1 new GCC HEAD@194625 regression
GCC regression tester
regress@geoffk.org
Thu Dec 20 02:10:00 GMT 2012
With your recent patch, GCC HEAD revision 194625 had problems on:
native: regress-3 (1 NEW failures)
The previous build was of revision 194620.
Regressions for native:
old g++.sum g++.dg/other/anon5.C
old gfortran.sum gfortran.dg/lto/pr45586
fixed gfortran.sum gfortran.dg/pr26246_2.f90
NEW gfortran.sum gfortran.dg/same_type_as_1.f03
Log information for changes since the last build:
------------------------------------------------------------------------
r194621 | burnus | 2012-12-19 15:05:49 -0800 (Wed, 19 Dec 2012) | 7 lines
Changed paths:
M /trunk/gcc/fortran/ChangeLog
M /trunk/gcc/fortran/trans-decl.c
2012-12-19 Tobias Burnus <burnus@net-b.de>
PR fortran/55733
* trans-decl.c (gfc_create_string_length): Avoid setting
TREE_STATIC for automatic variables with -fno-automatic.
------------------------------------------------------------------------
r194622 | pault | 2012-12-19 16:15:00 -0800 (Wed, 19 Dec 2012) | 131 lines
Changed paths:
M /trunk/gcc/fortran/ChangeLog
M /trunk/gcc/fortran/array.c
M /trunk/gcc/fortran/check.c
M /trunk/gcc/fortran/class.c
M /trunk/gcc/fortran/decl.c
M /trunk/gcc/fortran/expr.c
M /trunk/gcc/fortran/gfortran.h
M /trunk/gcc/fortran/interface.c
M /trunk/gcc/fortran/match.c
M /trunk/gcc/fortran/misc.c
M /trunk/gcc/fortran/module.c
M /trunk/gcc/fortran/resolve.c
M /trunk/gcc/fortran/simplify.c
M /trunk/gcc/fortran/symbol.c
M /trunk/gcc/fortran/trans-decl.c
M /trunk/gcc/fortran/trans-expr.c
M /trunk/gcc/fortran/trans-intrinsic.c
M /trunk/gcc/fortran/trans-stmt.c
M /trunk/gcc/fortran/trans-types.c
M /trunk/gcc/fortran/trans.c
M /trunk/gcc/testsuite/ChangeLog
A /trunk/gcc/testsuite/gfortran.dg/unlimited_polymorphic_1.f03
A /trunk/gcc/testsuite/gfortran.dg/unlimited_polymorphic_2.f03
A /trunk/gcc/testsuite/gfortran.dg/unlimited_polymorphic_3.f03
M /trunk/libgfortran/ChangeLog
M /trunk/libgfortran/intrinsics/extends_type_of.c
2012-12-19 Paul Thomas <pault@gcc.gnu.org>
* array.c (resolve_array_list): Apply C4106.
* check.c (gfc_check_same_type_as): Exclude polymorphic
entities from check for extensible types. Improved error
for disallowed argument types to name the offending type.
* class.c : Update copyright date.
(gfc_class_null_initializer): Add argument for initialization
expression and deal with unlimited polymorphic typespecs.
(get_unique_type_string): Give unlimited polymorphic
entities a type string.
(gfc_intrinsic_hash_value): New function.
(gfc_build_class_symbol): Incorporate unlimited polymorphic
entities.
(gfc_find_derived_vtab): Deal with unlimited polymorphic
entities.
(gfc_find_intrinsic_vtab): New function.
* decl.c (gfc_match_decl_type_spec): Match typespec for
unlimited polymorphic type.
(gfc_match_data_decl): Skip to 'ok' if unlimited polymorphic.
expr.c (gfc_check_pointer_assign): Apply C717. If unlimited
polymorphic lvalue, find rvalue vtable for all typespecs,
except unlimited polymorphic expressions.
(gfc_check_vardef_context): Handle unlimited polymorphic
entities.
* gfortran.h : Add unlimited polymorphic attribute. Add
second arg to gfc_class_null_initializer primitive and
primitive for gfc_find_intrinsic_vtab. Add UNLIMITED_POLY
to detect unlimited polymorphic expressions.
* interface.c (gfc_compare_types): If expr1 is unlimited
polymorphic, always return 1. If expr2 is unlimited polymorphic
enforce C717.
(gfc_compare_interfaces): Skip past conditions that do not
apply for unlimited polymorphic entities.
(compare_parameter): Make sure that an unlimited polymorphic,
allocatable or pointer, formal argument is matched by an
unlimited polymorphic actual argument.
(compare_actual_formal): Ensure that an intrinsic vtable exists
to match an unlimited polymorphic formal argument.
* match.c (gfc_match_allocate): Type kind parameter does not
need to match an unlimited polymorphic allocate-object.
(alloc_opt_list): An unlimited polymorphic allocate-object
requires a typespec or a SOURCE tag.
(select_intrinsic_set_tmp): New function.
(select_type_set_tmp): Call new function. If it returns NULL,
build a derived type or class temporary instead.
(gfc_match_type_is): Remove restriction to derived types only.
Bind(C) or sequence derived types not permitted.
* misc (gfc_typename): Printed CLASS(*) for unlimited
polymorphism.
* module.c : Add AB_UNLIMITED_POLY to pass unlimited
polymorphic attribute to and from modules.
* resolve.c (resolve_common_vars): Unlimited polymorphic
entities cannot appear in common blocks.
(resolve_deallocate_expr): Deallocate unlimited polymorphic
enities.
(resolve_allocate_expr): Likewise for allocation. Make sure
vtable exists.
(gfc_type_is_extensible): Unlimited polymorphic entities are
not extensible.
(resolve_select_type): Handle unlimited polymorphic selectors.
Ensure that length type parameters are assumed and that names
for intrinsic types are generated.
(resolve_fl_var_and_proc): Exclude select type temporaries
from test of extensibility of type.
(resolve_fl_variable): Likewise for test that assumed character
length must be a dummy or a parameter.
(resolve_fl_derived0): Return SUCCESS unconditionally for
unlimited polymorphic entities. Also, allow unlimited
polymorphic components.
(resolve_fl_derived): Return SUCCESS unconditionally for
unlimited polymorphic entities.
(resolve_symbol): Return early with unlimited polymorphic
entities.
* simplifiy.c : Update copyright year.
(gfc_simplify_extends_type_of): No simplification possible
for unlimited polymorphic arguments.
* symbol.c (gfc_use_derived): Nothing to do for unlimited
polymorphic "derived type".
(gfc_type_compatible): Return unity if ts1 is unlimited
polymorphic.
* trans-decl.c (create_function_arglist) Formal arguments
without a character length should be treated in the same way
as passed lengths.
(gfc_trans_deferred_vars): Nullify the vptr of unlimited
polymorphic pointers. Avoid unlimited polymorphic entities
triggering gcc_unreachable.
* trans-expr.c (gfc_conv_intrinsic_to_class): New function.
(gfc_trans_class_init_assign): Make indirect reference of
src.expr.
(gfc_trans_class_assign): Expression NULL of unknown type
should set NULL vptr on lhs. Treat C717 cases where lhs is
a derived type and the rhs is unlimited polymorphic.
(gfc_conv_procedure_call): Handle the conversion of a non-class
actual argument to match an unlimited polymorphic formal
argument. Suppress the passing of a character string length
in this case. Make sure that calls to the character __copy
function have two character string length arguments.
(gfc_conv_initializer): Pass the initialization expression to
gfc_class_null_initializer.
(gfc_trans_subcomponent_assign): Ditto.
(gfc_conv_structure): Move handling of _size component.
trans-intrinsic.c: (gfc_conv_same_type_as): Handle conditions
where unlimited polymorphic arguments have null vptr.
* trans-stmt.c (trans_associate_var): Correctly treat array
temporaries associated with unlimited polymorphic selectors.
Recover the overwritten dtype for the descriptor. Use the _size
field of the vptr for character string lengths.
(gfc_trans_allocate): Cope with unlimited polymorphic allocate
objects; especially with character source tags.
(reset_vptr): New function.
(gfc_trans_deallocate): Call it.
* trans-types.c (gfc_get_derived_type): Detect unlimited
polymorphic types and deal with cases where the derived type of
components is null.
* trans.c : Update copyright year.
(trans_code): Call gfc_trans_class_assign for C717 cases where
the lhs is not unlimited polymorphic.
2012-12-19 Paul Thomas <pault@gcc.gnu.org>
* intrinsics/extends_type_of.c : Return correct results for
null vptrs.
2012-12-19 Paul Thomas <pault@gcc.gnu.org>
* gfortran.dg/unlimited_polymorphic_1.f03: New test.
* gfortran.dg/unlimited_polymorphic_2.f03: New test.
* gfortran.dg/unlimited_polymorphic_3.f03: New test.
* gfortran.dg/same_type_as.f03: Correct for improved message.
------------------------------------------------------------------------
r194625 | gccadmin | 2012-12-19 16:18:51 -0800 (Wed, 19 Dec 2012) | 1 line
Changed paths:
M /trunk/gcc/DATESTAMP
Daily bump.
------------------------------------------------------------------------
For more information, see <http://glutton.geoffk.org/HEAD/>.
-------------- next part --------------
--
Geoffrey Keating <geoffk@geoffk.org>
(via an automated GCC regression-testing script.)
More information about the Gcc-regression
mailing list