Jakub Jelinek [Thu, 23 Aug 2007 23:29:57 +0000 (01:29 +0200)]
re PR tree-optimization/32573 (ice for legal code with -O3)
PR tree-optimization/32573
PR middle-end/32946
* tree-data-ref.c (initialize_data_dependence_relation): Clear
DDR_SUBSCRIPTS, DDR_DIR_VECTS and DDR_DIST_VECTS at the beginning.
(finalize_ddr_dependent): Clear DDR_SUBSCRIPTS after freeing it.
(build_classic_dist_vector): Return false rather than true if
DDR_ARE_DEPENDENT is non-NULL at the beginning. Return false
if either subscript_dependence_tester_1 or build_classic_dist_vector_1
returned false. Don't call save_dist_v before calling
build_classic_dist_vector_1.
(free_dependence_relation): Don't guard freeing DDR_SUBSCRIPTS
with NULL DDR_ARE_DEPENDENT. Also free DDR_DIST_VECTS and/or
DDR_DIR_VECTS vectors.
Uros Bizjak [Thu, 23 Aug 2007 14:23:40 +0000 (16:23 +0200)]
re PR target/17390 (missing floating point compare optimization)
PR target/17390
* config/i386/i386.c (ix86_expand_fp_compare): Expand fp comparison to
fake fcomi i387 instruction for !TARGET_CMOVE.
(ix86_expand_branch): Expand natural sequence with one jump for
all targets, not only TARGET_CMOVE.
* config/i386/i386.md (*cmpfp_0_cc): New define_insn_and_split
pattern to implement fake fcomi sequence. Split instruction after
reload to correct compare sequences.
(*cmpfp_xf_cc): Ditto.
(*cmpfp_<mode>_cc): Ditto.
(*cmpfp_u_cc): Ditto.
(*cmpfp_<mode>_cc): Ditto.
testsuite/ChangeLog:
PR target/17390
* gcc.target/i386/pr17390.c: New test.
re PR libfortran/23138 ([mingw32] real(16) values are printed incorrectly)
PR libfortran/23138
* acinclude.m4 (LIBGFOR_CHECK_MINGW_SNPRINTF): New check.
* configure.ac: Use LIBGFOR_CHECK_MINGW_SNPRINTF.
* libgfortran.h: If HAVE_MINGW_SNPRINTF is true, use __mingw_snprintf
instead of snprintf.
* config.h.in: Regenerate.
* configure: Regenerate.
re PR fortran/33020 (Bind(C): c_f_pointer: type/rank mismatch error with integer(8) SHAPE)
2007-08-22 Christopher D. Rickett <crickett@lanl.gov>
PR fortran/33020
* resolve.c (gfc_iso_c_sub_interface): Remove setting of type and
kind for optional SHAPE parameter of C_F_POINTER.
2007-08-22 Christopher D. Rickett <crickett@lanl.gov>
PR fortran/33020
* gfortran.dg/c_f_pointer_shape_tests_2.f03: Update test to
include multiple kinds for SHAPE parameter within a single
namespace.
* gfortran.dg/c_f_pointer_shape_tests_2_driver.c: Ditto.
* gfortran.dg/c_f_pointer_shape_tests_3.f03: New test case.
Janus Weil [Wed, 22 Aug 2007 21:11:13 +0000 (21:11 +0000)]
decl.c (match_attr_spec): Pass on errors from gfc_match_bind_c.
2007-08-22 Janus Weil <jaydub66@gmail.com>
* decl.c (match_attr_spec): Pass on errors from gfc_match_bind_c.
(gfc_match_bind_c): Bugfix in check for NAME= with abstract interfaces.
(gfc_match_mopdproc): Bugfix to reject module procedures in
abstract interfaces.
2007-08-22 Janus Weil <jaydub66@gmail.com>
* interface_abstract_1.f90: Extended test case.
* interface_abstract_3.f90: New test.
Jason Merrill [Wed, 22 Aug 2007 17:23:37 +0000 (13:23 -0400)]
re PR c++/29365 (Unnecessary anonymous namespace warnings)
PR c++/29365
* pt.c (outermost_tinst_level): New function.
* lex.c (in_main_input_context): New function.
* decl2.c (constrain_class_visibility): Use it to avoid warning
about uses of the anonymous namespace in the main input file.
Maxim Kuvyrkov [Wed, 22 Aug 2007 15:07:10 +0000 (15:07 +0000)]
target.h (struct gcc_target.sched: dfa_pre_advance_cycle, [...]): New scheduler hooks.
* target.h (struct gcc_target.sched: dfa_pre_advance_cycle,
dfa_post_advance_cycle): New scheduler hooks.
* target-def.h (TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE,
TARGET_SCHED_DFA_POST_ADVANCE_CYCLE): New macros to initialize
new hooks.
(TARGET_SCHED): Use them.
* doc/tm.texi (TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE,
TARGET_SCHED_DFA_POST_ADVANCE_CYCLE): Document new hooks.
* haifa-sched.c (advance_one_cycle): Invoke new hooks.
* genautomata.c (insn_has_dfa_reservation_p): New DFA interface
function to facilitate debugging.
(INSN_HAS_DFA_RESERVATION_P_FUNC_NAME): New macro.
(output_insn_has_dfa_reservation_p): New static function to output
insn_has_dfa_reservation_p ().
(write_automata): Use it.
* genattr.c (main): Output declaration for
insn_has_dfa_reservation_p ().
Paul Brook [Tue, 21 Aug 2007 15:35:30 +0000 (15:35 +0000)]
configure.ac: Add --with-pkgversion and --with-bugurl.
gcc:
2007-08-21 Paul Brook <paul@codesourcery.com>
Nathan Sidwell <nathan@codesourcery.com>
Mark Mitchell <mark@codesourcery.com>
Joseph Myers <joseph@codesourcery.com>
* configure.ac: Add --with-pkgversion and --with-bugurl.
* configure: Regenerate.
* doc/install.texi: Document them.
* version.c (version_string): Remove VERSUFFIX.
(VERSUFFIX): Remove.
(pkgversion_string): New.
(bug_report_url): Do not hard-code initializer.
* version.h (pkgversion_string): Declare.
* Makefile.in (PKGVERSION_s, BUGURL_s, PKGVERSION, BUGURL_TEXI):
Define.
(version.o): Define PKGVERSION and BUGURL.
(gcc-vers.texi): Define VERSION_PACKAGE and BUGURL.
(%.pod): Define BUGURL.
* gcc.c (process_command, main): Use pkgversion_string.
* toplev.c (compile_file, print_version): Likewise.
* protoize.c (main): Likewise.
* gcov.c (print_version): Likewise. Update copyright date.
* gcov-dump.c (print_version): Likewise. Update copyright date.
* mips-tdump.c (main): Likewise. Update copyright date.
* mips-tfile.c (main): Likewise. Update copyright date.
* doc/include/gcc-common.texi: Include VERSION_PACKAGE as subtitle.
* doc/bugreport.texi: Use BUGURL for bug-reporting instructions;
shorten description.
* doc/gcc.texi: Include VERSION_PACKAGE in version description.
* doc/gccint.texi: Likewise.
* doc/invoke.texi: Use BUGURL for bug-reporting instructions.
Update copyright date.
gcc/fortran:
2007-08-21 Paul Brook <paul@codesourcery.com>
Nathan Sidwell <nathan@codesourcery.com>
Mark Mitchell <mark@codesourcery.com>
Joseph Myers <joseph@codesourcery.com>
* gfortranspec.c (lang_specific_driver): Use pkgversion_string.
* Make-lang.in (gfortran.pod): Define BUGURL.
* invoke.texi: Use BUGURL for bug-reporting instructions.
gcc/java:
2007-08-21 Paul Brook <paul@codesourcery.com>
Nathan Sidwell <nathan@codesourcery.com>
Mark Mitchell <mark@codesourcery.com>
Joseph Myers <joseph@codesourcery.com>
* jcf-dump.c (version): Use pkgversion_string. Update copyright
date.
gcc/treelang:
2007-08-21 Paul Brook <paul@codesourcery.com>
Nathan Sidwell <nathan@codesourcery.com>
Mark Mitchell <mark@codesourcery.com>
Joseph Myers <joseph@codesourcery.com>
* tree1.c (treelang_handle_option): Use pkgversion_string.
* treelang.texi (which-treelang): Define to use VERSION_PACKAGE.
maintainer-scripts:
2007-08-21 Paul Brook <paul@codesourcery.com>
Nathan Sidwell <nathan@codesourcery.com>
Mark Mitchell <mark@codesourcery.com>
Joseph Myers <joseph@codesourcery.com>
* gcc_release: Do not include copies of bugs.html in releases.
* update_web_docs_svn: Set VERSION_PACKAGE and BUGURL in
gcc-vers.texi.
Co-Authored-By: Joseph Myers <joseph@codesourcery.com> Co-Authored-By: Mark Mitchell <mark@codesourcery.com> Co-Authored-By: Nathan Sidwell <nathan@codesourcery.com>
From-SVN: r127667
Uros Bizjak [Tue, 21 Aug 2007 12:26:12 +0000 (14:26 +0200)]
i386.c (ix86_expand_clear): Use FLAGS_REG.
* config/i386/i386.c (ix86_expand_clear): Use FLAGS_REG.
(ix86_expand_strlensi_unroll_1): Ditto.
(ix86_expand_branch): Use FLAGS_REG and FPSR_REG.
(ix86_expand_carry_flag_compare): Update comment.
Jakub Jelinek [Tue, 21 Aug 2007 09:22:14 +0000 (11:22 +0200)]
re PR middle-end/32912 (ICE with vector code)
PR middle-end/32912
* fold-const.c (fold_unary): Optimize BIT_NOT_EXPR of VECTOR_CST.
(fold_binary): Handle vectors in X | ~X and X ^ ~X optimizations.
* gcc.dg/pr32912-1.c: New test.
* gcc.dg/pr32912-2.c: New test.
* gcc.dg/pr32912-3.c: New test.
Chao-ying Fu [Mon, 20 Aug 2007 22:26:47 +0000 (22:26 +0000)]
c-common.h (enum rid): Add new enumeration values of RID_SAT, RID_FRACT, and RID_ACCUM.
* c-common.h (enum rid): Add new enumeration values of RID_SAT,
RID_FRACT, and RID_ACCUM. RID_SAT needs to be inserted before
RID_ONEWAY, so that it can be checked in declspecs_add_type.
(c_common_fixed_point_type_for_size): Declare.
* c-parser.c (reswords): Add _Fract, _Accum, and _Sat.
(c_token_starts_typename): Handle RID_FRACT, RID_ACCUM, and RID_SAT.
(c_token_starts_declspecs): Likewise.
(c_parser_declspecs): Likewise.
(c_parser_attributes): Likewise.
* c-tree.h (enum c_typespec_keyword): Add cts_fract and cts_accum.
(c_declspecs): Add saturating_p.
* c-decl.c (build_null_declspecs): Initialize saturating_p.
(declspecs_add_type): Avoid using complex with _Fract, _Accum, or _Sat.
Handle RID_SAT.
Avoid using void, bool, char, int, float, double, _Decimal32,
_Decimal64, _Decimal128, and complex with _Sat.
Handle RID_FRACT and RID_ACCUM.
Make sure _Sat is used with _Fract or _Accum.
(finish_declspecs): Handle cts_fract and cts_accum.
* c-common.c (fixed-value.h): New include.
(constant_expression_warning): Handle FIXED_CST.
(overflow_warning): Likewise.
(warnings_for_convert_and_check): Likewise.
(c_common_fixed_point_type_for_size): New.
(c_common_type_for_mode): Handle fixed-point modes to
return various saturating/non-saturating, signed/unsigned types.
(c_common_signed_or_unsigned_type): Support fixed-point types.
(shorten_compare): Check fixed-point zero.
Handle FIXED_POINT_TYPE.
(c_common_truthvalue_conversion): Handle FIXED_CST.
Handle FIXED_POINT_TYPE.
(c_common_nodes_and_builtins): Record builtin types for fixed-point
types.
(handle_mode_attribute): Handle fixed-point modes. Need to check
if the signness of base type and fixed-point modes are consistent.
(handle_vector_size_attribute): Handle fixed-point modes.
(same_scalar_type_ignoring_signedness): Handle FIXED_POINT_TYPE.
(warn_for_div_by_zero): Check fixed-point zero.
* c-typeck.c (c_common_type): Check FIXED_POINT_TYPE. Build
a common fixed-point type based on fbit, ibit, sign, and saturation.
(build_unary_op): Allow FIXED_POINT_TYPE for CONVERT_EXPR,
NEGATE_EXPR, TRUTH_NOT_EXPR, PREINCREMENT_EXPR, POSTINCREMENT_EXPR,
PREDECREMENT_EXPR, and POSTDECREMENT_EXPR.
(convert_for_assignment): Support FIXED_POINT_TYPE.
(digest_init): Handle FIXED_POINT_TYPE.
(build_binary_op): Support FIXED_POINT_TYPE in *_DIV_EXPR,
TRUTH_ANDIF_EXPR, TRUTH_ORIF_EXPR, TRUTH_AND_EXPR, TRUTH_OR_EXPR,
TRUTH_XOR_EXPR, RSHIFT_EXPR, LSHIFT_EXPR, EQ_EXPR, NE_EXPR, LE_EXPR,
GE_EXPR, LT_EXPR, GT_EXPR.
* target-def.h (TARGET_FIXED_POINT_SUPPORTED_P): New.
(TARGET_INITIALIZER): Add TARGET_FIXED_POINT_SUPPORTED_P.
* target.h (gcc_target): Add fixed_point_supported_p.
* targhooks.c (default_scalar_mode_supported_p): Handle MODE_FRACT,
MODE_UFRACT, MODE_ACCUM, and MODE_UACCUM.
(default_fixed_point_supported_p): Define.
* targhooks.h (default_fixed_point_supported_p): Declare.
* doc/tm.texi (TARGET_FIXED_POINT_SUPPORTED_P): Add.
* doc/install.texi (Configuration): Add --enable-fixed-point.
* configure.ac (--enable-fixed-point): New to enable fixed-point
arithmetic extension to C. For mips targets, we enable it by default.
* configure, config.in: Regenerate.
Richard Guenther [Mon, 20 Aug 2007 12:31:44 +0000 (12:31 +0000)]
re PR c++/22369 (C++ produces mis-matched types with pointers to member functions)
2007-08-20 Richard Guenther <rguenther@suse.de>
PR c++/22369
PR c++/22451
* call.c (build_new_method_call): Convert initializer to
the basetype.
* init.c (build_aggr_init): Do not fiddle with types.
(build_vec_delete_1): Use correct type for POINTER_PLUS_EXPR.
* except.c (build_throw): Do not drop qualifiers for the
pointer type.
* typeck.c (get_member_function_from_ptrfunc): Do not
fiddle with types, instead convert.
(build_ptrmemfunc1): Convert to the target type for
initialization.
(gfc_trans_allocate): Convert result to target type.
* cp-objcp-common.c (cxx_get_alias_set): Pointers to
pointer-to-member structures shall have alias set zero as well.
Richard Guenther [Mon, 20 Aug 2007 11:19:22 +0000 (11:19 +0000)]
tree.c (WALK_SUBTREE): Call walk_tree_1.
2007-08-20 Richard Guenther <rguenther@suse.de>
* tree.c (WALK_SUBTREE): Call walk_tree_1.
(walk_type_fields): Take lh parameter.
(walk_tree): Rename to ...
(walk_tree_1): ... this. Do not call the walk_subtrees
langhook but the now passed callback. Pass lh on recursion.
(walk_tree_without_duplicates): Rename to ...
(walk_tree_without_duplicates_1): ... this. Take lh parameter
and call walk_tree_1.
* tree.h (walk_tree_lh): New typedef.
(walk_tree_1): Declare.
(walk_tree_without_duplicates_1): Likewise.
(walk_tree): New define to walk_tree_1 with NULL lh parameter.
(walk_tree_without_duplicates): New define to
walk_tree_without_duplicates_1 with NULL lh parameter.
* langhooks.c (lhd_tree_inlining_walk_subtrees): Remove.
* langhooks.h (lang_hooks_for_tree_inlining): Remove walk_subtrees
langhook.
* langhooks-def.h (lhd_tree_inlining_walk_subtrees): Remove.
(LANG_HOOKS_TREE_INLINING_WALK_SUBTREES): Likewise.
(LANG_HOOKS_TREE_INLINING_INITIALIZER): Remove walk_subtrees
initializer.
Andrew Pinski [Mon, 20 Aug 2007 07:42:55 +0000 (07:42 +0000)]
re PR tree-optimization/30564 (ice for legal code with -O3)
2007-08-20 Andrew Pinski <andrew_pinski@playstation.sony.com>
PR middle-end/30564
* tree-inline.c (optimize_inline_calls): Move the cgraph checking
code in front of the compacting of basic blocks.
Move the folding of statements inbetween the cgraph checking
and compacting of basic blocks.
2007-08-20 Andrew Pinski <andrew_pinski@playstation.sony.com>
PR middle-end/30564
* gcc.c-torture/compile/pr30564-1.c: New test.
* gcc.c-torture/compile/pr30564-2.c: New test.
Andrew Pinski [Mon, 20 Aug 2007 06:53:05 +0000 (06:53 +0000)]
re PR target/32522 (Bootstrap failure on Alpha due to pointer-plus changes)
2007-08-19 Andrew Pinski <andrew_pinski@playstation.sony.com>
Serge Belyshev <belyshev@depni.sinp.msu.ru>
PR target/32522
* config/alpha/alpha.c (va_list_skip_additions): Check for
POINTER_PLUS_EXPR in addition to PLUS_EXPR.
(alpha_stdarg_optimize_hook): Look for POINTER_PLUS_EXPR instead of
PLUS_EXPR when checking ap.__base.
(alpha_va_start): Create POINTER_PLUS_EXPR instead of PLUS_EXPR
when doing addition on pointer types. Use size_int instead of
build_int_cst.
(alpha_gimplify_va_arg_1): Likewise, but use sizetype instead of
ptr_type in the second operand.
Roger Sayle [Mon, 20 Aug 2007 06:11:15 +0000 (06:11 +0000)]
match.c (intrinsic_operators): Make static.
* match.c (intrinsic_operators): Make static.
(gfc_op2string): New function for converting a gfc_intrinsic_op to
to a "const char*", replacing the macro of the same name.
* gfortran.h (intrinsic_operators): Delete prototype.
(gfc_op2string): Replace macro with function prototype.
* tree-flow.h (struct stmt_ann_d): Remove makes_clobbering_call.
* tree-ssa-alias.c (init_transitive_clobber_worklist): Add
on_worklist argument and avoid adding things to worklist multiple
times.
(add_to_worklist): Ditto.
(mark_aliases_call_clobbered): Mark entire structure clobbered if
single SFT is clobbered.
(set_initial_properties): Ditto.
(compute_call_clobbered): Update for changes to function
arguments.
(create_overlap_variables_for): Always create SFT for offset 0.
(create_structure_vars): Handle PHI's, since we are in SSA form at
this point.
* tree-ssa-loop-ivopts.c (get_ref_tag): Don't return subvars.
* tree-ssa-operands.c (access_can_touch_variable): Don't handle
TARGET_MEM_REF.
(add_vars_for_offset): Figure out aliases from access + points-to.
(add_virtual_operand): Use add_vars-for_offset.
(get_tmr_operands): Update for NMT changes, rewrite to be correct.
(add_call_clobber_ops): Remove makes_clobbering_call set.
(get_expr_operands): Always pass through the INDIRECT_REF
reference.
* tree-ssa-structalias.c (struct constraint_graph): Remove
variables member.
Add pe, pe_rep, pointer_label, loc_label, pointed_by, points_to,
address_taken, pt_used, number_incoming.
(FIRST_ADDR_NODE): Removed.
(merge_graph_nodes): Remove broken code for the moment.
(init_graph): New function.
(build_pred_graph): Remove code to init_graph.
Add location equivalence support.
(struct scc_info): Rename roots to deleted.
(scc_visit): Ditto.
(init_scc_info): Ditto
(init_topo_info): Use graph->size.
(compute_topo_order): Ditto.
(do_da_constraint): Removed.
(do_sd_constraint): Remove calls to find().
set_union_with_increment should always get 0 as last arg here.
(do_complex_constraint): Replace do_da_constraint with assert.
Stop calling find.
(struct equiv_class_label): New.
(pointer_equiv_class_table): Ditto.
(location_equiv_class_table): Ditto.
(equiv_class_label_hash): Ditto.
(equiv_class_label_eq): Ditto
(equiv_class_lookup): Ditto.
(equiv_class_ladd): Ditto.
(pointer_equiv_class): Ditto.
(location_equiv_class): Ditto.
(condense_visit): Rename and rewrite from label_visit to do only
SCC related stuff for HU.
(label_visit): Do HU work for HU.
(perform_var_substitution): Update to do HU and location
equivalence.
(free_var_substitution_info): Update to free HU and location
equivalence structures. */
(find_equivalent_node): Update for pointer but not location
equivalence.
(unite_pointer_equivalences): New function.
(move_complex_constraints): Rewrite to only do moving.
(rewrite_constraints): Split out of move_complex_constraints.
(solve_graph): Use graph->size.
(process_constraint_1): Add from_call argument, use it.
Split *a = &b into two constraints.
(process_constraint): Use new process_constraint_1.
(get_constraint_for_component_ref): Handle bitmaxsize == -1 case.
(get_constraint_for): Handle non-pointer integers properly.
Remove code that used to handle structures.
(handle_ptr_arith): Fix a few bugs in pointer arithmetic handling
with unknown addends.
(handle_rhs_call): New function.
(find_func_aliases): Use handle_rhs_call.
(set_uids_in_ptset): Add an assert.
(set_used_smts): Fix bug in not considering unified vars.
(compute_tbaa_pruning): Stop initing useless iteration_obstack.
(compute_points_to_sets): Update for other function changes.
(delete_points_to_sets): Ditto.
(ipa_pta_execute): Ditto.
(pass_ipa_pta): We need to update SSA after ipa_pta.
Joseph Myers [Sun, 19 Aug 2007 23:07:15 +0000 (00:07 +0100)]
dwarf2out.c (text_section_used, [...]): New.
* dwarf2out.c (text_section_used, cold_text_section_used,
cold_text_section, dwarf2out_note_section_used): New.
(dwarf2out_init): Initialize cold_text_section.
(dwarf2out_switch_text_section, dwarf2out_begin_function): Call
dwarf2out_note_section_used.
(size_of_aranges): Only count entry for text section if it was
used. Count entry for cold text section if it was used.
(output_aranges): Only output entries for text section and cold
text section if they were used.
testsuite:
* gcc.dg/debug/dwarf2/aranges-fnsec-1.c: New test.
Tobias Burnus [Sun, 19 Aug 2007 20:08:14 +0000 (22:08 +0200)]
[multiple changes]
2007-08-18 Tobias Burnus <burnus@net-b.de>
* gfortran.h (gfc_is_intrinsic_typename): Add declaration.
* symbol.c (gfc_is_intrinsic_typename): New function.
* parse.c (decode_statement): Check for space in ABSTRACT INTERFACE.
(parse_interface): Use gfc_is_intrinsic_typename.
* decl.c (gfc_match_derived_decl): Ditto.
* module.c (gfc_match_use): Use gcc_unreachable() for
INTERFACE_ABSTRACT in switch().
* tree-vectorizer.h (dr_alignment_support): Renamed
dr_unaligned_software_pipeline to dr_explicit_realign_optimized.
Added a new value dr_explicit_realign.
(_stmt_vec_info): Added new fields: dr_base_address, dr_init,
dr_offset, dr_step, and dr_aligned_to, along with new access
functions for these fields: STMT_VINFO_DR_BASE_ADDRESS,
STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET, STMT_VINFO_DR_STEP, and
STMT_VINFO_DR_ALIGNED_TO.
* tree-vectorizer.c (vect_supportable_dr_alignment): Add
documentation.
In case of outer-loop vectorization with non-fixed misalignment - use
the dr_explicit_realign scheme instead of the optimized realignment
scheme.
(new_stmt_vec_info): Initialize new fields.
* tree-vect-analyze.c (vect_compute_data_ref_alignment): Handle the
'nested_in_vect_loop' case. Change verbosity level.
(vect_analyze_data_ref_access): Handle the 'nested_in_vect_loop' case.
Don't fail on zero step in the outer-loop for loads.
(vect_analyze_data_refs): Call split_constant_offset to calculate base,
offset and init relative to the outer-loop.
* tree-vect-transform.c (vect_create_data_ref_ptr): Replace the unused
BSI function argument with a new function argument - at_loop.
Simplify the condition that determines STEP. Takes additional argument
INV_P. Support outer-loop vectorization (handle the nested_in_vect_loop
case), including zero step in the outer-loop. Call
vect_create_addr_base_for_vector_ref with additional argument.
(vect_create_addr_base_for_vector_ref): Takes additional argument LOOP.
Updated function documentation. Handle the 'nested_in_vect_loop' case.
Fixed and simplified calculation of step.
(vectorizable_store): Call vect_create_data_ref_ptr with loop instead
of bsi, and with additional argument. Call bump_vector_ptr with
additional argument. Fix typos. Handle the 'nested_in_vect_loop' case.
(vect_setup_realignment): Takes additional arguments INIT_ADDR and
DR_ALIGNMENT_SUPPORT. Returns another value AT_LOOP. Handle the case
when the realignment setup needs to take place inside the loop. Support
the dr_explicit_realign scheme. Allow generating the optimized
realignment scheme for outer-loop vectorization. Added documentation.
(vectorizable_load): Support the dr_explicit_realign scheme. Handle the
'nested_in_vect_loop' case, including loads that are invariant in the
outer-loop and the realignment schemes. Handle the case when the
realignment setup needs to take place inside the loop. Call
vect_setup_realignment with additional arguments. Call
vect_create_data_ref_ptr with additional argument and with loop instead
of bsi. Fix 80-column overflow. Fix typos. Rename PHI_STMT to PHI.
(vect_gen_niters_for_prolog_loop): Call
vect_create_addr_base_for_vector_ref with additional arguments.
(vect_create_cond_for_align_checks): Likewise.
(bump_vector_ptr): Updated to support the new dr_explicit_realign
scheme: takes additional argument bump; argument ptr_incr is now
optional; updated documentation.
(vect_init_vector): Takes additional argument (bsi). Use it, if
available, to insert the vector initialization.
(get_initial_def_for_induction): Pass additional argument in call to
vect_init_vector.
(vect_get_vec_def_for_operand): Likewise.
(vect_setup_realignment): Likewise.
(vectorizable_load): Likewise.
Dorit Nuzman [Sun, 19 Aug 2007 09:39:50 +0000 (09:39 +0000)]
tree-vectorizer.h (vect_is_simple_reduction): Takes a loop_vec_info as argument instead of struct loop.
* tree-vectorizer.h (vect_is_simple_reduction): Takes a loop_vec_info
as argument instead of struct loop.
(nested_in_vect_loop_p): New function.
(vect_relevant): Add enum values vect_used_in_outer_by_reduction and
vect_used_in_outer.
(is_loop_header_bb_p): New. Used to differentiate loop-header phis
from other phis in the loop.
(destroy_loop_vec_info): Add additional argument to declaration.
* tree-vectorizer.c (supportable_widening_operation): Also check if
nested_in_vect_loop_p (don't allow changing the order in this case).
(vect_is_simple_reduction): Takes a loop_vec_info as argument instead
of struct loop. Call nested_in_vect_loop_p and don't require
flag_unsafe_math_optimizations if it returns true.
(new_stmt_vec_info): When setting def_type for phis differentiate
loop-header phis from other phis.
(bb_in_loop_p): New function.
(new_loop_vec_info): Inner-loop phis already have a stmt_vinfo, so just
update their loop_vinfo. Order of BB traversal now matters - call
dfs_enumerate_from with bb_in_loop_p.
(destroy_loop_vec_info): Takes additional argument to control whether
stmt_vinfo of the loop stmts should be destroyed as well.
(vect_is_simple_reduction): Allow the "non-reduction" use of a
reduction stmt to be defines by a non loop-header phi.
(vectorize_loops): Call destroy_loop_vec_info with additional argument.
* tree-vect-transform.c (vectorizable_reduction): Call
nested_in_vect_loop_p. Check for multitypes in the inner-loop.
(vectorizable_call): Likewise.
(vectorizable_conversion): Likewise.
(vectorizable_operation): Likewise.
(vectorizable_type_promotion): Likewise.
(vectorizable_type_demotion): Likewise.
(vectorizable_store): Likewise.
(vectorizable_live_operation): Likewise.
(vectorizable_reduction): Likewise. Also pass loop_info to
vect_is_simple_reduction instead of loop.
(vect_init_vector): Call nested_in_vect_loop_p.
(get_initial_def_for_reduction): Likewise.
(vect_create_epilog_for_reduction): Likewise.
(vect_init_vector): Check which loop to work with, in case there's an
inner-loop.
(get_initial_def_for_inducion): Extend to handle outer-loop
vectorization. Fix indentation.
(vect_get_vec_def_for_operand): Support phis in the case vect_loop_def.
In the case vect_induction_def get the vector def from the induction
phi node, instead of calling get_initial_def_for_inducion.
(get_initial_def_for_reduction): Extend to handle outer-loop
vectorization.
(vect_create_epilog_for_reduction): Extend to handle outer-loop
vectorization.
(vect_transform_loop): Change assert to just skip this case. Add a
dump printout.
(vect_finish_stmt_generation): Add a couple asserts.
(vect_estimate_min_profitable_iters): Multiply
cost of inner-loop stmts (in outer-loop vectorization) by estimated
inner-loop bound.
(vect_model_reduction_cost): Don't add reduction epilogue cost in case
this is an inner-loop reduction in outer-loop vectorization.
* tree-vect-analyze.c (vect_analyze_scalar_cycles_1): New function.
Same code as what used to be vect_analyze_scalar_cycles, only with
additional argument loop, and loop_info passed to
vect_is_simple_reduction instead of loop.
(vect_analyze_scalar_cycles): Code factored out into
vect_analyze_scalar_cycles_1. Call it for each relevant loop-nest.
Updated documentation.
(analyze_operations): Check for inner-loop loop-closed exit-phis during
outer-loop vectorization that are live or not used in the outerloop,
cause this requires special handling.
(vect_enhance_data_refs_alignment): Don't consider versioning for
nested-loops.
(vect_analyze_data_refs): Check that there are no datarefs in the
inner-loop.
(vect_mark_stmts_to_be_vectorized): Also consider vect_used_in_outer
and vect_used_in_outer_by_reduction cases.
(process_use): Also consider the case of outer-loop stmt defining an
inner-loop stmt and vice versa.
(vect_analyze_loop_1): New function.
(vect_analyze_loop_form): Extend, to allow a restricted form of nested
loops. Call vect_analyze_loop_1.
(vect_analyze_loop): Skip (inner-)loops within outer-loops that have
been vectorized. Call destroy_loop_vec_info with additional argument.
* tree-vect-patterns.c (vect_recog_widen_sum_pattern): Don't allow
in the inner-loop when doing outer-loop vectorization. Add
documentation and printout.
(vect_recog_dot_prod_pattern): Likewise. Also add check for
GIMPLE_MODIFY_STMT (in case we encounter a phi in the loop).
Roger Sayle [Sun, 19 Aug 2007 01:52:23 +0000 (01:52 +0000)]
primary.c (match_logical_constant_string): New function to match a ".true." or a ".false.".
* primary.c (match_logical_constant_string): New function to match
a ".true." or a ".false.".
(match_logical_constant): Use it instead of gfc_match_strings.