2009-03-30 Sebastian Pop <sebastian.pop@amd.com>
* doc/invoke.texi (-floop-interchange, -floop-strip-mine,
-floop-block): Document dependences on PPL, CLooG and Graphite.
From-SVN: r145273
PR rtl-optimization/323
* c-common.c (c_fully_fold, convert_and_check,
c_common_truthvalue_conversion): Handle EXCESS_PRECISION_EXPR.
(c_fully_fold_internal): Disallow EXCESS_PRECISION_EXPR.
* c-common.def (EXCESS_PRECISION_EXPR): New.
* c-cppbuiltin.c (builtin_define_float_constants): Define
constants with enough digits for long double.
* c-lex.c (interpret_float): Interpret constant with excess
precision where appropriate.
* c-opts.c (c_common_post_options): Set
flag_excess_precision_cmdline. Give an error for
-fexcess-precision=standard for C++ for processors where the
option is significant.
* c-parser.c (c_parser_conditional_expression): Handle excess
precision in condition.
* c-typeck.c (convert_arguments): Handle arguments with excess
precision.
(build_unary_op): Move excess precision outside operation.
(build_conditional_expr): Likewise.
(build_compound_expr): Likewise.
(build_c_cast): Do cast on operand of EXCESS_PRECISION_EXPR.
(build_modify_expr): Handle excess precision in RHS.
(convert_for_assignment): Handle excess precision in converted
value.
(digest_init, output_init_element, process_init_element): Handle
excess precision in initializer.
(c_finish_return): Handle excess precision in return value.
(build_binary_op): Handle excess precision in operands and add
excess precision as needed for operation.
* common.opt (-fexcess-precision=): New option.
* config/i386/i386.h (X87_ENABLE_ARITH, X87_ENABLE_FLOAT): New.
* config/i386/i386.md (float<SSEMODEI24:mode><X87MODEF:mode>2):
For standard excess precision, output explicit conversion to and
truncation from XFmode.
(*float<SSEMODEI24:mode><X87MODEF:mode>2_1,
*float<SSEMODEI24:mode><X87MODEF:mode>2_i387_with_temp,
*float<SSEMODEI24:mode><X87MODEF:mode>2_i387, two unnamed
define_splits, floatdi<X87MODEF:mode>2_i387_with_xmm, two unnamed
define_splits, *floatunssi<mode>2_1, two unnamed define_splits,
floatunssi<mode>2, add<mode>3, sub<mode>3, mul<mode>3, divdf3,
divsf3, *fop_<mode>_comm_i387, *fop_<mode>_1_i387,
*fop_<MODEF:mode>_2_i387, *fop_<MODEF:mode>_3_i387,
*fop_df_4_i387, *fop_df_5_i387, *fop_df_6_i387, two unnamed
define_splits, sqrt<mode>2): Disable where appropriate for
standard excess precision.
* convert.c (convert_to_real): Do not shorten arithmetic to type
for which excess precision would be used.
* defaults.h (TARGET_FLT_EVAL_METHOD_NON_DEFAULT): Define.
* doc/invoke.texi (-fexcess-precision=): Document option.
(-mfpmath=): Correct index entry.
* flags.h (enum excess_precision, flag_excess_precision_cmdline,
flag_excess_precision): New.
* langhooks.c (lhd_post_options): Set
flag_excess_precision_cmdline.
* opts.c (common_handle_option): Handle -fexcess-precision=.
* toplev.c (flag_excess_precision_cmdline, flag_excess_precision,
init_excess_precision): New.
(lang_dependent_init_target): Call init_excess_precision.
* tree.c (excess_precision_type): New.
* tree.h (excess_precision_type): Declare.
ada:
* gcc-interface/misc.c (gnat_post_options): Set
flag_excess_precision_cmdline. Give an error for
-fexcess-precision=standard for processors where the option is
significant.
fortran:
* options.c (gfc_post_options): Set
flag_excess_precision_cmdline. Give an error for
-fexcess-precision=standard for processors where the option is
significant.
java:
* lang.c (java_post_options): Set flag_excess_precision_cmdline.
Give an error for -fexcess-precision=standard for processors where
the option is significant.
testsuite:
* gcc.target/i386/excess-precision-1.c,
gcc.target/i386/excess-precision-2.c,
gcc.target/i386/excess-precision-3.c,
gcc.target/i386/excess-precision-4.c,
gcc.target/i386/excess-precision-5.c,
gcc.target/i386/excess-precision-6.c: New tests.
From-SVN: r145272
PR c/35235
* c-typeck.c (build_component_ref): Do not copy qualifiers from
non-lvalue to component.
testsuite:
* gcc.dg/c99-array-lval-8.c: New test.
From-SVN: r145271
2009-03-29 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/38823
* gfortran.dg/power1.f90: New test.
2009-03-29 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/38823
* gfortran.h: Add ARITH_PROHIBIT to arith enum.
expr.c (gfc_match_init_expr): Add global variable init_flag to
flag matching an initialization expression.
(check_intrinsic_op): Move no longer reachable error message to ...
* arith.c (arith_power): ... here. Remove gfc_ prefix in
gfc_arith_power. Use init_flag. Allow constant folding of x**y
when y is REAL or COMPLEX.
(eval_intrinsic): Remove restriction that y in x**y must be INTEGER
for constant folding.
* gfc_power: Update gfc_arith_power to arith_power
From-SVN: r145261
PR c/456
PR c/5675
PR c/19976
PR c/29116
PR c/31871
PR c/35198
fixincludes:
* inclhack.def (glibc_tgmath): New fix.
* fixincl.x: Regenerate.
* tests/base/tgmath.h: New.
gcc:
* builtins.c (fold_builtin_sincos): Build COMPOUND_EXPR in
void_type_node.
(fold_call_expr): Return a NOP_EXPR from folding rather than the
contained expression.
* c-common.c (c_fully_fold, c_fully_fold_internal, c_save_expr):
New.
(c_common_truthvalue_conversion): Use c_save_expr. Do not fold
conditional expressions for C.
(decl_constant_value_for_optimization): Move from
decl_constant_value_for_broken_optimization in c-typeck.c. Check
whether optimizing and that the expression is a VAR_DECL not of
array type instead of doing such checks in the caller. Do not
check pedantic. Call gcc_unreachable for C++.
* c-common.def (C_MAYBE_CONST_EXPR): New.
* c-common.h (c_fully_fold, c_save_expr,
decl_constant_value_for_optimization): New prototypes.
(C_MAYBE_CONST_EXPR_PRE, C_MAYBE_CONST_EXPR_EXPR,
C_MAYBE_CONST_EXPR_INT_OPERANDS, C_MAYBE_CONST_EXPR_NON_CONST,
EXPR_INT_CONST_OPERANDS): Define.
* c-convert.c (convert): Strip nops from expression.
* c-decl.c (groktypename): Take extra parameters expr and
expr_const_operands. Update call to grokdeclarator.
(start_decl): Update call to grokdeclarator. Add statement for
expressions used in type of decl.
(grokparm): Update call to grokdeclarator.
(push_parm_decl): Update call to grokdeclarator.
(build_compound_literal): Add parameter non_const and build a
C_MAYBE_COSNT_EXPR if applicable.
(grokdeclarator): Take extra parameters expr and
expr_const_operands. Track expressions used in declaration
specifiers and declarators. Fold array sizes and track whether
they are constant expressions and whether they are integer
constant expressions.
(parser_xref_tag): Set expr and expr_const_operands fields in
return value.
(grokfield): Update call to grokdeclarator.
(start_function): Update call to grokdeclarator.
(build_null_declspecs): Set expr and expr_const_operands fields in
return value.
(declspecs_add_type): Handle expressions in typeof specifiers.
* c-parser.c (c_parser_declspecs): Set expr and
expr_const_operands fields for declaration specifiers.
(c_parser_enum_specifier): Likewise.
(c_parser_struct_or_union_specifier): Likewise.
(c_parser_typeof_specifier): Likewise. Update call to
groktypename. Fold expression as needed. Return expressions with
type instead of adding statements.
(c_parser_attributes): Update calls to c_parser_expr_list.
(c_parser_statement_after_labels): Fold expression before passing
to objc_build_throw_stmt.
(c_parser_condition): Fold expression.
(c_parser_asm_operands): Fold expression.
(c_parser_conditional_expression): Use c_save_expr. Update call
to build_conditional_expr.
(c_parser_alignof_expression): Update call to groktypename.
(c_parser_postfix_expression): Preserve C_MAYBE_CONST_EXPR as
original_code. Fold expression argument of va_arg. Create
C_MAYBE_CONST_EXPR to preserve side effects of expressions in type
argument to va_arg. Update calls to groktypename. Fold array
index for offsetof. Verify that first argument to
__builtin_choose_expr has integer type.
(c_parser_postfix_expression_after_paren_type): Update calls to
groktypename and build_compound_literal. Handle expressions with
side effects in type name.
(c_parser_postfix_expression_after_primary): Update call to
c_parser_expr_list. Set original_code for calls to
__builtin_constant_p.
(c_parser_expr_list): Take extra parameter fold_p. Fold
expressions if requested.
(c_parser_objc_type_name): Update call to groktypename.
(c_parser_objc_synchronized_statement): Fold expression.
(c_parser_objc_receiver): Fold expression.
(c_parser_objc_keywordexpr): Update call to c_parser_expr_list.
(c_parser_omp_clause_num_threads, c_parser_omp_clause_schedule,
c_parser_omp_atomic, c_parser_omp_for_loop): Fold expressions.
* c-tree.h (CONSTRUCTOR_NON_CONST): Define.
(struct c_typespec): Add elements expr and expr_const_operands.
(struct c_declspecs): Add elements expr and expr_const_operands.
(groktypename, build_conditional_expr, build_compound_literal):
Update prototypes.
(in_late_binary_op): Declare.
* c-typeck.c (note_integer_operands): New function.
(in_late_binary_op): New variable.
(decl_constant_value_for_broken_optimization): Move to c-common.c
and rename to decl_constant_value_for_optimization.
(default_function_array_conversion): Do not strip nops.
(default_conversion): Do not call
decl_constant_value_for_broken_optimization.
(build_array_ref): Do not fold result.
(c_expr_sizeof_expr): Fold operand. Use C_MAYBE_CONST_EXPR for
result when operand is a VLA.
(c_expr_sizeof_type): Update call to groktypename. Handle
expressions included in type name. Use C_MAYBE_CONST_EXPR for
result when operand names a VLA type.
(build_function_call): Update call to build_compound_literal.
Only fold result for calls to __builtin_* functions. Strip
NOP_EXPR from INTEGER_CST returned from such functions. Fold
the function designator.
(convert_arguments): Fold arguments. Update call to
convert_for_assignment.
(build_unary_op): Handle increment and decrement of
C_MAYBE_CONST_EXPR. Move lvalue checks for increment and
decrement earlier. Fold operand of increment and decrement.
Handle address of C_MAYBE_CONST_EXPR. Only fold expression being
built for integer operand. Wrap returns that are INTEGER_CSTs
without being integer constant expressions or that have integer
constant operands without being INTEGER_CSTs.
(lvalue_p): Handle C_MAYBE_CONST_EXPR.
(build_conditional_expr): Add operand ifexp_bcp. Track whether
result is an integer constant expression or can be used in
unevaluated parts of one and avoid folding and wrap as
appropriate. Fold operands before possibly doing -Wsign-compare
warnings.
(build_compound_expr): Wrap result for C99 if operands can be used
in integer constant expressions.
(build_c_cast): Update call to digest_init. Do not ignore
overflow from casting floating-point constants to integers. Wrap
results that could be confused with integer constant expressions,
null pointer constants or floating-point constants.
(c_cast_expr): Update call to groktypename. Handle expressions
included in type name.
(build_modify_expr): Handle modifying a C_MAYBE_CONST_EXPR. Fold
lhs inside possible SAVE_EXPR. Fold RHS before assignment.
Update calls to convert_for_assignment.
(convert_for_assignment): Take new parameter
null_pointer_constant. Do not strip nops or call
decl_constant_value_for_broken_optimization. Set
in_late_binary_op for conversions to boolean.
(store_init_value): Update call to digest_init.
(digest_init): Take new parameter null_pointer_constant. Do not
call decl_constant_value_for_broken_optimization. pedwarn for
initializers not constant expressions. Update calls to
convert_for_assignment.
(constructor_nonconst): New.
(struct constructor_stack): Add nonconst element.
(really_start_incremental_init, push_init_level, pop_init_level):
Handle constructor_nonconst and nonconst element.
(set_init_index): Call constant_expression_warning for array
designators.
(output_init_element): Fold value. Set constructor_nonconst as
applicable. pedwarn for initializers not constant expressions.
Update call to digest_init. Call constant_expression_warning
where constant initializers are required.
(process_init_element): Use c_save_expr.
(c_finish_goto_ptr): Fold expression.
(c_finish_return): Fold return value. Update call to
convert_for_assignment.
(c_start_case): Fold switch expression.
(c_process_expr_stmt): Fold expression.
(c_finish_stmt_expr): Create C_MAYBE_CONST_EXPR as needed to
ensure statement expression is not evaluated in constant
expression.
(build_binary_op): Track whether results are integer constant
expressions or may occur in such, disable folding and wrap results
as applicable. Fold operands for -Wsign-compare warnings unless
in_late_binary_op.
(c_objc_common_truthvalue_conversion): Handle results folded to
integer constants that are not integer constant expressions.
* doc/extend.texi: Document when typeof operands are evaluated,
that condition of __builtin_choose_expr is an integer constant
expression, and more about use of __builtin_constant_p in
initializers.
gcc/objc:
* objc-act.c (objc_finish_try_stmt): Set in_late_binary_op.
gcc/testsuite:
* gcc.c-torture/compile/20081108-1.c,
gcc.c-torture/compile/20081108-2.c,
gcc.c-torture/compile/20081108-3.c, gcc.dg/bconstp-2.c,
gcc.dg/bconstp-3.c, gcc.dg/bconstp-4.c, gcc.dg/c90-const-expr-6.c,
gcc.dg/c90-const-expr-7.c, gcc.dg/c90-const-expr-8.c,
gcc.dg/c90-const-expr-9.c, gcc.dg/c90-const-expr-10.c,
gcc.dg/c90-const-expr-11.c, gcc.dg/c99-const-expr-6.c,
gcc.dg/c99-const-expr-7.c, gcc.dg/c99-const-expr-8.c,
gcc.dg/c99-const-expr-9.c, gcc.dg/c99-const-expr-10.c,
gcc.dg/c99-const-expr-11.c, gcc.dg/c99-const-expr-12.c,
gcc.dg/c99-const-expr-13.c, gcc.dg/compare10.c,
gcc.dg/gnu89-const-expr-1.c, gcc.dg/gnu89-const-expr-2.c,
gcc.dg/gnu99-const-expr-1.c, gcc.dg/gnu99-const-expr-2.c,
gcc.dg/gnu99-const-expr-3.c, gcc.dg/vla-12.c, gcc.dg/vla-13.c,
gcc.dg/vla-14.c, gcc.dg/vla-15.c, gcc.dg/vla-16.c: New tests.
* gcc.dg/c90-const-expr-1.c, gcc.dg/c90-const-expr-2.c,
gcc.dg/c90-const-expr-3.c, gcc.dg/c99-const-expr-2.c,
gcc.dg/c99-const-expr-3.c, gcc.dg/c99-static-1.c: Remove XFAILs.
* gcc.dg/c90-const-expr-2.c: Use ZERO in place of 0 in another
case.
* gcc.dg/overflow-warn-1.c, gcc.dg/overflow-warn-2.c,
gcc.dg/overflow-warn-3.c, gcc.dg/overflow-warn-4.c: Remove
XFAILs. Update expected messages.
* gcc.dg/pr14649-1.c, gcc.dg/pr19984.c, gcc.dg/pr25682.c: Update
expected messages.
* gcc.dg/real-const-1.c: Replace with test from original PR.
* gcc.dg/vect/pr32230.c: Use intermediate cast to __PTRDIFF_TYPE__
when casting from non-constant integer to pointer.
From-SVN: r145254
2009-03-29 Daniel Kraft <d@domob.eu>
PR fortran/37423
* gfortran.h (struct gfc_typebound_proc): Added new flag `deferred' and
added a comment explaining DEFERRED binding handling.
* decl.c (match_binding_attributes): Really match DEFERRED attribute.
(match_procedure_in_type): Really match PROCEDURE(interface) syntax
and do some validity checks for DEFERRED and this construct.
* module.c (binding_overriding): New string constant for DEFERRED.
(mio_typebound_proc): Module-IO DEFERRED flag.
* resolve.c (check_typebound_override): Ensure that a non-DEFERRED
binding is not overridden by a DEFERRED one.
(resolve_typebound_procedure): Allow abstract interfaces as targets
for DEFERRED bindings.
(ensure_not_abstract_walker), (ensure_not_abstract): New methods.
(resolve_fl_derived): Use new `ensure_not_abstract' method for
non-ABSTRACT types extending ABSTRACT ones to ensure each DEFERRED
binding is overridden.
(check_typebound_baseobject): New method.
(resolve_compcall), (resolve_typebound_call): Check base-object of
the type-bound procedure call.
* gfc-internals.texi (Type-bound procedures): Document a little bit
about internal handling of DEFERRED bindings.
2009-03-29 Daniel Kraft <d@domob.eu>
PR fortran/37423
* gfortran.dg/typebound_proc_4.f03: Remove not-implemented check for
DEFERRED bindings.
* gfortran.dg/typebound_proc_9.f03: New test.
* gfortran.dg/typebound_proc_10.f03: New test.
* gfortran.dg/typebound_proc_11.f03: New test.
* gfortran.dg/abstract_type_5.f03: New test.
From-SVN: r145248
* regmove.c (perhaps_ends_bb_p): Remove.
(optimize_reg_copy_1): Don't call perhaps_ends_bb_p. Get basic block
from INSN and check that the main loop stays within that basic block.
(optimize_reg_copy_1, optimize_reg_copy_3, fixup_match_2): Likewise.
(regmove_forward_pass): Split out from regmove_optimize. Use
FOR_EACH_BB and FOR_BB_INSNS instead of traversing the insns stream.
(regmove_backward_pass): Split out from regmove_optimize. Use
FOR_EACH_BB_REVERSE and FOR_BB_INSNS_REVERS_SAFE.
(regmove_optimize): Simplify.
From-SVN: r145247
PR middle-end/28850
* tree-pass.h (pass_cleanup_eh): New function.
(remove_unreachable_regions): Break code handling RTL
to rtl_remove_unreachable_regions; remove ERT_MUST_NOT_THROW
that can not be reached by runtime.
(can_be_reached_by_runtime): New function.
(label_to_region_map): New function.
(num_eh_regions): New function.
(rtl_remove_unreachable_regions): New function.
(convert_from_eh_region_ranges): Call rtl_remove_unreachable_regions.
(remove_eh_region): New function.
* except.h: Include sbitmap and vecprim.
(remove_eh_region, remove_unreachable_regions, label_to_region_map,
num_eh_regions): Declare.
* passes.c (init_optimization_passes): Schedule cleanup_eh.
* Makefile.in (EXCEPT_H): New; replace all uses of except.h
by it.
* tree-eh.c (tree_remove_unreachable_handlers): New function.
(tree_empty_eh_handler_p): New function.
(cleanup_empty_eh): New function.
(cleanup_eh): New function.
(pass_cleanup_eh): New function.
From-SVN: r145233
* except.c (duplicate_eh_regions_0): Handle AKA bitmap.
(duplicate_eh_regions_1): Likewise.
(duplicate_eh_regions): Likewise; cleanup code gorwing the region
vector; call EH verification.
(foreach_reachable_handler, can_throw_internal_1, can_throw_external_1):
Be ready for region being removed.
From-SVN: r145230
gcc/
2009-03-00 David Ayers <ayers@fsfe.org>
PR objc/27377
* c-typeck.c (build_conditional_expr): Emit ObjC warnings
by calling objc_compare_types and surpress warnings about
incompatible C pointers that are compatible ObjC pointers.
testsuite/
2009-03-00 David Ayers <ayers@fsfe.org>
PR objc/27377
* objc.dg/conditional-1.m: New tests.
From-SVN: r145225
* cgraphbuild.c (build_cgraph_edges, rebuild_cgraph_edges): Don't
call initialize_inline_failed.
(initialize_inline_failed): Move it from here ...
* cgraph.c (initialize_inline_failed): ... to here.
(cgraph_create_edge): Call initialize_inline_failed rather than
setting inline_failed directly.
testsuite/
* gcc.dg/winline-10.c: New test.
From-SVN: r145215
2009-03-28 Martin Jambor <mjambor@suse.cz>
* fold-const.c (get_pointer_modulus_and_residue): New parameter
allow_func_align.
(fold_binary): Allow function decl aligment consideration is the
second argument is integer constant one.
* tree-ssa-forwprop.c (simplify_bitwise_and): New function.
(tree_ssa_forward_propagate_single_use_vars): Handle assing statements
with BIT_AND_EXPR on the RHS by calling simplify_bitwise_and.
* g++.dg/tree-ssa/fwprop-align.C: New test.
From-SVN: r145203
* config/rs6000/rs6000-c.c (rs6000_macro_to_expand): If macro
following vector keyword has expansion starting with pixel or bool
keyword, expand vector to __vector and pixel or bool to __pixel or
__bool.
* gcc.target/powerpc/altivec-28.c: New test.
From-SVN: r145199
2009-03-28 Richard Guenther <rguenther@suse.de>
PR tree-optimization/38723
* tree-ssa-pre.c (compute_avail): Add all default definitions to
the entry block.
* gcc.dg/tree-ssa/ssa-fre-22.c: New testcase.
From-SVN: r145197
2009-02-13 Paul Thomas <pault@gcc.gnu.org>
PR fortran/36703
PR fortran/36528
* trans-expr.c (gfc_conv_function_val): Stabilize Cray-pointer
function references to ensure that a valid expression is used.
(gfc_conv_function_call): Pass Cray pointers to procedures.
2009-02-13 Paul Thomas <pault@gcc.gnu.org>
PR fortran/36528
* gfortran.dg/cray_pointers_8.f90: New test.
PR fortran/36703
* gfortran.dg/cray_pointers_9.f90: New test.
From-SVN: r145196
2009-03-28 Paul Thomas <pault@gcc.gnu.org>
PR fortran/38765
* parse.c (parse_derived): Do not break on finding pointer,
allocatable or private components.
2009-03-28 Paul Thomas <pault@gcc.gnu.org>
PR fortran/38765
* gfortran.dg/alloc_comp_assign_9.f90: New test.
From-SVN: r145195
2009-03-28 Daniel Kraft <d@domob.eu>
* intrinsics/string_intrinsics.c: #include <assert.h>
* intrinsics/string_intrinsics_inc.c (string_trim): Use string_len_trim
instead of calculating the length directly.
(string_len_trim): For KIND=1, speed search up.
2009-03-28 Daniel Kraft <d@domob.eu>
* gfortran.dg/trim_1.f90: New test.
From-SVN: r145192
* tree-pass.h (pass_fixup_cfg): New pass.
* ipa-inline.c (inline_transform): Set
always_inline_functions_inlined/after_inlining.
* tree-optimize.c (execute_fixup_cfg): Do not set them here.
(pass_fixup_cfg): New pass.
* passes.c (init_optimization_passes): Add fixup_cfg.
From-SVN: r145191
2009-03-28 Richard Guenther <rguenther@suse.de>
* gcc.dg/Warray-bounds.c: Do not use redundant stores.
* g++.dg/warn/Warray-bounds.C: Likewise.
From-SVN: r145189
