[PATCH][GCC] Make DR_TARGET_ALIGNMENT compile time variable

This patch enables targets to describe DR_TARGET_ALIGNMENT as a compile-time
variable.  It does so by turning the variable into a 'poly_uint64'.

gcc/ChangeLog:
2018-11-13  Andre Vieira  <andre.simoesdiasvieira@arm.com>

* config/aarch64/aarch64.c
(aarch64_vectorize_preferred_vector_alignment): Change return type to
poly_uint64.
(aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
alignment being a poly int.
* doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
return type to poly_uint64.
* target.def (default_preferred_vector_alignment): Likewise.
* targhooks.c (default_preferred_vector_alignment): Likewise.
* targhooks.h (default_preferred_vector_alignment): Likewise.
* tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
(vect_compute_data_ref_alignment): Adapt to vector alignment being a
poly int.
(vect_update_misalignment_for_peel): Likewise.
(vect_enhance_data_refs_alignment): Likewise.
(vect_find_same_alignment_drs): Likewise.
(vect_duplicate_ssa_name_ptr_info): Likewise.
(vect_setup_realignment): Likewise.
(vect_can_force_dr_alignment_p): Change alignment parameter type to
poly_uint64.
* tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
mask with a compile time variable vector alignment.
(vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
int.
(vect_do_peeling): Exit early if vector alignment is not constant.
* tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
a poly int.
(vectorizable_store): Likewise.
(vectorizable_load): Likweise.
* tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
poly_uint64.
(vect_known_alignment_in_bytes): Adapt to vector alignment being a
poly int.
(vect_can_force_dr_alignment_p): Change alignment parameter type to
poly_uint64.

From-SVN: r266072

re PR tree-optimization/87962 (ICE in vect_get_vec_def_for_operand_1, at tree-vect-stmts.c:1485)

2018-11-13  Richard Biener  <rguenther@suse.de>

PR tree-optimization/87962
* tree-vect-loop.c (vect_is_simple_reduction): More reliably
detect outer reduction for disqualifying in-loop uses.

* gcc.dg/pr87962.c: New testcase.

From-SVN: r266071

re PR tree-optimization/87967 (ICE in slpeel_duplicate_current_defs_from_edges)

2018-11-13  Richard Biener  <rguenther@suse.de>

PR tree-optimization/87967
* tree-vect-loop.c (vect_transform_loop): Also copy PHIs
for constants for the scalar loop.

* g++.dg/opt/pr87967.C: New testcase.

From-SVN: r266070

[RS6000] Don't put large integer constants in TOC for -mcmodel=medium

For -mcmodel=medium we can use toc-relative addressing to access
constants placed in read-only data, which is better since they can be
merged when in .rodata.cst8.

* config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
integer constants when -mcmodel=medium.

From-SVN: r266069

[ARC] Add support for profiling in glibc.

Use PROFILE_HOOK to add mcount library calls in each toolchain.

gcc/
xxxx-xx-xx  Claudiu Zissulescu  <claziss@synopsys.com>

* config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
* config/arc/elf.h (PROFILE_HOOK): Define.
* config/arc/linux.h (PROFILE_HOOK): Likewise.

From-SVN: r266068

[ARC] Do not emit ZOL in the presence of text jump tables.

Avoid emitting lp instruction when in its ZOL body we find a jump
table data in text section. One of the reason is the jump tables size
can be changed latter on, hence the total ZOL length may be wrongly
computed.

gcc/
xxxx-xx-xx  Claudiu Zissulescu  <claziss@synopsys.com>

* config/arc/arc.c (hwloop_optimize): Bailout when detecting a
jump table data in the text section.

From-SVN: r266067

[ARC] Update EH code.

Our ABI says the blink is pushed first on stack followed by an unknown
number of register saves, and finally by fp.  Hence we cannot use the
EH_RETURN_ADDRESS macro as the stack is not finalized at that moment.
The alternative is to use the eh_return pattern and to initialize all
the bits after register allocation when the stack layout is finalized.

gcc/
xxxx-xx-xx  Claudiu Zissulescu  <claziss@synopsys.com>

* config/arc/arc.c (arc_eh_return_address_location): Repurpose it
to fit the eh_return pattern.
* config/arc/arc.md (eh_return): Define.
(VUNSPEC_ARC_EH_RETURN): Likewise.
* config/arc/arc-protos.h (arc_eh_return_address_location): Match
new implementation.
* config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.

testsuite/
xxxx-xx-xx  Claudiu Zissulescu  <claziss@synopsys.com>

* gcc.target/arc/builtin_eh.c: New test.

From-SVN: r266066

[ARC] Refurbish and improve prologue/epilogue functions.

Reimplement how prologue and epilogue is emitted to accomodate
enter/leave instructions, as well as improving the size of the
existing techinques.

The following modifications are added:

- millicode thunk calls can be now selected regardless of the
  optimization level. However they are enabled for size optimizations
  by default.  Also, the millicode optimization is turned off when we
  compile for long jumps.

- the compiler is able to use enter/leave instructions for prologue
  and epilogue. As these instructions are not ABI compatible we gurad
  them under a switch (i.e., -mcode-density-frame). When this option
  is on, the compiler will try emitting enter/leave instructions, if
  not, then millicode thunk calls (if enabled), and latter the regular
  push/pop instructions.

- The prologue/epilogue is now optimized to use pointer walks, hence
  improving the chance to have push_s/pop_s instructions emitted. It
  also tries to combine the stack adjustments with load/store
  operations.

gcc/
xxxxx-xx-xx  Claudiu Zissulescu  <claziss@synopsys.com>

        * common/config/arc/arc-common.c (arc_option_optimization_table):
        Millicode optimization is default on for size optimizations.
        * config/arc/arc-protos.h (arc_check_multi): New function.
        * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
        (ENTER_LEAVE_START_REG): Define.
        (ENTER_LEAVE_END_REG): Likewise.
        (arc_override_options): Disable millicode when long calls option
        is on.
        (arc_frame_info): Change it from int to bool.
        (arc_compute_frame_size): Clean up.
        (arc_save_restore): Remove.
        (frame_save_reg): New function.
        (frame_restore_reg): Likewise.
        (arc_enter_leave_p): Likewise.
        (arc_save_callee_saves): Likewise.
        (arc_restore_callee_saves): Likewise.
        (arc_save_callee_enter): Likewise.
        (arc_restore_callee_leave): Likewise.
        (arc_save_callee_milli): Likewise.
        (arc_restore_callee_milli): Likewise.
        (arc_expand_prologue): Reimplement to emit enter/leave
        instructions.
        (arc_expand_epilogue): Likewise.
        (arc_check_multi): New function.
        * config/arc/arc.md (push_multi_fp): New pattern.
        (push_multi_fp_blink): Likewise.
        (pop_multi_fp): Likewise.
        (pop_multi_fp_blink): Likewise.
        (pop_multi_fp_ret): Likewise.
        (pop_multi_fp_blink_ret): Likewise.
        * config/arc/arc.opt (mmillicode): Update option.
        (mcode-density-frame): New option.
        * config/arc/predicates.md (push_multi_operand): New predicate.
        (pop_multi_operand): Likewise.
        * doc/invoke.texi (ARC): Update ARC options information.

gcc/testsuite
xxxxx-xx-xx  Claudiu Zissulescu  <claziss@synopsys.com>

        * gcc.target/arc/firq-1.c: Update test.
        * gcc.target/arc/firq-3.c: Likewise.
        * gcc.target/arc/firq-4.c: Likewise.
        * gcc.target/arc/interrupt-6.c: Likewise.

From-SVN: r266065

[ARC] Add peephole rules to combine store/loads into double store/loads

Simple peephole rules which combines multiple ld/st instructions into
64-bit load/store instructions. It only works for architectures which
are having double load/store option on.

gcc/
Claudiu Zissulescu  <claziss@synopsys.com>

* config/arc/arc-protos.h (gen_operands_ldd_std): Add.
* config/arc/arc.c (operands_ok_ldd_std): New function.
(mem_ok_for_ldd_std): Likewise.
(gen_operands_ldd_std): Likewise.
* config/arc/arc.md: Add peephole2 rules for std/ldd.

From-SVN: r266064

toplev.c (output_stack_usage): Turn test on flag_stack_usage into test on stack_usage_file.

* toplev.c (output_stack_usage): Turn test on flag_stack_usage into
test on stack_usage_file.
(lang_dependent_init): Do not open the .su file if generating LTO.

From-SVN: r266063

re PR rtl-optimization/87918 (ICE in simplify_binary_operation, at simplify-rtx.c:2153 since r264688)

PR rtl-optimization/87918
* simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
simplify_gen_relational rather than simplify_gen_binary.

* gcc.target/i386/pr87918.c: New test.

From-SVN: r266062

tree-ssanames.h (set_range_info): Use value_range_base.

2018-11-13  Richard Biener  <rguenther@suse.de>

* tree-ssanames.h (set_range_info): Use value_range_base.
(get_range_info): Likewise.
* tree-ssanames.c (set_range_info): Likewise.
(get_range_info): Likewise.
* tree-vrp.c (value_range_base::union_helper): Split
out common parts of value_range[_base]::union_.
(value_range_base::union_): Update.
(value_range::union_): Likewise.
(determine_value_range_1): Use value_range_base.
(determine_value_range): Likewise.
* tree-vrp.h (value_range_base::union_helper): Move ...
(value_range::union_helper): ... from here.

From-SVN: r266061

linux-unwind.h (_sig_ucontext_t): Remove.

2018-11-13  Xianmiao Qu  <xianmiao_qu@c-sky.com>

libgcc/
* config/csky/linux-unwind.h (_sig_ucontext_t): Remove.
(csky_fallback_frame_state): Modify the check of the
instructions to adapt to changes in the kernel

From-SVN: r266060

misc.c (gnat_init_gcc_eh): Set -fnon-call-exceptions for the runtime on platforms where...

* gcc-interface/misc.c (gnat_init_gcc_eh): Set -fnon-call-exceptions
for the runtime on platforms where System.Machine_Overflow is true.

From-SVN: r266057

Implement P0315R4, Lambdas in unevaluated contexts.

When lambdas were added in C++11 they were banned from unevaluated contexts
as a way to avoid needing to deal with them in mangling or SFINAE.  This
proposal avoids that with a more narrow proposal: lambdas never compare as
equivalent (so we don't need to mangle them), and substitution failures
within a lambda are hard errors.  Lambdas appearing in places that types
couldn't previously have been declared introduces various complications; in
particular, it seems likely to mean types with no linkage being used more
broadly, risking ODR violations.  I want to follow up this patch with some
related diagnostics.

* decl2.c (min_vis_expr_r): Handle LAMBDA_EXPR.
* mangle.c (write_expression): Handle LAMBDA_EXPR.
* parser.c (cp_parser_lambda_expression): Allow lambdas in
unevaluated context.  Start the tentative firewall sooner.
(cp_parser_lambda_body): Use cp_evaluated.
* pt.c (iterative_hash_template_arg): Handle LAMBDA_EXPR.
(tsubst_function_decl): Substitute a lambda even if it isn't
dependent.
(tsubst_lambda_expr): Use cp_evaluated.  Always complain.
(tsubst_copy_and_build) [LAMBDA_EXPR]: Do nothing if tf_partial.
* semantics.c (begin_class_definition): Allow in template parm list.
* tree.c (strip_typedefs_expr): Pass through LAMBDA_EXPR.
(cp_tree_equal): Handle LAMBDA_EXPR.

From-SVN: r266056

Avoid double substitution with complete explicit template arguments.

Previously, when we got a function template with explicit arguments for all
of the template parameters, we still did "deduction", which of course
couldn't deduce anything, but did other deduction-time checking of
non-dependent conversions and such.  This broke down with the unevaluated
lambdas patch (to follow): substituting into the lambda multiple times, once
to get the function type for deduction and then again to generate the actual
decl, doesn't work, since different substitutions of a lambda produce
different types.  I believe that skipping the initial substitution when we
have all the arguments is still conformant, and produces better diagnostics
for some testcases.

* pt.c (fn_type_unification): If we have a full set of explicit
arguments, go straight to substitution.

From-SVN: r266055

decl2.c (min_vis_expr_r, [...]): New.

* decl2.c (min_vis_expr_r, expr_visibility): New.

We weren't properly constraining visibility based on names that appear in
the mangled representation of expressions.  This was made more obvious
by the upcoming unevaluated lambdas patch.

(min_vis_r): Call expr_visibility.
(constrain_visibility_for_template): Likewise.

From-SVN: r266054

Implement P0722R3, destroying operator delete.

A destroying operator delete takes responsibility for calling the destructor
for the object it is deleting; this is intended to be useful for sized
delete of a class allocated with a trailing buffer, where the compiler can't
know the size of the allocation, and so would pass the wrong size to the
non-destroying sized operator delete.

gcc/c-family/
* c-cppbuiltin.c (c_cpp_builtins): Define
__cpp_impl_destroying_delete.
gcc/cp/
* call.c (std_destroying_delete_t_p, destroying_delete_p): New.
(aligned_deallocation_fn_p, usual_deallocation_fn_p): Use
destroying_delete_p.
(build_op_delete_call): Handle destroying delete.
* decl2.c (coerce_delete_type): Handle destroying delete.
* init.c (build_delete): Don't call dtor with destroying delete.
* optimize.c (build_delete_destructor_body): Likewise.
libstdc++-v3/
* libsupc++/new (std::destroying_delete_t): New.

From-SVN: r266053

Implement P0780R2, pack expansion in lambda init-capture.

Mostly this was straightforward; the tricky bit was finding, in the
instantiation, the set of capture proxies built when instantiating the
init-capture.  The comment in lookup_init_capture_pack goes into detail.

* parser.c (cp_parser_lambda_introducer): Parse pack init-capture.
* pt.c (tsubst_pack_expansion): Handle init-capture packs.
(lookup_init_capture_pack): New.
(tsubst_expr) [DECL_EXPR]: Use it.
(tsubst_lambda_expr): Remember field pack expansions for
init-captures.

From-SVN: r266052

cp-tree.h (struct cp_evaluated): New.

* cp-tree.h (struct cp_evaluated): New.

This patch simplifies the saving/clearing/restoring of
cp_unevaluated_operand and c_inhibit_evaluation_warnings in the presence of
mid-block returns.

* init.c (get_nsdmi): Use it.
* parser.c (cp_parser_enclosed_template_argument_list): Use it.
* pt.c (coerce_template_parms, tsubst_aggr_type): Use it.

From-SVN: r266051

Change __cpp_explicit_bool to __cpp_conditional_explicit.

People objected to the old macro name as unclear, so it was changed.

* c-cppbuiltin.c (c_cpp_builtins): Change __cpp_explicit_bool to
__cpp_conditional_explicit.

From-SVN: r266050

[RS6000] secondary_reload and find_replacement

This patch removes a call only necessary when using reload.  It also
corrects a PRE_DEC address offset.

* config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
offset for PRE_DEC.
(rs6000_secondary_reload_gpr): Don't call find_replacement.

From-SVN: r266049

re PR middle-end/59634 (Documentation (info/man page): lack of information for cache size parameters (--param))

2018-11-13  Sandra Loosemore  <sandra@codesourcery.com>

PR middle-end/59634

gcc/
* doc/invoke.texi (Optimize Options): Clarify that the
l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
apply to data cache size.

From-SVN: r266048

[RS6000] Comment fixes

* config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.

From-SVN: r266047

[RS6000] Rotate testcase

The testcase exercises one of the rotate patterns.

gcc/
* config/rs6000/predicates.md (logical_const_operand),
(logical_operand): Correct comment.
gcc/testsuite/
* gcc.target/powerpc/rotmask.c: New.

From-SVN: r266046

[RS6000] Hide insn not needing to be public

* config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.

From-SVN: r266045

[RS6000] Ignore "c", "l" and "h" for register preference

This catches a few places where move insn patterns don't slightly
disparage CTR, LR and VRSAVE regs.  Also fixes the doc for the rs6000
h constraint, and removes an r->cl alternative covered by r->h.

* gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
description.
* config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
covered by alternative.
(movcc_internal1): Ignore h for register preference.
(mov<mode>_hardfloat64): Likewise.
(mov<mode>_softfloat): Ignore c, l, h for register preference.

From-SVN: r266044

re PR preprocessor/47823 (#pragma once is documented in a weird spot)

2018-11-12  Sandra Loosemore  <sandra@codesourcery.com>

PR preprocessor/47823

gcc/
* doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
documentation to...
(Pragmas): ...here.
* doc/extend.texi (Pragmas): Note additional pragmas documented
in the CPP manual.

From-SVN: r266043

Daily bump.

From-SVN: r266042

re PR target/86677 (popcount builtin detection is breaking some kernel build)

gcc/ChangeLog:

2018-11-13  Kugan Vivekanandarajah  <kuganv@linaro.org>

PR middle-end/86677
PR middle-end/87528
* tree-scalar-evolution.c (expression_expensive_p): Make BUILTIN POPCOUNT
as expensive when backend does not define it.

gcc/testsuite/ChangeLog:

2018-11-13  Kugan Vivekanandarajah  <kuganv@linaro.org>

PR middle-end/86677
PR middle-end/87528
* g++.dg/tree-ssa/pr86544.C: Run only for target supporting popcount
pattern.
* gcc.dg/tree-ssa/popcount.c: Likewise.
* gcc.dg/tree-ssa/popcount2.c: Likewise.
* gcc.dg/tree-ssa/popcount3.c: Likewise.
* gcc.target/aarch64/popcount4.c: New test.
* lib/target-supports.exp (check_effective_target_popcountl): New.

From-SVN: r266039

MIPS: Default to --with-llsc for the R5900 Linux target as well

The Linux kernel requires and emulates LL and SC for the R5900 too.  The
special --without-llsc default for the R5900 is therefore not applicable
in that case.

Reviewed-by: Maciej W. Rozycki <macro@linux-mips.org>
2018-11-12  Fredrik Noring  <noring@nocrew.org>

gcc/
* config.gcc: Update with-llsc defaults for MIPS r5900.

From-SVN: r266038

Instrument only selected files (PR gcov-profile/87442).

2018-11-12  Martin Liska  <mliska@suse.cz>

PR gcov-profile/87442
* common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
options.
* doc/invoke.texi: Document them.
* tree-profile.c (parse_profile_filter): New.
(parse_profile_file_filtering): Likewise.
(release_profile_file_filtering): Likewise.
(include_source_file_for_profile): Likewise.
(tree_profiling): Filter source files based on the
newly added options.
2018-11-12  Martin Liska  <mliska@suse.cz>

PR gcov-profile/87442
* gcc.dg/profile-filtering-1.c: New test.
* gcc.dg/profile-filtering-2.c: New test.

From-SVN: r266037

Fix documentation of __builtin_cpu_is and __builtin_cpu_supports for x86.

2018-11-12  Martin Liska  <mliska@suse.cz>

PR target/87903
* doc/extend.texi: Add missing values for __builtin_cpu_is and
__builtin_cpu_supports for x86 target.

From-SVN: r266036

re PR middle-end/21110 (incorrect documentat for high and lo_sum RTL operators)

2018-11-12  Sandra Loosemore  <sandra@codesourcery.com>

PR middle-end/21110

gcc/
* doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
have to be Pmode.
(Arithmetic): Likewise for "lo_sum".

From-SVN: r266035

PR c/81824 - Warn for missing attributes with function aliases

gcc/testsuite/ChangeLog:
* gcc.dg/Wattribute-alias.c: Require ifunc support.

From-SVN: r266034

[PR87815]Don't generate shift sequence for load replacement in DSE when the mode size is not compile-time constant

The patch adds a check if the gap is compile-time constant.

This happens when dse decides to replace the load with previous store value.
The problem is that, shift sequence could not accept compile-time non-constant
mode operand.

gcc/

2018-11-12  Renlin Li  <renlin.li@arm.com>

PR target/87815
* dse.c (get_stored_val): Add check for compile-time
constantness of gap.

gcc/testsuite/

2018-11-12  Renlin Li  <renlin.li@arm.com>

PR target/87815
* gcc.target/aarch64/sve/pr87815.c: New.

From-SVN: r266033

PR libstdc++/87963 fix build for 64-bit mingw

PR libstdc++/87963
* src/c++17/memory_resource.cc (chunk::_M_bytes): Change type from
unsigned to uint32_t.
(chunk): Fix static assertion for 64-bit targets that aren't LP64.
(bigblock::all_ones): Fix undefined shift.

From-SVN: r266032

[GCC, ARM] Enable armv8.5-a and add +sb and +predres for previous ARMv8-a in ARM

This patch adds -march=armv8.5-a to the Arm backend.
Armv8.5-A also adds two new security features:
- Speculation Barrier instruction
- Execution and Data Prediction Restriction Instructions
These are made optional to all older Armv8-A versions. Thus we are adding two
new options "+sb" and "+predres" to all older Armv8-A. These are passed on to
the assembler and have no code generation effects and have already gone in the
trunk of binutils.

*** gcc/ChangeLog ***

2018-11-12  Sudakshina Das  <sudi.das@arm.com>

* config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
(ARMv8_5a): New fgroup.
(armv8.5-a): New arch.
(armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
sb and predres.
* config/arm/arm-tables.opt: Regenerate.
* config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
* config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
* config/arm/t-multilib (v8_5_a_simd_variants): New variable.
Add matching rules for -march=armv8.5-a and extensions.
* doc/invoke.texi (ARM options): Document -march=armv8.5-a.
Add sb and predres to all armv8-a except armv8.5-a.

*** gcc/testsuite/ChangeLog ***

2018-11-12  Sudakshina Das  <sudi.das@arm.com>

* gcc.target/arm/multilib.exp: Add some -march=armv8.5-a
combination tests.

From-SVN: r266031

2018-11=12  Richard Biener  <rguenther@suse.de>

* tree-vrp.h (value_range[_base]::set): Make public.  Provide
overload for single value.
(value_range[_base]::set_nonnull): New.
(value_range[_base]::set_null): Likewise.
(value_range): Document bitmap copying behavior, mark
copy constructor and assignment operator deleted.
(value_range::move): New.
(value_range::set_and_canonicalize): Default bitmap to zero.
(set_value_range_to_nonnull): Remove.
(set_value_range_to_null): Likewise.
(set_value_range): Likewise.
(set_value_range_to_value): Likewise.
(extract_range_from_unary_expr): Work on value_range_base.
(extract_range_from_binary_expr_1): Likewise.  Rename to...
(extract_range_from_binary_expr): ... this.
* tree-vrp.c (value_range::update): Clear equiv bitmap
if required.
(value_range::move): New, move equiv bitmap.
(value_range_base::set_undefined): Avoid assignment.
(value_range::set_undefined): Likewise.
(value_range_base::set_varying): Likewise.
(value_range::set_varying): Likewise.
(set_value_range): Remove.
(value_range_base::set): New overload for value.
(value_range::set): Likewise.
(set_value_range_to_nonnull): Remove.
(value_range_base::set_nonnull): New.
(value_range::set_nonnull): Likewise.
(set_value_range_to_null): Remove.
(value_range_base::set_null): New.
(value_range::set_null): Likewise.
(range_is_null): Work on value_range_base.
(range_is_nonnull): Likewise.
(ranges_from_anti_range): Likewise.
(extract_range_into_wide_ints): Likewise.
(extract_range_from_multiplicative_op): Likewise.
(extract_range_from_binary_expr): Likewise.  Update for API changes.
(extract_range_from_unary_expr): Likewise.  Remove OBJ_TYPE_REF
handling.
(value_range::intersect_helper): Avoid copy and assignment.
(value_range::union_helper): Likewise.
(determine_value_range_1): Adjust.
* gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
Avoid assignment by using move.
(evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
* tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
Likewise.
* tree-ssanames.c (get_range_info): Likewise.
* vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
* vr-values.c (vr_values::get_value_range): Adjust.
(vr_values::update_value_range): Likewise.
(symbolic_range_based_on_p): Work on value_range_base.
(vr_values::extract_range_from_binary_expr): Use value_range_base.
(vr_values::extract_range_from_unary_expr): Likewise.
(vr_values::extract_range_from_cond_expr): Avoid assignment.
(vr_values::extract_range_from_comparison): Adjust.
(vr_values::check_for_binary_op_overflow): Use value_range_base.
(vr_values::extract_range_basic): Adjust.
(vr_values::adjust_range_with_scev): Likewise.
(vr_values::vrp_visit_assignment_or_call): Likewise.
(vr_values::get_vr_for_comparison): Change API to avoid
assignment and copy construction.
(vr_values::compare_name_with_value): Adjust accordingly.
(vr_values::compare_names): Likewise.
(vr_values::extract_range_from_phi_node): Avoid assignment and
bogus in-place modify of equiv bitmap.
(vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
* ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
for extract_range_from_unary_expr API change.
* ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.

From-SVN: r266030

* config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.

From-SVN: r266029

tree-vrp.h (value_range_base::symbolic_p, [...]): Move from value_range.

2018-11-12  Richard Biener  <rguenther@suse.de>

* tree-vrp.h (value_range_base::symbolic_p,
value_range_base::constant_p, value_range_base::zero_p,
value_range_base::singleton_p): Move from value_range.
(value_range::dump): Add.
* gimple-ssa-evrp-analyze.c
(evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
* ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
* tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
Use set_varying.
* tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
(value_range::constant_p): Likewise.
(value_range::singleton_p): Likewise.
(value_range_base::dump): Add.
(set_value_range_to_undefined): Remove.
(set_value_range_to_varying): Likewise.
(range_int_cst_p): Take value_range_base argument.
(range_int_cst_singleton_p): Likewise.
(value_range_constant_singleton): Likewise.
(vrp_set_zero_nonzero_bits): Likewise.
(extract_range_from_multiplicative_op): Use set_varying.
(extract_range_from_binary_expr_1): Likewise. Use set_undefined.
(extract_range_from_unary_expr): Likewise.
(dump_value_range_base): Change to overload of dump_value_range.
(vrp_prop::vrp_initialize): Use set_varying and set_undefined.
(vrp_prop::visit_stmt): Likewise.
(value_range::intersect_helper): Likewise.
(value_range::union_helper): Likewise.
(determine_value_range_1): Likewise.

From-SVN: r266028

tree-vrp.c (set_value_range_to_nonnull): Clear equiv.

2018-11-12  Richard Biener  <rguenther@suse.de>

* tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
(set_value_range_to_null): Likewise.
* vr-values.c (vr_values::extract_range_from_comparison):
Clear equiv for constant singleton ranges.

From-SVN: r266027

sse.md: Combine VFIXUPIMM* patterns

2018-11-12 Wei Xiao <wei3.xiao@intel.com>

        * config/i386/sse.md: Combine VFIXUPIMM* patterns
        (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
        (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
        (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
        (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
        (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
        (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.

From-SVN: r266026

re PR c/69502 (attribute aligned reduces alignment contrary to documentation)

2018-11-11  Sandra Loosemore  <sandra@codesourcery.com>

PR c/69502

gcc/
* doc/extend.texi (Common Type Attributes): For the align type
attribute, copy language about decreasing alignment from the
corresponding variable attribute.

From-SVN: r266025

csky.md (*fpuv2_nmulsf3_1, [...]): Handle -frounding-math.

2018-11-11  Xianmiao Qu  <xianmiao_qu@c-sky.com>

gcc/
* config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
-frounding-math.

gcc/testsuite
* gcc.target/csky/fnmul-1.c: New.
* gcc.target/csky/fnmul-2.c: New.
* gcc.target/csky/fnmul-3.c: New.
* gcc.target/csky/fnmul-4.c: New.

From-SVN: r266024

Daily bump.

From-SVN: r266023

re PR c++/43105 (Document restrictions on mixing -fno-rtti and -frtti)

2018-11-11  Sandra Loosemore  <sandra@codesourcery.com>

PR c++/43105

gcc/
* doc/invoke.texi (C++ Dialect Options): Add warning about mixing
-frtti and -fno-rtti code.

From-SVN: r266020

Fix copypasto in las ChangeLog.

From-SVN: r266019

re PR libstdc++/54005 (Use __atomic_always_lock_free in libstdc++ is_lock_free instead of __atomic_is_lock_free)

PR libstdc++-v3/54005
* include/bits/atomic_base.h (__atomic_base<_TTp>::is_lock_free(),
__atomic_base<_PTp*>::is_lock_free()): Call __atomic_always_lock_free
with the type-derived _S_alignment instead of __alignof the object.
* include/std/atomic (atomic<T>::is_lock_free()): Likewise.

From-SVN: r266018

re PR c/26366 (__builtin_expect needs better documentation)

2018-11-11  Sandra Loosemore  <sandra@codesourcery.com>

PR c/26366

gcc/
* doc/extend.texi (Other Builtins): Document probability associated
with __builtin_expect.

From-SVN: r266017

re PR target/87928 (ICE in ix86_compute_frame_layout, at config/i386/i386.c:11161 since r228607)

PR target/87928
* config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
instead of (TARGET_64BIT && ix86_abi == MS_ABI).
* config/i386/darwin.h (STACK_BOUNDARY): Ditto.
* config/i386/cygming.h (STACK_BOUNDARY): Remove.

testsuite /Changelog:

PR target/87928
* gcc.target/i386/pr87928.c: New test.

From-SVN: r266016

csky-linux-elf.h (CC1_SPEC): Support -profile.

2018-11-11  Xianmiao Qu  <xianmiao_qu@c-sky.com>

* config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.

From-SVN: r266013

csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.

2018-11-11  Xianmiao Qu  <xianmiao_qu@c-sky.com>

* config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.

From-SVN: r266012

tree-vrp.h (class value_range_base): New base class for value_range containing all but the m_equiv member.

2018-11-11  Richard Biener  <rguenther@suse.de>

* tree-vrp.h (class value_range_base): New base class for
value_range containing all but the m_equiv member.
(dump_value_range_base): Add.
(range_includes_zero_p): Work on value_range_base.
* tree-vrp.c (value_range_base::set): Split out base handling
from...
(value_range::set): this.
(value_range::set_equiv): New.
(value_range_base::value_range_base): New constructors.
(value_range_base::check): Split out base handling from...
(value_range::check): this.
(value_range::equal_p): Refactor in terms of
ignore_equivs_equal_p which is now member of the base.
(value_range_base::set_undefined): New.
(value_range_base::set_varying): Likewise.
(value_range_base::dump):Split out base handling from...
(value_range::dump): this.
(value_range_base::set_and_canonicalize): Split out base handling
from...
(value_range::set_and_canonicalize): this.
(value_range_base::union_): New.
* ipa-prop.h (struct ipa_jump_func): Use value_range_base *
for m_vr.
* ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
instead of value_range everywhere.
(ipcp_vr_lattice::print): Use dump_value_range_base.
(ipcp_vr_lattice::meet_with): Adjust.
(ipcp_vr_lattice::meet_with_1): Likewise.
(ipa_vr_operation_and_type_effects): Likewise.
(propagate_vr_across_jump_function): Likewise.
* ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
(ipa_get_value_range): Likewise.
(ipa_set_jfunc_vr): Likewise.
(ipa_compute_jump_functions_for_edge): Likewise.

From-SVN: r266011

Implement P0318R1 unwrap_ref_decay and unwrap_reference

Implement P0318R1 unwrap_ref_decay and unwrap_reference
* include/std/type_traits (unwrap_reference, unwrap_reference_t)
(unwrap_ref_decay, unwrap_ref_decay_t): New traits and aliases.
* testsuite/20_util/unwrap_reference/1.cc: New test.
* testsuite/20_util/unwrap_reference/2.cc: New test.

From-SVN: r266010

Implement P1007R3 std::assume_aligned

Implement P1007R3 std::assume_aligned
* include/std/memory (assume_aligned): Implement for C++17.
* testsuite/20_util/assume_aligned/1.cc: New test.
* testsuite/20_util/assume_aligned/2_neg.cc: New test.
* testsuite/20_util/assume_aligned/3.cc: New test.

From-SVN: r266009

re PR middle-end/65703 (-fdefer-pop documentation is confusing)

2018-11-10  Sandra Loosemore  <sandra@codesourcery.com>

PR middle-end/65703

gcc/
* doc/invoke.texi (Optimize Options): Add @opindex entries
for the positive forms of -fno-xxx and -mno-xxx options
that were lacking them.

From-SVN: r266008

Daily bump.

From-SVN: r266007

combine: More make_more_copies

This makes make_more_copies do what its documentation says, that is,
only make an intermediate pseudo if copying to a pseudo.

This regressed generated code quality when we didn't keep the original
notes that were on the copy, but since r265582 we do, and only allowing
pseudos now is a win.  It also simplifies the code.

* combine.c (make_more_copies): Only make an intermediate copy if the
dest of a move is a pseudo.

From-SVN: r266004

Daily bump.

From-SVN: r266001

re PR testsuite/87965 (Test cases c-c++-common/pr60226.c and gcc.dg/pr39323-1.c fail starting with r265977)

gcc/testsuite/ChangeLog:

PR testsuite/87965
* c-c++-common/pr60226.c: Adjust text of expected diagnostic.
* gcc.dg/pr39323-1.c: Ditto.

From-SVN: r265998

MAINTAINERS: add myself as or1k maintainer

ChangeLog:

yyyy-mm-dd  Stafford Horne  <shorne@gmail.com>

* MAINTAINERS (CPU Port Maintainers): Add myself for or1k.
(Write After Approval): Remove myself.

From-SVN: r265996

Walloca-16.c: Ignore conflicting types for built-in warnings.

* gcc.dg/Walloca-16.c: Ignore conflicting types for built-in
warnings.
* gcc.dg/Wrestrict-4.c: Ditto.
* gcc.dg/Wrestrict-5.c: Ditto.
* gcc.dg/pr83463.c: Ditto.
* gcc.dg/torture/pr55890-2.c: Ditto.
* gcc.dg/torture/pr55890-3.c: Ditto.
* gcc.dg/torture/pr71816.c: Ditto.

From-SVN: r265995

re PR target/87221 (cannot build with -pie)

PR target/87221
* config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
(NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.

From-SVN: r265994

re PR driver/41179 (Documentation for "-fno-toplevel-reorder" is confusing (and wrong))

2018-11-09  Sandra Loosemore  <sandra@codesourcery.com>

PR driver/41179
PR middle-end/65703

gcc/
* doc/invoke.texi (Optimize Options): Clarify default behavior
for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.

From-SVN: r265993

re PR tree-optimization/87940 (FAIL: gcc.dg/warn-strlen-no-nul.c)

2018-11-09  Bernd Edlinger  <bernd.edlinger@hotmail.de>

        PR tree-optimization/87940
        * expr.c (string_constant): Don't strip NOPS in subexpressions.
        Fold PLUS_EXPR correctly.

From-SVN: r265992

S/390: Allow relative addressing of literal pool entries

r265490 allowed the compiler to choose in a more flexible way whether to
use load or load-address-relative-long (LARL) instruction.  When it
chose LARL for literal pool references, the latter ones were rewritten
by pass_s390_early_mach to use UNSPEC_LTREF, which assumes base register
usage, which in turn is not compatible with LARL.  The end result was an
ICE because of unrecognizable insn.

UNSPEC_LTREF and friends are necessary in order to communicate the
dependency on the base register to pass_sched2.  When relative
addressing is used, no base register is necessary, so in such cases the
rewrite must be avoided.

gcc/ChangeLog:

2018-11-09  Ilya Leoshkevich  <iii@linux.ibm.com>

PR target/87762
* config/s390/s390.c (s390_safe_relative_long_p): New function.
(annotate_constant_pool_refs): Skip insns which support
relative addressing.
(annotate_constant_pool_refs_1): New helper function.
(find_constant_pool_ref): Skip insns which support relative
addression.
(find_constant_pool_ref_1): New helper function.
(replace_constant_pool_ref): Skip insns which support
relative addressing.
(replace_constant_pool_ref_1): New helper function.
(s390_mainpool_start): Adapt to the new signature.
(s390_mainpool_finish): Likewise.
(s390_chunkify_start): Likewise.
(s390_chunkify_finish): Likewise.
(pass_s390_early_mach::execute): Likewise.
(s390_prologue_plus_offset): Likewise.
(s390_emit_prologue): Likewise.
(s390_emit_epilogue): Likewise.

From-SVN: r265991

c-parser.c (c_parser_omp_clause_final): Use c_parser_expr_no_commas...

c/
* c-parser.c (c_parser_omp_clause_final): Use
c_parser_expr_no_commas, convert_lvalue_to_rvalue,
c_objc_common_truthvalue_conversion, c_fully_fold and parentheses
parsing instead of c_parser_paren_condition.
(c_parser_omp_clause_if): Use c_parser_expr_no_commas,
convert_lvalue_to_rvalue, c_objc_common_truthvalue_conversion and
c_fully_fold instead of c_parser_condition.
(c_parser_omp_clause_num_threads, c_parser_omp_clause_num_tasks,
c_parser_omp_clause_grainsize, c_parser_omp_clause_priority,
c_parser_omp_clause_hint, c_parser_omp_clause_num_teams,
c_parser_omp_clause_thread_limit, c_parser_omp_clause_linear): Use
c_parser_expr_no_commas instead of c_parser_expression.
cp/
* parser.c (cp_parser_omp_clause_final, cp_parser_omp_clause_if): Use
cp_parser_assignment_expression instead of cp_parser_condition.
(cp_parser_omp_clause_num_threads, cp_parser_omp_clause_num_tasks,
cp_parser_omp_clause_grainsize, cp_parser_omp_clause_priority,
cp_parser_omp_clause_num_teams, cp_parser_omp_clause_thread_limit,
cp_parser_omp_clause_linear, cp_parser_omp_clause_device): Use
cp_parser_assignment_expression instead of cp_parser_expression.
(cp_parser_omp_clause_hint): Likewise.  Formatting fix.
testsuite/
* c-c++-common/gomp/clauses-5.c: New test.

From-SVN: r265990

c-parser.c (c_parser_omp_clause_reduction): Call sorry_at on reduction clause with inscan modifier.

* c-parser.c (c_parser_omp_clause_reduction): Call sorry_at on
reduction clause with inscan modifier.

* parser.c (cp_parser_omp_clause_reduction): Call sorry_at on
reduction clause with inscan modifier.

From-SVN: r265989

c-parser.c (c_parser_omp_requires): Call sorry_at on requires clauses other than atomic_default_mem_order.

* c-parser.c (c_parser_omp_requires): Call sorry_at on requires
clauses other than atomic_default_mem_order.

* parser.c (cp_parser_omp_requires): Call sorry_at on requires
clauses other than atomic_default_mem_order.

* c-c++-common/gomp/requires-1.c: Prune not supported yet messages.
* c-c++-common/gomp/requires-2.c: Likewise.
* c-c++-common/gomp/requires-4.c: Likewise.

From-SVN: r265988

gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid but unsupported lastprivate with conditional modifier.

* gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
but unsupported lastprivate with conditional modifier.

* c-c++-common/gomp/lastprivate-conditional-1.c: New test.
* c-c++-common/gomp/lastprivate-conditional-2.c: New test.

From-SVN: r265987

mips.c (mips_loongson_ext2_prefetch_cookie): Handle unused argument better.

* config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
unused argument better.  Add gcc_unreachable to silence warning.

From-SVN: r265986

affinity-fmt.c: Include inttypes.h if HAVE_INTTYPES_H.

* affinity-fmt.c: Include inttypes.h if HAVE_INTTYPES_H.
(gomp_display_affinity): Use __builtin_choose_expr to handle
properly handle argument having integral, or pointer or some other
type.  If inttypes.h is available and PRIx64 is defined, use PRIx64
with uint64_t type instead of %llx and unsigned long long.

From-SVN: r265985

PR libstdc++/87787 fix UBsan error in std::vector

PR libstdc++/87787
* include/bits/stl_uninitialized.h (__relocate_a_1): Do not call
memmove when there's nothing to copy (and pointers could be null).

From-SVN: r265984

Fix whitespace in ChangeLog

From-SVN: r265983

gcc.dg/torture/stackalign/builtin-apply-2.c: Skip on v850.

From-SVN: r265982

cmd/cgo: fix typo in gccgo name mangling recipe

    The code to implement new-style gccgo name mangling had a recipe that
    didn't quite match the one in the compiler (incorrect handling for
    '.'). This showed up as a failure in the gotools cgo test if the
    directory containing the test run included a "." character.

    Reviewed-on: https://go-review.googlesource.com/c/147917

From-SVN: r265981

PR middle-end/81824 - Warn for missing attributes with function aliases

gcc/c-family/ChangeLog:

PR middle-end/81824
* c-attribs.c (handle_copy_attribute): New function.

gcc/cp/ChangeLog:

PR middle-end/81824
* pt.c (warn_spec_missing_attributes): Move code to attribs.c.
Call decls_mismatched_attributes.

gcc/ChangeLog:

PR middle-end/81824
* attribs.c (has_attribute): New helper function.
(decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
* attribs.h (decls_mismatched_attributes): Declare.
* cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
(maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
* common.opt (-Wattribute-alias): Take an argument.
(-Wno-attribute-alias): New option.
* doc/extend.texi (Common Function Attributes): Document copy.
(Common Variable Attributes): Same.
* doc/invoke.texi (-Wmissing-attributes): Document enhancement.
(-Wattribute-alias): Document new option argument.

gcc/testsuite/ChangeLog:

PR middle-end/81824
* gcc.dg/Wattribute-alias.c: New test.
* gcc.dg/Wmissing-attributes.c: New test.
* gcc.dg/attr-copy.c: New test.
* gcc.dg/attr-copy-2.c: New test.
* gcc.dg/attr-copy-3.c: New test.
* gcc.dg/attr-copy-4.c: New test.

From-SVN: r265980

re PR fortran/78351 (comma not terminating READ of formatted input field - ok in 4.1.7, not 4.4.7- maybe related to 25419?)

2018-11-09  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR libfortran/78351
* io/transfer.c (read_sf_internal): Delete leftover
debug code.

From-SVN: r265979

arm - minor cleanups for CPU alias support

A couple of very minor issues with the new support for CPU
aliases.

* config/arm/parsecpu.awk (/alias/): Tighten invisible alias
matching criteria.  Remove unused array initializer.

From-SVN: r265978

PR c/87795 - Excessive alignment permitted for functions and labels

gcc/c-family/ChangeLog:

PR c/87795
* c-common.c (check_user_alignment): Use MAX_OFILE_ALIGNMENT.

gcc/testsuite/ChangeLog:

PR c/87795
* gcc.dg/attr-aligned.c: New test.

From-SVN: r265977

xmmintrin.h (_mm_cvtss_si32): Fix incorrect constraints by introducing a new temporary.

2018-11-09  Bill Schmidt  <wschmidt@linux.ibm.com>
    Jinsong Ji  <jji@us.ibm.com>

* config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
constraints by introducing a new temporary.
(_mm_cvtss_si64): Likewise.

Co-Authored-By: Jinsong Ji <jji@us.ibm.com>
From-SVN: r265975

syscall: change RLIM_INFINITY from 0xffffffffffffffff to -1

    For compatibility with the gc toolchain's syscall package.

    Fixes golang/go#28665

    Reviewed-on: https://go-review.googlesource.com/c/148697

From-SVN: r265974

Update ChangeLog entry

From-SVN: r265971

Come up with the flag -fipa-stack-alignment.

2018-11-09  Martin Liska  <mliska@suse.cz>

* common.opt: Add -fipa-stack-alignment flag.
* doc/invoke.texi: Document it.
* final.c (rest_of_clean_state): Guard stack
shrinking with flag.
2018-11-09  Martin Liska  <mliska@suse.cz>

* gcc.target/i386/ipa-stack-alignment.c: New test.

From-SVN: r265970

Come up with -fipa-reference-addressable flag.

2018-11-09  Martin Liska  <mliska@suse.cz>

* cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
to ...
(ipa_discover_variable_flags): ... this.
* common.opt: Come up with new flag -fipa-reference-addressable.
* doc/invoke.texi: Document it.
* ipa-reference.c (propagate): Call the renamed fn.
* ipa-visibility.c (whole_program_function_and_variable_visibility):
Likewise.
* ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
...
(ipa_discover_variable_flags): ... this.  Discover
non-addressable variables only with the newly added flag.
* opts.c: Enable the newly added flag with -O1 and higher
optimization level.
2018-11-09  Martin Liska  <mliska@suse.cz>

* gcc.dg/tree-ssa/writeonly-2.c: New test.

From-SVN: r265969

json.cc: fix comment

gcc/ChangeLog:
* json.cc (selftest::test_writing_literals): Fix comment.

From-SVN: r265968

workshare-reduction-1.c: New test.

2018-11-09  Jakub Jelinek  <jakub@redhat.com>

* gcc.dg/gomp/workshare-reduction-1.c: New test.
* gcc.dg/gomp/workshare-reduction-2.c: New test.
* gcc.dg/gomp/workshare-reduction-3.c: New test.
* gcc.dg/gomp/workshare-reduction-4.c: New test.
* gcc.dg/gomp/workshare-reduction-5.c: New test.
* gcc.dg/gomp/workshare-reduction-6.c: New test.
* gcc.dg/gomp/workshare-reduction-7.c: New test.
* gcc.dg/gomp/workshare-reduction-8.c: New test.
* gcc.dg/gomp/workshare-reduction-9.c: New test.
* gcc.dg/gomp/workshare-reduction-10.c: New test.
* gcc.dg/gomp/workshare-reduction-11.c: New test.
* gcc.dg/gomp/workshare-reduction-12.c: New test.
* gcc.dg/gomp/workshare-reduction-13.c: New test.
* gcc.dg/gomp/workshare-reduction-14.c: New test.
* gcc.dg/gomp/workshare-reduction-15.c: New test.
* gcc.dg/gomp/workshare-reduction-16.c: New test.
* gcc.dg/gomp/workshare-reduction-17.c: New test.
* gcc.dg/gomp/workshare-reduction-18.c: New test.
* gcc.dg/gomp/workshare-reduction-19.c: New test.
* gcc.dg/gomp/workshare-reduction-20.c: New test.
* gcc.dg/gomp/workshare-reduction-21.c: New test.
* gcc.dg/gomp/workshare-reduction-22.c: New test.
* gcc.dg/gomp/workshare-reduction-23.c: New test.
* gcc.dg/gomp/workshare-reduction-24.c: New test.
* gcc.dg/gomp/workshare-reduction-25.c: New test.
* gcc.dg/gomp/workshare-reduction-26.c: New test.
* gcc.dg/gomp/workshare-reduction-27.c: New test.
* gcc.dg/gomp/workshare-reduction-28.c: New test.
* gcc.dg/gomp/workshare-reduction-29.c: New test.
* gcc.dg/gomp/workshare-reduction-30.c: New test.
* gcc.dg/gomp/workshare-reduction-31.c: New test.
* gcc.dg/gomp/workshare-reduction-32.c: New test.
* gcc.dg/gomp/workshare-reduction-33.c: New test.
* gcc.dg/gomp/workshare-reduction-34.c: New test.
* gcc.dg/gomp/workshare-reduction-35.c: New test.
* gcc.dg/gomp/workshare-reduction-36.c: New test.
* gcc.dg/gomp/workshare-reduction-37.c: New test.
* gcc.dg/gomp/workshare-reduction-38.c: New test.
* gcc.dg/gomp/workshare-reduction-39.c: New test.
* gcc.dg/gomp/workshare-reduction-40.c: New test.
* gcc.dg/gomp/workshare-reduction-41.c: New test.
* gcc.dg/gomp/workshare-reduction-42.c: New test.
* gcc.dg/gomp/workshare-reduction-43.c: New test.
* gcc.dg/gomp/workshare-reduction-44.c: New test.
* gcc.dg/gomp/workshare-reduction-45.c: New test.
* gcc.dg/gomp/workshare-reduction-46.c: New test.
* gcc.dg/gomp/workshare-reduction-47.c: New test.
* gcc.dg/gomp/workshare-reduction-48.c: New test.
* gcc.dg/gomp/workshare-reduction-49.c: New test.
* gcc.dg/gomp/workshare-reduction-50.c: New test.
* gcc.dg/gomp/workshare-reduction-51.c: New test.
* gcc.dg/gomp/workshare-reduction-52.c: New test.
* gcc.dg/gomp/workshare-reduction-53.c: New test.
* gcc.dg/gomp/workshare-reduction-54.c: New test.
* gcc.dg/gomp/workshare-reduction-55.c: New test.
* gcc.dg/gomp/workshare-reduction-56.c: New test.
* gcc.dg/gomp/workshare-reduction-57.c: New test.
* gcc.dg/gomp/workshare-reduction-58.c: New test.
libgomp/
* testsuite/libgomp.c-c++-common/task-reduction-13.c: New test.
* testsuite/libgomp.c-c++-common/task-reduction-14.c: New test.

From-SVN: r265967

Remove extra memory allocation of strings.

2018-11-09  Martin Liska  <mliska@suse.cz>

* config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
string to a stack buffer.
(aarch64_parse_cpu): Likewise.
(aarch64_parse_tune): Likewise.

From-SVN: r265966

re PR c/87953 (asan: stack-buffer-overflow in vectorizable_reduction)

2018-11-09  Richard Biener  <rguenther@suse.de>

PR tree-optimization/87953
* tree-vect-loop.c (vectorizable_reduction): For analysis
always pass ops[0] to vectorizable_condition.

From-SVN: r265964

or1k: gcc: initial support for openrisc

2018-11-09  Stafford Horne  <shorne@gmail.com>
    Richard Henderson  <rth@twiddle.net>
    Joel Sherrill  <joel@rtems.org>

* common/config/or1k/or1k-common.c: New file.
* config/or1k/*: New.
* config.gcc (or1k*-*-*): New.
* configure.ac (or1k*-*-*): New test for openrisc tls.
* configure: Regenerated.
* doc/install.texi: Document OpenRISC triplets.
* doc/invoke.texi: Document OpenRISC arguments.
* doc/md.texi: Document OpenRISC.

Co-Authored-By: Joel Sherrill <joel@rtems.org>
Co-Authored-By: Richard Henderson <rth@twiddle.net>
From-SVN: r265963

or1k: testsuite: initial support for openrisc

gcc/testsuite/ChangeLog:

2018-11-09  Stafford Horne  <shorne@gmail.com>
    Richard Henderson  <rth@twiddle.net>

* gcc.c-torture/execute/20101011-1.c: Adjust for OpenRISC.
* gcc.dg/20020312-2.c: Likewise.
* gcc.dg/attr-alloc_size-11.c: Likewise.
* gcc.dg/builtin-apply2.c: Likewise.
* gcc.dg/nop.h: Likewise.
* gcc.dg/torture/stackalign/builtin-apply-2.c: Likewise.
* gcc.dg/tree-ssa/20040204-1.c: Likewise.
* gcc.dg/tree-ssa/reassoc-33.c: Likewise.
* gcc.dg/tree-ssa/reassoc-34.c: Likewise.
* gcc.dg/tree-ssa/reassoc-35.c: Likewise.
* gcc.dg/tree-ssa/reassoc-36.c: Likewise.
* lib/target-supports.exp
(check_effective_target_logical_op_short_circuit): Add or1k*-*-*.
* gcc.target/or1k/*: New.

Co-Authored-By: Richard Henderson <rth@twiddle.net>
From-SVN: r265962

or1k: libgcc: initial support for openrisc

libgcc/ChangeLog:

2018-11-09  Stafford Horne  <shorne@gmail.com>
    Richard Henderson  <rth@twiddle.net>

* config.host: Add OpenRISC support.
* config/or1k/*: New.

Co-Authored-By: Richard Henderson <rth@twiddle.net>
From-SVN: r265961

arm - use the new CPU alias option to simplify the list of CPUs.

This patch simplifies the table of CPUs supported in GCC by making
use of the new alias feature.  Most of the changes are fairly
straight-forward:
- arm7tdmi and arm7tdmi-s are the same thing.
- arm710t, arm720t and arm740t differ only in features external to the core
- arm920 and arm920t are the same thing; arm922t and arm940t differ from
  arm920t only in features external to the core; ep9312 is an arm920t-derived
  core that we continue to recognize for legacy reasons.
- arm10tdmi and arm1020t differ only in features external to the core.
- arm9e, arm946te-s, arm966e-s and arm968e-s differ only in features external
  to the core.
- arm10e, arm1020e and arm1022e differ only in features external to the core.

The arm10e/arm1020e/arm1022e change is the only one which changes
behaviour of the compiler slightly.  Previously, and for no reason
that I can remember, the scheduler for arm1020e/arm1022e was not used
for arm10e: this was probably an oversight.  The unification means
that the same scheduler is now used for all three cores.

* config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
(arm7tdmi-s): Delete CPU.
(arm710t): Add aliases for arm720t and arm740t.
(arm720t, arm740t): Delete CPUs.
(arm920t): Add aliases for arm920, arm922t and arm940t.
(arm920, arm922t, arm940t): Delete CPUs.
(arm10tdmi): Add alias for arm1020t.
(arm1020t): Delete CPU.
(arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
(arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
(arm10e): Add aliases for arm1020e and arm1022e.
(arm1020e, arm1022e): Delete CPU.
* config/arm/arm.md (generic_sched): Remove entries that are now
handled by aliases.
(generic_vfp): Likewise.
* config/arm/arm1020e.md: Simplify tuning selection based on alias
changes.
* config/arm/arm-tune.md: Regenerated.
* config/arm/arm-tables.opt: Regenerated.

From-SVN: r265960

re PR tree-optimization/87621 (outer loop auto-vectorization fails for exponentiation code)

2018-11-09  Richard Biener  <rguenther@suse.de>

PR tree-optimization/87621
* tree-vect-loop.c (vectorizable_reduction): Handle reduction
op with only phi inputs.
* tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
(ch_base::copy_headers): Run CSE on copied loop headers.
(pass_ch_vect::process_loop_p): Simplify.

* g++.dg/vect/pr87621.cc: New testcase.

From-SVN: r265959

large-addr-aware for biarch non-w64 mingw32

for  gcc/ChangeLog

* config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
        for non-w64 x86_64 biarch.

From-SVN: r265958

[PR86438] compare-elim: cope with set of in_b

When in_a resolves to a register set in the prev_clobber insn, we may
use the SET_SRC for the compare instead.  However, when in_b so
resolves, we proceed to use the reg with its earlier value.  When both
resolve to the same register and prev_clobber is an insn that modifies
the register, this arrangement may cause the compare to match (when it
shouldn't) and the elimination of the compare to incorrectly succeed.

(set (reg 1) (plus (reg 1) (const_int N)))
(set (reg 2) (reg 1))
(set (reg flags) (compare (reg 1) (reg 2)))

in_a: (reg 1)            --> (plus (reg 1) (const_int N))
in_b: (reg 2) -> (reg 1) -/> oops

(parallel [
 (set (reg flags) (compare (plus (reg 1) (const_int N))
                           (reg 1))) ;; should be (plus...)
 (set (reg 1) (plus (reg 1) (const_int N)))])
(set (reg 2) (reg 1))

This patch arranges for in_b to also undergo SET_SRC substitution
when appropriate, with a shortcut for when in_a and in_b are the same
rtx.

for  gcc/ChangeLog

PR rtl-optimization/86438
* compare-elim.c (try_eliminate_compare): Use SET_SRC instead
of in_b for the compare if in_b is SET_DEST.

for  gcc/testsuite/ChangeLog

PR rtl-optimization/86438
* gcc.dg/torture/pr86438.c: New.

From-SVN: r265957

[PR87793] reject non-toplevel unspecs in debug loc exprs on x86

Before revision 254025, we'd reject UNSPECs in debug loc exprs.
TARGET_CONST_NOT_OK_FOR_DEBUG_P still rejects that by default, on all
ports that override it, except for x86, that accepts @gotoff unspecs.
We can indeed accept them in top-level expressions, but not as
subexpressions: the assembler rejects the difference between two
@gotoff symbols, for example.

We could simplify such a difference and drop the @gotoffs, provided
that the symbols are in the same section; we could also accept
@gotoffs plus literal constants.  However, accepting those but
rejecting such combinations as subexpressions would be ugly, and most
likely not worth the trouble: sym@gotoff+litconst hardly makes sense
as a standalone expression, and the difference between @gotoffs should
be avoided to begin with, as follows.

Ideally, the debug loc exprs would use the symbolic data in
REG_EQUIV/REG_EQUAL notes, or delegitimized addresses, instead of
simplifying the difference between two legitimized addresses so that
the occurrences of the GOT register cancel each other.  That would
require some more elaborate surgery in var-tracking and cselib than
would be appropriate at this stage.

for  gcc/ChangeLog

PR target/87793
* config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
non-toplevel UNSPEC.

for  gcc/testsuite/ChangeLog

PR target/87793
* gcc.dg/pr87793.c: New.

From-SVN: r265956

tree-vrp.c (value_range::check): Do not access internals directly.

* tree-vrp.c (value_range::check): Do not access internals
directly.
(value_range::singleton_p): Same.
(value_range::type): Same.
(vrp_finalize): Use value_range API.

From-SVN: r265955

* tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.

From-SVN: r265954