* tree-parloops.c (loop_parallel_p): New argument
parloop_obstack. Pass it down.
(parallelize_loops): New variable parloop_obstack. Initialize it,
pass it down, free it.
* tree-loop-linear.c (linear_transform_loops): Pass down
lambda_obstack.
* tree-data-ref.h (lambda_compute_access_matrices): New argument
of type struct obstack *.
* tree-data-ref.c (analyze_subscript_affine_affine): New variable
scratch_obstack. Initialize it, pass down, free it.
* lambda.h (lambda_loop_new): Remove.
(lambda_matrix_new, lambda_matrix_inverse)
(lambda_trans_matrix_new, lambda_trans_matrix_inverse): New
argument of type struct obstack *.
* lambda-trans.c (lambda_trans_matrix_new): New argument
lambda_obstack. Pass it down, use obstack allocation for ret.
(lambda_trans_matrix_inverse): New argument lambda_obstack. Pass
it down.
* lambda-mat.c (lambda_matrix_get_column)
(lambda_matrix_project_to_null): Remove.
(lambda_matrix_new): New argument lambda_obstack. Use obstack
allocation for mat.
(lambda_matrix_inverse_hard, lambda_matrix_inverse): New argument
lambda_obstack.
* lambda-code.c (lambda_loop_new): New function.
(lambda_lattice_new, compute_nest_using_fourier_motzkin)
(lambda_compute_auxillary_space, lambda_compute_target_space)
(lambda_loopnest_transform, gcc_loop_to_lambda_loop)
(lambda_loopnest_to_gcc_loopnest): Pass down lambda_obstack.
(build_access_matrix): New argument lambda_obstack. Use obstack
allocation for am.
(lambda_compute_step_signs, lambda_compute_access_matrices): New
argument lambda_obstack. Pass it down.
* ifcvt.c (dead_or_predicable): Use df_simulate_find_defs and
df_simulate_find_noclobber_defs as appropriate. Keep track of an
extra set merge_set_noclobber, and use it to relax the final test
slightly.
* df.h (df_simulate_find_noclobber_defs): Declare.
* df-problems.c (df_simulate_find_defs): Don't ignore partial or
conditional defs.
(df_simulate_find_noclobber_defs): New function.
* include/parallel/partition.h (__parallel_partition):
Improve scalability by:
-introducing new variables __leftold, __rightold, __dist, thus
-getting rid of omp lock by using atomic operations
-getting rid of two omp barriers
* c-tree.h (push_init_level, pop_init_level, set_init_index)
(process_init_element): New argument of type struct obstack *.
* c-typeck.c (push_init_level, pop_init_level, set_designator)
(set_init_index, set_init_label, set_nonincremental_init)
(set_nonincremental_init_from_string, find_init_member)
(output_init_element, output_pending_init_elements)
(process_init_element): New argument braced_init_obstack. Pass it
down.
(push_range_stack, add_pending_init): New argument
braced_init_obstack. Use obstack allocation.
* c-parser.c (c_parser_initelt, c_parser_initval): New argument
braced_init_obstack. Pass it down.
(c_parser_braced_init): New variables ret, braced_init_obstack.
Initialize obstack, pass it down and finally free it.
gcc/
PR middle-end/29274
* optabs.h (expand_widening_mult): Declare.
* tree-pass.h (pass_optimize_widening_mul): Declare.
* tree-ssa-math-opts.c (execute_optimize_widening_mul,
gate_optimize_widening_mul): New static functions.
(pass_optimize_widening_mul): New.
* expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: New
case.
<case MULT_EXPR>: Remove support for widening multiplies.
* tree.def (WIDEN_MULT_EXPR): Tweak comment.
* cfgexpand.c (expand_debug_expr) <case WIDEN_MULT_EXPR>: Use
simplify_gen_unary rather than directly building extensions.
* tree-cfg.c (verify_gimple_assign_binary): Add tests for
WIDEN_MULT_EXPR.
* expmed.c (expand_widening_mult): New function.
* passes.c (init_optimization_passes): Add pass_optimize_widening_mul.
gcc/testsuite/
PR middle-end/29274
* gcc.target/i386/wmul-1.c: New test.
* gcc.target/i386/wmul-2.c: New test.
* gcc.target/bfin/wmul-1.c: New test.
* gcc.target/bfin/wmul-2.c: New test.
PR fortran/43829
* resolve.c (gfc_resolve_index): Wrap around ...
(gfc_resolve_index_1): ... this. Add parameter to allow
any integer kind index type.
(resolve_array_ref): Allow any integer kind for the start
index of an array ref.
* gfortran.dg/vector_subscript_6.f90: New testcase.
* gfortran.dg/assign_10.f90: Adjust.
* lto.c (globalize_cross_file_statics): When function has address taken,
it needs to be public.
* varpool.c (decide_is_variable_needed): Variable is always needed
during ltrans.
* cgraph.c (dump_cgraph_node): Dump also assembler name.
* ipa-inline.c (cgraph_decide_inlining_of_small_functions): Do not ice
at WPA dumping.
(cgraph_decide_inlining): Do not expect callee to be removed in all
cases.
jakub [Wed, 21 Apr 2010 11:57:42 +0000 (11:57 +0000)]
PR middle-end/43570
* omp-low.c (scan_sharing_clauses): Don't scan_omp_op
OMP_CLAUSE_DECL for OMP_CLAUSE_COPYPRIVATE.
(lower_copyprivate_clauses): Use private var in outer
context instead of original var. Make sure the types
are correct for VLAs.
jakub [Wed, 21 Apr 2010 07:05:00 +0000 (07:05 +0000)]
* dwarf2out.c (add_var_loc_to_decl): Add LABEL argument. Drop
last chain entry if it starts with the still current label.
(add_location_or_const_value_attribute): Check that
loc_list->first->next is NULL instead of comparing ->first with
->last.
(dwarf2out_var_location): Pass last_label resp. last_postcall_label
to add_var_loc_to_decl.
jakub [Wed, 21 Apr 2010 06:57:37 +0000 (06:57 +0000)]
* dwarf2out.c (output_call_frame_info): For dw_cie_version
>= 4 add also address size and segment size fields into CIE
header.
* unwind-dw2.c (extract_cie_info): Handle CIE version 4, as
long as address size is the same as sizeof (void *) and
segment size is 0.
* unwind-dw2-fde.c (get_cie_encoding): Likewise. If
address size or segment size is unexpected, return DW_EH_PE_omit.
(classify_object_over_fdes): If get_cie_encoding returned
DW_EH_PE_omit, return -1.
(init_object): If classify_object_over_fdes returned -1,
pretend there were no FDEs at all.
* config/i386/i386.md (bswap<mode>2): Macroize expander from
bswap{si,di}2 using SWI48 mode iterator.
(*bswap<mode>2_movbe): Macroize insn from *bswap{si,di}_movbe using
SWI48 mode iterator. Set type attribute of bswap insn to bitmanip,
set modrm attribute of bswap insn to 0 and remove length attribute.
(*bswap<mode>2_1): Macroize insn from *bswap{si,di}_1 using SWI48 mode
iterator. Set type attribute to bitmanip, set modrm attribute to 0,
set mode attribute to <MODE> and remove length attribute.
gcc/cp/ChangeLog:
PR c++/43800
PR c++/43704
* typeck.c (incompatible_dependent_types_p): If one of the
compared types if not a typedef then honour their main variant
equivalence.
PR fortran/43227
* resolve.c (resolve_fl_derived): If a component character
length has not been resolved, do so now.
(resolve_symbol): The same as above for a symbol character
length.
* trans-decl.c (gfc_create_module_variable): A 'length' decl is
not needed for a character valued, procedure pointer.
PR fortran/43266
* resolve.c (ensure_not_abstract_walker): If 'overriding' is
not found, return FAILURE rather than ICEing.
2010-04-20 Paul Thomas <pault@gcc.gnu.org>
PR fortran/43227
* gfortran.dg/proc_decl_23.f90: New test.
PR fortran/43266
* gfortran.dg/abstract_type_6.f03: New test.
* config/i386/i386.md (ffs<mode>2): Macroize expander from ffs_cmove
and ffsdi2 using SWI48 mode iterator. Expand SImode insn through
ffsi2_no_cmove for !TARGET_CMOVE.
(ffssi2_no_cmove): Rename from *ffs_no_cmove. Make public.
(ffssi2): Remove expander.
(*ffs<mode>_1): Macroize insn from *ffs{si,di} using SWI48
mode iterator.
(ctz<mode>2): Ditto from ctz{si,di}2.
(clz<mode>2): Macroize expander from ctz{hi,si,di}2 using SWI248
mode iterator.
(clz<mode>2_abm): Macroize insn from clz{hi,si,di}2_abm using SWI248
mode iterator.
jakub [Tue, 20 Apr 2010 15:37:51 +0000 (15:37 +0000)]
PR libgomp/43706
* config/linux/affinity.c (gomp_init_affinity): Decrease
gomp_available_cpus if affinity mask confines the process to fewer
CPUs.
* config/linux/proc.c (get_num_procs): If gomp_cpu_affinity is
non-NULL, just return gomp_available_cpus.
* cgraph.c (cgraph_remove_node): Kill bodies in other partitoin.
(dump_cgraph_node): Dump new flags.
* cgraph.h (struct cgraph_node): Add flags reachable_from_other_partition
and in_other_partition.
(cgraph_can_remove_if_no_direct_calls_p): Functions used by other partition
can not be removed.
* cgraphunit.c (cgraph_mark_functions_to_output): Functions used by the other
partition must be output; silence sanity checking on leaking functions
bodies from other paritition.
* lto-cgraph.c (reachable_from_other_partition_p): New function.
(lto_output_node): Output new flags; do not sanity check that inline
clones are output; drop lto_forced_extern_inline_p code; do not mock
visibility flags at partition boundaries.
(add_node_to): New function.
(output_cgraph): Use it to sort functions so masters appear before
clones.
(input_overwrite_node): Input new flags.
* passes.c (ipa_write_summaries): Do not call
lto_new_extern_inline_states.
* lto-section-out.c (forced_extern_inline, lto_new_extern_inline_states,
lto_delete_extern_inline_states, lto_force_functions_extern_inline,
lto_forced_extern_inline_p): Kill.
* lto-streamer.h (lto_new_extern_inline_states,
* lto_delete_extern_inline_states, lto_force_functions_extern_inline,
lto_forced_extern_inline_p): Kill.
* lto.c (lto_add_inline_clones): Do not track inlined_decls.
(lto_add_all_inlinees): Likewise.
(lto_wpa_write_files): Likewise.
* tree-ssa-structalias.c (do_structure_copy): Properly handle
DEREF.
(dump_sa_points_to_info): Remove asserts.
(init_base_vars): nothing_id isn't an escape point nor does it
have pointers.
wilson [Tue, 20 Apr 2010 01:16:59 +0000 (01:16 +0000)]
partial fix, make all 'e' class regs fixed
PR rtl-optimization/43520
* config/ia64/ia64.h (FIXED_REGISTERS, CALL_USED_REGISTERS): Make
ar.lc fixed and call-used.
* opts.c (decode_options): Disable whpr incompatible passes.
* lto/lto.c (lto_1_to_1_map): Skip clones.
(read_cgraph_and_symbols): Do not mark everything as needed.
(do_whole_program_analysis): Do map only after optimizing;
set proper cgraph_state; use passmanager.
gcc/cp/ChangeLog:
PR c++/43704
* typeck.c (structural_comptypes): Test dependent typedefs
incompatibility before testing for their main variant based
equivalence.
gcc/testsuite/ChangeLog:
PR c++/43704
* g++.dg/template/typedef32.C: New test.
PR tree-optimization/37027
* tree-vectorizer.h (struct _loop_vec_info): Add new field reductions
and macro to access it.
(vectorizable_reduction): Add argument.
(vect_get_slp_defs): Likewise.
* tree-vect-loop.c (vect_analyze_scalar_cycles_1): Collect reduction
statements for possible use in SLP.
(new_loop_vec_info): Initialize LOOP_VINFO_REDUCTIONS.
(destroy_loop_vec_info): Free LOOP_VINFO_REDUCTIONS.
(vect_create_epilog_for_reduction): Handle SLP. Modify documentation,
add new argument.
(vectorizable_reduction): Likewise.
* tree-vect-stmts.c (vect_get_vec_defs): Update call to
vect_get_slp_defs.
(vectorizable_type_demotion, vectorizable_type_promotion,
vectorizable_store): Likewise.
(vect_analyze_stmt): Update call to vectorizable_reduction.
(vect_transform_stmt): Likewise.
* tree-vect-slp.c (vect_get_and_check_slp_defs): Handle reduction.
(vect_build_slp_tree): Fix indentation. Check that there are no loads
from different interleaving chains in same node.
(vect_slp_rearrange_stmts): New function.
(vect_supported_load_permutation_p): Allow load permutations for
reductions. Call vect_slp_rearrange_stmts() to rearrange statements
inside SLP nodes if necessary.
(vect_analyze_slp_instance): Handle reductions.
(vect_analyze_slp): Try to build SLP instances originating from groups
of reductions.
(vect_detect_hybrid_slp_stmts): Skip reduction statements.
(vect_get_constant_vectors): Create initial vectors for reductions
according to reduction code. Add new argument.
(vect_get_slp_defs): Add new argument, pass it to
vect_get_constant_vectors.
(vect_schedule_slp_instance): Remove SLP tree root statements.
* fold-const.c (fold_comparison): Use ssizetype.
* gimple-fold.c (maybe_fold_offset_to_array_ref): Likewise.
* ipa-prop.c (ipa_modify_call_arguments): Use sizetype.
* tree-loop-distribution.c (build_size_arg_loc): Likewise.
* tree-object-size.c (compute_object_sizes): Use size_type_node.
* tree.h (initialize_sizetypes): Remove parameter.
(build_common_tree_nodes): Remove second parameter.
* stor-layout.c (initialize_sizetypes): Remove parameter.
Always create an unsigned type.
(set_sizetype): Assert that the passed type is unsigned and simplify.
* tree.c (build_common_tree_nodes): Remove second parameter.
Adjust call to initialize_sizetypes.
* c-decl.c (c_init_decl_processing): Remove second argument in call to
build_common_tree_nodes.
cp/
* decl.c (cxx_init_decl_processing): Remove second argument in call to
build_common_tree_nodes.
java/
* decl.c (java_init_decl_processing): Remove argument in call to
initialize_sizetypes
fortran/
* f95-lang.c (gfc_init_decl_processing): Remove second argument in call
to build_common_tree_nodes.
ada/
* gcc-interface/misc.c (gnat_init): Remove second argument in call to
build_common_tree_nodes.
lto/
* lto-lang.c (lto_init): Remove second argument in call to
build_common_tree_nodes.
* i386.md (UNSPEC_SSE_PROLOGUE_SAVE_LOW): New.
(sse_prologue_save_insn expander): Use new pattern.
(sse_prologue_save_insn1): New pattern and splitter.
(sse_prologue_save_insn): Update to deal also with 64bit aligned
blocks.
* i386.c (setup_incoming_varargs_64): Do not compute jump destination here.
(ix86_gimplify_va_arg): Update alignment needed.
(ix86_local_alignment): Do not align all local arrays
to 128bit.
PR fortran/31538
* fortran/trans-array.c (gfc_conv_ss_startstride): Remove the use of
gfc_msg_bounds by using 'Array bound mismatch' directly.
(gfc_trans_dummy_array_bias): Remove the use of gfc_msg_bounds. Reword
error message to include the mismatch in the extent of array bound.
* fortran/trans.c: Remove gfc_msg_bounds. It is only used in one place.
* fortran/trans.h: Remove extern definition of gfc_msg_bounds.
* uintp.h (UI_Lt): Declare.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Array_Type>: Do the size
computation in sizetype.
<E_Array_Subtype>: Use unified handling for all index types. Do not
generate MAX_EXPR-based expressions, only COND_EXPR-based ones. Add
bypass for PATs.
(annotate_value): Change test for negative values.
(validate_size): Apply test for negative values on GNAT nodes.
(set_rm_size): Likewise.
* gcc-interface/misc.c (gnat_init): Set unsigned types for sizetypes.
* gcc-interface/utils.c (rest_of_record_type_compilation): Change test
for negative values.
(max_size) <MINUS_EXPR>: Do not reassociate a COND_EXPR on the LHS.
(builtin_type_for_size): Adjust definition of signed_size_type_node.
* gcc-interface/utils2.c (compare_arrays): Optimize comparison of
lengths against zero.
* arm-protos.h (tune_params): New structure.
* arm.c (current_tune): New variable.
(arm_constant_limit): Delete.
(struct processors): Add pointer to the tune parameters.
(arm_slowmul_tune): New tuning option.
(arm_fastmul_tune, arm_xscale_tune, arm_9e_tune): Likewise.
(all_cores): Adjust to pick up the tuning model.
(arm_constant_limit): New function.
(arm_override_options): Select the appropriate tuning model. Delete
initialization of arm_const_limit.
(arm_split_constant): Use the new constant-limit model.
(arm_rtx_costs): Pick up the current tuning model.
* arm.md (is_strongarm, is_xscale): Delete.
* arm-generic.md (load_ldsched_x, load_ldsched): Test explicitly
for Xscale variant architectures.
(mult_ldsched_strongarm, mult_ldsched): Similarly for StrongARM.
* gcc-interface/gigi.h (enum standard_datatypes): Add new values
ADT_sbitsize_one_node and ADT_sbitsize_unit_node.
(sbitsize_one_node): New macro.
(sbitsize_unit_node): Likewise.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Array_Subtype>: Fix
latent bug in the computation of subrange_p. Fold wider_p predicate.
(cannot_be_superflat_p): Use an explicitly signed 64-bit type to do
the final comparison.
(make_aligning_type): Build real negation and use sizetype throughout
the offset computation.
(maybe_pad_type): Do not issue the warning when the new size expression
is too complex.
(annotate_value) <INTEGER_CST>: Simplify code handling negative values.
* gcc-interface/misc.c (gnat_init): Initialize sbitsize_one_node and
sbitsize_unit_node.
* gcc-interface/trans.c (Attribute_to_gnu) <Attr_Pool_Address>: Fold
double negation.
(gnat_to_gnu) <N_Free_Statement>: Likewise.
* gcc-interface/utils.c (convert): Use sbitsize_unit_node.
* gcc-interface/utils2.c (compare_arrays): Compute real lengths and use
constants in sizetype. Remove dead code and tweak comments. Generate
equality instead of inequality comparisons for zero length tests.
projects / thirdparty / gcc.git / log
