ada/
* gcc-interface/utils.c (convert) <CONSTRUCTOR case>: When converting
to the packable version of the type, clear TREE_STATIC/TREE_CONSTANT
on the result if at least one of the input fields couldn't be output
as a static constant any more.
testsuite/
* gnat.dg/nested_float_packed.ads: New test.
Co-Authored-By: Eric Botcazou <ebotcazou@adacore.com>
From-SVN: r148049
* dbxout.c (dbxout_range_type): Add LOW and HIGH parameters. Use them
for bounds.
(print_int_cst_bounds_in_octal_p): Likewise.
(dbxout_type): Adjust calls to above functions. Be prepared to deal
with subtypes.
* dwarf2out.c (base_type_die): Likewise.
(is_subrange_type): Delete.
(subrange_type_die): Add LOW and HIGH parameters. Use them for bounds.
(modified_type_die): Call subrange_type_for_debug_p on subtypes.
* fold-const.c (fold_truth_not_expr) <CONVERT_EXPR>: Do not strip it if
the destination type is boolean.
(build_range_check): Do not special-case subtypes.
(fold_sign_changed_comparison): Likewise.
(fold_unary): Likewise.
* langhooks-def.h (LANG_HOOKS_GET_SUBRANGE_BOUNDS): Define.
(LANG_HOOKS_FOR_TYPES_INITIALIZER): Add LANG_HOOKS_GET_SUBRANGE_BOUNDS.
* langhooks.h (lang_hooks_for_types): Add get_subrange_bounds.
* tree.c (subrange_type_for_debug_p): New predicate based on the former
is_subrange_type.
* tree.h (subrange_type_for_debug_p): Declare.
* tree-chrec.c (avoid_arithmetics_in_type_p): Delete.
(convert_affine_scev): Remove call to above function.
(chrec_convert_aggressive): Likewise.
* tree-ssa.c (useless_type_conversion_p_1): Do not specifically return
false for conversions involving subtypes.
* tree-vrp.c (vrp_val_max): Do not special-case subtypes.
(vrp_val_min): Likewise.
(needs_overflow_infinity): Likewise.
(extract_range_from_unary_expr): Likewise.
ada/
* gcc-interface/ada-tree.h (TYPE_GCC_MIN_VALUE, TYPE_GCC_MAX_VALUE):
New macros.
(TYPE_RM_VALUES): Likewise.
(TYPE_RM_SIZE): Rewrite in terms of TYPE_RM_VALUES.
(SET_TYPE_RM_SIZE): New macro.
(TYPE_RM_MIN_VALUE, TYPE_RM_MAX_VALUE): Likewise.
(SET_TYPE_RM_SIZE, SET_TYPE_RM_MAX_VALUE): Likewise.
(TYPE_MIN_VALUE, TYPE_MAX_VALUE): Redefine.
* gcc-interface/gigi.h (create_range_type): Declare.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Modular_Integer_Type>
Use SET_TYPE_RM_MAX_VALUE to set the upper bound on the UMT type.
<E_Signed_Integer_Subtype>: Build a regular integer type first and
then set the RM bounds. Use SET_TYPE_RM_SIZE to set the RM size.
<E_Floating_Point_Subtype>: Build a regular floating-point type first
and then set the RM bounds.
<E_Array_Type>: Use create_range_type instead of build_range_type.
<E_Array_Subtype>: Build a regular integer type first and then set
the RM bounds for the extra subtype.
<E_String_Literal_Subtype>: Use create_range_type instead of
build_range_type.
<all>: Set the RM bounds for enumeration types and the GCC bounds for
floating-point types.
(set_rm_size): Use SET_TYPE_RM_SIZE to set the RM size.
(make_type_from_size) <INTEGER_TYPE>: Use SET_TYPE_RM_{MIN,MAX}_VALUE
to set the bounds. Use SET_TYPE_RM_SIZE to set the RM size.
(substitute_in_type) <INTEGER_TYPE>: Deal with GCC bounds for domain
types and with RM bounds for subtypes.
* gcc-interface/misc.c (LANG_HOOKS_GET_SUBRANGE_BOUNDS): Define.
(gnat_print_type) <REAL_TYPE>: New case.
<ENUMERAL_TYPE>: Fall through to above case.
(gnat_get_subrange_bounds): New function.
* gcc-interface/trans.c (add_decl_expr): Mark the trees rooted as
TYPE_RM_MIN_VALUE and TYPE_RM_MAX_VALUE, if any.
* gcc-interface/utils.c (gnat_init_decl_processing): Use precision 8
for booleans. Adjust and use SET_TYPE_RM_SIZE to set the RM size.
(create_range_type): New function.
(create_param_decl): Build a regular integer type first and then set
the RM bounds for the extra subtype.
(unchecked_convert): Remove kludge for 'Valid.
* gcc-interface/utils2.c (build_binary_op) <ARRAY_RANGE_REF>: Convert
the index to sizetype instead of TYPE_DOMAIN.
From-SVN: r147563
* ttypes.ads (Target_Double_Float_Alignment): New variable.
(Target_Double_Scalar_Alignment): Likewise.
* get_targ.ads (Get_Strict_Alignment): Adjust external name.
(Get_Double_Float_Alignment): New imported function.
(Get_Double_Scalar_Alignment): Likewise.
* layout.adb (Set_Elem_Alignment): Take into account specific caps for
the alignment of "double" floating-point types and "double" or larger
scalar types, as parameterized by Target_Double_Float_Alignment and
Target_Double_Scalar_Alignment respectively.
* gcc-interface/gigi.h (double_float_alignment): Declare.
(double_scalar_alignment): Likewise.
(is_double_float_or_array): Likewise.
(is_double_scalar_or_array): Likewise.
(get_target_double_float_alignment): Likewise.
(get_target_double_scalar_alignment): Likewise.
* gcc-interface/targtyps.c (get_strict_alignment): Rename into...
(get_target_strict_alignment): ...this.
(get_target_double_float_alignment): New function.
(get_target_double_scalar_alignment): Likewise.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Signed_Integer_Subtype>:
Test the presence of an alignment clause for under-aligned integer
types. Take into account specific caps for the alignment of "double"
floating-point types and "double" or larger scalar types, as
parameterized by Target_Double_Float_Alignment and
Target_Double_Scalar_Alignment respectively.
(validate_alignment): Likewise.
* gcc-interface/trans.c (Attribute_to_gnu) <Attr_Alignment>: Likewise.
(gigi): Initialize double_float_alignment and double_scalar_alignment.
* gcc-interface/utils.c (double_float_alignment): New global variable.
(double_scalar_alignment): Likewise.
(is_double_float_or_array): New predicate.
(is_double_scalar_or_array): Likewise.
From-SVN: r146675
* gcc-interface/gigi.h (create_index_type): Adjust head comment.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Signed_Integer_Subtype>:
Use front-end predicates to compute signedness and precision.
<E_String_Literal_Subtype>: Fold range type.
Make sure to set longest_float_type_node to a scalar type.
(elaborate_entity): Use consistent Constraint_Error spelling.
(substitute_in_type) <INTEGER_TYPE>: Always copy the type.
* gcc-interface/misc.c (gnat_print_type) <INTEGER_TYPE>: Use brief
output for the modulus, if any.
<ENUMERAL_TYPE>: Likewise for the RM size.
* gcc-interface/trans.c (gnat_to_gnu): Use consistent Constraint_Error
spelling.
* gcc-interface/utils.c (finish_record_type): Really test the alignment
of BLKmode bit-fields to compute their addressability.
(create_index_type): Adjust comments.
(create_param_decl): Create the biased subtype manually.
* gcc-interface/utils2.c (build_component_ref): Use consistent
Constraint_Error spelling.
From-SVN: r146644
* gcc-interface/decl.c (gnat_to_gnu_entity): Compute is_type predicate
on entry. Defer common processing for types. Reorder and clean up.
Compute the equivalent GNAT node and the default size for types only.
<E_Modular_Integer_Type>: Directly use Esize for the type's precision.
<E_Access_Type>: For an unconstrained designated type, do not pretend
that a dummy type is always made.
<all> Fix nits in comments.
(validate_size): Fix formatting nits and comments.
(set_rm_size): Likewise.
* gcc-interface/utils.c (create_param_decl): Replace bogus argument
passed to TARGET_PROMOTE_PROTOTYPES hook.
From-SVN: r146549
* fe.h (Get_External_Name): Declare.
* gcc-interface/gigi.h (concat_id_with_name): Rename to...
(concat_name): ...this.
* gcc-interface/decl.c (gnat_to_gnu_entity): Rename gnu_entity_id to
gnu_entity_name and adjust for above renaming.
<E_Access_Type>: Use create_concat_name to get the name of the various
types associated with unconstrained array types.
(make_aligning_type): Adjust for above renaming.
(maybe_pad_type): Likewise.
(components_to_record): Likewise. Use get_identifier_with_length for
the encoding of the variant.
(get_entity_name): Use get_identifier_with_length.
(create_concat_name): Likewise. Use Get_External_Name if no suffix.
Do not fiddle with Name_Buffer.
(concat_id_with_name): Rename to...
(concat_name): ...this. Use get_identifier_with_length. Do not fiddle
with Name_Buffer.
* gcc-interface/utils.c (rest_of_record_type_compilation): Adjust for
above renaming.
From-SVN: r146547
* ada-tree.h (TYPE_RM_SIZE_NUM): Delete.
(TYPE_RM_SIZE): Access TYPE_LANG_SLOT_1 directly for integral types.
* decl.c (gnat_to_gnu_entity) <E_Modular_Integer_Type>: Remove useless
support code for packed array types and assert its uselessness.
<E_Signed_Integer_Subtype>: Reuse entity identifier in more places and
adjust for TYPE_RM_SIZE change.
<all> Fix nits in comments. Use Original_Array_Type accessor instead
of Associated_Node_For_Itype accessor for packed array types.
(make_packable_type): Likewise.
(maybe_pad_type): Likewise.
(set_rm_size): Likewise. Rework conditional statement. Adjust for
TYPE_RM_SIZE change.
(make_type_from_size): Adjust for TYPE_RM_SIZE change.
(rm_size): Fix nits in comments. Rework conditional statements.
* misc.c (gnat_print_type): Adjust for TYPE_RM_SIZE change.
* trans.c (Attribute_to_gnu): Fix nits in comments.
* utils.c (gnat_init_decl_processing): Use more appropriate function
to initialize the size_type_node. Adjust for TYPE_RM_SIZE change.
From-SVN: r146530
* gcc-interface/ada-tree.h (DECL_HAS_REP_P): Delete.
* gcc-interface/decl.c (gnat_to_gnu_entity): Add support for extension
of types with unknown discriminants.
(substitute_in_type): Rewrite and restrict to formal substitutions.
* gcc-interface/utils.c (create_field_decl): Do not set DECL_HAS_REP_P.
(update_pointer_to): Update comment.
From-SVN: r146447
2009-04-10 Tristan Gingold <gingold@adacore.com>
* gcc-interface/Makefile.in, gcc-interface/utils.c: Include "rtl.h" to
avoid compile time warnings.
Do not add gcc/config in include search list while compiling the RTS.
Pragma Thread_Local_Storage is available on any target.
From-SVN: r145891
* gcc-interface/gigi.h (standard_datatypes): Remove ADT_void_type_decl.
(void_type_decl_node): Remove.
(init_gigi_decls): Likewise.
(gnat_install_builtins): Declare.
(record_builtin_type): Likewise.
(create_type_stub_decl): Likewise.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Void>: Use void_type.
(gnat_to_gnu_entity) <E_Array_Type>: Make fat and thin pointer types
artificial.
<E_Array_Subtype>: Use the index types, not only their name, in the
record giving the names of the bounds, if any.
For a packed array type, make it artificial only if the base type
was artificial as well. Remove redundant statement.
(gnat_to_gnu_entity) <E_Incomplete_Type>: Do not create TYPE_DECL for
dummy types.
Use create_type_stub_decl to build the TYPE_STUB_DECL of types.
(rest_of_type_decl_compilation_no_defer): Likewise.
* gcc-interface/misc.c (gnat_printable_name): Add missing guard.
* gcc-interface/utils.c (make_dummy_type): Always create TYPE_STUB_DECL
and use create_type_stub_decl to build it.
(gnat_pushdecl): Rewrite condition.
(gnat_install_builtins): Remove bogus declaration.
(record_builtin_type): New function.
(finish_record_type): Use create_type_stub_decl to build TYPE_STUB_DECL
of types.
(create_type_stub_decl): New function.
(create_type_decl): Assert that the type is not dummy. If the type
hasn't been named yet, equate the TYPE_STUB_DECL to the created node.
(build_vms_descriptor32): Do not create TYPE_DECL for the descriptor.
(build_vms_descriptor): Likewise.
(init_gigi_decls): Delete and move bulk of code to...
* gcc-interface/trans.c (gigi): ...here. Use record_builtin_type.
(emit_range_check): Add gnat_node parameter.
(emit_index_check): Likewise.
(emit_check): Likewise.
(build_unary_op_trapv): Likewise.
(build_binary_op_trapv): Likewise.
(convert_with_check): Likewise.
(Attribute_to_gnu): Adjust calls for above changes.
(call_to_gnu): Likewise.
(gnat_to_gnu): Likewise.
(assoc_to_constructor): Likewise.
(pos_to_constructor): Likewise.
(Sloc_to_locus): Set BUILTINS_LOCATION for Standard_Location nodes.
(process_type): Do not create TYPE_DECL for dummy types.
From-SVN: r145660
2008-10-06 Eric Botcazou <ebotcazou@adacore.com>
* gcc-interface/utils.c (can_fold_for_view_convert_p): New predicate.
(unchecked_convert): Use it to disable problematic folding with
VIEW_CONVERT_EXPR in the general case. Always disable it for the
special VIEW_CONVERT_EXPR built for integral types and cope with
its addressability issues by preserving the first conversion.
From-SVN: r140902
PR ada/37585
* gcc-interface/utils.c (create_subprog_decl): Disable inlining for
inlined external functions if they contain a nested function not
declared inline.
From-SVN: r140511
* gcc-interface/ada-tree.h (DECL_PARM_ALT): Now DECL_PARM_ALT_TYPE.
* gcc-interface/decl.c (gnat_to_gnu_param): Fix formatting, simplify
and adjust for above renaming.
* gcc-interface/utils.c (convert_vms_descriptor): Likewise. Add new
gnu_expr_alt_type parameter. Convert the expression to it instead
of changing its type in place.
(build_function_stub): Adjust call to above function.
From-SVN: r138518
2008-08-01 Eric Botcazou <ebotcazou@adacore.com>
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Constant>: Remove dead
code. Do not get full definition of deferred constants with address
clause for a use. Do not ignore deferred constant definitions with
address clause. Ignore constant definitions already marked with the
error node.
<object>: Remove obsolete comment. For a deferred constant with
address clause, get the initializer from the full view.
* gcc-interface/trans.c (gnat_to_gnu) <N_Attribute_Definition_Clause>:
Rework and remove obsolete comment.
<N_Object_Declaration>: For a deferred constant with address clause,
mark the full view with the error node.
* gcc-interface/utils.c (convert_to_fat_pointer): Rework and fix
formatting nits.
From-SVN: r138513
2008-08-01 Eric Botcazou <ebotcazou@adacore.com>
* gcc-interface/utils.c (convert_vms_descriptor): Add gnu_expr_alt_type
parameter.
Convert the expression to it instead of changing its type in place.
(build_function_stub): Adjust call to above function.
From-SVN: r138492
2008-07-31 Geert Bosch <bosch@adacore.com>
* arit64.c:
New file implementing __gnat_mulv64 signed integer multiplication with
overflow checking
* fe.h (Backend_Overflow_Checks_On_Target): Define for use by Gigi
* gcc-interface/gigi.h:
(standard_types): Add ADT_mulv64_decl
(mulv64_decl): Define subprogram declaration for __gnat_mulv64
* gcc-interface/utils.c:
(init_gigi_decls): Add initialization of mulv64_decl
* gcc-interface/trans.c:
(build_unary_op_trapv): New function
(build_binary_op_trapv): New function
(gnat_to_gnu): Use the above functions instead of
build_{unary,binary}_op
* gcc-interface/Makefile.in
(LIBGNAT_SRCS): Add arit64.c
(LIBGNAT_OBJS): Add arit64.o
From-SVN: r138384
PR ada/36554
* dwarf2out.c (is_subrange_type): Deal with BOOLEAN_TYPE.
ada/
* back_end.adb (Call_Back_End): Pass Standard_Boolean to gigi.
* gcc-interface/gigi.h (gigi): Take new standard_boolean parameter.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Enumeration_Subtype>:
Set precision to 1 for subtype of BOOLEAN_TYPE.
(set_rm_size): Set TYPE_RM_SIZE_NUM for BOOLEAN_TYPE.
(make_type_from_size): Deal with BOOLEAN_TYPE.
* gcc-interface/misc.c (gnat_print_type): Likewise.
* gcc-interface/trans.c (gigi): Take new standard_boolean parameter.
Set boolean_type_node as its translation in the table, as well
as boolean_false_node for False and boolean_true_node for True.
* gcc-interface/utils.c (gnat_init_decl_processing): Create custom
8-bit boolean_type_node and set its TYPE_RM_SIZE_NUM.
(create_param_decl): Deal with BOOLEAN_TYPE.
(build_vms_descriptor): Likewise.
(build_vms_descriptor64): Likewise.
(convert): Deal with BOOLEAN_TYPE like with ENUMERAL_TYPE.
From-SVN: r138348
2008-07-30 Doug Rupp <rupp@adacore.com>
* gigi.h (build_vms_descriptor64): New function prototype.
(fill_vms_descriptor): Modified function prototype.
* utils.c (build_vms_descriptor64): New function.
* utils2.c (fill_vms_descriptor): Fix handling on 32bit systems.
* trans.c (call_to_gnu): Call fill_vms_descriptor with new third
argument.
* decl.c (gnat_to_gnu_tree): For By_Descriptor mech, build both a
64bit and 32bit descriptor and save the 64bit version as an alternate
TREE_TYPE in the parameter.
(make_type_from_size) <RECORD_TYPE>: Use the appropriate mode for the
thin pointer.
* ada-tree.h (DECL_PARM_ALT, SET_DECL_PARM_ALT): New macros.
From-SVN: r138307
