Alexandre Oliva [Tue, 21 May 2019 17:09:13 +0000 (17:09 +0000)]
preserve more debug stmts in gimple jump threading
Gimple jump threading does not duplicate forwarder blocks that might
be present before or after the second copied block. This silently
drops debug binds and markers that might be present in them. This
patch attempts to preserve them.
For blocks after either copied block, we attempt to append debug stmts
to the copied block, if it does not end with a block-ending stmt.
Failing that, for blocks between both copied blocks, we prepend its
debug stmts to the copy of the second block.
If everything fails, we still drop debug stmts on the floor, though
preexisting code consolidates debug binds in the block that threading
flows into, so only markers are really lost. We can't do much better
than that without conditional binds and markers, or debug stmts in
edges, or somesuch.
If we append debug stmts to a reusable template block, we copy it
after splitting out the debug stmts, and before putting them back.
for gcc/ChangeLog
* tree-ssa-threadupdate.c (struct ssa_local_info_t): Add
field template_last_to_copy.
(ssa_create_duplicates): Set it, and use it. Attempt to
preserve more debug stmts.
Jonathan Wakely [Tue, 21 May 2019 13:50:41 +0000 (14:50 +0100)]
PR libstdc++/90252 fix effective-target check for TBB
PR libstdc++/90252
* testsuite/lib/libstdc++.exp (check_effective_target_tbb-backend):
Use "additional_flags" to pass -ltbb to v3_target_compile command.
Use check_v3_target_prop_cached to cache the result of the test.
Richard Biener [Tue, 21 May 2019 12:01:00 +0000 (12:01 +0000)]
re PR tree-optimization/90510 (Unnecessary permutation)
2019-05-21 Richard Biener <rguenther@suse.de>
PR middle-end/90510
* fold-const.c (fold_read_from_vector): New function.
* fold-const.h (fold_read_from_vector): Declare.
* match.pd (VEC_PERM_EXPR): Build BIT_INSERT_EXPRs for
single-element insert permutations. Canonicalize selector
further and fix issue with last commit.
* tree-cfg.h (split_critical_edges): Add for_edge_insertion_p
parameter with default value false to declaration.
(split_edges_for_insertion): New inline function. Wrapper for
split_critical_edges with for_edge_insertion_p = true.
* tree-cfg.c (split_critical_edges): Don't split non-critical
edges if for_edge_insertion_p is false. Fix whitespace.
* tree-ssa-pre.c (pass_pre::execute): Call
split_edges_for_insertion instead of split_critical_edges.
* gcc/tree-ssa-tail-merge.c (tail_merge_optimize): Ditto.
* gcc/tree-ssa-sink.c (pass_sink_code::execute): Ditto.
(pass_data_sink_code): Update function name in the comment.
* tree-ssa-uninit.c (value_sat_pred_p): This new function is a wrapper
around is_value_included_in that knows how to handle BIT_AND_EXPR.
(is_pred_expr_subset_of): Use the new function. Handle more cases where
code1 == EQ_EXPR and where code1 == BIT_AND_EXPR and thus fix some false
positives.
testsuite/
* gcc.dg/uninit-28-gimple.c: New test.
* gcc.dg/uninit-29-gimple.c: New test.
* gcc.dg/uninit-30-gimple.c: New test.
* gcc.dg/uninit-31-gimple.c: New test.
c-family/
* c-ada-spec.h (enum cpp_operation): Add IS_ASSIGNMENT_OPERATOR.
* c-ada-spec.c (print_assignment_operator): New function.
(dump_ada_declaration) <FUNCTION_DECL>: Call it do dump explicit copy
assignment operators declared as methods and filter out the others.
cp/
* decl2.c (cpp_check) <IS_ASSIGNMENT_OPERATOR>: New case.
Janne Blomqvist [Tue, 21 May 2019 06:59:16 +0000 (09:59 +0300)]
Convert contrib/mklog script to Python 3
Upstream will drop support for Python 2.x on January 1, 2020. This
patch converts the contrib/mklog script to Python 3. The process for
the conversion was as follows.
- Use the futurize tool (https://python-future.org ) to apply the
print_with_import, except, and dict transformations.
- Remove the "from __future__ import print_function".
- Change the shebang line to search for python3 in the environment.
- Modify the run() function to return a str instead of bytes.
Martin Liska [Tue, 21 May 2019 06:55:56 +0000 (08:55 +0200)]
fix diagnostic quoting/spelling in rs6000
2019-05-21 Martin Liska <mliska@suse.cz>
* config/rs6000/driver-rs6000.c (elf_platform): Do not use
an extra newline.
* config/rs6000/rs6000-c.c (rs6000_pragma_longcal): Wrap pragma in %<%>.
(altivec_resolve_overloaded_builtin): Likewise for vec_lvsl and
vec_lvsr.
* config/rs6000/rs6000.c (rs6000_option_override_internal):
Quote a C type.
(rs6000_function_arg): Likewise.
(rs6000_expand_set_fpscr_drn_builtin): Remove trailing dot.
(rs6000_expand_ternop_builtin): Use interval syntax.
(get_element_number): Likewise.
(altivec_expand_builtin): Likewise.
(rs6000_get_function_versions_dispatcher): Quote target_clones.
Fix test-suite.
2019-05-21 Martin Liska <mliska@suse.cz>
* gcc.target/powerpc/ppc64-abi-warn-1.c: Wrap a type.
* gcc.target/powerpc/pr80315-1.c: Use new interval format.
* gcc.target/powerpc/pr80315-2.c: Likewise.
* gcc.target/powerpc/pr80315-3.c: Likewise.
* gcc.target/powerpc/pr80315-4.c: Likewise.
* gcc.target/powerpc/warn-lvsl-lvsr.c: Wrap builtin names.
Uros Bizjak [Mon, 20 May 2019 22:24:38 +0000 (00:24 +0200)]
re PR testsuite/90503 (gcc.target/i386/pr22076.c FAILs)
PR testsuite/90503
* gcc.target/i386/pr22076.c (dg-options): Add -mno-sse2.
Remove -flax-vector-conversions.
(dg-additional-options): Remove.
(test): Change to void. Declare m0 and m1 as __m64 and
cast initializer in a proper way. Do not return result.
(dg-final): Scan for 2 instances of movq.
Mark Eggleston [Mon, 20 May 2019 21:49:33 +0000 (15:49 -0600)]
gfortran.texi: Remove reference to the ASSIGN statement...
* gfortran.texi: Remove reference to the ASSIGN statement, capitalise
complex, state that padding is with spaces and modify the Hollerith
constant examples.
Jan Hubicka [Mon, 20 May 2019 13:37:11 +0000 (15:37 +0200)]
tree-ssa-alias.c (refs_may_alias_p_2): Break out from ...
* tree-ssa-alias.c (refs_may_alias_p_2): Break out from ...
(refs_may_alias_p_1): ... here; update stats.
(refs_may_alias_p): Do not update stats here.
Iain Sandoe [Mon, 20 May 2019 12:28:18 +0000 (12:28 +0000)]
darwin, testsuite - fix PR58321
Darwin doesn't emit a .file directive by default
and one of the scan-asm hits for ELF targets comes from
this directive. Adjust for Darwin and explain.
2019-05-20 Iain Sandoe <iain@sandoe.co.uk>
PR testsuite/58321
* gcc.target/i386/memcpy-strategy-3.c: Adjust count for Darwin and
add a comment as to the reason for the difference.
* gcc.target/i386/memset-strategy-1.c: Likewise.
Richard Biener [Mon, 20 May 2019 12:02:35 +0000 (12:02 +0000)]
tree-ssa-structalias.c (find_func_aliases): POINTER_DIFF_EXPR doesn't produce pointers.
2019-05-20 Richard Biener <rguenther@suse.de>
* tree-ssa-structalias.c (find_func_aliases): POINTER_DIFF_EXPR
doesn't produce pointers.
{TRUNC,CEIL,FLOOR,ROUND,EXACT}_{DIV,MOD}_EXPR points to what
the first operand points to.
Jan Hubicka [Mon, 20 May 2019 12:01:40 +0000 (14:01 +0200)]
tree-ssa-alias.c (compare_sizes): New function.
* tree-ssa-alias.c (compare_sizes): New function.
(sompare_type_sizes): New function
(aliasing_component_refs_p): Use it.
(indirect_ref_may_alias_decl_p): Likewise.
Jonathan Wakely [Mon, 20 May 2019 11:32:51 +0000 (12:32 +0100)]
PR c++/90532 Ensure __is_constructible(T[]) is false
An array of an unknown bound is an incomplete type, so no object of such
a type can be constructed. This means __is_constructible should always
be false for an array of unknown bound.
This patch also changes the std::is_default_constructible trait to use
std::is_constructible, which now gives the right answer for arrays of
unknown bound.
gcc/cp:
PR c++/90532 Ensure __is_constructible(T[]) is false
* method.c (is_xible_helper): Return error_mark_node for construction
of an array of unknown bound.
gcc/testsuite:
PR c++/90532 Ensure __is_constructible(T[]) is false
* g++.dg/ext/90532.C: New test.
Pádraig Brady [Mon, 20 May 2019 11:15:03 +0000 (11:15 +0000)]
std::allocator::deallocate support sized-deallocation
Pass the size to the allocator so that it may optimize deallocation.
This was seen to significantly reduce the work required in jemalloc,
with about 40% reduction in CPU cycles in the free path.
Note jemalloc >= 5.2 is required to fix a crash with 0 sizes.
2019-05-20 Pádraig Brady <pbrady@fb.com>
* libstdc++-v3/include/ext/new_allocator.h (deallocate): Pass the size
to the deallocator with -fsized-deallocation.
Jakub Jelinek [Mon, 20 May 2019 09:49:07 +0000 (11:49 +0200)]
cfgloop.h (struct loop): Add simdlen member.
* cfgloop.h (struct loop): Add simdlen member.
* cfgloopmanip.c (copy_loop_info): Copy simdlen as well.
* omp-expand.c (expand_omp_simd): Set it if simdlen clause is present.
* tree-vect-loop.c (vect_analyze_loop): Pass loop->simdlen != 0
as new argument to autovectorize_vector_sizes target hook. If
loop->simdlen, pick up vector size where the vectorization factor
is equal to loop->simd, and if there is none, fall back to the first
successful one.
(vect_transform_loop): Adjust autovectorize_vector_sizes target hook
caller.
* omp-low.c (omp_clause_aligned_alignment): Likewise.
* omp-general.c (omp_max_vf): Likewise.
* optabs-query.c (can_vec_mask_load_store_p): Likewise.
* tree-vect-slp.c (vect_slp_bb): Likewise.
* target.def (autovectorize_vector_sizes): Add ALL argument and
document it.
* doc/tm.texi: Adjust documentation.
* targhooks.c (default_autovectorize_vector_sizes): Add bool argument.
* targhooks.h (default_autovectorize_vector_sizes): Likewise.
* config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes): Add
bool argument.
* config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
* config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
* config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
* config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise. If
true and TARGET_AVX512F or TARGET_AVX, push 3 or 2 sizes even if
preferred vector size is not 512-bit or 256-bit, just put those
unpreferred ones last.
Martin Liska [Mon, 20 May 2019 07:55:00 +0000 (09:55 +0200)]
Come up with hook libc_has_fast_function (PR middle-end/90263).
2019-05-20 Martin Liska <mliska@suse.cz>
PR middle-end/90263
* builtins.c (expand_builtin_memory_copy_args): When having a
target with fast mempcpy implementation do now use memcpy.
* config/i386/i386.c (ix86_libc_has_fast_function): New.
(TARGET_LIBC_HAS_FAST_FUNCTION): Likewise.
* doc/tm.texi: Likewise.
* doc/tm.texi.in: Likewise.
* target.def:
* expr.c (emit_block_move_hints): Add 2 new arguments.
* expr.h (emit_block_move_hints): Bail out when libcall
to memcpy would be used.
2019-05-20 Martin Liska <mliska@suse.cz>
PR middle-end/90263
* gcc.c-torture/compile/pr90263.c: New test.
* lib/target-supports.exp: Add check_effective_target_glibc.
This replaces the "wu" constraint by "v", with isa "p8v". Or, in most
cases, use "wa", since the instructions allow all VSX registers, and it
does not change how GCC behaves, so it is clearer that way.
This also delete the unused <Fa>.
* config/rs6000/constraints.md (define_register_constraint "wu"):
Delete.
* config/rs6000/rs6000.h (enum r6000_reg_class_enum): Delete
RS6000_CONSTRAINT_wu.
* config/rs6000/rs6000.c (rs6000_debug_reg_global): Adjust.
(rs6000_init_hard_regno_mode_ok): Adjust.
* config/rs6000/rs6000.md: Replace "wu" constraint by "v" or "wa",
both with "p8v".
(define_mode_attr Fa): Delete.
* config/rs6000/vsx.md: Ditto.
* doc/md.texi (Machine Constraints): Adjust.
Janne Blomqvist [Sun, 19 May 2019 19:38:11 +0000 (22:38 +0300)]
libfortran/90038 Reap dead children when wait=.false.
When using posix_spawn or fork to launch a child process, the parent
needs to wait for the child, otherwise the dead child is left as a
zombie process. For this purpose one can install a signal handler for
SIGCHLD.
2019-05-19 Janne Blomqvist <jb@gcc.gnu.org>
PR libfortran/90038
* intrinsics/execute_command_line (sigchld_handler): New function.
(execute_command_line): Install handler for SIGCHLD.
* configure.ac: Check for presence of sigaction and waitpid.
* config.h.in: Regenerated.
* configure: Regenerated.
Thomas Koenig [Sun, 19 May 2019 10:21:06 +0000 (10:21 +0000)]
re PR fortran/88821 (Inline packing of non-contiguous arguments)
2019-05-19 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/88821
* expr.c (gfc_is_simply_contiguous): Return true for
an EXPR_ARRAY.
* trans-array.c (is_pointer): New function.
(gfc_conv_array_parameter): Call gfc_conv_subref_array_arg
when not optimizing and not optimizing for size if the formal
arg is passed by reference.
* trans-expr.c (gfc_conv_subref_array_arg): Add arguments
fsym, proc_name and sym. Add run-time warning for temporary
array creation. Wrap argument if passing on an optional
argument to an optional argument.
* trans.h (gfc_conv_subref_array_arg): Add optional arguments
fsym, proc_name and sym to prototype.
2019-05-19 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/88821
* gfortran.dg/alloc_comp_auto_array_3.f90: Add -O0 to dg-options
to make sure the test for internal_pack is retained.
* gfortran.dg/assumed_type_2.f90: Split compile and run time
tests into this and
* gfortran.dg/assumed_type_2a.f90: New file.
* gfortran.dg/c_loc_test_22.f90: Likewise.
* gfortran.dg/contiguous_3.f90: Likewise.
* gfortran.dg/internal_pack_11.f90: Likewise.
* gfortran.dg/internal_pack_12.f90: Likewise.
* gfortran.dg/internal_pack_16.f90: Likewise.
* gfortran.dg/internal_pack_17.f90: Likewise.
* gfortran.dg/internal_pack_18.f90: Likewise.
* gfortran.dg/internal_pack_4.f90: Likewise.
* gfortran.dg/internal_pack_5.f90: Add -O0 to dg-options
to make sure the test for internal_pack is retained.
* gfortran.dg/internal_pack_6.f90: Split compile and run time
tests into this and
* gfortran.dg/internal_pack_6a.f90: New file.
* gfortran.dg/internal_pack_8.f90: Likewise.
* gfortran.dg/missing_optional_dummy_6: Split compile and run time
tests into this and
* gfortran.dg/missing_optional_dummy_6a.f90: New file.
* gfortran.dg/no_arg_check_2.f90: Split compile and run time tests
into this and
* gfortran.dg/no_arg_check_2a.f90: New file.
* gfortran.dg/typebound_assignment_5.f90: Split compile and run time
tests into this and
* gfortran.dg/typebound_assignment_5a.f90: New file.
* gfortran.dg/typebound_assignment_6.f90: Split compile and run time
tests into this and
* gfortran.dg/typebound_assignment_6a.f90: New file.
* gfortran.dg/internal_pack_19.f90: New file.
* gfortran.dg/internal_pack_20.f90: New file.
* gfortran.dg/internal_pack_21.f90: New file.
Thomas Koenig [Sun, 19 May 2019 08:22:41 +0000 (08:22 +0000)]
re PR fortran/90329 (Incompatibility between gfortran and C lapack calls)
2019-05-19 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/90329
* invoke.texi: Document -fbroken-callers.
* lang.opt: Add -fbroken-callers.
* trans-decl.c (create_function_arglist): Only set
DECL_HIDDEN_STRING_LENGTH if flag_broken_callers is set.
Darwin is able to use two runtimes for objective-c; the
default is its native "NeXT" runtime, but also it can build
code using the "gnu-runtime". In order to do this, we have to
be able to find the gnu-runtime headers (which are installed
into the compiler's tree).
The process to do this is erroneously prepending the sysroot
to this when a sysroot is in force. The gnu-runtime headers have
never been installed in a Darwin (macOS) SDK so we must make
sure that they are found local to the compiler.
gcc/
2019-05-18 Iain Sandoe <iain@sandoe.co.uk>
* config/darwin-c.c (darwin_register_objc_includes): Do not
prepend the sysroot when building gnu-runtime header search
paths.
Iain Sandoe [Sat, 18 May 2019 08:27:24 +0000 (08:27 +0000)]
objective-c - add instancetype.
The instancetype has been added as a typedef alias to id
in order to allow diagnosis of cases where a class is used
or returned where an instance is expected.
This adds the typedef, and tests that we can parse it.
It doesn't alter the diagnostics yet.
gcc/objc/
2019-05-18 Iain Sandoe <iain@sandoe.co.uk>
* objc/objc-act.h (OCTI_INSTANCE_TYPE, OCTI_INSTANCETYPE_NAME): New.
(objc_global_trees): Add instance type and name.
(INSTANCE_TYPEDEF_NAME): New.
* objc/objc-act.c (synth_module_prologue): Build decls for
objc_instancetype_type and objc_instancetype_name.
Iain Sandoe [Sat, 18 May 2019 08:07:56 +0000 (08:07 +0000)]
darwin - use switch_to_section instead of asm output.
We have been emitting two section swiches in the Darwin's
file end function. This means that varasm is not updated
which could matter if we elect to reorder some of the file
end operations in support of LTO actions.
2019-05-18 Iain Sandoe <iain@sandoe.co.uk>
* config/darwin.c (darwin_file_end): Use switch_to_section ()
instead of direct output of the asm.
Jonathan Wakely [Fri, 17 May 2019 23:08:00 +0000 (00:08 +0100)]
PR libstdc++/90520 adjust Xmethod for recent unique_ptr changes
PR libstdc++/90520
* python/libstdcxx/v6/printers.py (UniquePointerPrinter.__init__):
Raise exception if unique_ptr tuple member has unknown structure.
* python/libstdcxx/v6/xmethods.py (UniquePtrGetWorker.__call__):
Adjust worker to support new __uniq_ptr_data base class. Do not
assume field called _M_head_impl is the first tuple element.
(different because calls_eh_return can be true for sibcalls). This is
a bugfix. The code was never exercised.
One place in the epilogue still uses crtl->calls_eh_return. If that
is changed the prologue has to have a corresponding change, and the
emit_prologue function does not have an epilogue_type parameter, so
bail on changing this for now. We might want to do this (saving the
CR fields to separate stack slots) always, not just for functions
calling eh_return, but that will require more investigation.
* config/rs6000/rs6000.c (restore_saved_cr): Change a boolean
argument to be type bool (was int before).
(rs6000_emit_epilogue): Simplify some code. Declare some variables
at first use. Use type bool for some variables. Fix a theoretical
eh_return bug for svr4.
This adds the "enabled" attribute to the rs6000 backend. It uses the
(new) "isa" attribute to automatically select which instruction
alternatives should be enabled.
For now it allows isa strings of "p5", "p6", "p7", meaning the
instructions introduced on that CPU, not requiring vectors; and "p7v",
"p8v", "p9v" for the same, but with vectors.
These are currently mapped to TARGET_POPCNTB, TARGET_CMPB,
TARGET_POPCNTD, TARGET_VSX, TARGET_P8_VECTOR, and TARGET_P9_VECTOR;
that will change to something a bit saner later.
* config/rs6000/rs6000.md (isa): New attribute.
(enabled): New attribute.