Eric Botcazou [Thu, 3 Nov 2005 19:57:55 +0000 (19:57 +0000)]
dwarf2asm.c (dw2_force_const_mem): Add new parameter 'public'.
* dwarf2asm.c (dw2_force_const_mem): Add new parameter 'public'.
On USE_LINKONCE_INDIRECT platforms, build a DECL_ONE_ONLY indirect
reference only if 'public' is true.
(dw2_output_indirect_constant_1): On USE_LINKONCE_INDIRECT platforms,
emit the .hidden directive only if the indirect reference is public.
(dw2_asm_output_encoded_addr_rtx): Add new parameter 'public'.
Pass it to dw2_force_const_mem.
* dwarf2asm.h (dw2_asm_output_encoded_addr_rtx): New param 'public'.
* dwarf2out.c (output_cfi): Adjust calls to above function.
(output_call_frame_info): Likewise.
* except.c (output_ttype): Pass TREE_PUBLIC of the type_info object
as 'public' argument to dw2_asm_output_encoded_addr_rtx.
Andrew Pinski [Thu, 3 Nov 2005 16:15:53 +0000 (16:15 +0000)]
re PR middle-end/23155 (Gimplification failed for union cast)
2005-11-03 Andrew Pinski <pinskia@physics.uc.edu>
PR middle-end/23155
* g++.dg/ext/c99struct1.C: New test.
* gcc.dg/union-cast-1.c: New test.
* gcc.dg/union-cast-2.c: New test.
* gcc.dg/union-cast-3.c: New test.
2005-11-03 Andrew Pinski <pinskia@physics.uc.edu>
PR middle-end/23155
* gimplifier.c (gimplify_expr): Create a temporary for lvalue
CONSTRUCTOR.
Daniel Berlin [Thu, 3 Nov 2005 15:39:48 +0000 (15:39 +0000)]
re PR tree-optimization/24351 (ICE in do_simple_structure_copy with some C++ code)
2005-11-03 Daniel Berlin <dberlin@dberlin.org>
Fix PR tree-optimization/24351
* tree-ssa-structalias.c (struct variable_info): Add
collapsed_into.
(get_varinfo_fc): New function to follow collapsing.
(new_var_info): Set collapsed_to to NULL.
(dump_constraint): Follow collapsing.
(build_constraint_graph): Handle collapsing.
(do_simple_structure_copy): Return false if something bad
happened.
(collapse_rest_of_var): New function.
(do_structure_copy): Collapse if do_simple_structure_copy returns
false.
Paolo Carlini [Thu, 3 Nov 2005 13:13:21 +0000 (13:13 +0000)]
sso_string_base.h (__sso_string_base<>::_M_swap): Change the various traits_type::copy call to always copy the entire local buffer...
2005-11-03 Paolo Carlini <pcarlini@suse.de>
* include/ext/sso_string_base.h (__sso_string_base<>::_M_swap):
Change the various traits_type::copy call to always copy the
entire local buffer; return early and don't do a full swap on
the lengths for two common cases; change two _S_copy to plain
traits_type::copy.
Eric Botcazou [Thu, 3 Nov 2005 11:31:46 +0000 (12:31 +0100)]
re PR rtl-optimization/23585 (mem_fun* code fine with -O1, bus error with -O2)
PR rtl-optimization/23585
* rtlanal.c (rtx_addr_can_trap_p_1) <PLUS>: Return 0 for an address
that can't trap plus a constant integer, if the mode has zero size.
Mark Mitchell [Wed, 2 Nov 2005 23:36:05 +0000 (23:36 +0000)]
re PR c++/24560 ("insufficient contextual information to determine type" is not a helpful error message)
PR c++/24560
* parser.c (cp_parser_postfix_dot_deref_expression): Improve error
message for use of overloaded functions on LHS of "." operator.
PR c++/24560
* g++.dg/parse/dot1.C: New test.
Ulrich Weigand [Wed, 2 Nov 2005 23:16:31 +0000 (23:16 +0000)]
re PR target/24615 (internal compiler error: in print_shift_count_operand, at config/s390/s390.c:4025)
ChangeLog:
PR target/24615
* config/s390/s390-protos.h (s390_decompose_shift_count): Declare.
* config/s390/s390.c (s390_decompose_shift_count): New function.
(s390_extra_constraint_str) ['Y']: Use s390_decompose_shift_count.
(print_shift_count_operand): Use s390_decompose_shift_count.
* config/s390/predicates.md ("setmem_operand", "shift_count_operand"):
Use s390_decompose_shift_count. Do not accept any non-base hard regs.
PR 22429
* fold-const.c (build_range_check): Use unsigned when signed
overflow is undefined also. If etype is subtype, make sure that
the subtraction is in the supertype.
Mark Mitchell [Wed, 2 Nov 2005 21:34:51 +0000 (21:34 +0000)]
re PR c++/19253 (bad error message / ICE for invalid template parameter)
PR c++/19253
* parser.c (cp_parser_postfix_expression): Use
cp_parser_elaborated_type_specifier to handle typename-types in
functional casts.
(cp_parser_enclosed_argument_list): Skip ahead to the end of the
template argument list if the closing ">" is not found.
PR c++/19253
* g++.dg/parse/typename8.C: Compile with -w -fpermissive.
* g++.dg/parse/typename9.C: New test.
* g++/dg/parse/typename10.C: Likewise.
Andrew Pinski [Wed, 2 Nov 2005 21:01:54 +0000 (21:01 +0000)]
re PR fortran/18157 (ice-on-valid code, pointer to user-defined type, fold-struct.c)
2005-11-02 Andrew Pinski <pinskia@physics.uc.edu>
PR fortran/18157
* gfortran.fortran-torture/compile/defined_type_1.f90: New test.
* gfortran.fortran-torture/compile/defined_type_2.f90: New
test.
* gfortran.fortran-torture/compile/defined_type_3.f90:
New test.
2005-11-02 Andrew Pinski <pinskia@physics.uc.edu>
PR fortran/18157
* trans-array.c (gfc_conv_resolve_dependencies): Use the correct
type for the temporary array.
* trans-expr.c (gfc_trans_assignment): Pass lss
instead of lss_section
to gfc_conv_resolve_dependencies to get the
correct type.
Tobias Schlüter [Wed, 2 Nov 2005 00:04:28 +0000 (01:04 +0100)]
decl.c (gfc_match_entry): Function entries don't need an argument list if there's no RESULT clause.
fortran/
* decl.c (gfc_match_entry): Function entries don't need an argument
list if there's no RESULT clause.
testsuite/
* gfortran.fortran-torture/execute/entry_9.f90: Revert previous
change (r106358).
Erik Edelmann [Tue, 1 Nov 2005 21:40:06 +0000 (21:40 +0000)]
re PR fortran/24245 (-fdump-parse-tree gives ICE for CONTAINED functions)
fortran/
2005-11-01 Erik Edelmann <eedelman@gcc.gnu.org>
PR 24245
* trans.c (gfc_generate_code): Move code to create a main
program symbol from here ...
* parse.c (main_program_symbol): ... to this new
function, setting the locus from gfc_current_locus
instead of ns->code->loc.
(gfc_parse_file): Call main_program_symbol for main programs.
Bob Wilson [Tue, 1 Nov 2005 18:07:13 +0000 (18:07 +0000)]
xtensa.h (TRAMPOLINE_TEMPLATE): Add a byte of padding.
* config/xtensa/xtensa.h (TRAMPOLINE_TEMPLATE): Add a byte of padding.
(TRAMPOLINE_SIZE): Round up to 60.
* config/xtensa/lib2funcs.S (TRAMPOLINE_SIZE): Likewise.
Paul Thomas [Tue, 1 Nov 2005 05:53:29 +0000 (05:53 +0000)]
re PR fortran/21565 (namelist in block data is illegal)
2005-11-01 Paul Thomas <pault@gcc.gnu.org>
PR fortran/21565
* symbol.c (check_conflict): An object cannot be in a namelist and in
block data.
PR fortran/18737
* resolve.c (resolve_symbol): Set the error flag to
gfc_set_default_type, in the case of an external symbol, so that
an error message is emitted if IMPLICIT NONE is set.
PR fortran/14994
* gfortran.h (gfc_generic_isym_id): Add GFC_ISYM_SECNDS to enum.
* check.c (gfc_check_secnds): New function.
* intrinsic.c (add_functions): Add call to secnds.
* iresolve.c (gfc_resolve_secnds): New function.
* trans-intrinsic (gfc_conv_intrinsic_function): Add call to
secnds via case GFC_ISYM_SECNDS.
* intrinsic.texi: Add documentation for secnds.
2005-11-01 Paul Thomas <pault@gcc.gnu.org>
PR fortran/14994
* libgfortran/intrinsics/date_and_time.c: Add interface to
the functions date_and_time for the intrinsic function secnds.
2005-11-01 Paul Thomas <pault@gcc.gnu.org>
PR fortran/21565
gfortran.dg/namelist_blockdata.f90: New test.
PR fortran/18737
gfortran.dg/external_implicit_none.f90: New test.
PR fortran/14994
* gfortran.dg/secnds.f: New test.
James E Wilson [Mon, 31 Oct 2005 23:39:29 +0000 (15:39 -0800)]
Restore old AT_frame_base code for targets that don't use dwarf2 unwind info.
Partial fix.
PR debug/24444
* dwarf2out.c (convert_cfa_to_loc_list): Put inside DWARF2_UNWIND_INFO
ifdef. Put ifdefs around call in gen_subprogram_die.
(compute_frame_pointer_to_cfa_displacement): Likewise.
(gen_subprogram_die): Restore old code for when DWARF2_UNWIND_INFO is
not defined.
Jan Hubicka [Mon, 31 Oct 2005 14:48:57 +0000 (15:48 +0100)]
re PR gcov-profile/24487 (Basic block frequencies inaccurate)
PR profile/24487
* predict.c (predict_loops): Do not estimate more than
MAX_PRED_LOOP_ITERATIONS in PRED_LOOP_ITERATIONS heuristic.
* predict.def (MAX_PRED_LOOP_ITERATIONS): Define.
Andrew Pinski [Mon, 31 Oct 2005 13:22:20 +0000 (13:22 +0000)]
re PR middle-end/23492 (ACATS c48009e SEGV in set_bb_for_stmt tree-cfg.c:2673 on x86_64)
2005-10-31 Andrew Pinski <pinskia@physics.uc.edu>
PR middle-end/23492
* tree-inline.c (setup_one_parameter): If the init_stmt
is NULL, don't insert the statement.
2005-10-31 Andrew Pinski <pinskia@physics.uc.edu>
PR middle-end/23492
* gcc.c-torture/compile/zero-strct-5.c: New test.
Erik Edelmann [Sun, 30 Oct 2005 22:43:45 +0000 (22:43 +0000)]
re PR fortran/18883 (ICE in gfc_finish_var_decl)
fortran/
2005-10-30 Erik Edelmann <eedelman@gcc.gnu.org>
PR fortran/18883
* trans-decl.c (gfc_finish_var_decl): Add decl to the
current function, rather than the parent. Make
assertion accept fake result variables.
* trans-expr.c (gfc_conv_variable): If the character
length of an ENTRY isn't set, get the length from
the master function instead.
testsuite
2005-10-30 Erik Edelmann <eedelman@gcc.gnu.org>
* check.c (gfc_check_malloc, gfc_check_free): New functions.
* gfortran.h (gfc_generic_isym_id): Add GFC_ISYM_MALLOC.
* intrinsic.c (add_functions): Add symbols for MALLOC function.
(add_subroutines): Add symbol for FREE subroutine.
* intrinsic.h: Prototypes for gfc_check_malloc, gfc_check_free,
gfc_resolve_malloc and gfc_resolve_free.
* intrinsic.texi: Add doc for FREE and MALLOC intrinsics.
* iresolve.c (gfc_resolve_malloc, gfc_resolve_free): New
functions.
* trans-intrinsic.c (gfc_conv_intrinsic_function): Add case for
GFC_ISYM_MALLOC.
* Makefile.am: Add intrinsics/malloc.c file.
* Makefile.in: Regenerate.
* intrinsics/malloc.c: New file, with implementations for free
and malloc library functions.
* config/ms1/ms1.md ("decrement_and_branch_until_zero"): Rewrite.
("*decrement_and_branch_until_zero_no_clobber"): New.
Add corresponding splitter for decrement_and_branch_until_zero
instruction.
Key all decrement_and_branch_until_zero patterns off of
TARGET_MS1_16_003.
check.c (gfc_check_alarm_sub, [...]): New functions.
* check.c (gfc_check_alarm_sub, gfc_check_signal,
gfc_check_signal_sub): New functions.
* gfortran.h (gfc_generic_isym_id): Add GFC_ISYM_SIGNAL.
* intrinsic.c (add_functions): Add signal intrinsic.
(add_subroutines): Add signal and alarm intrinsics.
* intrinsic.texi: Document the new intrinsics.
* iresolve.c (gfc_resolve_signal, gfc_resolve_alarm_sub,
gfc_resolve_signal_sub): New functions.
* trans-intrinsic.c (gfc_conv_intrinsic_function): Add case
for GFC_ISYM_SIGNAL.
* intrinsic.h: Add prototypes for gfc_check_alarm_sub,
gfc_check_signal, gfc_check_signal_sub, gfc_resolve_signal,
gfc_resolve_alarm_sub, gfc_resolve_signal_sub.
* Makefile.am (intrinsics): Add signal.c.
* Makefile.in: Regenerate.
* configure.ac: Checks for signal and alarm.
* config.h.in: Regenerate.
* configure: Regenerate.
* intrinsics/signal.c: New file for SIGNAL and ALARM intrinsics.
Andrew Pinski [Fri, 28 Oct 2005 20:25:55 +0000 (20:25 +0000)]
re PR c++/23229 (g++ gives incorrect error message with void main() and a void function)
2005-10-28 Andrew Pinski <pinskia@physics.uc.edu>
PR C++/23229
* decl.c (grokfndecl): Create a new function type
after erroring out about main not returning int.
2005-10-28 Andrew Pinski <pinskia@physics.uc.edu>
acinclude.m4 (LIBGFOR_CHECK_FPSETMASK): New check.
* acinclude.m4 (LIBGFOR_CHECK_FPSETMASK): New check.
* configure.ac: Check for floatingpoint.h, fptrap.h and float.h
headers. Use LIBGFOR_CHECK_FPSETMASK. Check for fp_trap and
fp_enable functions.
* configure.host: Add case for systems with fpsetmask and systems
with fp_trap/fp_enable.
* config/fpu-sysv.h: New file, FPU code using fpsetmask.
* config/fpu-aix.h: New file, FPU code for AIX using fp_trap and
fp_enable.
* aclocal.m4: Regenerate.
* configure: Regenerate.
* config.h.in: Regenerate.