Strenghten assumption about function start and end line (PR gcov-profile/88045).

2018-11-19  Martin Liska  <mliska@suse.cz>

PR gcov-profile/88045
* coverage.c (coverage_begin_function): Add assert.
2018-11-19  Martin Liska  <mliska@suse.cz>

PR gcov-profile/88045
* g++.dg/gcov/pr88045.C: New test.

From-SVN: r266262

Detect mixed usage of spaces and tabs.

2018-11-19  Martin Liska  <mliska@suse.cz>

* check_GNU_style_lib.py: Detect mixed usage
of spaces and tabs.

From-SVN: r266261

re PR other/40498 (no_instrument_function attribute not documented to prevent -pg instrumentation)

2018-11-18  Sandra Loosemore  <sandra@codesourcery.com>

PR other/40498

gcc/
* doc/extend.texi (Common Function Attributes): Document that
no_instrument_function applies to -p and -pg, too.
* doc/invoke.texi (Instrumentation Options): Add cross-references
to docs for -p, -pg, and -finstrument-functions.

From-SVN: r266260

Daily bump.

From-SVN: r266259

Remove unused modules from libphobos std.internal package.

libphobos/ChangeLog:

2018-11-19  Iain Buclaw  <ibuclaw@gdcproject.org>

* src/Makefile.am: Remove std.internal.digest.sha_SSSE3 and
std.internal.math.biguintx86 modules.
* src/Makefile.in: Rebuild.
* src/std/internal/digest/sha_SSSE3.d: Remove.
* src/std/internal/math/biguintx86.d: Remove.

From-SVN: r266256

PR libstdc++/83566 - cyl_bessel_j returns wrong result for x>1000

2018-11-16  Michele Pezzutti <mpezz@tiscali.it>
    Edward Smith-Rowland  <3dw4rd@verizon.net>

PR libstdc++/83566 - cyl_bessel_j returns wrong result for x>1000
for high orders.
* include/tr1/bessel_function.tcc: Perform no fewer than nu/2 iterations
of the asymptotic series (nu is the Bessel order).
* testsuite/tr1/5_numerical_facilities/special_functions/
09_cyl_bessel_j/check_value.cc: Add tests at nu=100, 1000<=x<=2000.
* testsuite/tr1/5_numerical_facilities/special_functions/
11_cyl_neumann/check_value.cc: Ditto.
* testsuite/special_functions/08_cyl_bessel_j/check_value.cc: Ditto.
* testsuite/special_functions/10_cyl_neumann/check_value.cc: Ditto.

Co-Authored-By: Edward Smith-Rowland <3dw4rd@verizon.net>
From-SVN: r266252

re PR fortran/88073 (Internal compiler error  compiling WHERE construct with -O or -O2)

2018-11-18  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/88073
* frontend-passes.c (combine_array_constructor): Do not do
anything if in a WHERE statement.

2018-11-18  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/88073
* gfortran.dg/where_7.f90: New test.

From-SVN: r266251

Allow target to override gnu-user.h crti and crtn

Also give target access to the gnu-user.h LINK_GCC_C_SEQUENCE_SPEC.

* config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
(GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
(GNU_USER_TARGET_CRTN): Define.
(GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
(GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.

From-SVN: r266250

Delete !HAVE_LD_PIE variants of startfile/endfile specs

The HAVE_LD_PIE variant doesn't contain anything that will break
linking when !HAVE_LD_PIE that isn't already broken if you choose to
build PIEs with a linker that doesn't support PIE.  All this
HAVE_LD_PIE protects is the choice of different crt files, which is
more about libc capability than linker capability.

* config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
!HAVE_LD_PIE variant.
(GNU_USER_TARGET_ENDFILE_SPEC): Likewise.

From-SVN: r266249

re PR fortran/70260 (ICE: gimplification failed)

2018-11-18  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/70260
    * expr.c (gfc_check_assign): Reject assigning to an external
    symbol.
    (gfc_check_pointer_assign): Add suppress_type_test
    argument. Insert line after if. A non-proc pointer can not point
    to a constant.  Only check types if suppress_type_test is false.
    * gfortran.h (gfc_check_pointer_assign): Add optional
    suppress_type_test argument.
    * resolve.c (gfc_resolve_code):  Move up gfc_check_pointer_assign
    and give it the extra argument.
    (resolve_fl_procedure): Set error on value for a function with
    an inizializer.

2018-11-18  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/70260
    * gfortran.dg/proc_ptr_result_5.f90:  Add dg-error directive.
    * gfortran.dg/protected_4.f90: Split line to allow for extra error.
    * gfortran.dg/protected_6.f90: Likewise.
    * gfortran.dg/assign_11.f90: New test.
    * gfortran.dg/pointer_assign_12.f90: New test.

From-SVN: r266248

Correct ChangeLog for PR31357 fix

From-SVN: r266247

re PR driver/31357 (document that --help and --help=<value> options cannot be combined.)

2018-11-17  Eric Gallager  <egallager@gcc.gnu.org>
    Sandra Loosemore  <sandra@codesourcery.com>

PR driver/31357

gcc/
* doc/invoke.texi (Overall Options): Document that --help and
--help=<value> options cannot be combined.

Co-Authored-By: Sandra Loosemore <sandra@codesourcery.com>
From-SVN: r266246

Daily bump.

From-SVN: r266245

Implement std::pmr::synchronized_pool_resource

Define the thread-safe pool resource, using a shared_mutex to allow
multiple threads to concurrently allocate from thread-specific pools.

Define new weak symbols for the pthread_rwlock_t functions, to avoid
making libstdc++.so depend on libpthread.so

When the necessary Gthread support is absent only define the
feature-test macro to 1, rather than 201603. This is intended to imply
incomplete support, because everything except synchronized_pool_resource
works.

Implement std::pmr::synchronized_pool_resource
* config/abi/pre/gnu.ver: Add new symbols.
* include/std/memory_resource [_GLIBCXX_HAS_GTHREADS]
(__cpp_lib_memory_resource): Define to expected value, 201603.
(synchronized_pool_resource): New class.
[!_GLIBCXX_HAS_GTHREADS] (__cpp_lib_memory_resource): Define to 1.
* include/std/shared_mutex (__glibcxx_rwlock_rdlock)
(__glibcxx_rwlock_tryrdlock, __glibcxx_rwlock_wrlock)
(__glibcxx_rwlock_trywrlock, __glibcxx_rwlock_unlock)
(__glibcxx_rwlock_destroy, __glibcxx_rwlock_init)
(__glibcxx_rwlock_timedrdlock, __glibcxx_rwlock_timedwrlock): Define
weak symbols for POSIX rwlock functions.
(__shared_mutex_pthread): Use weak symbols.
* include/std/version (__cpp_lib_memory_resource): Define.
* src/c++17/memory_resource.cc [_GLIBCXX_HAS_GTHREADS]
(synchronized_pool_resource::_TPools): New class.
(destroy_TPools): New function for pthread_key_create destructor.
(synchronized_pool_resource::synchronized_pool_resource)
(synchronized_pool_resource::~synchronized_pool_resource)
(synchronized_pool_resource::release)
(synchronized_pool_resource::do_allocate)
(synchronized_pool_resource::do_deallocate): Define public members.
(synchronized_pool_resource::_M_thread_specific_pools)
(synchronized_pool_resource::_M_alloc_tpools)
(synchronized_pool_resource::_M_alloc_shared_tpools): Define private
members.
* testsuite/20_util/synchronized_pool_resource/allocate.cc: New test.
* testsuite/20_util/synchronized_pool_resource/cons.cc: New test.
* testsuite/20_util/synchronized_pool_resource/is_equal.cc: New test.
* testsuite/20_util/synchronized_pool_resource/multithreaded.cc: New
test.
* testsuite/20_util/synchronized_pool_resource/release.cc: New test.
* testsuite/performance/20_util/memory_resource/pools.cc: Add
multithreaded tests using pmr::synchronized_pool_resource.

From-SVN: r266242

RISC-V: Fix epilogue unwind info with fp and single sp adjust.

gcc/
* config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
(riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
(riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset.  Set it
to step2 if frame_pointer_needed and step1 is 0.

From-SVN: r266241

re PR c++/4225 (documentation about CPLUS_INCLUDE_PATH vs. system headers incomplete)

2018-11-17  Sandra Loosemore  <sandra@codesourcery.com>

PR c++/4225

gcc/
* doc/cpp.texi (System Headers): Add note about implicit
extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.

From-SVN: r266240

re PR c++/4025 (Document use/non-use of  -fno-implicit-templates in bootflags)

2018-11-17  Sandra Loosemore  <sandra@codesourcery.com>

PR c++/4025

gcc/
* doc/invoke.texi (C++ Dialect Options): Clarify usage of
-fno-implicit-templates.

From-SVN: r266239

libphobos: Add IEEE quadruple support for std.conv

Backport from upstream phobos 2.079 for AArch64.

Reviewed-on: https://github.com/dlang/phobos/pull/5965

From-SVN: r266238

re PR tree-optimization/87546 (Gcc miscompiles at -O3 on valid code)

PR tree-optimization/87546
* tree-vect-patterns.c (vect_look_through_possible_promotion): Add
min_precision variable, initially set it to orig_precision, only does
something if op_type's precision is <= min_precision and update
min_precision whenever calling set_op.

* gcc.dg/vect/O3-pr87546.c: New test.

From-SVN: r266237

re PR rtl-optimization/88018 (ICE in insert_insn_on_edge at cfgrtl.c:1952 since r255066)

PR rtl-optimization/88018
* cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
on the presence of fallthru edge, rather than if it is a USE or not.

* g++.dg/tsan/pr88018.C: New test.

From-SVN: r266236

re PR ipa/87957 (ICE tree check: expected tree that contains ‘decl minimal’ structure, have ‘identifier_node’ in warn_odr, at ipa-devirt.c:1051 since r265519)

PR ipa/87957
* ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.

From-SVN: r266235

Fix wrong alignment returned by .alignof property.

The D language expects the value to be the minimum alignment required
for the type, not the preferred alignment.

gcc/d/ChangeLog:

2018-11-17  Iain Buclaw  <ibuclaw@gdcproject.org>

PR d/87824
* d-target.cc (Target::alignsize): Return min_align_of_type.

From-SVN: r266234

invoke.texi (Option Summary): Fix whitespace and line breaks in @gccoptlist environments.

2018-11-16  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* doc/invoke.texi (Option Summary): Fix whitespace and line
breaks in @gccoptlist environments.
(Warning Options): Likewise.
(Optimize Options): Likewise.
(PowerPC SPE Options): Likewise.
(RS/6000 and PowerPC Options): Likewise.

From-SVN: r266233

mn10300.md (adddi3_degenerate): Remove bogus gcc_assert.

* config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
gcc_assert.

From-SVN: r266232

re PR middle-end/23197 (Improve documentation on -fprofile-generate, -fprofile-use)

2018-11-16  Sandra Loosemore  <sandra@codesourcery.com>

PR middle-end/23197

gcc/
* doc/invoke.texi (Optimize Options): Update options enabled by
fprofile-generate, -fprofile-use, and -fauto-profile.

From-SVN: r266231

re PR middle-end/87854 (gcc.c-torture/compile/pr46534.c ICE for 16-bit size_t)

2018-11-17  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

PR middle-end/87854
* g++.dg/parse/concat1.C: Add dg-error for targets with size_t < 32
bits. Update comment.
* gcc.dg/concat2.c: Likewise.
* gcc.c-torture/compile/pr46534.c: Likewise.

From-SVN: r266230

Daily bump.

From-SVN: r266229

re PR target/87927 (ICE: segmentation fault with patchable_function_entry attribute for msp430-elf -mlarge)

2018-11-16  Jozef Lawrynowicz  <jozef.l@mittosystems.com>

PR target/87927
* target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
* target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
the byte_op hook.
* target.h: Add psi, pdi, pti to struct asm_int_op definition.
* targhooks.c (default_print_patchable_function_entry): Assert
asm_int_op does not return a NULL string.
* varasm.c (integer_asm_op): Return the op for a partial int type
when the requested size does not correspond to an integer type.
* config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
* doc/tm.texi: Regenerate.

From-SVN: r266226

re PR c++/87269 (ICE in tsubst_copy, at cp/pt.c:15475 starting from r261802)

PR c++/87269
* g++.dg/lookup/pr87269.C (std::size_t): New typedef.
(operator"" _a) Change unsigned long type to std::size_t.

From-SVN: r266225

re PR c++/52869 ([DR 1207] "this" not being allowed in noexcept clauses)

PR c++/52869

DR 1207
* parser.c (cp_parser_noexcept_specification_opt): Call
inject_this_parameter.

From-SVN: r266224

Implement P0479R5, [[likely]] and [[unlikely]].

[[likely]] and [[unlikely]] are equivalent to the GNU hot/cold attributes,
except that they can be applied to arbitrary statements as well as labels;
this is most likely to be useful for marking if/else branches as likely or
unlikely.  Conveniently, PREDICT_EXPR fits the bill nicely as a
representation.

I also had to fix marking case labels as hot/cold, which didn't work before.
Which then required me to force __attribute ((fallthrough)) to apply to the
statement rather than the label.

gcc/
* gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
gcc/c-family/
* c-lex.c (c_common_has_attribute): Handle likely/unlikely.
* c-attribs.c (attr_cold_hot_exclusions): Make public.
gcc/cp/
* tree.c (handle_likeliness_attribute): New.
(std_attribute_table): Add likely/unlikely.
* cp-gimplify.c (lookup_hotness_attribute, remove_hotness_attribute)
(process_stmt_hotness_attribute, first_stmt): New.
(genericize_if_stmt): Check for duplicate predictions.
* parser.c (cp_parser_statement): Call
process_stmt_hotness_attribute.
(cp_parser_label_for_labeled_statement): Apply attributes to case.
* decl.c (finish_case_label): Give label in template type void.
* pt.c (tsubst_expr) [CASE_LABEL_EXPR]: Copy attributes.
[PREDICT_EXPR]: Handle.

From-SVN: r266223

libphobos: Add IEEE quadruple support to core.internal.convert

Backport from upstream druntime 2.083 for AArch64.

Reviewed-on: https://github.com/dlang/druntime/pull/2257

From-SVN: r266222

constraints.md (wF constraint): Remove power9 fusion documentation.

[gcc]
2018-11-16  Michael Meissner  <meissner@linux.ibm.com>

* config/rs6000/constraints.md (wF constraint): Remove power9
fusion documentation.  Just document wF's use for power8 fusion.
* config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
(fusion_gpr_addis): Delete power9 fusion support.  Change power8
fusion support to require the upper 12 bits to be all 0's or all
1's.
(fusion_gpr_mem_load): Add comment.
(fusion_addis_mem_combo_load): Remove power9 fusion support.
(fusion_addis_mem_combo_store): Delete.
(fusion_offsettable_mem_operand): Delete.
* config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
power8 fusion here.
(ISA_3_0_MASKS_SERVER): Delete power9 fusion.
(POWERPC_MASKS): Delete power9 fusion.
* config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
(fusion_p9_p): Delete.
(expand_fusion_p9_load): Delete.
(expand_fusion_p9_store): Delete.
(emit_fusion_p9_load): Delete.
(emit_fusion_p9_store): Delete.
* config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
fusion support.
(rs6000_option_override_internal): Set power8 fusion based on
whether we are tuning for power8.  Delete power9 fusion support.
(rs6000_opt_masks): Delete -mpower9-fusion switch.
(emit_fusion_load): Rename emit_fusion_load_store to
emit_fusion_load, and drop fusion store support.  Update callers.
(emit_fusion_load_store): Likewise.
(emit_fusion_gpr_load): Likewise.
(fusion_p9_p): Delete.
(expand_fusion_p9_load): Delete.
(expand_fusion_p9_store): Delete.
(emit_fusion_p9_load): Delete.
(emit_fusion_p9_store): Delete.
* config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
(GPR_FUSION): Delete.
(FPR_FUSION): Delete.
(power9 fusion peephole2s): Delete.
(fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
(fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
(fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
(fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
(fusion_p9_<mode>_constant): Delete.
* config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
power9 fusion switch.
* doc/md.texi (PowerPC constraints): Update wF constraint
documentation for power8 fusion only.

[gcc/testsuite]
2018-11-16  Michael Meissner  <meissner@linux.ibm.com>

* gcc.target/powerpc/fusion3.c: Delete.
* gcc.target/powerpc/fusion4.c: Delete.

From-SVN: r266220

re PR rtl-optimization/87475 (ICE in patch_jump_insn, at cfgrtl.c:1275)

PR rtl-optimization/87475
* cfgrtl.c (patch_jump_insn): Allow redirection failure for
CROSSING_JUMP_P insns.
(cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.

* g++.dg/opt/pr87475.C: New test.

From-SVN: r266219

re PR target/88051 (internal compiler error: in add_clobbers, at config/i386/sync.md:1762)

PR target/88051
* config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
* config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
(movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.

From-SVN: r266218

re PR middle-end/87854 (gcc.c-torture/compile/pr46534.c ICE for 16-bit size_t)

PR middle-end/87854
* c-common.c (fix_string_type): Reject string literals larger than
TYPE_MAX_VALUE (ssizetype) bytes.

From-SVN: r266217

re PR middle-end/88032 (ICE in operand_subword_force, at emit-rtl.c:1793)

PR middle-end/88032
* optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
is VOIDmode, in which case use int_mode.  Similarly for op1_mode.

From-SVN: r266216

When running the testsuite on boards that can't report an error status DejaGNU...

When running the testsuite on boards that can't report an error status
DejaGNU uses a special wrapper to print the exit code on stdout and
parses stdout to find whether an execution failed or passed.

In testcases that use "freopen (..., ..., stdout)" this special line is
printed to the alternate location described in the freopen call and
DejaGNU can't find the error code.

This results in DejaGNU using a default return status of 2 and the test
failing.

This patch skips the two testcases that use freopen on stdout when
testing a board that requires this wrapper.

Testing done by running these two tests on arm-none-eabi cross build and
observing that they are unsupported.

gcc/testsuite/ChangeLog:

2018-11-16  Matthew Malcomson  <matthew.malcomson@arm.com>

* gcc.c-torture/execute/printf-2.c: Skip on wrapped boards.
* gcc.c-torture/execute/user-printf.c: Likewise.

From-SVN: r266214

tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add location_t parameter.

* tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
location_t parameter.
(switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
(switch_decision_tree::do_jump_if_equal): Likewise.
* tree-switch-conversion.c (switch_decision_tree::emit): Pass location
of switch statement to emit_case_nodes.
(switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
set it on the newly built GIMPLE comparison statement.
(switch_decision_tree::do_jump_if_equal): Likewise.
(switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
in calls to do_jump_if_equal as well as recursive calls.

From-SVN: r266213

Remove ovl_used, it is no longer needed

Remove ovl_used, it is no longer needed
* cp-tree.h (OVL_USED_P): Delete.
(lookup_keep): Delete.
* friend.c (add_friend): Don't call it.
* parser.c (lookup_literal_operator): Likewise.
(cp_parser_primary_expression): Likewise.
* semantics.c (perform_koenig_lookup): Likewise.
* pt.c (tsubst_copy <OVERLOAD>): Don't assert OVL_USED_P.
* tree.c (ovl_copy): Delete.
(ovl_insert): Remove OVL_USED_P checks.
(ovl_iterator::reveal_node): Likewise.
(ovl_iterator::remove__node): Likewise.
(ovl_used, lookup_keep): Delete.

From-SVN: r266212

Allow memory operands for PTWRITE

The earlier PTWRITE builtin definition was unnecessarily restrictive,
only allowing register input to PTWRITE. The instruction actually
supports memory operands too, so allow that too.

gcc/:

2018-11-16  Andi Kleen  <ak@linux.intel.com>

* config/i386/i386.md: Allow memory operands to ptwrite.

From-SVN: r266211

[PR c++/87269] Mark string operator overload in template defn.

https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01458.html
PR c++/87269
* parser.c (lookup_literal_operator): Mark overload for keeping
when inside template.  Refactor.

* g++.dg/lookup/pr87269.C: New.

From-SVN: r266210

[Patch][libstdc++.exp] Update the usage of cached result.

This patch replaces the usage of cached results with a global dictionary.

Additionally, check_v3_target_namedlocale is updated to check on every variant.
Originally, it is only checked once.

gcc/libstdc++-v3/:
2018-11-16  Renlin Li  <renlin.li@arm.com>
    Tejas Belagod  <tejas.belagod@arm.com>

testsuite/lib/libstdc++.exp (check_v3_target_prop_cached): New proc.
(check_v3_target): Use check_v3_target_prop_cached.

Co-Authored-By: Tejas Belagod <tejas.belagod@arm.com>
From-SVN: r266209

Don't use %z printf length specifier

* system.h (PRsa): New macro.
(SIZE_AMOUNT): Cast number to uint64_t.
* alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
(pool_usage::dump_footer): Likewise and also use PRIu64.
* bitmap.h (bitmap_usage::dump): Likewise.
* ggc-common.c (ggc_usage::dump): Likewise.
* ggc-page.c (ggc_print_statistics): Likewise.
* input.c (dump_line_table_statistics): Likewise.
* mem-stats.h (mem_usage::dump): Likewise.
(mem_usage::dump_footer): Likewise.
* rtl.c (dump_rtx_statistics): Likewise.
* tree-cfg.c (dump_cfg_stats): Likewise.
* tree-dfa.c (dump_dfa_stats): Likewise.
* tree-phinodes.c (phinodes_print_statistics): Likewise.
* tree-ssanames (ssanames_print_statistics): Likewise.
* vec.c (vec_usage::dump): Likewise.
(vec_usage::dump_footer): Likewise.

From-SVN: r266208

Fix ICE in lto_symtab_merge_symbols_1 (PR lto/88004).

2018-11-16  Martin Liska  <mliska@suse.cz>

PR lto/88004
* lto-symtab.c (lto_symtab_merge_symbols_1): Do not call
lto_symtab_symbol_p as it does checking of transparent alias.
These needs to be also merged in the function.

From-SVN: r266207

re PR testsuite/88053 (g++.dg/lto/pr54625-1.C fails with r266194)

2018-11-16  Richard Biener  <rguenther@suse.de>

PR testsuite/88053
* g++.dg/lto/pr54625-2_0.c: Add -w.

From-SVN: r266206

re PR tree-optimization/88011 (r266028 causes a bunch of go failures)

2018-11-16  Richard Biener  <rguenther@suse.de>

PR tree-optimization/88011
* tree-vrp.c (extract_range_from_binary_expr): Fix error in
replacing set_value_range_to_undefined and
set_value_range_to_varying with method calls.

From-SVN: r266205

S/390: Add a new pattern for r{o,x}sbg

Fixes rXsbg_mode_sXl test failures.

Combine used to give us

(set (reg:SI 65)
    (ior:SI (lshiftrt:SI (reg:SI 3 %r3 [ bD.2238 ])
            (const_int 2 [0x2]))
        (reg:SI 2 %r2 [ aD.2237 ])))

but now we get

(set (reg:SI 65)
    (ior:SI (subreg:SI (zero_extract:DI (reg:DI 69)
                (const_int 32 [0x20])
                (const_int 30 [0x1e])) 4)
        (subreg:SI (reg:DI 68) 4)))

or

(set (reg:SI 65)
    (ior:SI (subreg:SI (and:DI (lshiftrt:DI (reg:DI 69)
                    (const_int 2 [0x2]))
                (const_int 4294967295 [0xffffffff])) 4)
        (subreg:SI (reg:DI 68) 4)))

with an extra subreg, which appears because pseudos, unlike hard
registers, can be accessed only using their natural mode.

This patch adds a special case for that.  Also, it performs r*sbg
bit index computations during gcc run, so that expectations do not
depend on which concrete pattern was matched.

gcc/ChangeLog:

2018-11-16  Ilya Leoshkevich  <iii@linux.ibm.com>

* config/s390/s390.md
(*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
assembler.
(*r<noxa>sbg_<mode>_sll): Likewise.
(*r<noxa>sbg_<mode>_srl): Likewise.
(*r<noxa>sbg_sidi_srl): New pattern.

gcc/testsuite/ChangeLog:

2018-11-16  Ilya Leoshkevich  <iii@linux.ibm.com>

* gcc.target/s390/md/rXsbg_mode_sXl.c: Do not use arithmetic in
r{o,x}sbg expectations.
* gcc.target/s390/risbg-ll-2.c: Likewise.

From-SVN: r266203

re PR testsuite/88053 (g++.dg/lto/pr54625-1.C fails with r266194)

2018-11-16  Richard Biener  <rguenther@suse.de>

PR testsuite/88053
* g++.dg/lto/pr54625-1_0.c: Add -w.

From-SVN: r266202

Allow target overrides of ctors/dtors support advertisement on VxWorks

2018-11-16  Jerome Lambourg  <lambourg@adacore.com>

        * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
        Default to TARGET_VXWORKS_RTP.
        (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
        of TARGET_VXWORKS_RTP.
        * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
        * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
        of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.

From-SVN: r266201

linux-unwind.h (sc_pt_regs): Update for kernel.

2018-11-15  Xianmiao Qu  <xianmiao_qu@c-sky.com>

libgcc/
* config/csky/linux-unwind.h (sc_pt_regs): Update for kernel.
(sc_pt_regs_lr): Update for kernel.
(sc_pt_regs_tls): Update for kernel.

From-SVN: r266200

re PR c++/25759 (Documentation not clear, attribute packed on class in C++)

2018-11-15  Sandra Loosemore  <sandra@codesourcery.com>

PR c++/25759

gcc/
* doc/extend.texi (Common Type Attributes): Make it explicit
that attribute "packed" can apply to C++ classes.

From-SVN: r266199

Daily bump.

From-SVN: r266198

PR c++/87541 - ICE using a constant decl as an attribute alloc_size argument

PR c++/87541 - ICE using a constant decl as an attribute alloc_size argument
PR c++/87542 - bogus error on attribute format with a named constant argument

gcc/ChangeLog:

PR c++/87541
PR c++/87542
* tree.c (type_argument_type): New function.
* tree.h (type_argument_type): Declare it.
* gcc/doc/extend.texi (alloc_align): Update and clarify.
(alloc_size, nonnull, sentinel): Same.

gcc/c-family/ChangeLog:

PR c++/87541
PR c++/87542
* c-attribs.c (positional_argument): New function.
(handle_alloc_size_attribute): Use it and simplify.
(handle_alloc_align_attribute): Same.
(handle_assume_aligned_attribute): Same.
(handle_nonnull_attribute): Same.
* c-common.c (check_function_arguments): Pass fntype to
check_function_format.
* c-common.h (check_function_format): Add an argument.
(PosArgFlags, positional_argument): Declare new type and function.
* c-format.c (decode_format_attr): Add arguments.
(check_format_string, get_constant): Same.
(convert_format_name_to_system_name): Adjust.

gcc/testsuite/ChangeLog:

PR c++/87541
PR c++/87542
* g++.dg/ext/attr-alloc_size.C: New test.
* c-c++-common/pr71574.c: Adjust diagnostics.
* c-c++-common/attributes-1.c: Same.
* gcc.dg/attr-alloc_align-2.c: Same.
* gcc.dg/attr-alloc_align-4.c: New test.
* gcc.dg/attr-alloc_size-2.c: Adjust diagnostics.
* gcc.dg/attr-alloc_size.c: Same.
* gcc.dg/attr-assume_aligned-4.c: New test.
* gcc.dg/format/attr-3.c: Adjust diagnostics.
* gcc.dg/nonnull-2.c: Same.
* gcc.dg/torture/pr80612.c: Same.
* obj-c++.dg/attributes/method-format-1.mm: Same.
* obj-c++.dg/attributes/method-nonnull-1.mm: Same.
* objc.dg/attributes/method-format-1.m: same.
* objc.dg/attributes/method-nonnull-1.m: Same.

From-SVN: r266195

PR c/83656 - missing -Wbuiltin-declaration-mismatch on declaration without prototype

gcc/c/ChangeLog:

PR c/83656
* c-decl.c (header_for_builtin_fn): Declare.
(diagnose_mismatched_decls): Diagnose declarations of built-in
functions without a prototype.
* c-typeck.c (maybe_warn_builtin_no_proto_arg): New function.
(convert_argument): Same.
(convert_arguments): Factor code out into convert_argument.
Detect mismatches between built-in formal arguments in calls
to built-in without prototype.
(build_conditional_expr): Same.
(type_or_builtin_type): New function.
(convert_for_assignment): Add argument.  Conditionally issue
warnings instead of errors for mismatches.

gcc/testsuite/ChangeLog:

PR c/83656
* gcc.dg/20021006-1.c
* gcc.dg/Wbuiltin-declaration-mismatch.c: New test.
* gcc.dg/Wbuiltin-declaration-mismatch-2.c: New test.
* gcc.dg/Wbuiltin-declaration-mismatch-3.c: New test.
* gcc.dg/Wbuiltin-declaration-mismatch-4.c: New test.
* gcc.dg/Walloca-16.c: Adjust.
* gcc.dg/Wrestrict-4.c: Adjust.
* gcc.dg/Wrestrict-5.c: Adjust.
* gcc.dg/atomic/stdatomic-generic.c: Adjust.
* gcc.dg/atomic/stdatomic-lockfree.c: Adjust.
* gcc.dg/initpri1.c: Adjust.
* gcc.dg/pr15698-1.c: Adjust.
* gcc.dg/pr69156.c: Adjust.
* gcc.dg/pr83463.c: Adjust.
* gcc.dg/redecl-4.c: Adjust.
* gcc.dg/tls/thr-init-2.c: Adjust.
* gcc.dg/torture/pr55890-2.c: Adjust.
* gcc.dg/torture/pr55890-3.c: Adjust.
* gcc.dg/torture/pr67741.c: Adjust.
* gcc.dg/torture/stackalign/sibcall-1.c: Adjust.
* gcc.dg/torture/tls/thr-init-1.c: Adjust.
* gcc.dg/tree-ssa/builtins-folding-gimple-ub.c: Adjust.

From-SVN: r266194

[PR c++/86246] ICE tsubst explicit operator call

https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01405.html
PR c++/86246
PR c++/87989
* typeck.c (finish_class_member_access_expr): Conversion operator
to dependent type is dependent.

* g++.dg/template/pr86246.C: New.
* g++.dg/template/pr87989.C: New.

From-SVN: r266193

constexpr.c (ensure_literal_type_for_constexpr_object): Use DECL_SOURCE_LOCATION in error_at calls.

/cp
2018-11-15  Paolo Carlini  <paolo.carlini@oracle.com>

* constexpr.c (ensure_literal_type_for_constexpr_object): Use
DECL_SOURCE_LOCATION in error_at calls.

/testsuite
2018-11-15  Paolo Carlini  <paolo.carlini@oracle.com>

* g++.dg/cpp0x/constexpr-diag3.C: Check locations too.
* g++.dg/cpp0x/constexpr-ice19.C: Likewise.
* g++.dg/cpp0x/constexpr-nonlit2.C: Likewise.
* g++.dg/cpp1z/constexpr-lambda15.C: Likewise.
* g++.dg/ext/constexpr-vla5.C: Likewise.
* g++.dg/gomp/pr85134.C: Likewise.

From-SVN: r266192

Handle vectors that don't fit in an integer.

GCN vector sizes range between 64 and 512 bytes, none of which have
correspondingly sized integer modes.  This breaks a number of assumptions
throughout the compiler, but I don't really want to create modes just for this
purpose.

Instead, this patch fixes up the cases that I've found, so far, such that the
compiler tries something else, or fails to optimize, rather than just ICE.

2018-11-15  Andrew Stubbs  <ams@codesourcery.com>
            Kwok Cheung Yeung  <kcy@codesourcery.com>

gcc/
* tree-vect-stmts.c (vectorizable_store): Don't ICE when
int_mode_for_size fails.
(vectorizable_load): Likewise.

Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
From-SVN: r266190

[C++ DR 2336] virtual dtors, exception specs & abstract classes

https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01389.html
DR 2336
* cp-tree.h (enum special_function_kind): Add sfk_virtual_destructor.
* method.c (type_has_trivial_fn): Add it.
(SFK_DTOR_P): Likewise.
(synthesized_method_base_walk): Don't check access of vbases of
abstract classes when sfk_virtual_destructor.
(synthesized_method_walk): Skip vbases of abstract classes except
when sfk_virtual_destructor.
(get_defaulted_eh_spec): Set sfk_virtual_destructor as needed.

* g++.dg/cpp1y/pr79393-3.C: New.

From-SVN: r266188

doc/ux.texi: auto_diagnostic_group now does something

As of r266186, auto_diagnostic_group is now user-visible
(via -fdiagnostics-format=json), so update the diagnostic
guidelines accordingly.

gcc/ChangeLog:
* doc/ux.texi (Group logically-related diagnostics): Move
discussion of auto_diagnostic_group into this new subsection.
Give an example of where this grouping is used.

From-SVN: r266187

Machine-readable diagnostic output (PR other/19165)

This patch implements a -fdiagnostics-format=json option which
converts the diagnostics to be output to stderr in a JSON format;
see the documentation in invoke.texi.

Logically-related diagnostics are nested at the JSON level, using
the auto_diagnostic_group mechanism.

gcc/ChangeLog:
PR other/19165
* Makefile.in (OBJS): Move json.o to...
(OBJS-libcommon): ...here and add diagnostic-format-json.o.
* common.opt (fdiagnostics-format=): New option.
(diagnostics_output_format): New enum.
* diagnostic-format-json.cc: New file.
* diagnostic.c (default_diagnostic_final_cb): New function, taken
from start of diagnostic_finish.
(diagnostic_initialize): Initialize final_cb to
default_diagnostic_final_cb.
(diagnostic_finish): Move "being treated as errors" messages to
default_diagnostic_final_cb.  Call any final_cb.
(default_diagnostic_finalizer): Add diagnostic_t param.
(diagnostic_report_diagnostic): Pass "orig_diag_kind" to
diagnostic_finalizer callback.
* diagnostic.h (enum diagnostics_output_format): New enum.
(diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
(struct diagnostic_context): Add "final_cb".
(default_diagnostic_finalizer): Add diagnostic_t param.
(diagnostic_output_format_init): New decl.
* doc/invoke.texi (-fdiagnostics-format): New option.
* dwarf2out.c (gen_producer_string): Ignore
OPT_fdiagnostics_format_.
* gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
* lto-wrapper.c (append_diag_options): Ignore it.
* opts.c (common_handle_option): Handle it.

gcc/c-family/ChangeLog:
PR other/19165
* c-opts.c (c_diagnostic_finalizer): Add diagnostic_t param.

gcc/fortran/ChangeLog:
PR other/19165
* error.c (gfc_diagnostic_finalizer): Add diagnostic_t param.

gcc/jit/ChangeLog:
PR other/19165
* dummy-frontend.c (jit_begin_diagnostic): Add diagnostic_t param.

gcc/testsuite/ChangeLog:
PR other/19165
* c-c++-common/diagnostic-format-json-1.c: New test.
* c-c++-common/diagnostic-format-json-2.c: New test.
* c-c++-common/diagnostic-format-json-3.c: New test.
* c-c++-common/diagnostic-format-json-4.c: New test.
* c-c++-common/diagnostic-format-json-5.c: New test.
* gcc.dg/plugin/diagnostic_plugin_test_show_locus.c
(custom_diagnostic_finalizer): Add diagnostic_t param.
* gcc.dg/plugin/location_overflow_plugin.c
(verify_unpacked_ranges): Likewise.
(verify_no_columns): Likewise.
* gfortran.dg/diagnostic-format-json-1.F90: New test.
* gfortran.dg/diagnostic-format-json-2.F90: New test.
* gfortran.dg/diagnostic-format-json-3.F90: New test.

From-SVN: r266186

graphite: add missing dump_enabled_p checks (PR tree-optimization/88015)

gcc/ChangeLog:
PR tree-optimization/88015
* graphite-isl-ast-to-gimple.c
(translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
for dump_enabled_p.
* graphite-sese-to-poly.c (build_poly_scop): Likewise.

From-SVN: r266184

re PR tree-optimization/88029 (ICE in execute_todo, at passes.c:1974)

2018-11-15  Richard Biener  <rguenther@suse.de>

        PR middle-end/88029
        * gimple.c (gimple_call_flags): Union flags from decl, type
        and call fntype.
        * trans-mem.c (is_tm_pure_call): Simplify.

        * gcc.dg/tree-ssa/pr88029.c: New testcase.

From-SVN: r266183

re PR tree-optimization/88031 (ICE in vectorizable_reduction, at tree-vect-loop.c:6953)

2018-11-15  Richard Biener  <rguenther@suse.de>

PR tree-optimization/88031
* tree-vect-loop.c (vectorizable_reduction): Move check
for multiple types earlier so we get the expected dump.
Simplify calls to vectorizable_condition.
* tree-vect-stmts.h (vectorizable_condition): Update prototype.
* tree-vect-stmts.c (vectorizable_condition): Instead of
reduc_def and reduc_index take just a flag.  Simplify
code-generation now that we can rely on the defs being set up.
(vectorizable_comparison): Remove unused argument.

* gcc.dg/pr88031.c: New testcase.

From-SVN: r266182

[C++ DR 2336] Clean up synth walkers first

https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01376.html
* cp-tree.h (enum special_function_kind): Reorder and comment.
* method.c (SFK_CTOR_P, SFK_DTOR_P, SFK_ASSIGN_P, SFK_COPY_P)
(SFK_MOVE_P): New predicates.
(walk_field_subobs, synthesized_method_base_walk): Drop
copy_arg_p, move_p, assign_p args.  Use new SFK predicates.  Order
parameters consistently.
(synthesized_method_walk): Drop ctor_p, copy_arg_p, move_p,
assign_p calculations.  Use new SFK predicates.  Adjust calls to
worker functions.

From-SVN: r266180

Fix spaces in PR62178 test

Fix spaces in scan assembler tests.

    testsuite/
* gcc.target/aarch64/pr62178.c: Fix spaces.

From-SVN: r266179

misc.c (gnat_init_gcc_eh): Do not override the switch -fnon-call-exceptions passed on the command line in...

* gcc-interface/misc.c (gnat_init_gcc_eh): Do not override the switch
-fnon-call-exceptions passed on the command line in -gnatp mode.

From-SVN: r266176

re PR tree-optimization/88030 (ICE in calc_dfs_tree, at dominance.c:458)

2018-11-15  Richard Biener  <rguenther@suse.de>

PR tree-optimization/88030
* tree-complex.c (need_eh_cleanup): New global.
(update_complex_assignment): Mark blocks that need EH update.
(expand_complex_comparison): Likewise.
(tree_lower_complex): Allocate and deallocate need_eh_cleanup,
perform EH cleanup and schedule CFG cleanup if that did anything.

* gcc.dg/tsan/pr88030.c: New testcase.

From-SVN: r266175

re PR rtl-optimization/88018 (ICE in insert_insn_on_edge at cfgrtl.c:1952 since r255066)

PR rtl-optimization/88018
* cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
on the presence of fallthru edge, rather than if it is a USE or not.

* g++.dg/tsan/pr88018.C: New test.

From-SVN: r266174

re PR tree-optimization/87917 (ICE in initialize_matrix_A at gcc/tree-data-ref.c:3150)

2018-11-15  Richard Biener  <rguenther@suse.de>

PR middle-end/87917
* tree-data-ref.c (analyze_miv_subscript): Guard calls to
analyze_subscript_affine_affine properly.

* gcc.dg/tree-ssa/pr87917.c: New testcase.

From-SVN: r266173

csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.

2018-11-15  Xianmiao Qu  <xianmiao_qu@c-sky.com>

gcc/
* config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
(GLIBC_DYNAMIC_LINKER): Define.
(LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.

libgcc/
* config/csky/linux-unwind.h: Fix coding style.

From-SVN: r266172

re PR tree-optimization/84648 (Missed optimization : loop not removed.)

PR tree-optimization/84648
* tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
(number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
by calling adjust_cond_for_loop_until_wrap.

gcc/testsuite
* gcc.dg/tree-ssa/pr84648.c: New test.
* gcc.dg/pr68317.c: Add warning check on overflow.

From-SVN: r266171

re PR other/56334 (__attribute__((aligned)) documentation is misleading)

2018-11-15  Sandra Loosemore  <sandra@codesourcery.com>

PR other/56334

gcc/
* doc/extend.texi (Common Function Attributes): Clarify linker
restrictions on "aligned" attribute.
(Common Variable Attributes): Likewise.  Mention that linker
restrictions don't apply to stack-allocated variables.

From-SVN: r266170

gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT as an lvalue.

        * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
        as an lvalue.

From-SVN: r266169

re PR lto/55102 (The options -flto and -On do not behave as described in GCC docs)

2018-11-14  Sandra Loosemore  <sandra@codesourcery.com>

PR lto/55102
PR lto/56700

gcc/
* doc/invoke.texi (Optimize Options): Remove bad example about
interaction between -flto and -O.  Replace it with a note that
you need to compile with -O and not just link.  Copy-edit -flto
discussion to reduce verbiage and improve flow.

From-SVN: r266168

Daily bump.

From-SVN: r266167

Optimize pool resource allocation

A recent change caused a performance regression. This restores the
previous performance and adds a performance test.

* scripts/check_performance: Allow tests to choose a -std flag.
* src/c++17/memory_resource.cc (bitset::get_first_unset()): Use local
variables of the right types. Call update_next_word() unconditionally.
* testsuite/20_util/unsynchronized_pool_resource/cons.cc: New test.
* testsuite/performance/20_util/memory_resource/pools.cc: New test.
* testsuite/util/testsuite_performance.h (time_counter): Allow
timer to be restarted.

From-SVN: r266164

Fix test that does undefined shifts greater than width of size_t

* testsuite/20_util/unsynchronized_pool_resource/allocate.cc: Fix
test for 32-bit targets. Test additional allocation sizes.

From-SVN: r266163

re PR middle-end/59658 (Document -f* flags enabled by -Og)

2018-11-14  Sandra Loosemore  <sandra@codesourcery.com>

PR middle-end/59658

gcc/
* doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
also suppress many optimizations.  Alphabetize option lists for
-O1, -O2, and -Os.  Add list of options disabled with -Og, and
correct documentation for those options to say that.
* opts.c (default_options_table): Sort table by level and option
name, to make it easier to correlate to the manual.

From-SVN: r266162

i386.c (ix86_print_operand_address_as): Simplify printing of the names of segment registers.

* config/i386/i386.c (ix86_print_operand_address_as): Simplify
printing of the names of segment registers.

From-SVN: r266161

[debug/88006] -fdebug-types-section gives undefined ref

https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01280.html
PR debug/88006
PR debug/87462
* dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
type list.

* g++.dg/debug/dwarf2/pr87462.C: New.
* g++.dg/debug/dwarf2/pr88006.C: New.

From-SVN: r266158

Add missing ZLIBINC to CFLAGS-optinfo-emit-json.o

gcc/ChangeLog:
* Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).

From-SVN: r266156

parser.c (make_id_declarator): Add location_t parameter.

/cp
2018-11-14  Paolo Carlini  <paolo.carlini@oracle.com>

* parser.c (make_id_declarator): Add location_t parameter.
(cp_parser_lambda_declarator_opt): Adjust call.
(cp_parser_decomposition_declaration): Likewise.
(cp_parser_alias_declaration): Likewise.
(cp_parser_direct_declarator): Likewise.
(cp_parser_member_declaration): Likewise.
(cp_parser_objc_class_ivars): Likewise.
* decl.c (grokdeclarator): Use declarator->id_loc in two error
messages.

/testsuite
2018-11-14  Paolo Carlini  <paolo.carlini@oracle.com>

* g++.dg/cpp0x/nsdmi-union6.C: Test locations too.
* g++.dg/cpp0x/nsdmi6.C: Likewise.
* g++.dg/ext/flexary4.C: Likewise.
* g++.dg/ext/flexary9.C: Likewise.
* g++.dg/other/incomplete2.C: Likewise.
* g++.dg/parse/friend12.C: Likewise.

From-SVN: r266155

Handle libphobos in contrib/gcc_update

* gcc_update (files_and_dependencies): Handle libphobos.

From-SVN: r266154

P1236R1 - Signed integers are two's complement

P1236R1 - Signed integers are two's complement
gcc/cp/
* constexpr.c (cxx_eval_check_shift_p): Disable the signed LSHIFT_EXPR
checks for c++2a.
gcc/c-family/
* c-warn.c (maybe_warn_shift_overflow): Don't warn for c++2a.
* c-ubsan.c (ubsan_instrument_shift): Make signed shifts
with in-range second operand well defined for -std=c++2a.
gcc/
* doc/invoke.texi (Wshift-overflow): Adjust documentation for
c++2a.
gcc/testsuite/
* g++.dg/cpp2a/constexpr-shift1.C: New test.
* g++.dg/warn/permissive-1.C (enum A, enum D): Don't expect
diagnostics here for c++2a.
* g++.dg/cpp0x/constexpr-shift1.C (fn3, i3, fn4, i4): Don't expect
diagnostics here for c++2a.
* g++.dg/cpp0x/constexpr-60049.C (f3, x3, y3): Likewise.
* g++.dg/ubsan/cxx11-shift-1.C (main): Add some further tests.
* g++.dg/ubsan/cxx11-shift-2.C (main): Likewise.
* g++.dg/ubsan/cxx2a-shift-1.C: New test.
* g++.dg/ubsan/cxx2a-shift-2.C: New test.

From-SVN: r266153

re PR bootstrap/86739 (Bootstrap broken with host GCC 4.1.2)

PR bootstrap/86739
* hash-map.h (hash_map::iterator::reference_pair): New class.
(hash_map::iterator::operator*): Return it rather than std::pair.

From-SVN: r266152

optabs.c (expand_binop): Pass INT_MODE to operand_subword_force iff the operand is a constant.

* optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
iff the operand is a constant.

* gcc.c-torture/compile/20181114.c: New test.

2018-11-14  Ilya Leoshkevich  <iii@linux.ibm.com>

From-SVN: r266151

* gimple-ssa-evrp-analyze.c
(evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
ignore_equivs_equal_p to equal_p.
* ipa-cp.c (meet_with_1): Use equal_p instead of
ignore_equivs_equal_p.
* ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
* tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
(value_range::operator==): Remove.
(value_range::operator!=): Remove.
(vrp_prop::visit_stmt): Use equal_p.
* tree-vrp.h (value_range): Remove operator==, operator!=,
ignore_equivs_equal_p.
* vr-values.c (update_value_range): Use equal_p.

From-SVN: r266150

Add comment

From-SVN: r266149

re PR middle-end/86575 (-Wimplicit-fallthrough affects code generation)

Fix PR middle-end/86575

PR middle-end/86575
* gimplify.c (collect_fallthrough_labels): Add new argument,
return location via that, don't modify statements.
(warn_implicit_fallthrough_r): Adjust call, don't use
statement location directly.

From-SVN: r266148

re PR tree-optimization/87985 (Compile-time and memory hog w/ -O1 -ftree-slp-vectorize)

2018-11-14  Richard Biener  <rguenther@suse.de>

PR middle-end/87985
* tree-data-ref.c (split_constant_offset): Add wrapper
allocating a cache hash-map.
(split_constant_offset_1): Cache results of expanding
expressions from SSA def stmts.

* gcc.dg/pr87985.c: New testcase.

From-SVN: r266147

Add missing dir to create_testsuite_files script

* scripts/create_testsuite_files: Add special_functions to the list
of directories to search. Add comment referring to conformance.exp.
* testsuite/libstdc++-dg/conformance.exp: Add comment referring
to create_testsuite_files.

From-SVN: r266146

[PATCH 2/2] S/390: Fix expectation in mrecord-mcount test for 31-bit
 mode

The emitted address is .long, not .quad, in that case.

gcc/testsuite/ChangeLog:

2018-11-14  Ilya Leoshkevich  <iii@linux.ibm.com>

* gcc.target/s390/mrecord-mcount.c (profileme): Expect .long in
31-bit mode.

From-SVN: r266145

[PATCH 1/2] S/390: Disable 3 global-array-* tests for 31-bit mode

These tests rely on larl->movdi merge, which is not implemented for
31-bit mode.

gcc/testsuite/ChangeLog:

2018-11-14  Ilya Leoshkevich  <iii@linux.ibm.com>

* gcc.target/s390/global-array-almost-huge-element.c: Run only
in 64-bit mode.
* gcc.target/s390/global-array-almost-negative-huge-element.c:
Likewise.
* gcc.target/s390/global-array-even-element.c: Likewise.

From-SVN: r266144

re PR testsuite/88021 (aarch64 Busy hang running testcase pr60183.c since revision 265914)

2018-11-14  Richard Biener  <rguenther@suse.de>

PR middle-end/88021
* tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
to lambda_int.
(lambda_vector_mult_const): Likewise.
(lambda_matrix_right_hermite): Use lambda_int temporaries.

From-SVN: r266143

Simplify floating point comparisons

This patch implements some of the optimizations discussed in PR71026.

Simplify (C / x >= 0.0) into x >= 0.0 with -funsafe-math-optimizations
(since C / x can underflow to zero if x is huge, it's not safe otherwise).
If C is negative the comparison is reversed.

Simplify (x * C1) > C2 into x > (C2 / C1) with -funsafe-math-optimizations.
If C1 is negative the comparison is reversed.

    gcc/
PR 71026/tree-optimization
* match.pd: Simplify floating point comparisons.

    gcc/testsuite/
PR 71026/tree-optimization
* gcc.dg/div-cmp-1.c: New test.
* gcc.dg/div-cmp-2.c: New test.

Co-Authored-By: Jackson Woodruff <jackson.woodruff@arm.com>
From-SVN: r266142

re PR other/88007 (ICE in gt_ggc_m_S, at ggc-page.c:1474)

PR other/88007
* c-common.c (parse_optimize_options): Allocate option string from
opts_obstack rather than as GC memory.  Move the allocation after
warning for invalid option.

* gcc.dg/pr88007.c: New test.

From-SVN: r266141

re PR rtl-optimization/87817 (gcc.target/i386/bmi2-bzhi-2.c execution test)

PR rtl-optimization/87817
* config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
last argument has low 8 bits clear, fold to 0.

* gcc.target/i386/bmi2-bzhi-3.c (main): Add a couple of new tests.

From-SVN: r266140

[AArch64] Fix PR62178 testcase failures

The testcase for PR62178 has been failing for a while due to the pass
conditions being too tight, resulting in failures with -mcmodel=tiny:

ldr q2, [x0], 124
ld1r {v1.4s}, [x1], 4
cmp x0, x2
mla v0.4s, v2.4s, v1.4s
bne .L7

-mcmodel=small generates the slightly different:

ldr q1, [x0], 124
ldr s2, [x1, 4]!
cmp x0, x2
mla v0.4s, v1.4s, v2.s[0]
bne .L7

This is due to Combine merging a DUP instruction with either a load
or MLA - we can't force it to prefer one over the other.  However the
generated vector loop is fast either way since it generates MLA and
merges the DUP either with a load or MLA.  So relax the conditions
slightly and check we still generate MLA and there is no DUP or FMOV.

The testcase now passes - committed as obvious.

    testsuite/
* gcc.target/aarch64/pr62178.c: Relax scan-assembler checks.

From-SVN: r266139

fix Darwin bootstrap.

gcc/

* tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
MAX_OFILE_ALIGNMENT as needed.

From-SVN: r266138