Kazu Hirata [Wed, 17 May 2006 21:34:57 +0000 (21:34 +0000)]
re PR rtl-optimization/27477 (The H8 port doesn't build)
PR rtl-optimization/27477
* combine.c (try_combine): Don't split a parallel consisting
of two sets into two individual sets if both sets reference
cc0.
unwind-dw2.c: Add declarations for uw_update_context and uw_frame_state_for.
* unwind-dw2.c: Add declarations for uw_update_context and
uw_frame_state_for.
* pa/hpux-unwind.h (pa_fallback_frame_state): When an export stub is
detected, advance frame state and context once to skip over stub.
Sebastian Pop [Wed, 17 May 2006 14:25:59 +0000 (16:25 +0200)]
re PR tree-optimization/20256 (Perfect nest transformation not conservative enough)
PR middle-end/20256
PR middle-end/26435
* tree-loop-linear.c (linear_transform_loops): Don't test perfect_nest_p.
Call rewrite_into_loop_closed_ssa only when something changed.
* lambda.h (gcc_loopnest_to_lambda_loopnest): Update declaration.
* lambda-code.c (can_convert_to_perfect_nest): Declared.
(gcc_loopnest_to_lambda_loopnest): Removed need_perfect_nest parameter.
Test for perfect_nest_p here. Fix formating.
(replace_uses_equiv_to_x_with_y): Fix formating.
(stmt_uses_op): Removed.
(can_convert_to_perfect_nest): Removed loopivs parameter.
Complete the test by checking the scalar dependences.
(perfect_nestify): Remove the test for can_convert_to_perfect_nest.
Fix formating.
re PR fortran/26551 (gfortran compiles recursive subroutines declared without the RECURSIVE attribute)
PR fortran/26551
* resolve.c (resolve_call, resolve_function): Issue an error
if a function or subroutine call is recursive but the function or
subroutine wasn't declared as such.
Zdenek Dvorak [Wed, 17 May 2006 11:33:00 +0000 (13:33 +0200)]
re PR tree-optimization/27548 (ICE: SSA corruption - Conflict across an abnormal edge)
PR tree-optimization/27548
* tree-scalar-evolution.c (scev_const_prop): Do not prolong life
range of ssa names that appear on abnormal edges.
* tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Export.
* tree-flow.h (contains_abnormal_ssa_name_p): Declare.
Richard Guenther [Tue, 16 May 2006 15:34:12 +0000 (15:34 +0000)]
re PR tree-optimization/22303 (CCP does not handle STRING_CSTs)
2006-05-16 Richard Guenther <rguenther@suse.de>
PR tree-optimization/22303
* tree-ssa-ccp.c (fold_const_aggregate_ref): Handle reads
from STRING_CSTs.
(evaluate_stmt): Fall back to fold_const_aggregate_ref, if
ccp_fold did not simplify the statement.
Jakub Jelinek [Tue, 16 May 2006 10:16:36 +0000 (12:16 +0200)]
re PR middle-end/27573 (ICE with -fopenmp -fprofile-generate)
PR middle-end/27573
* omp-low.c (expand_omp_parallel): Don't assert
.OMP_DATA_I = &.OMP_DATA_O is the first statement in the block,
instead search for it.
* gcc.dg/gomp/pr27573.c: New test.
* gfortran.dg/gomp/pr27573.f90: New test.
Mark Mitchell [Tue, 16 May 2006 05:39:49 +0000 (05:39 +0000)]
re PR c++/27339 (out-of-class definition of value template parameter with private type)
PR c++/27339
* cp-tree.h (perform_access_checks): New function.
* semantics.c (perform_access_checks): New function.
(perform_deferred_access_checks): Use it.
* parser.c (cp_parser_simple_declaration): Adjust call to
cp_parser_init_declarator.
(cp_parser_type_parameter): Do not defer checks in default
arguments.
(cp_parser_explicit_specialization): Adjust call to
cp_parser_single_declaration.
(cp_parser_init_declarator): Perform template-parameter access
checks.
(cp_parser_parameter_declaration): Do not defer checks for
template parameter default arguments.
(cp_parser_template_declaration_after_export): Gather access
checks for template parameters, and pass them to
cp_parser_single_declaration.
(cp_parser_template_parameter_access_checks): New function.
(cp_parser_single_declaration): Add checks parameter.
PR c++/27339
* g++.dg/parser/access8.C: Adjust error marker.
* g++.dg/template/access17.C: New test.
* g++.dg/template/access18.C: Likewise.
Roger Sayle [Tue, 16 May 2006 04:16:00 +0000 (04:16 +0000)]
re PR target/26600 (internal compiler error: in push_reload, at reload.c:1303)
PR target/26600
* config/i386/i386.c (legitimate_constant_p) <CONST_DOUBLE>: TImode
integer constants other than zero are only legitimate on TARGET_64BIT.
<CONST_VECTOR> Only zero vectors are legitimate.
(ix86_cannot_force_const_mem): Integral and vector constants can
always be put in the constant pool.
Mark Mitchell [Mon, 15 May 2006 22:54:19 +0000 (22:54 +0000)]
re PR c++/27505 (ICE in const folding with bitfields)
PR c++/27505
* call.c (convert_like_real): Convert bitfields to their declared
types when forming an rvalue.
* tree.c (convert_bitfield_to_declared_type): New function.
(rvalue): Use it.
* cp-tree.h (convert_bitfield_to_declare_type): Declare it.
PR c++/27505
* g++.dg/expr/bitfield1.C: New test.
Zdenek Dvorak [Mon, 15 May 2006 18:24:55 +0000 (20:24 +0200)]
re PR tree-optimization/26830 (Repeated SSA update during loop header copying)
PR tree-optimization/26830
* tree-into-ssa.c (struct ssa_name_info): Add age field.
(info_for_ssa_name, current_info_for_ssa_name_age,
blocks_to_update): New variables.
(get_ssa_name_ann): Use info_for_ssa_name instead of SSA_NAME_AUX.
(clear_ssa_name_info, initialize_flags_in_bb,
mark_block_for_update): New functions.
(mark_def_sites, rewrite_stmt): Assert that blocks_to_update is NULL.
(insert_phi_nodes_for, mark_use_interesting, prepare_block_for_update,
prepare_def_site_for): Use mark_block_for_update.
(mark_def_interesting): Assert that the processed block is marked in
blocks_to_update. Do not take blocks argument.
(prepare_use_sites_for, prepare_names_to_update): Do not take blocks
argument.
(rewrite_update_init_block, rewrite_update_stmt): Only process
blocks with statements to rewrite.
(delete_update_ssa): Do not clear SSA_NAME_AUX.
(update_ssa): Initialize and free blocks_to_update. Do not
clear flags on statements. Do not use blocks bitmap.
* tree.h (SSA_NAME_AUX): Removed.
(struct tree_ssa_name): Removed aux field.
* print-tree.c (print_node): Do not print SSA_NAME_AUX.
Paul Thomas [Mon, 15 May 2006 17:16:26 +0000 (17:16 +0000)]
re PR fortran/25090 (Bad automatic character length)
2006-05-15 Paul Thomas <pault@gcc.gnu.org>
PR fortran/25090
* resolve.c: Static resolving_index_expr initialized.
(entry_parameter): New function to emit errors for variables
that are not entry parameters.
(gfc_resolve_expr): Call entry_parameter, when resolving
variables, if the namespace has entries and resolving_index_expr
is set.
(resolve_charlen): Set resolving_index_expr before the call to
resolve_index_expr and reset it afterwards.
(resolve_fl_variable): The same before and after the call to
is_non_constant_shape_array, which ultimately makes a call to
gfc_resolve_expr.
PR fortran/25082
* resolve.c (resolve_code): Add error condition that the return
expression must be scalar.
PR fortran/24711
* matchexp.c (gfc_get_parentheses): New function.
(match_primary): Remove inline code and call above.
* gfortran.h: Provide prototype for gfc_get_parentheses.
* resolve.c (resolve_array_ref): Call the above, when start is a
derived type variable array reference.
2006-05-15 Paul Thomas <pault@gcc.gnu.org>
PR fortran/25090
* gfortran.dg/entry_dummy_ref_1.f90: New test.
PR fortran/25082
* gfortran.dg/scalar_return_1.f90: New test.
PR fortran/24711
* gfortran.dg/derived_comp_array_ref_1.f90: New test.
Mircea Namolaru [Mon, 15 May 2006 10:36:39 +0000 (10:36 +0000)]
see.c: Code style changes such as redundant paranthesis...
2006-05-15 Mircea Namolaru <namolaru@il.ibm.com>
* see.c: Code style changes such as redundant paranthesis,
redundant intialization of local variables etc.
(see_main): Declared now as static.
* doc/invoke.texi: Update that -fsee is not enabled by
default at -O3.
Roger Sayle [Sun, 14 May 2006 15:48:11 +0000 (15:48 +0000)]
re PR rtl-optimization/22563 (performance regression for gcc newer than 2.95)
PR rtl-optimization/22563
* expmed.c (store_fixed_bit_field): When using AND and IOR to store
a fixed width bitfield, always force the intermediates into psuedos.
Steven Bosscher [Sat, 13 May 2006 23:14:47 +0000 (23:14 +0000)]
df-core.c (df_bb_regno_last_use_find): Do not look for dataflow information attached to non-INSNs such as NOTEs.
* df-core.c (df_bb_regno_last_use_find): Do not look for dataflow
information attached to non-INSNs such as NOTEs.
(df_bb_regno_first_def_find, df_bb_regno_last_def_find): Likewise.
Bryce McKinlay [Thu, 11 May 2006 20:53:04 +0000 (20:53 +0000)]
NameFinder.java (blacklist): New static field.
2006-05-11 Bryce McKinlay <mckinlay@redhat.com>
* gnu/gcj/runtime/NameFinder.java (blacklist): New static field.
(lookup): If addr2line fails to find an address, flag the binary as
having no debug info and avoid calling addr2line on it again.
Richard Guenther [Thu, 11 May 2006 08:29:40 +0000 (08:29 +0000)]
re PR middle-end/27529 (Does not fold (char *)(size_t)char_ptr or (size_t)(char *)size_t_var)
2006-05-11 Richard Guenther <rguenther@suse.de>
PR middle-end/27529
* fold-const.c (fold_unary): Handle intermediate conversion
to a pointer type like intermediate conversion to an integer
type in folding of (T1)(T2)var to var.
Match the code to the comment in the final conversion for
(T1)(T2)var to (T1)var regarding to type precision. Rather
than disallow T1 being of pointer type, assert that both T1
and var are of pointer type or not. Make sure not to fall
over the frontends lazyness wrt array to pointer decay though.