rtl-optimization/34808 (ICE in prescan_insns_for_dce)

PR rtl-optimization/34808
* emit-rtl.c (try_split): Handle REG_RETVAL notes.
* gcc.c-torture/compile/pr34808.c: New test.

From-SVN: r131680

re PR fortran/34861 (ICE in function with entry (and result?))

2008-01-20  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34861
* resolve.c (resolve_entries): Do not do an array bounds check
if the result symbols are the same.

PR fortran/34854
* module.c (read_module) : Hide the symtree of the previous
version of the symbol if this symbol is renamed.

2008-01-20  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34784
* gfortran.dg/mapping_2.f90: Correct ubound expression for h4.

PR fortran/34861
* gfortran.dg/entry_array_specs_3.f90: New test.

PR fortran/34854
* gfortran.dg/use_rename_1.f90: New test.

From-SVN: r131679

* sv.po: Update.

From-SVN: r131678

trans-array.c (gfc_add_loop_ss_code): Correct whitespace.

2008-01-20  Paul Thomas  <pault@gcc.gnu.org>

* trans-array.c (gfc_add_loop_ss_code) :  Correct whitespace.

From-SVN: r131677

global.c (find_reg): Only compute EH_RETURN_DATA_REGNO once per input.

gcc/
* global.c (find_reg): Only compute EH_RETURN_DATA_REGNO once per
input.

From-SVN: r131676

re PR fortran/34784 (implicit character(s) hides type of selected_int_kind intrinsic)

2008-01-20  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34784
* array.c (gfc_check_constructor_type): Clear the expression ts
so that the checking starts from the deepest level of array
constructor.
* primary.c (match_varspec): If an unknown type is changed to
default character and the attempt to match a substring fails,
change it back to unknown.

PR fortran/34785
* trans-array.c (gfc_add_loop_ss_code) : If ss->string_length is
NULL for an array constructor, use the cl.length expression to
build it.
(gfc_conv_array_parameter): Change call to gfc_evaluate_now to
a tree assignment.

2008-01-20  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/34784
* gfortran.dg/array_constructor_20.f90: New test.
* gfortran.dg/mapping_2.f90: Correct ubound expression for h4.

PR fortran/34785
* gfortran.dg/array_constructor_21.f90: New test.

From-SVN: r131675

Fix it.

From-SVN: r131674

re PR fortran/34795 (inquire statement , direct= specifier incorrectly returns YES)

2008-01-20  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR libfortran/34795
* gfortran.dg/inquire_13.f90: New test.

PR libfortran/34659
* gfortran.dg/continuation_9.f90: New test.

From-SVN: r131673

inquire.c (inquire_via_unit): If a unit is opened...

2008-01-19  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

* io/inquire.c (inquire_via_unit): If a unit is opened, return values
according to the open action for DIRECT, FORMATTED, and UNFORMATTED.
(inquire_via_filename): Return "UNKNOWN" for SEQUENTIAL, DIRECT,
FORAMATTED, and UNFORMATTED inquiries.
* io/unix.c (inquire_sequential): Return "UNKNOWN" when appropriate
for files that are not opened. (inquire_direct): Same.
(inquire_formatted): Same.

From-SVN: r131672

configure: Regenerate.

2008-01-19  Paolo Carlini  <pcarlini@suse.de>

* configure: Regenerate.
* Makefile.in: Likewise.
* src/Makefile.in: Likewise.
* doc/Makefile.in: Likewise.
* po/Makefile.in: Likewise.
* libmath/Makefile.in: Likewise.
* include/Makefile.in: Likewise.
* libsupc++/Makefile.in: Likewise.
* testsuite/Makefile.in: Likewise.
* aclocal.m4: Likewise.

From-SVN: r131671

re PR tree-optimization/26854 (Inordinate compile times on large routines)

2008-01-19  Kenneth Zadeck <zadeck@naturalbridge.com>

PR rtl-optimization/26854
PR rtl-optimization/34400
* ddg.c (create_ddg_dep_from_intra_loop_link): Do not use
DF_RD->gen.
* df.h (df_changeable_flags.DF_RD_NO_TRIM): New.
(df_rd_bb_info.expanded_lr_out): New.
* loop_invariant.c (find_defs): Added DF_RD_NO_TRIM flag.
* loop_iv.c (iv_analysis_loop_init): Ditto.
* df-problems.c (df_rd_free_bb_info, df_rd_alloc, df_rd_confluence_n,
df_rd_bb_local_compute, df_rd_transfer_function, df_rd_free):
Added code to allocate, initialize or free expanded_lr_out.
(df_rd_bb_local_compute_process_def): Restructured to make
more understandable.
(df_rd_confluence_n): Add code to do nothing with fake edges and
code to no apply invalidate_by_call sets if the sets are being trimmed.
(df_lr_local_finalize): Renamed to df_lr_finalize.
(df_live_local_finalize): Renamed to df_live_finalize.

From-SVN: r131670

* de.po: Restore to current version from TP site.

From-SVN: r131669

Daily bump.

From-SVN: r131667

* sv.po: Update.

From-SVN: r131664

* sv.po: Update.

From-SVN: r131663

re PR target/34831 (ICE on gcc.dg/pr34233.c for MIPS)

gcc/
PR target/34831
* config/mips/mips.md (div<mode>3): Use <recip_condition> when
deciding whether to use reciprocal instructions.

gcc/testsuite/
PR target/34831
* gcc.target/mips/pr34831.c: New test.

From-SVN: r131662

* sv.po: Update.

From-SVN: r131661

re PR fortran/34817 (mixed-kind "any" and "all" intrinsics with expressions)

2008-01-19  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/34817
PR fortran/34838
* iresolve.c (gfc_resolve_all):  Remove conversion of mask
argument to kind=1 by removing call to resolve_mask_arg().
(gfc_resolve_any):  Likewise.

2008-01-19  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/34817
PR fortran/34838
* gfortran.dg/any_all_1.f90:  New test.
* gfortran.dg/any_all_2.f90:  New test.

From-SVN: r131660

ia64-2.C: Place "dg-do run" statement before "dg-require-weak" statement.

* g++.dg/eh/ia64-2.C: Place "dg-do run" statement before
"dg-require-weak" statement.
* g++.dg/eh/weak1.C: Likewise.

From-SVN: r131657

dwarf2out.c (dwarf2out_switch_text_section): Do not call dwarf2out_note_section_used if cold_text_section is NULL.

        * dwarf2out.c (dwarf2out_switch_text_section): Do not call
        dwarf2out_note_section_used if cold_text_section is NULL.

From-SVN: r131654

re PR gcov-profile/34610 (ICE with "-fprofile-arcs -fopenmp")

PR gcov-profile/34610
* tree-cfg.c (make_edges): Mark both outgoing edges from
OMP_CONTINUE and from OMP_FOR as EDGE_ABNORMAL.
* omp-low.c (expand_omp_for): Clear EDGE_ABNORMAL bits
from OMP_FOR and OMP_CONTINUE outgoing edges.

* tree-profile.c (tree_profiling): Return early if
cfun->after_tree_profile != 0.  Set cfun->after_tree_profile
at the end.
* omp-low.c (expand_omp_parallel): Copy after_tree_profile
from cfun to child_cfun.
* function.h (struct function): Add after_tree_profile bit.

* gcc.dg/gomp/pr34610.c: New test.

From-SVN: r131653

re PR fortran/34760 (PRIVATE variable not allowed as STAT variable in ALLOCATE)

2008-01-19  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34760
        * primary.c (match_variable): Handle FL_UNKNOWN without
        uneducated guessing.
        (match_variable): Improve error message.

2008-01-19  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34760
        * gfortran.dg/implicit_11.f90: New.
        * gfortran.dg/allocate_stat.f90: Update dg-error pattern.
        * gfortran.dg/entry_15.f90: Ditto.
        * gfortran.dg/func_assign.f90: Ditto.
        * gfortran.dg/gomp/reduction3.f90: Ditto.
        * gfortran.dg/proc_assign_1.f90: Ditto.

        * gfortran.dg/interface_proc_end.f90: Use dg-error instead
        of dg-excess-errors.

From-SVN: r131652

* config/avr/avr.S (_exit): Disable interrupt.

From-SVN: r131651

re PR other/33768 (splay-tree.c typo)

2008-01-19  Manuel Lopez-Ibanez  <manu@gcc.gnu.org>

PR other/33768
* splay-tree.c (rotate_left): Fix minor typo in comment.
(rotate_right): Likewise.

From-SVN: r131650

re PR tree-optimization/26854 (Inordinate compile times on large routines)

2008-01-18  Kenneth Zadeck  <zadeck@naturalbridge.com>
    Steven Bosscher  <stevenb.gcc@gmail.com>

PR rtl-optimization/26854
PR rtl-optimization/34400
* df-problems.c (df_live_scratch): New scratch bitmap.
(df_live_alloc): Allocate df_live_scratch when doing df_live.
(df_live_reset): Clear the proper bitmaps.
(df_live_bb_local_compute): Only process the artificial defs once
since the order is not important.
(df_live_init): Init the df_live sets only with the variables
found live by df_lr.
(df_live_transfer_function): Use the df_lr sets to prune the
df_live sets as they are being computed.
(df_live_free): Free df_live_scratch.

Co-Authored-By: Steven Bosscher <stevenb.gcc@gmail.com>
From-SVN: r131649

Daily bump.

From-SVN: r131647

re PR fortran/32616 ("Too short actual argument" for array element storage sequence)

2008-01-18  Tobias Burnus  <burnus@net-b.de>

        PR fortran/32616
        * interface.c (get_expr_storage_size): Return storage size
        for array element designators.
        (compare_actual_formal): Reject unequal string sizes for
        assumed-shape dummy arguments. And fix error message for
        array-sections with vector subscripts.

2008-01-18  Tobias Burnus  <burnus@net-b.de>

        PR fortran/32616
        * gfortran.dg/argument_checking_15.f90: New.
        * gfortran.dg/argument_checking_5.f90: Change TODO into
        dg-warning.

From-SVN: r131643

enum_4.f90: Replace dg-excess-errors by dg-error.

2008-01-18  Tobias Burnus  <burnus@net-b.de>

        * gfortran.dg/enum_4.f90: Replace dg-excess-errors by dg-error.
        * gfortran.dg/enum_5.f90: Ditto.
        * gfortran.dg/enum_6.f90: Ditto.
        * gfortran.dg/bom_error.f90: Ditto.
        * gfortran.dg/func_decl_4.f90: Ditto.

From-SVN: r131642

re PR fortran/34782 (tab format failure to display properly (regression vs. g77))

2007-01-18  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR target/34782
* gfortran.dg/fmt_t_6.f: New test.

From-SVN: r131641

transfer.c (formatted_transfer_scalar): Set max_pos to the greater of the current max_pos or the newly calculated position.

2008-01-18  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

* io/transfer.c (formatted_transfer_scalar): Set max_pos to the greater
of the current max_pos or the newly calculated position.

From-SVN: r131640

write.c (write_real): Increase default precision for

2008-01-18  Tobias Burnus  <burnus@net-b.de>

        * io/write.c (write_real): Increase default precision for
        * REAL(16) by one.

2008-01-18  Tobias Burnus  <burnus@net-b.de>

        * gfortran.dg/large_real_kind_form_io_1.f90: Enlarge string for
        * internal I/O.

From-SVN: r131639

two-stage2.C: fix comment.

2008-01-18  Jonathan Wakely  <jwakely.gcc@gmail.com>

* g++.dg/lookup/two-stage2.C: fix comment.

From-SVN: r131636

re PR c++/33887 (Reference to bitfield gets wrong value when optimizing)

PR c++/33887
* link.cc (_Jv_Linker::prepare_constant_time_tables):
has_interfaces is boolean, treat it as such.

From-SVN: r131632

Document -fno-merge-debug-strings rather than -fmerge-debug-strings.

From-SVN: r131631

common.opt: Add fmerge-debug-strings.

./: * common.opt: Add fmerge-debug-strings.
* dwarf2out.c (DEBUG_STR_SECTION_FLAGS): Test
flag_merge_debug_strings rather than flag_merge_constants.
* doc/invoke.texi (Option Summary): Mention
-fmerge-debug-strings.
(Debugging Options): Document -fmerge-debug-strings.
testsuite/:
* gcc.dg/debug/dwarf2/var1.c: Add -fno-merge-debug-strings.

From-SVN: r131630

re PR c++/33407 (C++ operator new and new expression do not change dynamic type)

PR c++/33407
./: * tree.h (DECL_IS_OPERATOR_NEW): Define.
(struct tree_function_decl): Add new field operator_new_flag.
* tree-inline.c (expand_call_inline): When inlining a call to
operator new, force the return value to go into a variable, and
set DECL_NO_TBAA_P on that variable.
* c-decl.c (merge_decls): Merge DECL_IS_OPERATOR_NEW flag.
cp/:
* decl.c (duplicate_decls): Copy DECL_IS_OPERATOR_NEW flag.
(grok_op_properties): For NEW_EXPR and VEC_NEW_EXPR set
DECL_IS_OPERATOR_NEW flag.
testsuite/:
* g++.dg/init/new26.C: New test.

From-SVN: r131629

re PR middle-end/34801 (FAIL: gcc.dg/Warray-bounds.c)

2008-01-18  Richard Guenther   <rguenther@suse.de>

PR middle-end/34801
* gcc.dg/Warray-bounds.c: XFAIL two tests, remove one
redundant one.

From-SVN: r131628

pr34355.C: Correct order of dg-do and dg-require-effective-target.

* g++.dg/tree-ssa/pr34355.C: Correct order of dg-do and
dg-require-effective-target.

From-SVN: r131627

re PR target/34484 (libgcc should check if feclearexcept (and others) available for BID support on uclibc)

PR debug/34484
* dwarf2out.c (dwarf2out_switch_text_section): Do not guard with
DWARF2_DEBUGGING_INFO.
(dwarf2out_note_section_used): Ditto.  Add prototype.
(have_multiple_function_sections, text_section_used,
cold_text_section_used, *cold_text_sections): Move declarations
before their uses.

From-SVN: r131626

[multiple changes]

2008-01-18  Benjamin Kosnik  <bkoz@redhat.com>

* docs/*: To...
* doc/*: ...here.
* testsuite/Makefile.am: Move doc-performance to...
* Makefile.am: Add doc to SUBDIRS, move doxygen-* rules to...
* doc/Makefile.am: Consolidate documentation creation here.
(doc-doxygen-html): New.
(doc-doxygen-man): New.
(doc-performance): New.
* doc/Makefile.in: New.
* acinclude.m4 (glibcxx_SUBDIRS): Add doc directory.
* doc/doxygen/guide.html: Edit for unified html configuration.
* doc/doxygen/mainpage.html: Same.
* doc/doxygen/run_doxygen: Same, more namespace fixups for man
generation.
* doc/doxygen/user.cfg.in: Update for doxygen 1.5.4.

* include/tr1_impl/random: Remove maint from doxygen markup.
* include/tr1_impl/functional: Same.
* include/std/tuple: Same.
* include/std/streambuf: Same.
* include/std/bitset: Same.
* include/std/limits: Same.
* include/std/fstream: Same.
* include/std/istream: Same.
* include/std/sstream: Same.
* include/ext/pool_allocator.h: Same.
* include/ext/rc_string_base.h: Same.
* include/bits/basic_ios.h: Same.
* include/bits/stl_list.h: Same.
* include/bits/stl_map.h: Same.
* include/bits/locale_classes.h: Same.
* include/bits/stl_set.h: Same.
* include/bits/stl_iterator_base_types.h: Same.
* include/bits/basic_string.h: Same.
* include/bits/stl_multimap.h: Same.
* include/bits/stl_vector.h: Same.
* include/bits/ios_base.h: Same.
* include/bits/stl_deque.h: Same.
* include/bits/postypes.h: Same.
* include/bits/stl_multiset.h: Same.
* include/bits/stl_algo.h: Same.
* include/bits/stl_iterator.h: Same.
* include/bits/stl_tempbuf.h: Same.
* include/bits/stl_construct.h: Same.
* include/bits/stl_relops.h: Same.
* include/tr1/tuple: Same.
* include/backward/auto_ptr.h: Same.

* testsuite/23_containers/vector/requirements/dr438/assign_neg.cc:
Fixups for line number changes.
* testsuite/23_containers/vector/requirements/dr438/insert_neg.cc: Same.
* testsuite/23_containers/vector/requirements/dr438/
constructor_1_neg.cc: Same.
* testsuite/23_containers/vector/requirements/dr438/
constructor_2_neg.cc: Same.
* testsuite/23_containers/deque/requirements/dr438/assign_neg.cc: Same.
* testsuite/23_containers/deque/requirements/dr438/insert_neg.cc: Same.
* testsuite/23_containers/deque/requirements/dr438/
constructor_1_neg.cc: Same.
* testsuite/23_containers/deque/requirements/dr438/
constructor_2_neg.cc: Same.
* testsuite/23_containers/list/requirements/dr438/assign_neg.cc: Same.
* testsuite/23_containers/list/requirements/dr438/insert_neg.cc: Same.
* testsuite/23_containers/list/requirements/dr438/
constructor_1_neg.cc: Same.
* testsuite/23_containers/list/requirements/dr438/
constructor_2_neg.cc: Same.
* testsuite/20_util/auto_ptr/assign_neg.cc: Same.

* aclocal.m4: Regenerate.
* config.h.in: Regenerate.
* configure: Regenerate.
* Makefile.in: Regenerate.
* src/Makefile.in: Regenerate.
* po/Makefile.in: Regenerate.
* libmath/Makefile.in: Regenerate.
* include/Makefile.in: Regenerate.
* libsupc++/Makefile.in: Regenerate.
* testsuite/Makefile.in: Regenerate.

* scripts/make_graphs.py: Correct paths for new layout.

2008-01-17  Benjamin Kosnik  <bkoz@redhat.com>

* acinclude.m4 (AC_LC_MESSAGES): Remove serial.
* linkage.m4 (AC_REPLACE_MATHFUNCS): Same.
* configure: Regenerate.
* aclocal.m4: Regenerate.

From-SVN: r131625

re PR fortran/34556 (Rejects valid with bogus error message: parameter initalization)

2007-01-18  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR target/34556
* gfortran.dg/reshape_2.f90: New test.

From-SVN: r131624

re PR fortran/34556 (Rejects valid with bogus error message: parameter initalization)

2008-01-17  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR fortran/34556
* simplify.c (is_constant_array_expr): New static function that returns
true if the given expression is an array and is constant.
(gfc_simplify_reshape): Use new function.

From-SVN: r131623

unwind-dw2-xtensa.h (_Unwind_FrameState): Remove pc field and add signal_ra.

* config/xtensa/unwind-dw2-xtensa.h (_Unwind_FrameState): Remove pc
field and add signal_ra.
* config/xtensa/unwind-dw2-xtensa.c (uw_frame_state_for): Remove
assignments to frame state pc.  Move end of stack check after
MD_FALLBACK_FRAME_STATE_FOR.
(uw_update_context_1): Use frame state signal_regs if set, instead
of checking signal_frame flag.
(uw_update_context): Use frame state signal_ra if set.
* config/xtensa/linux.h (MD_UNWIND_SUPPORT): Define.
* config/xtensa/linux-unwind.h: New file.

From-SVN: r131622

re PR fortran/33375 (ICE (segfault) gfortran.dg/common_6.f90)

2008-01-17  H.J. Lu  <hongjiu.lu@intel.com>

PR fortran/33375
* symbol.c (free_common_tree): Renamed to ...
(gfc_free_common_tree): This.  Remove static.
(gfc_free_namespace): Updated.

* gfortran.h (gfc_free_common_tree): New.

* match.c (gfc_match_common): Call gfc_free_common_tree () with
gfc_current_ns->common_root and set gfc_current_ns->common_root
to NULL on syntax error.

From-SVN: r131621

re PR fortran/34686 (Aliasing bug when returning character pointers)

2008-01-18  Richard Sandiford  <rsandifo@nildram.co.uk>

       PR fortran/34686
       * trans-expr.c (gfc_conv_function_call): Use proper
       type for returned character pointers.

From-SVN: r131620