platform/upstream/gcc.git
17 years agolibstdc++.exp (v3-build_support): Link libtestc++.a, not a bunch of object files.
Benjamin Kosnik [Tue, 10 Oct 2006 11:35:42 +0000 (11:35 +0000)]
libstdc++.exp (v3-build_support): Link libtestc++.a, not a bunch of object files.

2006-10-10  Benjamin Kosnik  <bkoz@redhat.com>

* testsuite/lib/libstdc++.exp (v3-build_support): Link
libtestc++.a, not a bunch of object files.

From-SVN: r117601

17 years agore PR libstdc++/29118 (Timeouts in libstdc++, libjava and libgomp testsuites)
Benjamin Kosnik [Tue, 10 Oct 2006 10:14:13 +0000 (10:14 +0000)]
re PR libstdc++/29118 (Timeouts in libstdc++, libjava and libgomp testsuites)

2006-10-09  Benjamin Kosnik  <bkoz@redhat.com>

PR libstdc++/29118
* src/locale_init.cc (__get_locale_mutex): New.
(locale::locale): Use it.
(locale::global): Use it.

From-SVN: r117600

17 years agore PR middle-end/29272 (memcpy optimization causes wrong-code)
Jakub Jelinek [Tue, 10 Oct 2006 09:46:59 +0000 (11:46 +0200)]
re PR middle-end/29272 (memcpy optimization causes wrong-code)

PR middle-end/29272
* builtins.c (var_decl_component_p): New function.
(fold_builtin_memset, fold_builtin_memory_op): Restrict
single entry optimization to variables and components thereof.

* gcc.c-torture/execute/20060930-2.c: New test.

From-SVN: r117599

17 years agore PR rtl-optimization/29323 (set_nothrow_function_flags does invalid analysis on...
Richard Guenther [Tue, 10 Oct 2006 08:27:02 +0000 (08:27 +0000)]
re PR rtl-optimization/29323 (set_nothrow_function_flags does invalid analysis on weak functions)

2006-10-10  Richard Guenther  <rguenther@suse.de>

PR rtl-optimization/29323
* except.c (set_nothrow_function_flags): For functions
that do not bind local bail out early.

* decl.c (finish_function): Set TREE_NOTHROW only for
functions that bind local.

From-SVN: r117598

17 years ago* Add new port for score.
Chen Liqin [Tue, 10 Oct 2006 07:50:29 +0000 (07:50 +0000)]
* Add new port for score.

From-SVN: r117597

17 years agoinvoke.texi (RS/6000 and PowerPC Options): Add -m to the -mdlmzb option.
Ben Elliston [Tue, 10 Oct 2006 07:34:10 +0000 (07:34 +0000)]
invoke.texi (RS/6000 and PowerPC Options): Add -m to the -mdlmzb option.

* doc/invoke.texi (RS/6000 and PowerPC Options): Add -m to the
-mdlmzb option.

From-SVN: r117596

17 years agoDaily bump.
GCC Administrator [Tue, 10 Oct 2006 00:17:15 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117593

17 years agore PR libstdc++/29095 (cxxabi.h __cxa_cdtor_type not declared when included from...
Benjamin Kosnik [Mon, 9 Oct 2006 23:53:35 +0000 (23:53 +0000)]
re PR libstdc++/29095 (cxxabi.h __cxa_cdtor_type not declared when included from "C")

2006-10-09  Benjamin Kosnik  <bkoz@redhat.com>

PR libstdc++/29095
* libsupc++/cxxabi.h (__cxa_cdtor_type): Explicit "C" linkage.
* config/cpu/arm/cxxabi_tweaks.h: Same.
* config/cpu/generic/cxxabi_tweaks.h: Same.
* testsuite/abi: Add.
* testsuite/abi/header_cxxabi.cc: New.
* testsuite/demangle: Move...
* testsuite/abi/demangle: ...here.
* testsuite/libstdc++-dg/conformance.exp: Adjust testsuite file
calculation.
* scripts/create_testsuite_files: Same.
* testsuite/lib/libstdc++.exp (v3_target_compile_as_c): New.
(libstdc++-dg-test): Use it.

From-SVN: r117589

17 years agoChangelog: Undid erroneous tab-to-space conversion on last entry.
Brooks Moses [Mon, 9 Oct 2006 23:23:53 +0000 (16:23 -0700)]
Changelog: Undid erroneous tab-to-space conversion on last entry.

2006-10-09  Brooks Moses  <bmoses@stanford.edu>

* Changelog: Undid erroneous tab-to-space conversion on last entry.

From-SVN: r117588

17 years agoChangelog: Added spaces after * on last two entries.
Brooks Moses [Mon, 9 Oct 2006 23:19:58 +0000 (16:19 -0700)]
Changelog: Added spaces after * on last two entries.

2006-10-09  Brooks Moses  <bmoses@stanford.edu>

* Changelog: Added spaces after * on last two entries.

From-SVN: r117587

17 years agoMake-lang.in: Added intrinsic.texi to GFORTRAN_TEXI dependences.
Brooks Moses [Mon, 9 Oct 2006 23:17:52 +0000 (23:17 +0000)]
Make-lang.in: Added intrinsic.texi to GFORTRAN_TEXI dependences.

2006-10-09  Brooks Moses  <bmoses@stanford.edu>

*Make-lang.in: Added intrinsic.texi to GFORTRAN_TEXI
dependences.

From-SVN: r117586

17 years agointrinsic.texi: (MOVE_ALLOC) changed "Options" to "Standards".
Brooks Moses [Mon, 9 Oct 2006 22:33:55 +0000 (22:33 +0000)]
intrinsic.texi: (MOVE_ALLOC) changed "Options" to "Standards".

2006-10-09  Brooks Moses  <bmoses@stanford.edu>

*intrinsic.texi: (MOVE_ALLOC) changed "Options" to "Standards".

From-SVN: r117585

17 years agore PR fortran/15441 (RRSPACING broken for denormals)
Steven G. Kargl [Mon, 9 Oct 2006 20:55:29 +0000 (20:55 +0000)]
re PR fortran/15441 (RRSPACING broken for denormals)

2006-10-06  Steven G. Kargl  <kargl@gcc.gnu.org>

* gfortran.h: Define GFC_MPFR_TOO_OLD via mpfr version info.
* arith.c (arctangent, gfc_check_real_range): Use it.
* simplify.c (gfc_simplify_atan2, gfc_simplify_exponent,
gfc_simplify_log, gfc_simplify_nearest): Use it.

PR fortran/15441
PR fortran/29312
* iresolve.c (gfc_resolve_rrspacing): Give rrspacing library
routine hidden precision argument.
(gfc_resolve_spacing): Give spacing library routine hidden
precision, emin - 1, and tiny(x) arguments.
* simplify.c (gfc_simplify_nearest): Remove explicit subnormalization.
(gfc_simplify_rrspacing): Implement formula from Fortran 95 standard.
(gfc_simplify_spacing): Implement formula from Fortran 2003 standard.
* trans-intrinsic.c (gfc_intrinsic_map_t) Declare rrspacing and
spacing via LIBF_FUNCTION
(prepare_arg_info, call_builtin_clz, gfc_conv_intrinsic_spacing,
gfc_conv_intrinsic_rrspacing): Remove functions.
(gfc_conv_intrinsic_function): Remove calls to
gfc_conv_intrinsic_spacing and gfc_conv_intrinsic_rrspacing.
* f95-lang.c (gfc_init_builtin_functions): Remove __builtin_clz,
__builtin_clzl and __builtin_clzll

2006-10-06  Steven G. Kargl  <kargl@gcc.gnu.org>

PR fortran/15441
PR fortran/29312
* configure.ac: Add HAVE_LDEXPF, HAVE_LDEXP, and HAVE_LDEXPL
* m4/spacing.m4: New file.  Use new HAVE_* defines.
* m4/rrspacing.m4: Ditto.
* Makefile.am: Handle new files.
* configure: Regenerated.
* Makefile.in: Ditto.
* config.h.in: Ditto.
* generated/spacing_r4.c: Generated.
* generated/spacing_r8.c: Ditto.
* generated/spacing_r10.c: Ditto.
* generated/spacing_r16.c: Ditto.
* generated/rrspacing_r4.c: Ditto.
* generated/rrspacing_r8.c: Ditto.
* generated/rrspacing_r10.c: Ditto.
* generated/rrspacing_r16.c: Ditto.

From-SVN: r117584

17 years agoPR libstdc++/28277 (partial: __add_grouping)
Paolo Carlini [Mon, 9 Oct 2006 18:04:18 +0000 (18:04 +0000)]
PR libstdc++/28277 (partial: __add_grouping)

2006-10-09  Paolo Carlini  <pcarlini@suse.de>

PR libstdc++/28277 (partial: __add_grouping)
* include/bits/locale_facets.tcc (__add_grouping<>(_CharT*, _CharT,
const char*, size_t, const _CharT*, const _CharT*)): Rewrite in
non-recursive form.

From-SVN: r117581

17 years agoRevert emutls patch.
Richard Henderson [Mon, 9 Oct 2006 16:27:14 +0000 (09:27 -0700)]
Revert emutls patch.

From-SVN: r117578

17 years agore PR middle-end/29254 (verify_cgraph_node failed (inlined_to pointer is set but...
Richard Guenther [Mon, 9 Oct 2006 16:10:38 +0000 (16:10 +0000)]
re PR middle-end/29254 (verify_cgraph_node failed (inlined_to pointer is set but no predecessors found))

2006-10-09  Richard Guenther  <rguenther@suse.de>

PR middle-end/29254
* cgraphunit.c (verify_cgraph_node): Bail out on earlier
errors.

* gcc.dg/pr29254.c: New testcase.

From-SVN: r117577

17 years agore PR target/27880 (undefined reference to `_Unwind_GetIPInfo')
Steve Ellcey [Mon, 9 Oct 2006 15:55:38 +0000 (15:55 +0000)]
re PR target/27880 (undefined reference to `_Unwind_GetIPInfo')

PR target/27880
* unwind-c.c (PERSONALITY_FUNCTION): Ifdef use of _Unwind_GetIPInfo.
* configure.ac (HAVE_GETIPINFO): Check for _Unwind_GetIPInfo.
* configure: Regenerate.
* config.in: Regenerate.

From-SVN: r117576

17 years ago* tree-mudflap.c (mf_xform_derefs_1): Add parentheses around TREE_TYPE.
Aldy Hernandez [Mon, 9 Oct 2006 13:58:10 +0000 (13:58 +0000)]
* tree-mudflap.c (mf_xform_derefs_1): Add parentheses around TREE_TYPE.

From-SVN: r117575

17 years agoPR libstdc++/28277 (partial: money_put bits)
Paolo Carlini [Mon, 9 Oct 2006 10:49:50 +0000 (10:49 +0000)]
PR libstdc++/28277 (partial: money_put bits)

2006-10-09  Paolo Carlini  <pcarlini@suse.de>

PR libstdc++/28277 (partial: money_put bits)
* include/bits/locale_facets.tcc (money_put<>::_M_insert(iter_type,
ios_base&, char_type, const string_type&)): Avoid __builtin_alloca
with no limit, do the work in place.

* include/bits/locale_facets.tcc (money_put<>::do_put(iter_type,
bool, ios_base&, char_type, long double)): Avoid unnecessary
__builtin_alloca, do the work in place.

From-SVN: r117571

17 years agoglobals_io.cc (_GLIBCXX_once): Remove, unused.
Benjamin Kosnik [Mon, 9 Oct 2006 10:34:49 +0000 (10:34 +0000)]
globals_io.cc (_GLIBCXX_once): Remove, unused.

2006-10-09  Benjamin Kosnik  <bkoz@redhat.com>

* src/globals_io.cc (_GLIBCXX_once): Remove, unused.
(_GLIBCXX_mutex): Same.
(_GLIBCXX_mutex_addres): Same.
(_GLIBCXX_mutex_init): Same.
(_GLIBCXX_mutex_address_init): Same.

From-SVN: r117570

17 years agointrinsic.c (add_sym_0s, [...]): Use macro ACTUAL_NO...
Francois-Xavier Coudert [Mon, 9 Oct 2006 07:43:39 +0000 (09:43 +0200)]
intrinsic.c (add_sym_0s, [...]): Use macro ACTUAL_NO...

* intrinsic.c (add_sym_0s, add_sym_1s, add_sym_2s, add_sym_3s,
add_sym_4s, add_sym_5s, add_functions): Use macro ACTUAL_NO,
ACTUAL_YES, NOT_ELEMENTAL and ELEMENTAL instead of constants
0 and 1 as second and third arguments to add_sym* functions.

From-SVN: r117569

17 years agoDaily bump.
GCC Administrator [Mon, 9 Oct 2006 00:17:21 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117565

17 years agoMakefile.am: Add intrinsics/move_alloc.c, ommited in Paul's previous commit.
François-Xavier Coudert [Sun, 8 Oct 2006 19:07:29 +0000 (19:07 +0000)]
Makefile.am: Add intrinsics/move_alloc.c, ommited in Paul's previous commit.

* Makefile.am: Add intrinsics/move_alloc.c, ommited in Paul's
previous commit.
* Makefile.in: Regenerate.

From-SVN: r117561

17 years agohyper.c: Remove file.
Francois-Xavier Coudert [Sun, 8 Oct 2006 18:39:18 +0000 (20:39 +0200)]
hyper.c: Remove file.

* intrinsics/hyper.c: Remove file.
* intrinsics/c99_functions.c: Add fallback functions asinhf,
acoshf and atanhf.
* Makefile.am: Remove file intrinsics/hyper.c.
* Makefile.in: Regenerate.

From-SVN: r117560

17 years agoc99_protos.h: Correctly protect definitions of prototypes for asinhf...
Francois-Xavier Coudert [Sun, 8 Oct 2006 17:16:38 +0000 (19:16 +0200)]
c99_protos.h: Correctly protect definitions of prototypes for asinhf...

* c99_protos.h: Correctly protect definitions of prototypes for
asinhf, acoshf and atanhf to be the same as in intrinsics/hyper.c.

From-SVN: r117559

17 years ago[multiple changes]
Paul Thomas [Sun, 8 Oct 2006 16:21:55 +0000 (16:21 +0000)]
[multiple changes]

2006-10-05  Erik Edelmann <edelmann@gcc.gnu.org>
    Paul Thomas <pault@gcc.gnu.org>

PR fortran/20541
* interface.c (gfc_compare_derived_types): Add comparison of
the allocatable field.
* intrinsic.c (add_subroutines): Add MOVE_ALLOC.
* trans-expr.c (gfc_conv_aliased_arg, gfc_trans_subarray_assign,
gfc_trans_subcomponent_assign, gfc_conv_string_parameter,
gfc_trans_scalar_assign): Add extra arguments l_is_temp
and r_is_var to references to latter function.
(gfc_conv_function_call): Add enum for types of argument and
an associated variable parm_kind. Deallocate components of
INTENT(OUT) and non-variable arrays.
(gfc_trans_subcomponent_assign): Add block to assign arrays
to allocatable components.
(gfc_trans_scalar_assign): Add block to handle assignments of
derived types with allocatable components, using the above new
arguments to control allocation/deallocation of memory and the
copying of allocated arrays.
* trans-array.c (gfc_array_allocate): Remove old identification
of pointer and replace with that of an allocatable array. Add
nullify of structures with allocatable components.
(gfc_conv_array_initializer): Treat EXPR_NULL.
(gfc_conv_array_parameter): Deallocate allocatable components
of non-variable structures.
(gfc_trans_dealloc_allocated): Use second argument of library
deallocate to inhibit, without error, freeing NULL pointers.
(get_full_array_size): New function to return the size of a
full array.
(gfc_duplicate_allocatable): New function to allocate and copy
allocated data.
(structure_alloc_comps): New recursive function to deallocate,
nullify or copy allocatable components.
(gfc_nullify_alloc_comp, gfc_deallocate_alloc_comp,
gfc_copy_alloc_comp): New interface functions to call previous.
(gfc_trans_deferred_array): Add the code to nullify allocatable
components, when entering scope, and to deallocate them on
leaving. Do not call gfc_trans_static_array_pointer and return
for structures with allocatable components and default
initializers.
* symbol.c (gfc_set_component_attr): Set allocatable field.
(gfc_get_component_attr): Set the allocatable attribute.
* intrinsic.h : Prototype for gfc_check_move_alloc.
* decl.c (build_struct): Apply TR15581 constraints for
allocatable components.
(variable_decl): Default initializer is always NULL for
allocatable components.
(match_attr_spec): Allow, or not, allocatable components,
according to the standard in force.
* trans-array.h : Prototypes for gfc_nullify_alloc_comp,
gfc_deallocate_alloc_comp, gfc_copy_alloc_comp and
gfc_duplicate_allocatable.
* gfortran.texi : Add mention of TR15581 extensions.
* gfortran.h : Add attribute alloc_comp, add
gfc_components field allocatable and add the prototype
for gfc_expr_to_initialize.
* trans-stmt.c (generate_loop_for_temp_to_lhs,
generate_loop_for_rhs_to_temp, gfc_trans_where_assign,
gfc_trans_where_3): Add extra arguments to calls to
gfc_trans_scalar_assign and set appropriately.
(gfc_trans_allocate): Nullify allocatable components.
(gfc_trans_deallocate): Deallocate to ultimate allocatable
components but stop at ultimate pointer components.
* module.c (mio_symbol_attribute, mio_symbol_attribute,
mio_component): Add module support for allocatable
components.
* trans-types.c (gfc_get_derived_type): Treat allocatable
components.
* trans.h : Add two boolean arguments to
gfc_trans_scalar_assign.
* resolve.c (resolve_structure_cons): Check conformance of
constructor element and the component.
(resolve_allocate_expr): Add expression to nullify the
constructor expression for allocatable components.
(resolve_transfer): Inhibit I/O of derived types with
allocatable components.
(resolve_fl_derived): Skip check of bounds of allocatable
components.
* trans-decl.c (gfc_get_symbol_decl): Add derived types
with allocatable components to deferred variable.
(gfc_trans_deferred_vars): Make calls for derived types
with allocatable components to gfc_trans_deferred_array.
(gfc_generate_function_code): Nullify allocatable
component function result on entry.
* parse.c (parse_derived): Set symbol attr.allocatable if
allocatable components are present.
* check.c (gfc_check_allocated): Enforce attr.allocatable
for intrinsic arguments.
(gfc_check_move_alloc): Check arguments of move_alloc.
* primary.c (gfc_variable_attr): Set allocatable attribute.
* intrinsic.texi : Add index entry and section for
for move_alloc.

PR fortran/29115
* resolve.c (resolve_structure_cons): It is an error if the
pointer component elements of a derived type constructor are
not pointer or target.

PR fortran/29211
* trans-stmt.c (generate_loop_for_temp_to_lhs,
generate_loop_for_rhs_to_temp): Provide a string length for
the temporary by copying that of the other side of the scalar
assignment.

2006-10-05  Paul Thomas  <pault@gcc.gnu.org>
    Erik Edelmann  <edelmann@gcc.gnu.org>

PR libgfortran/20541
* Makefile.in : Add move_alloc.
* intrinsics/move_alloc.c: New function.
* Makefile.am : Add move_alloc.

2006-10-05  Erik Edelmann  <edelmann@gcc.gnu.org>
    Paul Thomas  <pault@gcc.gnu.org>

PR fortran/20541
* gfortran.dg/alloc_comp_basics_1.f90: New test.
* gfortran.dg/alloc_comp_basics_2.f90: New test.
* gfortran.dg/alloc_comp_assign_1.f90: New test.
* gfortran.dg/alloc_comp_assign_2.f90: New test.
* gfortran.dg/alloc_comp_assign_3.f90: New test.
* gfortran.dg/alloc_comp_assign_4.f90: New test.
* gfortran.dg/alloc_comp_constraint_1.f90: New test.
* gfortran.dg/alloc_comp_constraint_2.f90: New test.
* gfortran.dg/alloc_comp_constraint_3.f90: New test.
* gfortran.dg/alloc_comp_constructor_1.f90: New test.
* gfortran.dg/alloc_comp_constructor_2.f90: New test.
* gfortran.dg/alloc_comp_initializer_1.f90: New test.
* gfortran.dg/alloc_comp_std.f90: New test.
* gfortran.dg/move_alloc.f90: New test.

PR fortran/29115
* gfortran.dg/derived_constructor_comps_2.f90: New test.

PR fortran/29211
* gfortran.dg/forall_char_dependencies_1.f90: New test.

From-SVN: r117558

17 years agomips.c (mips_legitimize_tls_address): Implement all TLS models in the same way as...
Richard Sandiford [Sun, 8 Oct 2006 14:33:22 +0000 (14:33 +0000)]
mips.c (mips_legitimize_tls_address): Implement all TLS models in the same way as local exec if !TARGET_ABICALLS.

gcc/
* config/mips/mips.c (mips_legitimize_tls_address): Implement
all TLS models in the same way as local exec if !TARGET_ABICALLS.
Remove stray newline.

From-SVN: r117557

17 years agore PR libfortran/26540 ([4.1 only] intrinsics/signal.c warnings)
Francois-Xavier Coudert [Sun, 8 Oct 2006 14:27:22 +0000 (16:27 +0200)]
re PR libfortran/26540 ([4.1 only] intrinsics/signal.c warnings)

PR libfortran/26540
* intrinsics/signal.c (signal_sub, signal_sub_int): Use intptr_t
if available to cast function pointers to int and back.
* configure.ac: Check for intptr_t.
* config.h.in: Regenerate.
* configure: Regenerate.

From-SVN: r117556

17 years agore PR fortran/28585 (Fortran 2003: Support NEW_LINE intrinsic)
Tobias Burnus [Sun, 8 Oct 2006 13:21:42 +0000 (15:21 +0200)]
re PR fortran/28585 (Fortran 2003: Support NEW_LINE intrinsic)

PR fortran/28585

* intrinsic.c (add_functions): Add new_line Fortran 2003 intrinsic.
* intrinsic.h: Add gfc_simplify_new_line and gfc_check_new_line
prototypes.
* check.c (gfc_check_new_line): New function.
* simplify.c (gfc_simplify_new_line): New function.
* intrinsic.texi: Document new_line intrinsic.

* gfortran.dg/new_line.f90: New test.

From-SVN: r117555

17 years agomips.c (mips_classify_symbol): Do not return SYMBOL_SMALL_DATA if SYMBOL_REF_WEAK.
Richard Sandiford [Sun, 8 Oct 2006 07:56:53 +0000 (07:56 +0000)]
mips.c (mips_classify_symbol): Do not return SYMBOL_SMALL_DATA if SYMBOL_REF_WEAK.

gcc/
* config/mips/mips.c (mips_classify_symbol): Do not return
SYMBOL_SMALL_DATA if SYMBOL_REF_WEAK.
(mips_in_small_data_p): Tweak comments.

From-SVN: r117553

17 years agoelf.h (HANDLE_PRAGMA_PUSH_POP): Define.
Richard Sandiford [Sun, 8 Oct 2006 07:35:47 +0000 (07:35 +0000)]
elf.h (HANDLE_PRAGMA_PUSH_POP): Define.

gcc/
* config/mips/elf.h (HANDLE_PRAGMA_PUSH_POP): Define.

From-SVN: r117552

17 years agoPR libstdc++/28277 (partial: money_get bits)
Paolo Carlini [Sun, 8 Oct 2006 01:13:03 +0000 (01:13 +0000)]
PR libstdc++/28277 (partial: money_get bits)

2006-10-07  Paolo Carlini  <pcarlini@suse.de>

PR libstdc++/28277 (partial: money_get bits)
* include/bits/locale_facets.tcc (money_get<>::do_get(iter_type,
iter_type, bool, ios_base&, ios_base::iostate&, string_type&)):
Avoid __builtin_alloca with no limit, do the work in place.

From-SVN: r117549

17 years agobuiltins-config.h: Ensure we use -std=c99 on solaris2.
Kaveh R. Ghazi [Sun, 8 Oct 2006 00:48:26 +0000 (00:48 +0000)]
builtins-config.h: Ensure we use -std=c99 on solaris2.

* gcc.dg/builtins-config.h: Ensure we use -std=c99 on solaris2.
* gcc.dg/torture/builtin-convert-1.c,
gcc.dg/torture/builtin-convert-2.c,
gcc.dg/torture/builtin-convert-3.c,
gcc.dg/torture/builtin-power-1.c: Use -std=c99 on solaris2.

From-SVN: r117548

17 years agoDaily bump.
GCC Administrator [Sun, 8 Oct 2006 00:17:14 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117546

17 years agobool-1.c, [...]: Fix comment typos.
Kazu Hirata [Sat, 7 Oct 2006 21:06:03 +0000 (21:06 +0000)]
bool-1.c, [...]: Fix comment typos.

* gcc.dg/tree-ssa/bool-1.c, gcc.dg/tree-ssa/bool-2.c,
gcc.dg/tree-ssa/bool-4.c, gcc.dg/tree-ssa/bool-5.c,
gcc.dg/tree-ssa/bool-6.c, gcc.dg/tree-ssa/bool-8.c,
gcc.dg/tree-ssa/bool-9.c: Fix comment typos.

From-SVN: r117541

17 years agostruct-layout-encoding-1_generate.c: Xfail some tests for powerpc-darwin and powerpc...
Andrew Pinski [Sat, 7 Oct 2006 17:37:50 +0000 (10:37 -0700)]
struct-layout-encoding-1_generate.c: Xfail some tests for powerpc-darwin and powerpc-aix.

2006-10-07  Andrew Pinski  <pinskia@gmail.com>

        * objc.dg/gnu-encoding/struct-layout-encoding-1_generate.c:
        Xfail some tests for powerpc-darwin and powerpc-aix.

From-SVN: r117540

17 years agore PR c/29380 (FAIL: gcc.dg/pr29330.c (test for excess errors))
Jakub Jelinek [Sat, 7 Oct 2006 16:50:23 +0000 (18:50 +0200)]
re PR c/29380 (FAIL: gcc.dg/pr29330.c (test for excess errors))

PR c/29380
* gcc.dg/pr29330.c: Add -std=gnu99 to dg-options.

From-SVN: r117539

17 years agore PR target/29300 (FAIL: gcc.dg/pthread-init-[12].c (test for excess errors))
John David Anglin [Sat, 7 Oct 2006 16:42:29 +0000 (16:42 +0000)]
re PR target/29300 (FAIL: gcc.dg/pthread-init-[12].c (test for excess errors))

PR target/29300
* gcc.dg/pthread-init-2.c (dg-options): Define _POSIX_C_SOURCE=199506L
on hppa*-*-hpux*.

From-SVN: r117538

17 years agore PR target/29300 (FAIL: gcc.dg/pthread-init-[12].c (test for excess errors))
John David Anglin [Sat, 7 Oct 2006 16:35:11 +0000 (16:35 +0000)]
re PR target/29300 (FAIL: gcc.dg/pthread-init-[12].c (test for excess errors))

PR target/29300
* inclhack.def (hpux_pthread_initializers): New hack.
* tests/base/sys/pthread.h: New file.

* fixincl.x: Regenerate.

From-SVN: r117537

17 years agointrinsic.c (add_functions): Add comments for gfc_check_access_func and gfc_resolve_i...
Francois-Xavier Coudert [Sat, 7 Oct 2006 13:49:35 +0000 (15:49 +0200)]
intrinsic.c (add_functions): Add comments for gfc_check_access_func and gfc_resolve_index_func.

* gcc/fortran/intrinsic.c (add_functions): Add comments for
gfc_check_access_func and gfc_resolve_index_func.

From-SVN: r117535

17 years agore PR fortran/16580 ([4.1 only] gfortran ICE on test g77.f-torture/execute/intrinsic77.f)
François-Xavier Coudert [Sat, 7 Oct 2006 13:34:16 +0000 (13:34 +0000)]
re PR fortran/16580 ([4.1 only] gfortran ICE on test g77.f-torture/execute/intrinsic77.f)

PR fortran/16580
PR fortran/29288

* gcc/fortran/intrinsic.c (add_sym): Define the actual_ok when a
gfc_intrinsic_sym structure is filled.
(gfc_intrinsic_actual_ok): New function.
(add_sym_0s, add_sym_1s, add_sym_2s, add_sym_3s, add_sym_4s,
add_sym_5s): Intrinsic subroutines are not allowed as actual
arguments, so we remove argument actual_ok.
(add_functions): Correct the values for actual_ok of all intrinsics.
(add_subroutines): Remove the actual_ok argument, which was never used.
* gcc/fortran/intrinsic.h (gfc_intrinsic_actual_ok): New prototype.
* gcc/fortran/gfortran.h (gfc_resolve_index_func): New prototype.
* gcc/fortran/resolve.c (resolve_actual_arglist): Check whether
an intrinsic used as an argument list is allowed there.
* gcc/fortran/iresolve.c (gfc_resolve_index_func): New function.
(gfc_resolve_len): Change intrinsic function name to agree with
libgfortran.
* gcc/fortran/trans-decl.c (gfc_get_extern_function_decl): Add
new case, because some specific intrinsics take 3 arguments.
* gcc/fortran/intrinsic.texi: DIMAG is a GNU extension.

* libgfortran/Makefile.am: Add the new files to the build
process, and rules to build them.
* libgfortran/Makefile.in: Regenerate.
* libgfortran/m4/misc_specifics.m4: New file.
* libgfortran/m4/specific.m4: Add new special cases for function
with complex argument and real result, like abs_c* and aimag_c*.
* libgfortran/intrinsics/f2c_specifics.F90: Add specifics for
AIMAG, ASINH, ACOSH and ATANH.
* libgfortran/generated/_aimag_c4.F90: New file.
* libgfortran/generated/_aimag_c8.F90: New file.
* libgfortran/generated/_asinh_r10.F90: New file.
* libgfortran/generated/_acosh_r16.F90: New file.
* libgfortran/generated/_aimag_c10.F90: New file.
* libgfortran/generated/_atanh_r16.F90: New file.
* libgfortran/generated/_acosh_r4.F90: New file.
* libgfortran/generated/_acosh_r8.F90: New file.
* libgfortran/generated/_asinh_r4.F90: New file.
* libgfortran/generated/_asinh_r8.F90: New file.
* libgfortran/generated/_asinh_r16.F90: New file.
* libgfortran/generated/_atanh_r4.F90: New file.
* libgfortran/generated/_atanh_r8.F90: New file.
* libgfortran/generated/_acosh_r10.F90: New file.
* libgfortran/generated/misc_specifics.F90: New file.
* libgfortran/generated/_aimag_c16.F90: New file.
* libgfortran/generated/_atanh_r10.F90: New file.

* gcc/testsuite/gfortran.fortran-torture/execute/specifics.f90:
Add tests for using all possible intrinsics as actual arguments.
* gcc/testsuite/gfortran.dg/specifics_1.f90: Add tests for using
all possible intrinsics as actual arguments.
* gcc/testsuite/gfortran.dg/specifics_2.f90: New file.
* gcc/testsuite/gfortran.dg/specifics_3.f90: New file.

From-SVN: r117534

17 years ago* sched-int.h: Fix comment typos.
Kazu Hirata [Sat, 7 Oct 2006 11:20:13 +0000 (11:20 +0000)]
* sched-int.h: Fix comment typos.

From-SVN: r117532

17 years ago* gimplify.c, sched-int.h: Fix comment typos.
Kazu Hirata [Sat, 7 Oct 2006 11:07:14 +0000 (11:07 +0000)]
* gimplify.c, sched-int.h: Fix comment typos.

From-SVN: r117531

17 years agodebug-1.c: Use -fno-if-conversion MIPS targets.
Richard Sandiford [Sat, 7 Oct 2006 09:22:20 +0000 (09:22 +0000)]
debug-1.c: Use -fno-if-conversion MIPS targets.

gcc/testsuite/
* gcc.dg/debug/debug-1.c: Use -fno-if-conversion MIPS targets.
* gcc.dg/debug/debug-2.c: Likewise.

From-SVN: r117530

17 years agostl_tree.h (_Rb_tree<>::erase(const _Key&)): Don't call std::distance unnecessarily.
Ion Gaztanaga [Sat, 7 Oct 2006 08:20:35 +0000 (08:20 +0000)]
stl_tree.h (_Rb_tree<>::erase(const _Key&)): Don't call std::distance unnecessarily.

2006-10-07  Ion Gaztanaga  <igaztanaga@gmail.com>
    Paolo Carlini  <pcarlini@suse.de>

* include/bits/stl_tree.h (_Rb_tree<>::erase(const _Key&)): Don't
call std::distance unnecessarily.

Co-Authored-By: Paolo Carlini <pcarlini@suse.de>
From-SVN: r117529

17 years agoDaily bump.
GCC Administrator [Sat, 7 Oct 2006 00:17:14 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117525

17 years agosync-2.c: New test.
Uros Bizjak [Fri, 6 Oct 2006 22:24:04 +0000 (00:24 +0200)]
sync-2.c: New test.

* gcc.c-torture/compile/sync-2.c: New test.

From-SVN: r117519

17 years agoactual_array_substr_2.f90: Remove '\r's from line endings.
Tobias Schlüter [Fri, 6 Oct 2006 22:02:43 +0000 (00:02 +0200)]
actual_array_substr_2.f90: Remove '\r's from line endings.

* gfortran.dg/actual_array_substr_2.f90: Remove '\r's from
line endings.

From-SVN: r117518

17 years agoadd missed aprt of patch
Tobias Schlüter [Fri, 6 Oct 2006 21:50:18 +0000 (23:50 +0200)]
add missed aprt of patch

From-SVN: r117517

17 years agosched-int.h (IS_SPECULATION_CHECK_P, [...]): New macros.
Maxim Kuvyrkov [Fri, 6 Oct 2006 21:49:07 +0000 (21:49 +0000)]
sched-int.h (IS_SPECULATION_CHECK_P, [...]): New macros.

2006-10-06  Maxim Kuvyrkov  <mkuvyrkov@ispras.ru>

* sched-int.h (IS_SPECULATION_CHECK_P, IS_SPECULATION_SIMPLE_CHECK_P):
New macros.
* sched-ebb.c (begin_schedule_ready): Use them.
* haifa-sched.c (schedule_insn, move_insn, try_ready,
add_to_speculative_block, create_check_block_twin, speculate_insn,
fix_jump_move, move_block_after_check): Ditto.
* sched-rgn.c (new_ready): Ditto.

From-SVN: r117516

17 years agore PR rtl-optimization/29128 (ICE: in move_block_after_check, at haifa-sched.c:4337)
Maxim Kuvyrkov [Fri, 6 Oct 2006 21:45:13 +0000 (21:45 +0000)]
re PR rtl-optimization/29128 (ICE: in move_block_after_check, at haifa-sched.c:4337)

2006-10-06  Maxim Kuvyrkov  <mkuvyrkov@ispras.ru>

PR rtl-optimization/29128
* sched-int.h (IS_SPECULATION_BRANCHY_CHECK_P): New macro.
* sched-ebb.c (advance_target_bb): Use it to fix condition to
allow interblock movement of speculation checks.

* gcc.c-torture/compile/pr29128.c: New test.

From-SVN: r117515

17 years agoactual_array_constructor_2.f90: Remove '\r's from line endings.
Tobias Schlüter [Fri, 6 Oct 2006 21:40:14 +0000 (23:40 +0200)]
actual_array_constructor_2.f90: Remove '\r's from line endings.

        * gfortran.dg/actual_array_constructor_2.f90: Remove '\r's from
        line endings.

From-SVN: r117514

17 years agore PR target/29256 (loop performance regression)
Zdenek Dvorak [Fri, 6 Oct 2006 19:32:04 +0000 (21:32 +0200)]
re PR target/29256 (loop performance regression)

PR middle-end/29256
* tree-ssa-loop-ivopts.c (determine_base_object): Handle pointers
casted to integer type.
(get_address_cost): Decrease cost of [symbol + index] addressing modes
if they are significantly more expensive than [reg + index] ones.

* gcc.dg/tree-ssa/loop-19.c: New test.

From-SVN: r117513

17 years agore PR tree-optimization/29330 (-O -ftree-loop-linear --> virtual memory exhausted)
Jakub Jelinek [Fri, 6 Oct 2006 16:57:27 +0000 (18:57 +0200)]
re PR tree-optimization/29330 (-O -ftree-loop-linear   -->  virtual memory exhausted)

PR tree-optimization/29330
* tree-data-ref.c (free_data_ref): Use DR_FREE_ACCESS_FNS macro.
(initialize_data_dependence_relation): Clear DDR_LOOP_NEST pointer
on newly allocated ddrs.
(find_loop_nest_1, find_loop_nest): Change LOOP_NEST to a pointer
to VEC (loop_p, heap) pointer.
(compute_data_dependences_for_loop): Adjust caller.
(free_dependence_relations): Free DDR_LOOP_NEST.

* tree-loop-linear.c (linear_transform_loops): Don't forget to
free DEPENDENCE_RELATIONS and DATAREFS.

* gcc.dg/pr29330.c: New test.

From-SVN: r117509

17 years agore PR target/28924 (x86 sync builtins fail for char and short memory operands)
Jakub Jelinek [Fri, 6 Oct 2006 16:54:43 +0000 (18:54 +0200)]
re PR target/28924 (x86 sync builtins fail for char and short memory operands)

PR target/28924
* builtins.c (expand_builtin_sync_operation,
expand_builtin_compare_and_swap, expand_builtin_lock_test_and_set):
Use convert_to_mode to handle promoted arguments.

* gcc.c-torture/compile/20061005-1.c: New test.

From-SVN: r117508

17 years ago* print-tree.c (print_node_brief, print_node): Print sign of Inf.
J"orn Rennecke [Fri, 6 Oct 2006 15:59:47 +0000 (15:59 +0000)]
* print-tree.c (print_node_brief, print_node): Print sign of Inf.

From-SVN: r117506

17 years ago* config/sh/sh.h (LIBGCC2_DOUBLE_TYPE_SIZE): Define
J"orn Rennecke [Fri, 6 Oct 2006 14:44:18 +0000 (14:44 +0000)]
* config/sh/sh.h (LIBGCC2_DOUBLE_TYPE_SIZE): Define

From-SVN: r117505

17 years agosync.md ("sync_add<mode>", [...]): Use <modeconstraint> instead of "r" as operand...
Uros Bizjak [Fri, 6 Oct 2006 11:51:29 +0000 (13:51 +0200)]
sync.md ("sync_add<mode>", [...]): Use <modeconstraint> instead of "r" as operand 1 constraint.

* config/i386/sync.md ("sync_add<mode>", "sync_sub<mode>",
"sync_ior<mode>", "sync_and<mode>", "sync_xor<mode>"):
Use <modeconstraint> instead of "r" as operand 1 constraint.

From-SVN: r117499

17 years agore PR libstdc++/29368 (wrong STL docs for rfind())
Paolo Carlini [Fri, 6 Oct 2006 11:47:56 +0000 (11:47 +0000)]
re PR libstdc++/29368 (wrong STL docs for rfind())

2006-10-06  Paolo Carlini  <pcarlini@suse.de>

PR libstdc++/29368
* include/bits/basic_string.h: Adjust rfind documentation.
* include/ext/vstring.h: Likewise.

From-SVN: r117496

17 years agore PR libstdc++/29354 (Error when seeking on an ostringstream)
Paolo Carlini [Fri, 6 Oct 2006 09:57:43 +0000 (09:57 +0000)]
re PR libstdc++/29354 (Error when seeking on an ostringstream)

2006-10-06  Paolo Carlini  <pcarlini@suse.de>

PR libstdc++/29354
* include/bits/sstream.tcc (basic_stringbuf<>::seekpos(pos_type,
ios_base::openmode)): Allow for seek to pos_type(off_type(0))
when the stream is empty.
* testsuite/27_io/basic_stringbuf/seekpos/char/29354.cc: New.
* testsuite/27_io/basic_stringbuf/seekpos/wchar_t/29354.cc: New.

From-SVN: r117494

17 years agogimplify.c (gimplify_type_sizes): Don't recurse on the pointed-to type.
Olivier Hainque [Fri, 6 Oct 2006 09:12:09 +0000 (09:12 +0000)]
gimplify.c (gimplify_type_sizes): Don't recurse on the pointed-to type.

* gimplify.c (gimplify_type_sizes) [POINTER_TYPE, REFERENCE_TYPE]:
Don't recurse on the pointed-to type.
* c-decl.c (grokdeclarator) [cdk_pointer]: If we are in a NORMAL or
DECL context, attach an artificial TYPE_DECL to anonymous pointed-to
types with components of variable size.

* testsuite/gcc.dg/typename-vla-1.c: New case.
* testsuite/gnat.dg/forward_vla.adb: New case.

From-SVN: r117493

17 years ago* config/i386/mingw32.h (GOMP_SELF_SPECS): Add -mthreads for openmp.
Danny Smith [Fri, 6 Oct 2006 08:46:18 +0000 (08:46 +0000)]
* config/i386/mingw32.h (GOMP_SELF_SPECS): Add -mthreads for openmp.

From-SVN: r117492

17 years agobarrier-1.c: Change timestamp tests from '<' to '<='.
Danny Smith [Fri, 6 Oct 2006 08:32:33 +0000 (08:32 +0000)]
barrier-1.c: Change timestamp tests from '<' to '<='.

* testsuite/libgomp.c/barrier-1.c: Change timestamp tests from
'<' to '<='.

From-SVN: r117491

17 years agoacinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.
Danny Smith [Fri, 6 Oct 2006 08:25:32 +0000 (08:25 +0000)]
acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.

* acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.
* configure: Regenerate.
* fortran.c  (ialias_redirect): Add __USER_LABEL_PREFIX__ to alias.

From-SVN: r117490

17 years agoacinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.
Danny Smith [Fri, 6 Oct 2006 08:21:49 +0000 (08:21 +0000)]
acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.

* acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.
* configure: Regenerate.

From-SVN: r117489

17 years agoacinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.
Danny Smith [Fri, 6 Oct 2006 08:21:22 +0000 (08:21 +0000)]
acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.

* acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove __USER_LABEL_PREFIX__ from test.
* configure: Regenerate.

From-SVN: r117488

17 years agore PR tree-optimization/29290 (SPEC CPU2000 178.galgel ICE using -O3 -ftree-loop...
Jakub Jelinek [Fri, 6 Oct 2006 07:27:28 +0000 (09:27 +0200)]
re PR tree-optimization/29290 (SPEC CPU2000 178.galgel ICE using -O3 -ftree-loop-linear)

PR tree-optimization/29290
* tree-loop-linear.c (linear_transform_loops): Bail if loop_nest has
multiple exits.

* gfortran.dg/loop_nest_1.f90: New test.

From-SVN: r117484

17 years agore PR target/29198 (Incorrect reference to __thread array with -fPIC -O2 on x86)
Jakub Jelinek [Fri, 6 Oct 2006 07:25:02 +0000 (09:25 +0200)]
re PR target/29198 (Incorrect reference to __thread array with -fPIC -O2 on x86)

PR target/29198
* config/i386/i386.c (legitimize_pic_address): Reject TLS symbols.
* config/i386/predicates.md (local_symbolic_operand): Likewise.

* gcc.dg/tls/opt-12.c: New test.

From-SVN: r117483

17 years agore PR fortran/28415 (4.2.0 ICE when using automatic array and -fno-automatic)
Jakub Jelinek [Fri, 6 Oct 2006 07:23:00 +0000 (09:23 +0200)]
re PR fortran/28415 (4.2.0 ICE when using automatic array and -fno-automatic)

PR fortran/28415
* trans-decl.c (gfc_finish_var_decl): With -fno-automatic, don't
make artificial variables or pointer to variable automatic array
TREE_STATIC.

* gfortran.dg/save_2.f90: New test.

From-SVN: r117482

17 years agore PR c/29091 (vector constant not fully outputed)
Jakub Jelinek [Fri, 6 Oct 2006 07:15:48 +0000 (09:15 +0200)]
re PR c/29091 (vector constant not fully outputed)

PR c/29091
* varasm.c (output_constant): If TREE_VECTOR_CST_ELTS chain is shorter than
the number of vector elements fill the rest with zeros.

* gcc.dg/pr29091.c: New test.

From-SVN: r117481

17 years agoscanner.c (skip_free_comments): Return bool instead of void.
Jakub Jelinek [Fri, 6 Oct 2006 07:10:55 +0000 (09:10 +0200)]
scanner.c (skip_free_comments): Return bool instead of void.

* scanner.c (skip_free_comments): Return bool instead of void.
(gfc_next_char_literal): Don't return ' ' if & is missing after
!$omp or !$.  Use skip_{free,fixed}_comments directly instead
of gfc_skip_comments.

From-SVN: r117480

17 years ago* lib/target-supports.exp
Joseph Myers [Fri, 6 Oct 2006 00:47:35 +0000 (01:47 +0100)]
* lib/target-supports.exp
(check_effective_target_vect_cmdline_needed): Handle i?86-*-* the
same as x86_64-*-*; check for LP64.

From-SVN: r117477

17 years agoDaily bump.
GCC Administrator [Fri, 6 Oct 2006 00:17:15 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117475

17 years agoUpdate comment
Mike Stump [Thu, 5 Oct 2006 23:12:04 +0000 (23:12 +0000)]
Update comment

From-SVN: r117470

17 years agomips.c (mips_pass_by_reference): Do not return false for EABI if type is NULL.
Richard Sandiford [Thu, 5 Oct 2006 22:19:35 +0000 (22:19 +0000)]
mips.c (mips_pass_by_reference): Do not return false for EABI if type is NULL.

gcc/
* config/mips/mips.c (mips_pass_by_reference): Do not return false
for EABI if type is NULL.  Use a size check based on GET_MODE_SIZE
instead.

From-SVN: r117468

17 years agoMakefile.am: Use $(M4) instead of m4.
Steven G. Kargl [Thu, 5 Oct 2006 21:21:53 +0000 (21:21 +0000)]
Makefile.am: Use $(M4) instead of m4.

2006-10-05  Steven G. Kargl  <kargl@gcc.gnu.org>

* Makefile.am: Use $(M4) instead of m4.
* Makefile.in: Regenerated.

From-SVN: r117467

17 years ago* sv.po: Update.
Joseph Myers [Thu, 5 Oct 2006 17:03:18 +0000 (18:03 +0100)]
* sv.po: Update.

From-SVN: r117463

17 years agobfin.md (subsi3): Lose expander, change previously unnamed pattern into subsi3.
Bernd Schmidt [Thu, 5 Oct 2006 10:46:41 +0000 (10:46 +0000)]
bfin.md (subsi3): Lose expander, change previously unnamed pattern into subsi3.

* config/bfin/bfin.md (subsi3): Lose expander, change previously
unnamed pattern into subsi3.  Use correct constraints/predicates.
* config/bfin/bfin.h (CONST_OK_FOR_K): Handle "KN7".
* config/bfin/predicates.md (reg_or_neg7bit_operand_p): New.

From-SVN: r117455

17 years ago2006-10-05 Gary Benson <gbenson@redhat.com>
Gary Benson [Thu, 5 Oct 2006 09:32:57 +0000 (09:32 +0000)]
2006-10-05  Gary Benson  <gbenson@redhat.com>

* java/net/SocketPermission.java
(processHostport): Cope with IPv6 addresses with a
one-digit first component.

From-SVN: r117454

17 years agoloop.texi: Add data references analysis description.
Ira Rosen [Thu, 5 Oct 2006 08:47:11 +0000 (08:47 +0000)]
loop.texi: Add data references analysis description.

        * doc/loop.texi: Add data references analysis description.

From-SVN: r117453

17 years agogfortran.texi: (Current Status): update and rewrite to reflect actual status more...
Brooks Moses [Thu, 5 Oct 2006 04:58:18 +0000 (04:58 +0000)]
gfortran.texi: (Current Status): update and rewrite to reflect actual status more accurately.

2006-10-04  Brooks Moses  <bmoses@stanford.edu>

* gfortran.texi: (Current Status): update and rewrite to reflect
actual status more accurately.

From-SVN: r117449

17 years agoDaily bump.
GCC Administrator [Thu, 5 Oct 2006 00:17:15 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117446

17 years agogfortran.texi: Consistently refer to the compiler as "GNU Fortran".
Brooks Moses [Wed, 4 Oct 2006 23:07:39 +0000 (23:07 +0000)]
gfortran.texi: Consistently refer to the compiler as "GNU Fortran".

2006-10-04  Brooks Moses  <bmoses@stanford.edu>

* gfortran.texi: Consistently refer to the compiler as "GNU
Fortran".
* intrinsic.texi: Ditto.
* invoke.texi: Ditto.

From-SVN: r117442

17 years agogcov.texi: Add formatting to "Invoking gcov" section head.
Brooks Moses [Wed, 4 Oct 2006 22:40:45 +0000 (22:40 +0000)]
gcov.texi: Add formatting to "Invoking gcov" section head.

2006-10-04  Brooks Moses  <bmoses@stanford.edu>

* doc/gcov.texi: Add formatting to "Invoking gcov" section head.

From-SVN: r117441

17 years agoMakefile.in (libgcc.mk, [...]): Add emutls.c.
Richard Henderson [Wed, 4 Oct 2006 21:09:20 +0000 (14:09 -0700)]
Makefile.in (libgcc.mk, [...]): Add emutls.c.

gcc/
        * Makefile.in (libgcc.mk, LIBGCC_DEPS): Add emutls.c.
        * builtin-types.def (BT_WORD): Make unsigned.
        (BT_FN_VOID_PTR_WORD_WORD_PTR): New.
        * builtins.def (BUILT_IN_EMUTLS_GET_ADDRESS): New.
        (BUILT_IN_EMUTLS_REGISTER_COMMON): New.
        * c-decl.c (grokdeclarator): Don't error if !have_tls.
        * c-parser.c (c_parser_omp_threadprivate): Likewise.
        * cgraph.c (decide_is_variable_needed): Look at force_output.
        Recurse for emulated tls.
        * cgraphunit.c (cgraph_varpool_remove_unreferenced_decls): Remove
        checks redundant with decide_is_variable_needed.
        (cgraph_build_static_cdtor): Do cgraph_varpool_assemble_pending_decls.
        * dwarf2out.c (loc_descriptor_from_tree_1): Don't do anything for
        emulated tls.
        * expr.c (emutls_var_address): New.
        (expand_expr_real_1): Expand emulated tls.
        (expand_expr_addr_expr_1): Likewise.
        * libgcc-std.ver: Add __emutls_get_address, __emutls_register_common.
        * output.h (emutls_finish): Declare.
        * toplev.c (compile_file): Call it.
        * tree-ssa-address.c (gen_addr_rtx): Check for const-ness of the
        address before wrapping in CONST.
        * varasm.c (emutls_htab, emutls_object_type): New.
        (EMUTLS_VAR_PREFIX, EMUTLS_TMPL_PREFIX): New.
        (get_emutls_object_name, get_emutls_object_type): New.
        (get_emutls_init_templ_addr, emutls_decl): New.
        (emutls_common_1, emutls_finish): New.
        (assemble_variable): When emulating tls, swap decls; generate
        constructor for the emutls objects.
        (do_assemble_alias): When emulating tls, swap decl and target name.
        (default_encode_section_info): Don't add SYMBOL_FLAG_TLS_SHIFT
        for emulated tls.
        * emutls.c: New file.
        * config/sparc/sol2.h (ASM_DECLARE_OBJECT_NAME): Only emit
        tls_object for real tls.

gcc/cp/
        * decl.c (grokvardecl): Don't error if !have_tls.
        (grokdeclarator): Likewise.
        * parser.c (cp_parser_omp_threadprivate): Likewise.

gcc/fortran/
        * f95-lang.c (gfc_init_builtin_functions): Add __emutls_get_address
        and __emutls_register_common.
        * openmp.c (gfc_match_omp_threadprivate): Don't error if !have_tls.
        * trans-common.c (build_common_decl): Don't check have_tls.
        * trans-decl.c (gfc_finish_var_decl): Likewise.
        * types.def (BT_WORD, BT_FN_PTR_PTR): New.
        (BT_FN_VOID_PTR_WORD_WORD_PTR): New.

gcc/testsuite/
        * lib/target-supports.exp (check_effective_target_tls): Redefine
        to mean non-emulated tls.
        * gcc.dg/tls/alias-1.c: Remove tls requirement.
        * gcc.dg/tls/asm-1.c, gcc.dg/tls/debug-1.c, gcc.dg/tls/diag-1.c,
        gcc.dg/tls/diag-2.c, gcc.dg/tls/diag-3.c, gcc.dg/tls/diag-4.c,
        gcc.dg/tls/diag-5.c, gcc.dg/tls/init-1.c, gcc.dg/tls/nonpic-1.c,
        gcc.dg/tls/opt-10.c, gcc.dg/tls/opt-5.c, gcc.dg/tls/opt-6.c,
        gcc.dg/tls/opt-8.c, gcc.dg/tls/opt-9.c, gcc.dg/tls/pic-1.c,
        gcc.dg/tls/struct-1.c, gcc.dg/tls/trivial.c: Likewise.

From-SVN: r117440

17 years agoMAINTAINERS (Write After Approval): Add myself.
Brooks Moses [Wed, 4 Oct 2006 18:30:49 +0000 (18:30 +0000)]
MAINTAINERS (Write After Approval): Add myself.

2006-10-04  Brooks Moses  <bmoses@stanford.edu>

* MAINTAINERS (Write After Approval): Add myself.

From-SVN: r117437

17 years ago[multiple changes]
Paul Thomas [Wed, 4 Oct 2006 16:54:19 +0000 (16:54 +0000)]
[multiple changes]

2006-10-04  Paul Thomas <pault@gcc.gnu.org>

PR fortran/29343
* resolve.c (resolve_allocate_expr): Exclude derived types from
search for dependences between allocated variables and the
specification expressions for other allocations in the same
statement.

2006-10-04  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/29343
* gfortran.dg/alloc_alloc_expr_2.f90: New test.

From-SVN: r117435

17 years agoFix bug in constraint-check generator, reported by Rask Ingemann Lambertsen:
Zack Weinberg [Wed, 4 Oct 2006 16:41:55 +0000 (16:41 +0000)]
Fix bug in constraint-check generator, reported by Rask Ingemann Lambertsen:

Fix bug in constraint-check generator, reported by Rask Ingemann
Lambertsen:
* genpreds.c (add_constraint): Don't remove anything from the
expression here.
(write_tm_constrs_h): Detect whether "op" argument is used, and
mark it ARG_UNUSED if it isn't.
(write_insn_const_int_ok_for_constraint): Skip the part of the
expression that would test whether "op" (which is not available
here) is a CONST_INT.

From-SVN: r117433

17 years ago* gfortran.dg/bounds_check_fail_1.f90: New test.
Francois-Xavier Coudert [Wed, 4 Oct 2006 15:44:32 +0000 (17:44 +0200)]
* gfortran.dg/bounds_check_fail_1.f90: New test.

From-SVN: r117431

17 years agore PR c++/28448 (doc reference to -fdiagnostics-show-options looks like it should...
Ryan Mansfield [Wed, 4 Oct 2006 10:21:08 +0000 (10:21 +0000)]
re PR c++/28448 (doc reference to -fdiagnostics-show-options looks like it should be -fdiagnostics-show-option)

2006-10-02  Ryan Mansfield  <rmansfield@qnx.com>

PR c++/28448
* doc/invoke.texi (fdiagnostics-show-option): Fix typos.

From-SVN: r117429

17 years ago2.cc: Fail on Darwin 8.[0-4].* only.
Benjamin Kosnik [Wed, 4 Oct 2006 07:57:34 +0000 (07:57 +0000)]
2.cc: Fail on Darwin 8.[0-4].* only.

2006-10-04  Benjamin Kosnik  <bkoz@redhat.com>
    Mike Stump  <mrs@apple.com>

* testsuite/27_io/ios_base/storage/2.cc: Fail on Darwin 8.[0-4].* only.
* testsuite/23_containers/vector/resize/1.cc: Same.

Co-Authored-By: Mike Stump <mrs@apple.com>
From-SVN: r117428

17 years agopop_test.hpp: Correct typo.
Benjamin Kosnik [Wed, 4 Oct 2006 07:52:14 +0000 (07:52 +0000)]
pop_test.hpp: Correct typo.

2006-10-04  Benjamin Kosnik  <bkoz@redhat.com>

* testsuite/util/performance/priority_queue/mem_usage/
pop_test.hpp: Correct typo.
* testsuite/util/performance/assoc/mem_usage/
multimap_insert_test.hpp: Same.
* testsuite/util/performance/assoc/mem_usage/erase_test.hpp: Same.

From-SVN: r117427

17 years agore PR c++/29020 (ICE using A<N>::A instead of A<N> in friend declaration)
Mark Mitchell [Wed, 4 Oct 2006 06:53:55 +0000 (06:53 +0000)]
re PR c++/29020 (ICE using A<N>::A instead of A<N> in friend declaration)

PR c++/29020
* friend.c (do_friend): Improve comments; add assertion.
* parser.c (cp_parser_nested_name_specifier_opt): Resolve
typenames for qualified names used in declarations, even when
caching qualified name lookup.
PR c++/29020
* g++.dg/template/friend48.C: New test.

From-SVN: r117425

17 years agore PR fortran/29098 (allocation of a pointer to a derived type crashes)
Paul Thomas [Wed, 4 Oct 2006 04:48:35 +0000 (04:48 +0000)]
re PR fortran/29098 (allocation of a pointer to a derived type crashes)

2006-10-04  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/29098
* resolve.c (resolve_structure_cons): Do not return FAILURE if
component expression is NULL.

2006-10-04  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/29098
* gfortran.dg/default_initialization_2.f90: New test.

From-SVN: r117424

17 years agodarwin-libgcc.10.5.ver: Update to symbol list for current Leopard versions.
Geoffrey Keating [Wed, 4 Oct 2006 01:38:30 +0000 (01:38 +0000)]
darwin-libgcc.10.5.ver: Update to symbol list for current Leopard versions.

* config/i386/darwin-libgcc.10.5.ver: Update to symbol list for
current Leopard versions.
* config/rs6000/darwin-libgcc.10.5.ver: Likewise.

From-SVN: r117423

17 years agoDaily bump.
GCC Administrator [Wed, 4 Oct 2006 00:17:24 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r117421

17 years agore PR fortran/20779 (ALLOCATEing the STAT variable not detected)
Paul Thomas [Tue, 3 Oct 2006 21:40:24 +0000 (21:40 +0000)]
re PR fortran/20779 (ALLOCATEing the STAT variable not detected)

2006-10-03  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/20779
PR fortran/20891
* resolve.c (find_sym_in_expr): New function that returns true
if a symbol is found in an expression.
(resolve_allocate_expr): Check whether the STAT variable is
itself allocated in the same statement.  Use the call above to
check whether any of the allocated arrays are used in array
specifications in the same statement.

2006-10-03  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/20779
PR fortran/20891
* gfortran.dg/alloc_alloc_expr_1.f90: New test.

From-SVN: r117415

17 years agoarith.c (gfc_check_real_range): Use correct exponent range for subnormal numbers.
Steven G. Kargl [Tue, 3 Oct 2006 21:28:05 +0000 (21:28 +0000)]
arith.c (gfc_check_real_range): Use correct exponent range for subnormal numbers.

2006-10-03  Steven G. Kargl  <kargl@gcc.gnu.org>

* arith.c (gfc_check_real_range):  Use correct exponent range for
subnormal numbers.

From-SVN: r117414

17 years agore PR fortran/29284 (ICE for optional subroutine argument)
Paul Thomas [Tue, 3 Oct 2006 20:13:03 +0000 (20:13 +0000)]
re PR fortran/29284 (ICE for optional subroutine argument)

2006-10-03  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/29284
PR fortran/29321
PR fortran/29322
* trans-expr.c (gfc_conv_function_call): Check the expression
and the formal symbol are present when testing the actual
argument.

PR fortran/25091
PR fortran/25092
* resolve.c (resolve_entries): It is an error if the entries
of an array-valued function do not have the same shape.

2006-10-03  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/29284
* gfortran.dg/optional_assumed_charlen_1.f90: New test.

PR fortran/29321
PR fortran/29322
* gfortran.dg/missing_optional_dummy_2.f90: New test.

PR fortran/25091
PR fortran/25092
* gfortran.dg/entry_array_specs_1.f90: New test.

From-SVN: r117413

17 years agonearest_1.f90: Add -O0 because -ffloat-store is insufficient to inhibit some test...
Steven G. Kargl [Tue, 3 Oct 2006 19:18:09 +0000 (19:18 +0000)]
nearest_1.f90: Add -O0 because -ffloat-store is insufficient to inhibit some test failures at -O3.

2006-10-03  Steven G. Kargl  <kargl@gcc.gnu.org>

        * gfortran.dg/nearest_1.f90: Add -O0 because -ffloat-store is
        insufficient to inhibit some test failures at -O3.  Do testing
        on integers.

From-SVN: r117412