Janus Weil [Mon, 31 Aug 2009 19:08:03 +0000 (21:08 +0200)]
re PR fortran/40940 ([F03] CLASS statement)
2009-08-31 Janus Weil <janus@gcc.gnu.org>
Paul Thomas <pault@gcc.gnu.org>
PR fortran/40940
* array.c (gfc_match_array_constructor): Rename gfc_match_type_spec.
* decl.c (gfc_match_type_spec): Rename to gfc_match_decl_type_spec,
and reject CLASS with -std=f95.
(gfc_match_implicit, gfc_match_data_decl,gfc_match_prefix,
match_procedure_interface): Rename gfc_match_type_spec.
* gfortran.h (gfc_type_compatible): Add prototype.
* match.h (gfc_match_type_spec): Rename to gfc_match_decl_type_spec.
* match.c (match_intrinsic_typespec): Rename to match_type_spec, and
add handling of derived types.
(gfc_match_allocate): Rename match_intrinsic_typespec and check
type compatibility of derived types.
* symbol.c (gfc_type_compatible): New function to check if two types
are compatible.
Richard Guenther [Mon, 31 Aug 2009 12:52:17 +0000 (12:52 +0000)]
builtins.c (fold_builtin_memory_op): Use the alias oracle to query if the memory regions for memmove overlap.
2009-08-31 Richard Guenther <rguenther@suse.de>
* builtins.c (fold_builtin_memory_op): Use the alias oracle
to query if the memory regions for memmove overlap.
* tree-ssa-alias.c (ptr_deref_may_alias_decl_p): Relax the
asserts on pointers, instead deal with odd trees.
(ptr_derefs_may_alias_p): Likewise.
(refs_may_alias_p_1): Constructor bases also never alias.
Richard Guenther [Sun, 30 Aug 2009 11:52:13 +0000 (11:52 +0000)]
re PR tree-optimization/41186 (VN doesn't look through non-aliasing by offset memcpy)
2009-08-30 Richard Guenther <rguenther@suse.de>
PR tree-optimization/41186
* tree-ssa-alias.c (ptr_deref_may_alias_ref_p): Remove.
(ao_ref_init_from_ptr_and_size): New function.
(ref_maybe_used_by_call_p_1): Be more precise tracking
used ranges for builtin functions.
(ref_maybe_used_by_call_p): Adjust.
(call_may_clobber_ref_p_1): Be more precise tracking clobbered
ranges for builtin functions.
* tree-ssa-alias.h (ao_ref_init_from_ptr_and_size): Declare.
* g++.dg/torture/pr41186.C: New testcase.
* g++.dg/tree-ssa/pr41186.C: Likewise.
Daniel Kraft [Sun, 30 Aug 2009 08:26:38 +0000 (10:26 +0200)]
re PR fortran/37425 (Fortran 2003: GENERIC bindings as operators)
2009-08-30 Daniel Kraft <d@domob.eu>
PR fortran/37425
* dump-parse-tree.c (show_typebound_proc): Renamed from `show_typebound'
and accept gfc_typebound_proc and name instead of the symtree, needed
for intrinsic operator output.
(show_typebound_symtree): New method calling `show_typebound_proc'.
(show_f2k_derived): Output type-bound operators also.
(show_symbol): Moved output of `Procedure bindings:' label to
`show_f2k_derived'.
* gfortran.texi (Fortran 2003 status): Mention support of
array-constructors with explicit type specification, type-bound
procedures/operators, type extension, ABSTRACT types and DEFERRED.
Link to Fortran 2003 wiki page.
(Fortran 2008 status): Fix typo. Link to Fortran 2008 wiki page.
* gfc-internals.texi (Type-bound Procedures): Document the new
members/attributes of gfc_expr.value.compcall used for type-bound
operators.
(Type-bound Operators): New section documenting their internals.
Sebastian Pop [Fri, 28 Aug 2009 20:40:59 +0000 (20:40 +0000)]
graphite-interchange.c (pbb_interchange_profitable_p): Adjust the strides by multiplying by PDR_NB_REFS.
2009-08-28 Sebastian Pop <sebastian.pop@amd.com>
* graphite-interchange.c (pbb_interchange_profitable_p): Adjust
the strides by multiplying by PDR_NB_REFS.
* graphite-poly.c (can_collapse_pdr): New.
(pdr_find_duplicate): New.
(new_poly_dr): Call pdr_find_duplicate. Collapse duplicate PDRs.
Initialize PDR_NB_REFS.
* graphite-poly.h (struct poly_dr): Add field nb_refs.
(PDR_NB_REFS): New.
(new_poly_dr): Number of subscripts is a graphite_dim_t.
2009-08-28 Konrad Trifunovic <konrad.trifunovic@gmail.com>
Sebastian Pop <sebastian.pop@amd.com>
* graphite-blocking.c (pbb_strip_mine_loop_depth): Renamed
pbb_strip_mine_time_depth. Changed the implementation so that
transformation is expressed as a transformation on
time (scatttering) dimensions. Also, ensures that the 2d+1
scheduling format is preserved.
(pbb_strip_mine_profitable_p): Profitability is based on the
iteration number of a given time (scattering) dimension,
and not on a original loop depth dimension.
(pbb_strip_mine): Call pbb_number_of_iterations_at_time.
(pbb_do_strip_mine): Call psct_dynamic_dim.
* graphite-poly.c (pbb_number_of_iterations_at_time): New.
* graphite-poly.h (pbb_number_of_iterations_at_time): Declared.
(pbb_nb_dynamic_scattering_transform): New.
(psct_dynamic_dim): New.
Co-Authored-By: Sebastian Pop <sebastian.pop@amd.com>
From-SVN: r151186
Olivier Hainque [Fri, 28 Aug 2009 09:04:52 +0000 (09:04 +0000)]
collect2.c (DO_COLLECT_EXPORT_LIST): New internal macro, always defined.
* collect2.c (DO_COLLECT_EXPORT_LIST): New internal macro,
always defined. Reflect definition or absence of such for
COLLECT_EXPORT_LIST. Readability helper.
(scanfilter): New enum, to help control what symbols
are to be considered or ignored by scan_prog_file.
(enum pass): Rename as "scanpass", moved together with scanfilter
prior to scan_prog_file's prototype.
(scan_prog_file): Accept and honor scanpass and scanfilter arguments.
Group prototype with the scanpass/scanfilter definitions, factorize
head comments for the several implementations at the prototype.
(main): Reorganize the first pass link control to let AIX
drag only the needed frame tables in executables. Prevent
frame tables collection during the scan aimed at static ctors.
Pre-link and scan for frame tables later to compensate.
* doc/tm.texi (ASM_OUTPUT_DWARF_TABLE_REF): New macro.
A C statement to issue assembly directives that create a reference
to the given DWARF table identifier label from the current function
section.
* dwarf2out.c (switch_to_eh_frame_section): Add a BACK argument
to differentiate first time section entry. Only emit a .data
tables start identifier label the first time around.
(switch_to_frame_table_section): New function. Helper for
output_call_frame_info to switch possibly BACK into the eh_frame
or the debug_frame section depending on FOR_EH.
(output_call_frame_info): Use helper to first enter the proper
frame section.
(output_fde): Use ASM_OUTPUT_DWARF_TABLE_REF when defined to
emit a link to the frame table start label from each function
section.
* config/rs6000/rs6000.c (rs6000_aix_asm_output_dwarf_table_ref):
New function. Implementation of ASM_OUTPUT_DWARF_TABLE_REF.
* config/rs6000/rs6000-protos.h: Declare it.
* config/rs6000/aix.h (ASM_OUTPUT_DWARF_TABLE_REF): Define.
Uros Bizjak [Thu, 27 Aug 2009 20:54:28 +0000 (22:54 +0200)]
re PR rtl-optimization/40861 (ICE in simplify_subreg, at simplify-rtx.c:4981)
PR rtl-optimization/40861
* simplify-rtx.c (simplify_subreg): Do not call simplify_gen_subreg to
extract word from a multi-word subreg for negative byte positions.
testsuite/ChangeLog:
PR rtl-optimization/40861
* gcc.dg/pr40861.c: New test.
Janus Weil [Thu, 27 Aug 2009 19:48:46 +0000 (21:48 +0200)]
re PR fortran/40869 ([F03] PPC assignment checking)
2009-08-27 Janus Weil <janus@gcc.gnu.org>
PR fortran/40869
* expr.c (gfc_check_pointer_assign): Enable interface check for
pointer assignments involving procedure pointer components.
* gfortran.h (gfc_compare_interfaces): Modified prototype.
* interface.c (gfc_compare_interfaces): Add argument 'name2', to be
used instead of s2->name. Don't rely on the proc_pointer attribute,
but instead on the flags handed to this function.
(check_interface1,compare_parameter): Add argument for
gfc_compare_interfaces.
* resolve.c (check_generic_tbp_ambiguity): Ditto.
Daniel Kraft [Thu, 27 Aug 2009 11:42:56 +0000 (13:42 +0200)]
re PR fortran/37425 (Fortran 2003: GENERIC bindings as operators)
2009-08-27 Daniel Kraft <d@domob.eu>
PR fortran/37425
* gfortran.h (gfc_expr): Optionally store base-object in compcall value
and add a new flag to distinguish assign-calls generated.
(gfc_find_typebound_proc): Add locus argument.
(gfc_find_typebound_user_op), (gfc_find_typebound_intrinsic_op): Ditto.
(gfc_extend_expr): Return if failure was by a real error.
* interface.c (matching_typebound_op): New routine.
(build_compcall_for_operator): New routine.
(gfc_extend_expr): Handle type-bound operators, some clean-up and
return if failure was by a real error or just by not finding an
appropriate operator definition.
(gfc_extend_assign): Handle type-bound assignments.
* module.c (MOD_VERSION): Incremented.
(mio_intrinsic_op): New routine.
(mio_full_typebound_tree): New routine to make typebound-procedures IO
code reusable for type-bound user operators.
(mio_f2k_derived): IO of type-bound operators.
* primary.c (gfc_match_varspec): Initialize new fields in gfc_expr and
pass locus to gfc_find_typebound_proc.
* resolve.c (resolve_operator): Only output error about no matching
interface if gfc_extend_expr did not already fail with an error.
(extract_compcall_passed_object): Use specified base-object if present.
(update_compcall_arglist): Handle ignore_pass field.
(resolve_ordinary_assign): Update to handle extended code for
type-bound assignments, too.
(resolve_code): Handle EXEC_ASSIGN_CALL statement code.
(resolve_tb_generic_targets): Pass locus to gfc_find_typebound_proc.
(resolve_typebound_generic), (resolve_typebound_procedure): Ditto.
(resolve_typebound_intrinsic_op), (resolve_typebound_user_op): Ditto.
(ensure_not_abstract_walker), (resolve_fl_derived): Ditto.
(resolve_typebound_procedures): Remove not-implemented error.
(resolve_typebound_call): Handle assign-call flag.
* symbol.c (find_typebound_proc_uop): New argument to pass locus for
error message about PRIVATE, verify that a found procedure is not marked
as erraneous.
(gfc_find_typebound_intrinsic_op): Ditto.
(gfc_find_typebound_proc), (gfc_find_typebound_user_op): New locus arg.
2009-08-27 Daniel Kraft <d@domob.eu>
PR fortran/37425
* gfortran.dg/impure_assignment_1.f90: Change expected error message.
* gfortran.dg/typebound_operator_1.f03: Remove check for not-implemented
error and fix problem with recursive assignment.
* gfortran.dg/typebound_operator_2.f03: No not-implemented check.
* gfortran.dg/typebound_operator_3.f03: New test.
* gfortran.dg/typebound_operator_4.f03: New test.
Dodji Seketeli [Thu, 27 Aug 2009 07:05:24 +0000 (07:05 +0000)]
re PR debug/41170 (namespace DIE not generated when it contains only a typedef)
Fix for PR debug/41170
gcc/ChangeLog:
PR debug/41770
* dwarf2out.c (get_context_die): Declare this static function.
(gen_type_die_with_usage): Make sure a DIE is a generated for
the context of a typedef.
gcc/testsuite/ChangeLog:
PR debug/41770
* debug/dwarf2/namespace-1.C: New test.
Richard Guenther [Wed, 26 Aug 2009 13:23:04 +0000 (13:23 +0000)]
re PR c/41163 (verify_gimple fails)
2009-08-26 Richard Guenther <rguenther@suse.de>
PR middle-end/41163
* gimplify.c (gimplify_addr_expr): Canonicalize ADDR_EXPRs if
the types to not match.
* tree-cfg.c (verify_gimple_assign_single): Adjust ADDR_EXPR
verification.
* tree-ssa.c (useless_type_conversion_p): Conversions to
pointers to unprototyped functions are useless.
* gcc.c-torture/compile/pr41163.c: New testcase.
* gcc.dg/pr35899.c: Adjust.
Richard Guenther [Wed, 26 Aug 2009 09:02:01 +0000 (09:02 +0000)]
tree-ssa-structalias.c (create_variable_info_for): Remove strange whole-program condition, prepare to be called for non-globals.
2009-08-26 Richard Guenther <rguenther@suse.de>
* tree-ssa-structalias.c (create_variable_info_for): Remove
strange whole-program condition, prepare to be called for non-globals.
(intra_create_variable_infos): For restrict qualified DECL_BY_REFERENCE
params build a representative with known type and track its fields.
Jan Hubicka [Tue, 25 Aug 2009 21:44:20 +0000 (23:44 +0200)]
bmmintrin.h: Replace by #error.
* config/i386/bmmintrin.h: Replace by #error.
Revert:
Michael Meissner <michael.meissner@amd.com>
Dwarakanath Rajagopal <dwarak.rajagopal@amd.com>
Tony Linthicum <tony.linthicum@amd.com>
* config/i386/i386.h (TARGET_SSE5): New macro for SSE5.
(TARGET_ROUND): New macro for the round/ptest instructions which
are shared between SSE4.1 and SSE5.
(OPTION_MASK_ISA_ROUND): Ditto.
(OPTION_ISA_ROUND): Ditto.
(TARGET_FUSED_MADD): New macro for -mfused-madd swtich.
(TARGET_CPU_CPP_BUILTINS): Add SSE5 support.
* config/i386/i386.opt (-msse5): New switch for SSE5 support.
(-mfused-madd): New switch to give users control over whether the
compiler optimizes to use the multiply/add SSE5 instructions.
* config/i386/predicates.md (sse5_comparison_float_operator): New predicate to match the
comparison operators supported by the SSE5 com instruction.
(ix86_comparison_int_operator): New predicate to match just the
signed int comparisons.
(ix86_comparison_uns_operator): New predicate to match just the
unsigned int comparisons.
* config/i386/bmmintrin.h: New file, provide common x86 compiler
intrinisics for SSE5.
* config/i386/smmintrin.h: Move instructions shared with SSE5 to
mmintrin-common.h.
* config/i386/mmintrin-common.h: New file, to contain common
instructions between SSE4.1 and SSE5.
* config/i386/netware.c (gen_stdcall_or_fastcall_decoration): Use
FOREACH_FUNCTION_ARGS to iterate over the argument list.
(gen_regparm_prefix): Ditto.
* config/i386/winnt.c (gen_stdcall_or_fastcall_suffix): Use
FOREACH_FUNCTION_ARGS to iterate over the argument list. Use
prototype_p to determine if a function is prototyped.
Uros Bizjak [Tue, 25 Aug 2009 18:58:11 +0000 (20:58 +0200)]
alpha.md (*cmpdf_ieee_ext[123]): Remove.
* config/alpha/alpha.md (*cmpdf_ieee_ext[123]): Remove.
(*cmpdf_internal): Enable for all ALPHA_FPTM levels.
(*movdfcc_ext[1234]): Disable for IEEE mode.
Thomas Koenig [Tue, 25 Aug 2009 17:05:10 +0000 (17:05 +0000)]
re PR libfortran/34670 (bounds checking for array intrinsics)
2009-08-25 Thomas Koenig <tkoenig@gcc.gnu.org>
PR libfortran/34670
* runtime/bounds.c (count_0): New function.
* intrinsics/unpack_generic (unpack_bounds): New function.
(unpack_internal): Remove zero stride checks.
(unpack1): Use unpack_bounds.
(unpack1_char): Likeweise.
(unpack1_char4): Likewise
(unpack0): Likewise.
(unpack0_char): Likewise.
(unpack0_char4): Likewise.
2009-08-25 Thomas Koenig <tkoenig@gcc.gnu.org>
PR libfortran/34670
* gfortran.dg/unpack_bounds_1.f90: New test.
* gfortran.dg/unpack_bounds_2.f90: New test.
* gfortran.dg/unpack_bounds_3.f90: New test.
Eric Botcazou [Tue, 25 Aug 2009 15:02:19 +0000 (15:02 +0000)]
gimplify.c (prepare_gimple_addressable): New static function.
* gimplify.c (prepare_gimple_addressable): New static function.
(gimplify_modify_expr_to_memcpy): Invoke it on the RHS before marking
it addressable.
(gimplify_addr_expr): Invoke it similarly on the operand instead of
manually fiddling with it.
ada/
* gcc-interface/trans.c (call_to_gnu): Tidy.
(gnat_to_gnu) <N_Slice>: Set TYPE_ARRAY_MAX_SIZE if the slice has
non-constant size but the array itself has constant size.
* gcc-interface/utils.c (convert_vms_descriptor64): Fix type
consistency error.
(convert_vms_descriptor32): Likewise.
Janus Weil [Tue, 25 Aug 2009 14:26:44 +0000 (16:26 +0200)]
re PR fortran/41139 (a procedure pointer call as actual argument)
2009-08-25 Janus Weil <janus@gcc.gnu.org>
PR fortran/41139
* primary.c (gfc_match_varspec): Make sure EXPR_PPC is only used for
calls to procedure pointer components, other references to procedure
pointer components are EXPR_VARIABLE.
* resolve.c (resolve_actual_arglist): Bugfix (there can be calls without
actual arglist).
* trans-expr.c (gfc_get_proc_ptr_comp): Renamed to 'get_proc_ptr_comp',
removed argument 'se' and made static. Avoid inserting a temporary
variable for calling the PPC.
(conv_function_val): Renamed gfc_get_proc_ptr_comp.
(gfc_conv_procedure_call): Distinguish functions returning a procedure
pointer from calls to a procedure pointer. Distinguish calls to
procedure pointer components from procedure pointer components as
actual arguments.
* trans-stmt.h (gfc_get_proc_ptr_comp): Make it static.
Michael Matz [Tue, 25 Aug 2009 13:35:39 +0000 (13:35 +0000)]
expr.h (jumpifnot_1, [...]): Declare.
* expr.h (jumpifnot_1, jumpif_1, do_jump_1): Declare.
* dojump.c (do_jump_by_parts_greater): Take two operands instead of
full expression.
(do_jump_by_parts_equality, do_compare_and_jump): Ditto.
(jumpifnot_1, jumpif_1): New wrappers for do_jump_1.
(do_jump): Split out code for simple binary comparisons into ...
(do_jump_1): ... this, taking the individual operands and code.
Change callers to helper function above accordingly.
* expr.c (expand_expr_real_1): Use jumpifnot_1 for simple binary
comparisons.
Michael Matz [Tue, 25 Aug 2009 13:33:54 +0000 (13:33 +0000)]
expr.h (struct separate_ops, sepops): New type for passing around an exploded simple expression.
* expr.h (struct separate_ops, sepops): New type for passing
around an exploded simple expression.
* optabs.c (expand_widen_pattern_expr, expand_vec_shift_expr):
Use this structure instead of expression tree.
(get_vcond_icode, expand_vec_cond_expr_p): Don't take whole
expression, only its type.
(expand_vec_cond_expr): Take type and individual operands instead
of full expression.
* optabs.h (expand_widen_pattern_expr, expand_vec_cond_expr,
expand_vec_shift_expr): Change prototype accordingly.
* tree-vect-stmts.c (vectorizable_condition): Change call of
expand_vec_cond_expr_p to pass only type.
* expr.c (do_store_flags): Change prototype and implementation
to take an exploded expression.
(expand_expr_real_1): New local ops initialized with details
of the full expression. Use it instead of full
expression in calls to do_store_flags, expand_vec_cond_expr,
expand_widen_pattern_expr and expand_vec_shift_expr.
Michael Matz [Tue, 25 Aug 2009 13:31:56 +0000 (13:31 +0000)]
expr.c (expand_expr_real_1): New local treeop0...
* expr.c (expand_expr_real_1): New local treeop0, treeop1,
treeop2 initialized with first three operands of the full expression.
Substitute all TREE_OPERAND (exp, [012]) calls with them.
Kai Tietz [Tue, 25 Aug 2009 10:32:38 +0000 (10:32 +0000)]
gthr-win32.h (__UNUSED_PARAM): Define, if not already present.
2009-08-25 Kai Tietz <kai.tietz@onevision.com>
* gcc/gthr-win32.h (__UNUSED_PARAM): Define, if not already present.
(__gthread_objc_condition_allocate): Mark arguments as unused.
(__gthread_objc_condition_deallocate): Likewise.
(__gthread_objc_condition_wait): Likewise.
(__gthread_objc_condition_broadcast): Likewise.
(__gthread_objc_condition_signal): Likewise.
(__gthread_objc_thread_detach): Cast via INT_PTR to pointer.
(__gthread_objc_thread_id): Likewise.
projects / thirdparty / gcc.git / log
