Graham Markall [Wed, 31 May 2017 13:15:33 +0000 (13:15 +0000)]
[ARC] Recognise add_n and sub_n in combine again
Since the combine pass canonicalises shift-add insns using plus and
ashift (as opposed to plus and mult which it previously used to do), it
no longer creates *add_n or *sub_n insns, as the patterns match plus and
mult only. The outcome of this is that some opportunities to generate
add{1,2,3} and sub{1,2,3} instructions are missed.
This change adds additional *add_n and *sub_n insns that match the
plus-ashift pattern. The original *add_n and *sub_n insns are still left
in, as they are sometimes generated later on by constant propagation.
The idea of adding these insns is modelled on the changes in:
which addresses a similar issue for the PA target.
For the small test cases that are added, even if the combine pass misses
the opportunity to generate addN or subN, constant propagation manages
to do so, so the rtl of the combine pass is checked.
gcc/ChangeLog:
* config/arc/arc.c (arc_print_operand): Handle constant operands.
(arc_rtx_costs): Add costs for new patterns.
* config/arc/arc.md: Additional *add_n and *sub_n patterns.
* config/arc/predicates.md: Add _1_2_3_operand predicate.
gcc/testsuite/ChangeLog:
* gcc.target/arc/add_n-combine.c: New test.
* gcc.target/arc/sub_n-combine.c: New test.
This patch just adds a helper function for getting the next strinfo
in a chain, since part 2 adds another place where we do that.
2017-05-16 Richard Sandiford <richard.sandiford@linaro.org>
gcc/
* tree-ssa-strlen.c (get_next_strinfo): New function.
(get_stridx_plus_constant): Use it.
(zero_length_string): Likewise.
(adjust_related_strinfos): Likewise.
(adjust_last_stmt): Likewise.
Alternative check for vector refs with same alignment
vect_find_same_alignment_drs uses the ddr dependence distance
to tell whether two references have the same alignment. Although
that's safe with the current code, there's no particular reason
why a dependence distance of 0 should mean that the accesses start
on the same byte. E.g. a reference to a full complex value could
in principle depend on a reference to the imaginary component.
A later patch adds support for this kind of dependence.
On the other side, checking modulo vf is pessimistic when the step
divided by the element size is a factor of 2.
This patch instead looks for cases in which the drs have the same
base, offset and step, and for which the difference in their constant
initial values is a multiple of the alignment.
2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
gcc/
* tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
loop_vinfo argument and use of dependence distance vectors.
Check instead whether the two references differ only in their
initial value and assume that they have the same alignment if the
difference is a multiple of the vector alignment.
(vect_analyze_data_refs_alignment): Update call accordingly.
gcc/testsuite/
* gcc.dg/vect/vect-103.c: Update wording of dump message.
Tom de Vries [Tue, 30 May 2017 22:00:57 +0000 (22:00 +0000)]
Test if host compiler supports -std=c++11 in ms-sysv.exp
2017-05-30 Tom de Vries <tom@codesourcery.com>
PR testsuite/80910
* gcc.target/x86_64/abi/ms-sysv/ms-sysv.exp: Exit with status
unsupported if host compiler does not support c++11.
(host_supports_c++11): New proc.
David Malcolm [Tue, 30 May 2017 20:38:14 +0000 (20:38 +0000)]
C++ template type diff printing
gcc/ChangeLog:
* diagnostic-color.c (color_dict): Add "type-diff".
(parse_gcc_colors): Update comment.
* doc/invoke.texi (Diagnostic Message Formatting Options): Add
-fdiagnostics-show-template-tree and -fno-elide-type.
(GCC_COLORS): Add type-diff to example.
(type-diff=): New.
(-fdiagnostics-show-template-tree): New.
(-fno-elide-type): New.
* pretty-print.c (pp_format): Pass quote and formatters[argno] to
the pp_format_decoder callback. Call any m_format_postprocessor's
"handle" method.
(pretty_printer::pretty_printer): Initialize
m_format_postprocessor.
(pretty_printer::~pretty_printer): Delete any
m_format_postprocessor.
* pretty-print.h (printer_fn): Add bool and const char **
parameters.
(class format_postprocessor): New class.
(struct pretty_printer::format_decoder): Document the new
parameters.
(struct pretty_printer::m_format_postprocessor): New field.
* tree-diagnostic.c (default_tree_printer): Update for new
bool and const char ** params.
* tree-diagnostic.h (default_tree_printer): Likewise.
gcc/c/ChangeLog:
* c-objc-common.c (c_tree_printer): Gain bool and const char **
parameters.
gcc/c-family/ChangeLog:
* c-format.c (gcc_cxxdiag_char_table): Add 'H' and 'I' to
format_chars.
* c.opt (fdiagnostics-show-template-tree): New option.
(felide-type): New option.
gcc/cp/ChangeLog:
* call.c (perform_implicit_conversion_flags): Convert
"from %qT to %qT" to "from %qH to %qI" in diagnostic.
(print_conversion_rejection): Replace pairs of %qT with
%qH and %qI in various places.
(build_user_type_conversion_1): Likewise.
(build_integral_nontype_arg_conv): Likewise.
(build_conditional_expr_1): Likewise.
(convert_like_real): Likewise.
(convert_arg_to_ellipsis): Likewise.
(joust): Likewise.
(initialize_reference): Likewise.
* cvt.c (cp_convert_to_pointer): Likewise.
(cp_convert_to_pointer): Likewise.
(convert_to_reference): Likewise.
(ocp_convert): Likewise.
* error.c (cp_printer): Gain bool and const char ** parameters.
(struct deferred_printed_type): New struct.
(class cxx_format_postprocessor): New class.
(cxx_initialize_diagnostics): Wire up a cxx_format_postprocessor
to pp->m_format_postprocessor.
(comparable_template_types_p): New function.
(newline_and_indent): New function.
(arg_to_string): New function.
(print_nonequal_arg): New function.
(print_template_differences): New function.
(type_to_string_with_compare): New function.
(print_template_tree_comparison): New function.
(append_formatted_chunk): New function.
(add_quotes): New function.
(cxx_format_postprocessor::handle): New function.
(defer_phase_2_of_type_diff): New function.
(cp_printer): Add "quoted" and "buffer_ptr" params. Implement
%H and %I.
* typeck.c (cp_build_binary_op): Replace pairs of %qT with
%qH and %qI in various places.
(convert_member_func_to_ptr): Likewise.
(build_reinterpret_cast_1): Likewise.
(convert_for_assignment): Likewise.
* typeck2.c (check_narrowing): Likewise.
gcc/fortran/ChangeLog:
* error.c (gfc_format_decoder): Update for new bool and
const char ** params.
gcc/testsuite/ChangeLog:
* g++.dg/plugin/plugin.exp (plugin_test_list): Add...
* g++.dg/plugin/show-template-tree-color-no-elide-type.C: New
test case.
* g++.dg/plugin/show-template-tree-color.C: New test case.
* g++.dg/plugin/show_template_tree_color_plugin.c: New plugin.
* g++.dg/template/show-template-tree-2.C: New test case.
* g++.dg/template/show-template-tree-3.C: New test case.
* g++.dg/template/show-template-tree-4.C: New test case.
* g++.dg/template/show-template-tree-no-elide-type.C: New test case.
* g++.dg/template/show-template-tree.C: New test case.
-mno-gen-cell-microcode is a pain to handle correctly: it causes
different code generation for some very basic patterns, even patterns
specific to that option. It also requires marking up many patterns,
which is a pain when adding new patterns or modifying existing ones
(first non-trivial step is finding the Cell BE manual!)
-mwarn-cell-microcode is very expensive, even more so after my recent
fix for PR43763; and it used to ICE for seven years before that fix.
This patch removes both these command line options (it leaves the
positive form of -mgen-cell-microcode, doing nothing, for
compatibility).
Where cc_reg_not_micro_cr0_operand was used, we now get the regular
cc_reg_not_cr0_operand.
One testcase used -mgen-cell-microcode and one its negation; both
are adjusted.
* config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
(lwa_operand): Delete rs6000_gen_cell_microcode test.
* config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
rs6000_gen_cell_microcode code.
(rs6000_final_prescan_insn): Delete.
(rs6000_opt_vars): Delete the "gen-cell-microcode" and
"warn-cell-microcode" entries.
* config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
* config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
throughout. Change cc_reg_not_micro_cr0_operand to
cc_reg_not_cr0_operand throughout.
(*extendhi<mode>2_noload): Delete.
* config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
(mwarn-cell-microcode): Delete.
* doc/invoke.texi (RS/6000 and PowerPC Options): Delete
-mgen-cell-microcode and -mwarn-cell-microcode.
Nathan Sidwell [Tue, 30 May 2017 17:48:57 +0000 (17:48 +0000)]
cp-tree.def (OVERLOAD): Fix comment.
* cp-tree.def (OVERLOAD): Fix comment.
* cp-tree.h: Fix comments and whitespace.
* error.c (dump_decl): Use pp_cxx_colon_colon, ovl_scope.
* name-lookup.c (add_decl_to_level): Assert not class.
(check_local_shadow): Use OVL_P.
(pushdecl_with_scope_1): Rename to ...
(do_pushdecl_with_Scope): ... here.
(do_nonmember_using_decl): Use qualified_namespace_lookup return
value.
(push_class_level_binding_1): Use OVL_P.
(pushdecl_namespace_level): Use do_pushdecl_with_scope.
(pushtag_1): Rename to ...
(do_pushtag): ... here. Adjust do_pushdecl_with_scope call.
(pushtag): Adjust.
(store_class_bindings): Do not time here.
* name-lookup.h (pushdecl_outermost_localscope): Reorder.
* pt.c (listify): Declare argvec at point of initialization.
gimplify_modify_expr: avoid DECL_DEBUG_EXPR links across functions
An upcoming patch exposes a bug in gimplify_modify_expr. There, we try
not to create DECL_DEBUG_EXPR links across functions, however we don't
check that *FROM_P actually belongs to the current function before
modifying it. This patch fixes this oversight.
gcc/
* gimplify.c (gimplify_modify_expr): Don't create a
DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
function.
Wilco Dijkstra [Tue, 30 May 2017 14:29:55 +0000 (14:29 +0000)]
Fix ARM bootstrap failure due to an odd warning:
/src/gcc/gcc/config/arm/arm-builtins.c: In function 'rtx_def* arm_expand_builtin(tree, rtx, rtx, machine_mode, int)':
/src/gcc/gcc/config/arm/arm-builtins.c:3056:46: error: type qualifiers ignored on cast result type [-Werror=ignored-qualifiers]
if (d->code == (const enum arm_builtins) fcode)
^~~~~
Avoid the warning by removing const, and bootstrap is OK again.
Richard Biener [Tue, 30 May 2017 13:21:44 +0000 (13:21 +0000)]
tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type and reduc_def fields.
2017-05-30 Richard Biener <rguenther@suse.de>
* tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
and reduc_def fields.
(STMT_VINFO_REDUC_TYPE): New define.
(STMT_VINFO_REDUC_DEF): Likewise.
(vect_force_simple_reduction): Adjust prototype.
* tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
(vect_is_simple_reduction): Remove check_reduction argument.
(vect_force_simple_reduction): Adjust and set
STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
(vectorizable_reduction): Do not re-do reduction analysis
but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
* tree-parloops.c (gather_scalar_reductions): Adjust.
Jakub Jelinek [Tue, 30 May 2017 12:05:30 +0000 (14:05 +0200)]
re PR libgomp/80822 (libgomp incorrect affinity when OMP_PLACES=threads)
PR libgomp/80822
* config/linux/affinity.c (gomp_affinity_init_level_1): New function.
(gomp_affinity_init_level): Use it. Always analyze the core and thread
sibling lists, depending on level just pick up what CPUs to put
together into a place vs. whether add multiple ordered places.
Robin Dapp [Tue, 30 May 2017 10:59:07 +0000 (10:59 +0000)]
Vector peeling cost model 4/6
gcc/ChangeLog:
2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
* tree-vect-data-refs.c (vect_get_data_access_cost):
Workaround for SLP handling.
(vect_enhance_data_refs_alignment):
Compute costs for doing no peeling at all, compare to the best
peeling costs so far and avoid peeling if cheaper.
Robin Dapp [Tue, 30 May 2017 10:57:53 +0000 (10:57 +0000)]
Vector peeling cost model 3/6
gcc/ChangeLog:
2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
* tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
Return peeling info and set costs to zero for unlimited cost
model.
(vect_enhance_data_refs_alignment): Also inspect all datarefs
with unknown misalignment. Compute and costs for unknown
misalignment, compare them to the costs for known misalignment
and choose the cheapest for peeling.
Max Filippov [Mon, 29 May 2017 23:03:52 +0000 (23:03 +0000)]
gcc: xtensa: fix unused parameter warning
2017-05-28 Max Filippov <jcmvbkbc@gmail.com>
gcc/
* config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
Mark 'to' argument with ATTRIBUTE_UNUSED.
Max Filippov [Mon, 29 May 2017 23:02:36 +0000 (23:02 +0000)]
gcc: xtensa: fix fprintf format specifiers
HOST_WIDE_INT may not be long as assumed in print_operand and
xtensa_emit_call. Use HOST_WIDE_INT_PRINT_DEC/HOST_WIDE_INT_PRINT_HEX
format strings instead of %ld/0x%lx. This fixes incorrect assembly code
generation by the compiler running on armhf host.
2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
gcc/
* config/xtensa/xtensa.c (xtensa_emit_call): Use
HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
(print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
format string.
* check_GNU_style_lib.py (TrailingWhitespaceCheck.check): Assert no
trailing eol.
(TrailingWhitespaceTest): New unit test.
(check_GNU_style_file): Remove eol before checking.
Rainer Orth [Sun, 28 May 2017 19:12:45 +0000 (19:12 +0000)]
Fix libgfortran FMA3/FMA4 tests
* acinclude.m4 (LIBGFOR_CHECK_FMA3): Use test from
check_effective_target_fma in gcc.target/i386/i386.exp.
(LIBGFOR_CHECK_FMA4): Use test from check_effective_target_fma4.
* configure: Regenerate.
Ian Lance Taylor [Fri, 26 May 2017 21:48:49 +0000 (21:48 +0000)]
compiler: identify array index expressions in lvalue context
Update the Mark_lvalue_varexprs visitor class (used in
Assignment_statement::do_get_backend) to identify
Array_index_expression's that appear on the left hand side of
assignments, then pass this context info on when invoking
Array_type::get_value_pointer.
Martin Liska [Fri, 26 May 2017 14:45:59 +0000 (16:45 +0200)]
Simplify usage of some TDF_* flags.
2017-05-26 Martin Liska <mliska@suse.cz>
* cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
always leading ';; '.
(dump_bb_info): Likewise.
(brief_dump_cfg): Likewise.
* cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
* dumpfile.c: Remove usage of TDF_VERBOSE.
* dumpfile.h (enum dump_kind): Likewise.
(dump_gimple_bb_header): Do not use TDF_COMMENT.
* print-tree.c (debug_verbose): Remove.
* tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
(dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
* tree-diagnostic.c (default_tree_printer): Replace
TDF_DIAGNOSTIC with TDF_SLIM.
Bin Cheng [Fri, 26 May 2017 14:21:29 +0000 (14:21 +0000)]
tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass in parameter loop, rather than loop_vinfo.
* tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
in parameter loop, rather than loop_vinfo.
(create_intersect_range_checks): Ditto.
(vect_create_cond_for_alias_checks): Update call to above functions.
Bin Cheng [Fri, 26 May 2017 14:10:43 +0000 (14:10 +0000)]
tree-vect-data-refs.c (Operator==, [...]): Move from ...
* tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
Move from ...
* tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
* tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
out code pruning runtime alias checks.
* tree-data-ref.c (prune_runtime_alias_test_list): New function
factored out from above.
* tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
Move from ...
* tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
... to here.
(prune_runtime_alias_test_list): New decalaration.
PR ipa/80663
* g++.dg/ipa/pr80212.C: Remove the test as it does not longer
split at the problematic spot.
* gcc.dg/ipa/pr48195.c: Change 101 to 100 as 101 is no longer
a valid value of the param.
Jakub Jelinek [Fri, 26 May 2017 09:17:54 +0000 (11:17 +0200)]
cp-tree.h (struct lang_decl_decomp): New type.
* cp-tree.h (struct lang_decl_decomp): New type.
(struct lang_decl): Add u.decomp.
(LANG_DECL_DECOMP_CHECK): Define.
(DECL_DECOMPOSITION_P): Note it is set also on the vars
for user identifiers.
(DECL_DECOMP_BASE): Define.
(retrofit_lang_decl): Add extra int = 0 argument.
* lex.c (retrofit_lang_decl): Add SEL argument, if non-zero
use it to influence the selector choices and for selector
0 to non-zero transition copy old content.
(cxx_dup_lang_specific_decl): Handle DECL_DECOMPOSITION_P.
* decl.c (poplevel): For DECL_DECOMPOSITION_P, check
!DECL_DECOMP_BASE instead of !DECL_VALUE_EXPR. Adjust warning
wording if decl is a structured binding.
(cp_finish_decomp): Pass 4 as the new argument to retrofit_lang_decl.
Set DECL_DECOMP_BASE. Ignore DECL_READ_P sets from initialization
of individual variables for tuple structured bindings.
(grokdeclarator): Pass 4 as the new argument to retrofit_lang_decl.
Clear DECL_DECOMP_BASE.
* decl2.c (mark_used): Mark DECL_DECOMP_BASE TREE_USED as well.
* pt.c (tsubst_decomp_names): Assert DECL_DECOMP_BASE matches what
is expected.
* expr.c (mark_exp_read): Recurse on DECL_DECOMP_BASE instead of
DECL_VALUE_EXPR.
* g++.dg/cpp1z/decomp29.C (p): New variable.
(main): Add further tests.