aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double default to IBM.

[gcc]
2017-10-26  Michael Meissner  <meissner@linux.vnet.ibm.com>

* config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
default to IBM.
* config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
* config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
warning to rs6000.c.  Remove the Undocumented flag, since it has
been documented.
(-mabi=ibmlongdouble): Likewise.
* config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
already set, set the default format for long double.
(rs6000_debug_reg_global): Print whether long double is IBM or
IEEE.
(rs6000_option_override_internal): Rework setting long double
format.  Only warn if the user is changing the long double default
and they did not use -Wno-psabi.
* doc/invoke.texi (PowerPC options): Update the documentation for
-mabi=ieeelongdouble and -mabi=ibmlongdouble.

From-SVN: r254116

Add wider_subreg_mode helper functions

This patch adds helper functions that say which of the two modes
involved in a subreg is the larger, preferring the outer mode in
the event of a tie.  It also converts IRA and reload to track modes
instead of byte sizes, since this is slightly more convenient when
variable-sized modes are added later.

2017-10-26  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* rtl.h (wider_subreg_mode): New function.
* ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
rather than an unsigned int *.
* ira-color.c (regno_max_ref_width): Replace with...
(regno_max_ref_mode): ...this new variable.
(coalesced_pseudo_reg_slot_compare): Update accordingly.
Use wider_subreg_mode.
(ira_sort_regnos_for_alter_reg): Likewise.  Take a machine_mode *
rather than an unsigned int *.
* lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
(process_alt_operands): Likewise.
(invariant_p): Likewise.
* lra-spills.c (assign_mem_slot): Likewise.
(add_pseudo_to_slot): Likewise.
* lra.c (collect_non_operand_hard_regs): Likewise.
(add_regs_to_insn_regno_info): Likewise.
* reload1.c (regno_max_ref_width): Replace with...
(regno_max_ref_mode): ...this new variable.
(reload): Update accordingly.  Update call to
ira_sort_regnos_for_alter_reg.
(alter_reg): Update to use regno_max_ref_mode.  Call wider_subreg_mode.
(init_eliminable_invariants): Update to use regno_max_ref_mode.
(scan_paradoxical_subregs): Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r254115

Introduce emit_frame_chain

The current frame code combines the separate concepts of a frame chain
(saving old FP,LR in a record and pointing new FP to it) and a frame
pointer used to access locals.  Add emit_frame_chain to the aarch64_frame
descriptor and use it in the prolog and epilog code.  For now just
initialize it as before, so generated code is identical.

Also correctly set EXIT_IGNORE_STACK.  The current AArch64 epilog code
restores SP from FP if alloca is used.  If a frame pointer is used but
there is no alloca, SP must remain valid for the epilog to work correctly.

    gcc/
* config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
(aarch64_frame): Add emit_frame_chain boolean.
* config/aarch64/aarch64.c (aarch64_frame_pointer_required)
Move eh_return case to aarch64_layout_frame.
(aarch64_layout_frame): Initialize emit_frame_chain.
(aarch64_expand_prologue): Use emit_frame_chain.

From-SVN: r254114

Deduction guides for associative containers, debug mode deduction guide fixes.

* include/bits/stl_iterator.h (__iter_key_t)
(__iter_val_t, __iter_to_alloc_t): New.
* include/bits/stl_map.h: Add deduction guides.
* include/bits/stl_multimap.h: Likewise.
* include/bits/stl_multiset.h: Likewise.
* include/bits/stl_set.h: Likewise.
* include/bits/unordered_map.h: Likewise.
* include/bits/unordered_set.h: Likewise.
* include/debug/deque: Likewise.
* include/debug/forward_list: Likewise.
* include/debug/list: Likewise.
* include/debug/map.h: Likewise.
* include/debug/multimap.h: Likewise.
* include/debug/multiset.h: Likewise.
* include/debug/set.h: Likewise.
* include/debug/unordered_map: Likewise.
* include/debug/unordered_set: Likewise.
* include/debug/vector: Likewise.
* testsuite/23_containers/map/cons/deduction.cc: New.
* testsuite/23_containers/multimap/cons/deduction.cc: Likewise.
* testsuite/23_containers/multiset/cons/deduction.cc: Likewise.
* testsuite/23_containers/set/cons/deduction.cc: Likewise.
* testsuite/23_containers/unordered_map/cons/deduction.cc: Likewise.
* testsuite/23_containers/unordered_multimap/cons/deduction.cc:
Likewise.
* testsuite/23_containers/unordered_multiset/cons/deduction.cc:
Likewise.
* testsuite/23_containers/unordered_set/cons/deduction.cc: Likewise.

From-SVN: r254113

Simplify frame layout for stack probing

This patch makes some changes to the frame layout in order to simplify
stack probing.  We want to use the save of LR as a probe in any non-leaf
function.  With shrinkwrapping we may only save LR before a call, so it
is useful to define a fixed location in the callee-saves. So force LR at
the bottom of the callee-saves even with -fomit-frame-pointer.

Also remove a rarely used frame layout that saves the callee-saves first
with -fomit-frame-pointer.  Doing so allows the store of LR to be used as
a valid stack probe in all frames.

    gcc/
* config/aarch64/aarch64.c (aarch64_layout_frame):
        Ensure LR is always stored at the bottom of the callee-saves.
        Remove rarely used frame layout which saves callee-saves at top of
        frame, so the store of LR can be used as a valid probe in all cases.

From-SVN: r254112

Improve addressing of TI/TFmode

In https://gcc.gnu.org/ml/gcc-patches/2017-06/msg01125.html Jiong
pointed out some addressing inefficiencies due to a recent change in
regcprop (https://gcc.gnu.org/ml/gcc-patches/2017-04/msg00775.html).

This patch improves aarch64_legitimize_address_displacement to split
unaligned offsets of TImode and TFmode accesses.  The resulting code
is better and no longer relies on the original regcprop optimization.

For the test we now produce:

add x1, sp, 4
stp xzr, xzr, [x1, 24]

rather than:

        mov     x1, sp
        add     x1, x1, 28
        stp     xzr, xzr, [x1]

    gcc/
* config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
Improve unaligned TImode/TFmode base/offset split.

    testsuite/
* gcc.target/aarch64/ldp_stp_unaligned_2.c: New file.

From-SVN: r254111

Make more use of df_read_modify_subreg_p

This patch uses df_read_modify_subreg_p to check whether writing
to a subreg would preserve some of the existing contents.

This has the effect of putting more emphasis on the
REGMODE_NATURAL_SIZE-based definition of whether something can be
partially modified, instead of using UNITS_PER_WORD unconditionally.
This becomes important for SVE, where UNITS_PER_WORD has no
significance for subregs of multi-register LD2/ST2, LD3/ST3 and
LD4/ST4 tuples.

2017-10-26  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* caller-save.c (mark_referenced_regs):  Use read_modify_subreg_p.
* combine.c (find_single_use_1): Likewise.
(expand_field_assignment): Likewise.
(move_deaths): Likewise.
* lra-constraints.c (simplify_operand_subreg): Likewise.
(curr_insn_transform): Likewise.
* lra.c (collect_non_operand_hard_regs): Likewise.
(add_regs_to_insn_regno_info): Likewise.
* rtlanal.c (reg_referenced_p): Likewise.
(covers_regno_no_parallel_p): Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r254110

Stop print_hex from printing bits above the precision

2017-10-26  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* wide-int-print.cc (print_hex): Loop based on extract_uhwi.
Don't print any bits outside the precision of the value.
* wide-int.cc (test_printing): Add some new tests.

From-SVN: r254109

[obvious][arm testsuite] Fixup expected location in require-pic-register-loc.c

After r254010 we now add -gcolumn-info by default, that means the tests
in gcc.target/arm/require-pic-register-loc.c need adjusting to not expect
to see column zero.

gcc/testsuite/

* gcc.target/arm/require-pic-register-loc.c: Use wider regex for
column information.

From-SVN: r254106

[C++ PATCH] Kill IDENTIFIER_LABEL_VALUE

https://gcc.gnu.org/ml/gcc-patches/2017-10/msg01935.html
* decl.c (sort_labels): Restore function.
(pop_labels): Sort labels
(identify_goto): Add translation markup.

From-SVN: r254104

Use -xbrace_comment=no with recent Solaris/x86 as

* configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
supports -xbrace_comment option.
* configure: Regenerate.
* config.in: Regenerate.
* config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
(ASM_CPU_SPEC): Use it.

From-SVN: r254103

This patch adds a new hook that gives the preferred alignment for a static rtx...

TARGET_STATIC_RTX_ALIGNMENT

This patch adds a new hook that gives the preferred alignment for
a static rtx, so that we don't need to query the front end in
force_const_mem.

2017-10-26  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* target.def (static_rtx_alignment): New hook.
* targhooks.h (default_static_rtx_alignment): Declare.
* targhooks.c (default_static_rtx_alignment): New function.
* doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
* doc/tm.texi: Regenerate.
* varasm.c (force_const_mem): Use targetm.static_rtx_alignment
instead of targetm.constant_alignment.  Remove call to
set_mem_attributes.
* config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
(cris_preferred_mininum_alignment): New function, split out from...
(cris_constant_alignment): ...here.
(cris_static_rtx_alignment): New function.
* config/i386/i386.c (ix86_static_rtx_alignment): New function,
split out from...
(ix86_constant_alignment): ...here.
(TARGET_STATIC_RTX_ALIGNMENT): Redefine.
* config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
(mmix_static_rtx_alignment): New function.
* config/spu/spu.c (spu_static_rtx_alignment): New function.
(TARGET_STATIC_RTX_ALIGNMENT): Redefine.

From-SVN: r254102

2017-10-26  Tamar Christina  <tamar.christina@arm.com>

* gcc.dg/vect/vect-reduc-dot-s8a.c
(dg-additional-options, dg-require-effective-target): Add +dotprod.
* gcc.dg/vect/vect-reduc-dot-u8a.c
(dg-additional-options, dg-require-effective-target): Add +dotprod.

From-SVN: r254101

2017-10-26  Tamar Christina  <tamar.christina@arm.com>

* lib/target-supports.exp
(check_effective_target_arm_v8_2a_dotprod_neon_ok_nocache): New.
(check_effective_target_arm_v8_2a_dotprod_neon_ok): New.
(add_options_for_arm_v8_2a_dotprod_neon): New.
(check_effective_target_arm_v8_2a_dotprod_neon_hw): New.
(check_effective_target_vect_sdot_qi): Add ARM && AArch64.
(check_effective_target_vect_udot_qi): Likewise.
* gcc.target/arm/simd/vdot-exec.c: New.
* gcc.target/aarch64/advsimd-intrinsics/vdot-exec.c: New.
* gcc/doc/sourcebuild.texi: Document arm_v8_2a_dotprod_neon.

From-SVN: r254100

vect-multitypes-1.c: Correct target selector.

2017-10-26  Tamar Christina  <tamar.christina@arm.com>

* gcc.dg/vect/vect-multitypes-1.c: Correct target selector.

From-SVN: r254099

re PR target/81800 (On aarch64 ilp32 lrint should not be inlined as two instructions)

2017-10-26  Tamar Christina  <tamar.christina@arm.com>

PR target/81800
* config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): Add flag_trapping_math
and flag_fp_int_builtin_inexact.

gcc/testsuite/
2017-10-26  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/aarch64/inline-lrint_2.c (dg-options): Add -fno-trapping-math.

From-SVN: r254098

vect-dot-qi.h: New.

2017-10-26  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/aarch64/advsimd-intrinsics/vect-dot-qi.h: New.
* gcc.target/aarch64/advsimd-intrinsics/vdot-compile.c: New.
* gcc.target/aarch64/advsimd-intrinsics/vect-dot-s8.c: New.
* gcc.target/aarch64/advsimd-intrinsics/vect-dot-u8.c: New.

From-SVN: r254097

Daily bump.

From-SVN: r254096

C: detect more missing semicolons (PR c/7356)

c_parser_declaration_or_fndef has logic for parsing what might be
either a declaration or a function definition.

This patch adds a test to detect cases where a semicolon would have
terminated the decls as a declaration, where the token that follows
would start a new declaration specifier, and updates the error message
accordingly, with a fix-it hint.

This addresses PR c/7356, fixing the case of a stray token before a
#include that previously gave inscrutable output, and improving e.g.:

  int i
  int j;

from:

  t.c:2:1: error: expected '=', ',', ';', 'asm' or '__attribute__' before 'int'
   int j;
   ^~~

to:

  t.c:1:6: error: expected ';' before 'int'
   int i
        ^
        ;
   int j;
   ~~~

gcc.dg/noncompile/920923-1.c needs a slight update, as the output for
the first line changes from:

  920923-1.c:2:14: error: expected '=', ',', ';', 'asm' or
  '__attribute__' before 'unsigned'
   typedef BYTE unsigned char; /* { dg-error "expected" } */
                ^~~~~~~~

to:
  920923-1.c:2:13: error: expected ';' before 'unsigned'
   typedef BYTE unsigned char; /* { dg-error "expected" } */
               ^~~~~~~~~
               ;
  920923-1.c:2:1: warning: useless type name in empty declaration
   typedef BYTE unsigned char; /* { dg-error "expected" } */
   ^~~~~~~

The patch also adds a test for PR c/44515 as a baseline.

gcc/c/ChangeLog:
PR c/7356
* c-parser.c (c_parser_declaration_or_fndef): Detect missing
semicolons.

gcc/testsuite/ChangeLog:
PR c/7356
PR c/44515
* c-c++-common/pr44515.c: New test case.
* gcc.dg/pr7356-2.c: New test case.
* gcc.dg/pr7356.c: New test case.
* gcc.dg/spellcheck-typenames.c: Update the "singed" char "TODO"
case to reflect changes to output.
* gcc.dg/noncompile/920923-1.c: Add dg-warning to reflect changes
to output.

From-SVN: r254093

RISC-V: Add Sign/Zero extend patterns for PIC loads

Loads on RISC-V are sign-extending by default, but we weren't telling
GCC this in our PIC load patterns.  This corrects the problem, and adds
a zero-extending pattern as well.

gcc/ChangeLog

2017-10-25  Palmer Dabbelt  <palmer@dabbelt.com>

       * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
       * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
       mark as a sign-extending load.
       (local_pic_load_u): Define.

From-SVN: r254092

Makefile.am (check-go-tool): Output colon after ${fl}.

* Makefile.am (check-go-tool): Output colon after ${fl}.
(check-runtime, check-cgo-test, check-carchive-test): Likewise.
* Makefile.in: Rebuild.

From-SVN: r254090

re PR middle-end/82062 (simple conditional expressions no longer folded)

PR middle-end/82062
* fold-const.c (operand_equal_for_comparison_p): Also return true
if ARG0 is a simple variant of ARG1 with narrower precision.
(fold_ternary_loc): Always pass unstripped operands to the predicate.

From-SVN: r254089

[C++ PATCH] Kill IDENTIFIER_LABEL_VALUE

https://gcc.gnu.org/ml/gcc-patches/2017-10/msg01854.html
Kill IDENTIFIER_LABEL_VALUE.
* cp-tree.h (lang_identifier): Delete label_value slot.
(IDENTIFIER_LABEL_VALUE, SET_IDENTIFIER_LABEL_VALUE): Delete.
(struct named_label_hasher): Rename to ...
(struct named_label_hash): ... here.  Reimplement.
(struct language_function): Adjust x_named_labels.
* name-lookup.h (struct cp_label_binding): Delete.
(struct cp_binding_level): Delete shadowed_labels slot.
* decl.c (struct named_label_entry): Add name and outer slots.
(pop_label): Rename to ...
(check_label_used): ... here.  Don't pop.
(note_label, sort_labels): Delete.
(pop_labels, pop_local_label): Reimplement.
(poplevel): Pop local labels as any other decl. Remove
shadowed_labels handling.
(named_label_hash::hash, named_label_hash::equal): New.
(make_label_decl): Absorb into ...
(lookup_label_1): ... here.  Add making_local_p arg, reimplement.
(lookup_label, declare_local_label): Adjust.
(check_goto, define_label): Adjust.
* lex.c (make_conv_op_name): Don't clear IDENTIFIER_LABEL_VALUE.
* ptree.c (cxx_print_identifier): Don't print identifier binding.

From-SVN: r254087

[C++ PATCH] Label checking cleanups

https://gcc.gnu.org/ml/gcc-patches/2017-10/msg01847.html
* decl.c (identifier_goto): Reduce duplication.
(check_previous_goto_1): Likewise.
(check_goto): Move var decls to initialization.
(check_omp_return, define_label_1, define_label): Likewise.

From-SVN: r254086

i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather cost correctly.

* i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
cost correctly.
* i386.h (processor_costs): Add gather_static, gather_per_elt,
scatter_static, scatter_per_elt.
* x86-tune-costs.h: Add new cost entries.

From-SVN: r254083

match.c (gfc_match_type_is): Fix typo in error message

2017-10-25  Bernhard Reutner-Fischer  <aldot@gcc.gnu.org>

* match.c (gfc_match_type_is): Fix typo in error message.

From-SVN: r254082

Update C++17 library status documentation

* doc/xml/manual/status_cxx2017.xml: Update C++17 status, and
information on feature-test macros.
* doc/html/*: Regenerate.

From-SVN: r254078

PR libstdc++/82716 avoid stupid -Wmismatched-tags warnings

PR libstdc++/82716
* include/std/array (tuple_size, tuple_element): Change class-key
from class to struct, to avoid annoying Clang warnings.

From-SVN: r254077

PR libstdc++/79283 fix filesystem::read_symlink for /proc

PR libstdc++/79283
* src/filesystem/ops.cc (read_symlink): Handle st_size being zero.
* src/filesystem/std-ops.cc (read_symlink): Likewise.
(do_copy_file) [!NEED_DO_COPY_FILE]: Avoid multiple definitions.

From-SVN: r254076

Fix filesystem::path::lexically_normal algorithm

* src/filesystem/std-path.cc (path::lexically_normal): Add missing
step to algorithm, for removing dot-dot elements after root-directory.
* testsuite/27_io/filesystem/operations/canonical.cc: Use
compare_paths for more exhaustive checks.
* testsuite/27_io/filesystem/operations/proximate.cc: Likewise.
* testsuite/27_io/filesystem/path/append/path.cc: Likewise.
* testsuite/27_io/filesystem/path/concat/path.cc: Likewise.
* testsuite/27_io/filesystem/path/concat/strings.cc: Fix comment.
* testsuite/27_io/filesystem/path/construct/locale.cc: Likewise.
* testsuite/27_io/filesystem/path/decompose/root_directory.cc:
Likewise.
* testsuite/27_io/filesystem/path/generation/normal.cc: Use
compare_paths for more exhaustive checks. Add extra testcases.
* testsuite/27_io/filesystem/path/generation/proximate.cc: Use
compare_paths for more exhaustive checks.
* testsuite/27_io/filesystem/path/generation/relative.cc: Likewise.
* testsuite/27_io/filesystem/path/generic/generic_string.cc: Remove
unused header.
* testsuite/27_io/filesystem/path/modifiers/make_preferred.cc: Fix
comment.
* testsuite/27_io/filesystem/path/modifiers/remove_filename.cc: Use
compare_paths for more exhaustive checks.
* testsuite/27_io/filesystem/path/modifiers/replace_extension.cc:
Likewise.
* testsuite/27_io/filesystem/path/modifiers/replace_filename.cc:
Likewise.
* testsuite/util/testsuite_fs.h (compare_paths): Also compare native
strings.

From-SVN: r254075

tree-ssa-sccvn.h (vn_eliminate): Declare.

2017-10-25  Richard Biener  <rguenther@suse.de>

* tree-ssa-sccvn.h (vn_eliminate): Declare.
* tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
class pass_fre): Move to ...
* tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
class pass_fre): ... here and adjust for statistics.

From-SVN: r254074

re PR tree-optimization/82436 (465.tonto ICE in vect_get_slp_vect_defs, at tree-vect-slp.c:3410)

2017-10-25  Richard Biener  <rguenther@suse.de>

PR tree-optimization/82436
* gcc.dg/torture/pr82436-2.c: New testcase.

From-SVN: r254073

re PR c++/71820 (ICE on valid C++ code: in arg_assoc_type, at cp/name-lookup.c:5583)

2017-10-25  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/71820
* g++.dg/ext/typeof12.C: New.

From-SVN: r254072

Fix scan-assembler in tree-ssa/loop-1.c for nvptx

2017-10-25  Tom de Vries  <tom@codesourcery.com>

* gcc.dg/tree-ssa/loop-1.c: Add xfail for nvptx in scan-assembler-times
line, and add nvptx-specific version.

From-SVN: r254071

Fix gcc.target/i386/cet-sjlj-5.c on Solaris

* gcc.target/i386/cet-sjlj-5.c: Allow for emtpy user label prefix
in setjmp call.

From-SVN: r254070

re PR libstdc++/81706 (std::sin vectorization bug)

PR libstdc++/81706
* attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
(duplicate_one_attribute, copy_attributes_to_builtin): New functions.
* attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
declarations.

* c-decl.c (merge_decls): Copy "omp declare simd" attributes from
newdecl to corresponding __builtin_ if any.

* decl.c (duplicate_decls): Copy "omp declare simd" attributes from
newdecl to corresponding __builtin_ if any.

* gcc.target/i386/pr81706.c: New test.
* g++.dg/ext/pr81706.C: New test.

From-SVN: r254069

tree-ssa-pre.c (need_eh_cleanup, [...]): Move inside...

2017-10-25  Richard Biener  <rguenther@suse.de>

* tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
eliminate_push_avail, eliminate_insert): Move inside...
(class eliminate_dom_walker): ... this class in preparation
of move.
(fini_eliminate): Remove by merging with ...
(eliminate): ... this function.  Adjust for class changes.
(pass_pre::execute): Remove fini_eliminate call.
(pass_fre::execute): Likewise.

From-SVN: r254068

PR libstdc++/82706 fix test for case where operations succeed

PR libstdc++/82706
* testsuite/27_io/filesystem/operations/permissions.cc: Fix test.

From-SVN: r254067

Daily bump.

From-SVN: r254066

libstdc++.exp (check_v3_target_normal_mode): Add c++config.h include.

2017-10-24  François Dumont  <fdumont@gcc.gnu.org>

* testsuite/lib/libstdc++.exp (check_v3_target_normal_mode): Add
c++config.h include.
(check_v3_target_normal_namespace): New.
* testsuite/lib/dg-options.exp (dg-require-normal-namespace): New.
* testsuite/18_support/headers/limits/synopsis.cc: Add latter.
* testsuite/20_util/from_chars/requirements.cc: Likewise.
* testsuite/20_util/headers/functional/synopsis.cc: Likewise.
* testsuite/20_util/headers/memory/synopsis.cc: Likewise.
* testsuite/20_util/headers/utility/synopsis.cc: Likewise.
* testsuite/20_util/to_chars/requirements.cc: Likewise.
* testsuite/21_strings/headers/string/synopsis.cc: Likewise.
* testsuite/22_locale/headers/locale/synopsis.cc: Likewise.
* testsuite/23_containers/headers/bitset/synopsis.cc: Likewise.
* testsuite/23_containers/headers/deque/synopsis.cc: Likewise.
* testsuite/23_containers/headers/forward_list/synopsis.cc: Likewise.
* testsuite/23_containers/headers/list/synopsis.cc: Likewise.
* testsuite/23_containers/headers/map/synopsis.cc: Likewise.
* testsuite/23_containers/headers/queue/synopsis.cc: Likewise.
* testsuite/23_containers/headers/set/synopsis.cc: Likewise.
* testsuite/23_containers/headers/stack/synopsis.cc: Likewise.
* testsuite/23_containers/headers/vector/synopsis.cc: Likewise.
* testsuite/23_containers/map/modifiers/erase/abi_tag.cc: Likewise.
* testsuite/23_containers/multimap/modifiers/erase/abi_tag.cc: Likewise.
* testsuite/23_containers/multiset/modifiers/erase/abi_tag.cc: Likewise.
* testsuite/23_containers/set/modifiers/erase/abi_tag.cc: Likewise.
* testsuite/24_iterators/headers/iterator/synopsis.cc: Likewise.
* testsuite/24_iterators/headers/iterator/synopsis_c++11.cc: Likewise.
* testsuite/24_iterators/headers/iterator/synopsis_c++14.cc: Likewise.
* testsuite/24_iterators/headers/iterator/synopsis_c++17.cc: Likewise.
* testsuite/26_numerics/complex/abi_tag.cc: Likewise.
* testsuite/26_numerics/headers/complex/synopsis.cc: Likewise.
* testsuite/26_numerics/headers/valarray/synopsis.cc: Likewise.
* testsuite/27_io/headers/fstream/synopsis.cc: Likewise.
* testsuite/27_io/headers/ios/synopsis.cc: Likewise.
* testsuite/27_io/headers/istream/synopsis.cc: Likewise.
* testsuite/27_io/headers/ostream/synopsis.cc: Likewise.
* testsuite/27_io/headers/sstream/synopsis.cc: Likewise.
* testsuite/27_io/headers/streambuf/synopsis.cc: Likewise.

From-SVN: r254060

re PR target/82460 (AVX512: choose between vpermi2d and vpermt2d to save mov instructions.  Also, fails to optimize away shifts before shuffle)

PR target/82460
* config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
(VPERMI2, VPERMI2I): New mode iterators.
(<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
(<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
patterns.
(<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
mode iterator.  Remove 3 old define_insn patterns.
(*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
(<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
VPERMI2 mode iterator, remove the other two expanders.
(<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
to use VPERMI2 mode iterator, add another alternative for vpermi2*
instructions, remove the other two patterns.
(<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
mode iterator, remove the other two patterns.
* config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
(ix86_expand_vec_perm_vpermt2): ... this.  Swap mask and op0
arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
and adjust argument order accordingly.
(ix86_expand_vec_perm): Adjust caller.
(expand_vec_perm_1): Likewise.
(expand_vec_perm_vpermi2_vpshub2): Rename to ...
(expand_vec_perm_vpermt2_vpshub2): ... this.
(ix86_expand_vec_perm_const_1): Adjust caller.
(ix86_vectorize_vec_perm_const_ok): Adjust comments.

* gcc.target/i386/pr82460-1.c: New test.
* gcc.target/i386/pr82460-2.c: New test.
* gcc.target/i386/avx512f-vpermt2pd-1.c: Adjust scan-assembler*
regexps to allow vpermt2* to vpermi2* replacement or vice versa
where possible.
* gcc.target/i386/avx512vl-vpermt2pd-1.c: Likewise.
* gcc.target/i386/avx512f-vpermt2d-1.c: Likewise.
* gcc.target/i386/vect-pack-trunc-2.c: Likewise.
* gcc.target/i386/avx512vl-vpermt2ps-1.c: Likewise.
* gcc.target/i386/avx512vl-vpermt2q-1.c: Likewise.
* gcc.target/i386/avx512f-vpermt2ps-1.c: Likewise.
* gcc.target/i386/avx512vl-vpermt2d-1.c: Likewise.
* gcc.target/i386/avx512bw-vpermt2w-1.c: Likewise.
* gcc.target/i386/avx512vbmi-vpermt2b-1.c: Likewise.
* gcc.target/i386/avx512f-vpermt2q-1.c: Likewise.

From-SVN: r254059

re PR target/82370 (AVX512 can use a memory operand for immediate-count vpsrlw, but gcc doesn't.)

PR target/82370
* config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
(VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
(vec_shl_<mode>): Remove unused expander.
(avx512bw_<shift_insn><mode>3): New define_insn.
(<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
(<sse2_avx2>_<shift_insn><mode>3): ... this.  New define_insn.

* gcc.target/i386/pr82370.c: New test.

From-SVN: r254058

re PR c++/82466 (Missing warning for re-declaration of built-in function as variable)

2017-10-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82466
* doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
description.

/cp
2017-10-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82466
* decl.c (duplicate_decls): Warn for built-in functions declared as
non-function, use OPT_Wbuiltin_declaration_mismatch.

* decl.c (duplicate_decls): Avoid redundant '+' in warning_at.

/c
2017-10-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82466
* c-decl.c (diagnose_mismatched_decls): Use
OPT_Wbuiltin_declaration_mismatch.

/testsuite
2017-10-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82466
* c-c++-common/Wbuiltin-declaration-mismatch-1.c: New.
* c-c++-common/Wno-builtin-declaration-mismatch-1.c: Likewise.
* g++.dg/warn/Wbuiltin_declaration_mismatch-1.C: Likewise.
* g++.dg/parse/builtin2.C: Adjust.
* g++.old-deja/g++.mike/p811.C: Likewise.

From-SVN: r254057

Cleanup autopref scheduling

r253236 broke AArch64 bootstrap. Earlier revision r253071 changed scheduling
behaviour on AArch64 as autopref scheduling no longer checks the base.

This patch fixes the bootstrap failure and cleans up autopref scheduling.
The code is greatly simplified.  Sort accesses on the offset first, and
only if the offsets are the same fall back to other comparisons in
rank_for_schedule.  This doesn't at all restore the original behaviour
since we no longer compare the base address, but it now defines a total
sorting order.  More work will be required to improve the sorting so
that only loads/stores with the same base are affected.

    gcc/
PR rtl-optimization/82396
* gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
(autopref_multipass_init): Simplify initialization.
(autopref_rank_data): Simplify sort order.
* gcc/sched-int.h (autopref_multipass_data_): Remove
multi_mem_insn_p, min_offset and max_offset.

From-SVN: r254056

PR60580: Fix frame pointer option magic

To fix PR60580 simplify the logic in aarch64_override_options_after_change_1 ().
If the frame pointer is enabled, set it to a special value that behaves similar
to frame pointer omission.  If we don't do this all leaf functions will get a
frame pointer even if flag_omit_leaf_frame_pointer is set.

If flag_omit_frame_pointer has this special value, we must force the frame
pointer if not in a leaf function.  We also need to force it in a leaf function
if flag_omit_frame_pointer is not set or if LR is used.

Doing this allows both -fomit-frame-pointer and -fomit-leaf-frame-pointer to be
independently set and changed in each function with the expected behaviour.

    gcc/
PR middle-end/60580
* config/aarch64/aarch64.c (aarch64_frame_pointer_required)
Check special value of flag_omit_frame_pointer.
(aarch64_can_eliminate): Likewise.
(aarch64_override_options_after_change_1): Simplify handling of
-fomit-frame-pointer and -fomit-leaf-frame-pointer.

From-SVN: r254052

re PR c++/80991 (ICE with __is_trivially_constructible in template)

/cp
2017-10-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/80991
* pt.c (value_dependent_expression_p, [TRAIT_EXPR]): Handle
a TREE_LIST as TRAIT_EXPR_TYPE2.

/testsuite
2017-10-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/80991
* g++.dg/ext/is_trivially_constructible5.C: New.

From-SVN: r254051

Fix gcc.target/i386/387-ficom-[12].c on Solaris

* gcc.target/i386/387-ficom-1.c: Allow for ficomp without s
suffix.
* gcc.target/i386/387-ficom-2.c: Likewise.

From-SVN: r254050

Fix gcc.target/i386/cet-sjlj-3.c on Solaris

* gcc.target/i386/cet-sjlj-3.c: Allow for emtpy user label prefix
in setjmp call.

From-SVN: r254049

re PR tree-optimization/82697 (Wrong optimization with aliasing and "if")

2017-10-24  Richard Biener  <rguenther@suse.de>

PR tree-optimization/82697
* tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
zero for conditional load and unconditional store.

* gcc.dg/torture/pr82697.c: New testcase.

From-SVN: r254047

re PR c++/82307 (unscoped enum-base incorrect cast)

/cp
2017-10-24  Mukesh Kapoor  <mukesh.kapoor@oracle.com>
    Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82307
* cvt.c (type_promotes_to): Implement C++17, 7.6/4, about unscoped
enumeration type whose underlying type is fixed.

/testsuite
2017-10-24  Mukesh Kapoor  <mukesh.kapoor@oracle.com>
    Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82307
* g++.dg/cpp0x/enum35.C: New.
* g++.dg/cpp0x/enum36.C: Likewise.

Co-Authored-By: Paolo Carlini <paolo.carlini@oracle.com>
From-SVN: r254046

Avoid -Wattribute-alias warnings for long double compat symbols

* config/locale/gnu/c_locale.cc [_GLIBCXX_LONG_DOUBLE_COMPAT]: Ignore
-Wattribute-alias warnings.
* src/c++11/istream-inst.cc: Likewise.
* src/c++11/locale-inst.cc: Likewise.
* src/c++11/ostream-inst.cc: Likewise.
* src/c++11/wlocale-inst.cc: Likewise.
* src/c++98/hash-long-double-tr1-aux.cc: Likewise.

From-SVN: r254045

Refactor std::basic_string_view members and add noexcept

* include/bits/string_view.tcc (find_first_of, find_last_of)
(find_first_not_of, find_last_not_of): Add noexcept.
* include/std/string_view (basic_string_view(const _CharT*))
(basic_string_view(const _CharT*, size_type))
(front, back, remove_prefix, remove_suffix, find, rfind)
(find_first_of, find_first_not_of): Add noexcept.
(at(size_type), _S_compare(size_type, size_type)): Replace conditional
expressions with if statements.
(copy(_CharT*, size_type, size_type), substr(size_type, size_type)):
Use _M_check for length checks.
(compare(basic_string_view)): Reformat.
(_M_check(size_type, const char)): Add noexcept(false).
(_M_limit(size_type, size_type)): Use noexcept not _GLIBCXX_NOEXCEPT.

From-SVN: r254044

Add bootstrap-cet.mk to bootstrap GCC with Intel CET

Bootstrap GCC with Intel CET by configuring GCC with

--with-build-config="bootstrap-cet bootstrap-debug"

Tested on Linux/i686 and Linux/x86-64.

config/

* bootstrap-cet.mk: New file.

gcc/

* doc/install.texi: Document bootstrap-cet.

From-SVN: r254043

PR82687, g++.dg/asan/default-options-1.C fails with PR82575 fix

The problem with making discarded symbols hidden is that the
non-default visibility is sticky.  When symbols other than the
__gnu_lto ones are discarded that turns out to be a bad idea.

PR lto/82687
PR lto/82575
* simple-object-elf.c (simple_object_elf_copy_lto_debug_sections):
Only make __gnu_lto symbols hidden.  Delete outdated comment.
Silence ISO C warning.

From-SVN: r254042

PR libstdc++/82685 add 'noexcept' to string_view literals

PR libstdc++/82685
* include/experimental/string_view (operator""sv): Add noexcept.
* include/std/string_view (operator""sv): Likewise.

From-SVN: r254041

i386: Don't insert ENDBR at function entrance when called directly

There is no need to insert ENDBR instruction at function entrance if
function is only called directly.

gcc/

PR target/82659
* config/i386/i386.c (rest_of_insert_endbranch): Don't insert
ENDBR instruction at function entrance if function is only
called directly.

gcc/testsuite/

PR target/82659
* gcc.target/i386/cet-label-2.c: New test.
* gcc.target/i386/cet-sjlj-4.c: Likewise.
* gcc.target/i386/cet-sjlj-5.c: Likewise.
* gcc.target/i386/cet-switch-3.c: Likewise.
* gcc.target/i386/pr82659-1.c: Likewise.
* gcc.target/i386/pr82659-2.c: Likewise.
* gcc.target/i386/pr82659-3.c: Likewise.
* gcc.target/i386/pr82659-4.c: Likewise.
* gcc.target/i386/pr82659-5.c: Likewise.
* gcc.target/i386/pr82659-6.c: Likewise.

From-SVN: r254040

re PR rtl-optimization/82628 (wrong code at -Os on x86_64-linux-gnu in the 32-bit mode)

PR target/82628
* config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
patterns to better describe from which operation the CF is computed.
(addcarry<mode>_0, subborrow<mode>_0): New patterns.
* config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
one LTU with [DT]Imode and another one with [SD]Imode.  If arg0
is 0, use _0 suffixed expanders instead of emitting a comparison
before it.

From-SVN: r254039

Avoid 512-bit mode MOV for prefer-avx256 option in Intel AVX512 configuration

gcc/
* config/i386/i386.md(*movsf_internal, *movdf_internal):
Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.

From-SVN: r254038

re PR middle-end/82569 (failure in 177.mesa cpu2000 test case after r253530)

PR middle-end/82569
* tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
* expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
* loop-iv.c (iv_get_reaching_def): Likewise.
* cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
variable is promoted and the partition contains undefined values.

From-SVN: r254037

re PR tree-optimization/82672 ([GRAPHITE] ICE in verify_gimple_in_cfg)

2017-10-23  Richard Biener  <rguenther@suse.de>

PR tree-optimization/82672
* graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
Fold the stmt if we propagated into it.

* gfortran.dg/graphite/pr82672.f90: New testcase.

From-SVN: r254036

cdx-branch.c: Fix broken test.

2017-10-23  Sandra Loosemore  <sandra@codesourcery.com>

gcc/testsuite/
* gcc.target/nios2/cdx-branch.c:  Fix broken test.
* gcc.target/nios2/lo-addr-bypass.c: New.
* gcc.target/nios2/lo-addr-char.c: New.
* gcc.target/nios2/lo-addr-int.c: New.
* gcc.target/nios2/lo-addr-pic.c: New.
* gcc.target/nios2/lo-addr-short.c: New.
* gcc.target/nios2/lo-addr-tls.c: New.
* gcc.target/nios2/lo-addr-uchar.c: New.
* gcc.target/nios2/lo-addr-ushort.c: New.
* gcc.target/nios2/lo-addr-volatile.c: New.

From-SVN: r254035

nios2.c (nios2_rtx_costs): Make costs better reflect reality.

2017-10-23  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* config/nios2/nios2.c (nios2_rtx_costs): Make costs better
reflect reality.
(nios2_address_cost): Define.
(nios2_legitimize_address): Recognize (exp + constant) directly.
(TARGET_ADDRESS_COST): Define.

From-SVN: r254034

nios2-protos.h (nios2_large_constant_p): Declare.

2017-10-23  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
(nios2_symbolic_memory_operand_p): Declare.
(nios2_split_large_constant): Declare.
(nios2_split_symbolic_memory_operand): Declare.
* config/nios2/nios2.c: Adjust includes.
(nios2_symbolic_constant_allowed): New.
(nios2_symbolic_constant_p): New.
(nios2_plus_symbolic_constant_p): New.
(nios2_valid_addr_expr_p): Recognize addresses involving
symbolic constants.
(nios2_legitimate_address_p): Likewise, also LO_SUM.
(nios2_symbolic_memory_operand_p): New.
(nios2_large_constant_p): New.
(nios2_split_large_constant): New.
(nios2_split_plus_large_constant): New.
(nios2_split_symbolic_memory_operand): New.
(nios2_legitimize_address): Code refactoring.  Handle addresses
involving symbolic constants.
(nios2_emit_move_sequence): Likewise.
(nios2_print_operand): Improve error output.
(nios2_print_operand_address): Handle LO_SUM.
(nios2_cdx_narrow_form_p): Likewise.
* config/nios2/nios2.md (movqi_internal): Add splitter for memory
operands involving symbolic constants.
(movhi_internal, movsi_internal): Likewise.
(zero_extendhisi2, zero_extendqi<mode>2): Likewise.
(extendhisi2, extendqi<mode>2): Likewise.

From-SVN: r254033

tree-pass.h (PROP_rtl_split_insns): Define.

2017-10-23  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* tree-pass.h (PROP_rtl_split_insns): Define.
* recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.

From-SVN: r254032

nios2.c (TARGET_LRA_P): Don't override.

2017-10-23  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* config/nios2/nios2.c (TARGET_LRA_P): Don't override.

From-SVN: r254031

Daily bump.

From-SVN: r254030

re PR c++/80449 (ICE reporting failed partial class template specialization class template argument deduction)

/cp
2017-10-23  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/80449
* semantics.c (finish_compound_literal): Check do_auto_deduction
return value for error_mark_node.

/testsuite
2017-10-23  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/80449
* g++.dg/cpp1z/class-deduction46.C: New.

From-SVN: r254026

re PR debug/82630 (Bogus DW_AT_GNU_call_site_value)

PR debug/82630
* target.def (const_not_ok_for_debug_p): Default to
default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
* targhooks.h (default_const_not_ok_for_debug_p): New declaration.
* targhooks.c (default_const_not_ok_for_debug_p): New function.
* dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
which targetm.const_not_ok_for_debug_p returned true.
* config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
for UNSPECs.
* config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
Likewise.
* config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
* config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
if !base_term_p.
(ix86_const_not_ok_for_debug_p): New function.
(i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
(TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.

* g++.dg/guality/pr82630.C: New test.

From-SVN: r254025

Add INCLUDE_UNIQUE_PTR and use it (PR bootstrap/82610)

gcc/ChangeLog:
PR bootstrap/82610
* system.h: Conditionally include "unique-ptr.h" if
INCLUDE_UNIQUE_PTR is defined.
* unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
of defining INCLUDE_UNIQUE_PTR before including "system.h".

include/ChangeLog:
* unique-ptr.h: Make include of <memory> conditional on C++11 or
later.

From-SVN: r254024

PR c++/77369 - wrong noexcept handling in C++14 and below

* tree.c (strip_typedefs): Canonicalize TYPE_RAISES_EXCEPTIONS.

From-SVN: r254022

* g++.dg/opt/pr82577: Fix for C++17.

From-SVN: r254021

re PR target/82662 (ifcvt is not performed on double-word compare)

PR target/82662
* gcc.target/i386/pr82662.c: New test.

From-SVN: r254020

rl78.md: New define_expand "subdi3".

* config/rl78/rl78.md: New define_expand "subdi3".
* config/rl78/subdi3.S: New assembly file.
* config/rl78/t-rl78: Added subdi3.S to LIB2ADD.

From-SVN: r254019

Forgot to add the new file :-P 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>

Forgot to add the new file :-P
2017-10-13  Sebastian Perta  <sebastian.perta@renesas.com>
* config/rl78/adddi3.S: New assembly file.
* config/rl78/t-rl78: Added adddi3.S to LIB2ADD.

From-SVN: r254016

re PR c/82681 (c-warn.c:1218: typo in warning message)

PR c/82681
* c-warn.c (warnings_for_convert_and_check): Fix typos.

* gcc.dg/c90-const-expr-11.c: Fix typos in dg-warning.
* gcc.dg/overflow-warn-5.c: Likewise.
* gcc.dg/overflow-warn-8.c: Likewise.

From-SVN: r254014

i386: Skip DF_REF_INSN if DF_REF_INSN_INFO is false

We should check DF_REF_INSN_INFO before accessing DF_REF_INSN.

gcc/

PR target/82673
* config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
DF_REF_INSN if DF_REF_INSN_INFO is false.

gcc/testsuite/

PR target/82673
* gcc.target/i386/pr82673.c: New test.

From-SVN: r254013

i386.c (dimode_scalar_chain::compute_convert_gain): Use xmm_move instead of sse_move.

* i386.c (dimode_scalar_chain::compute_convert_gain): Use
xmm_move instead of sse_move.
(sse_store_index): New function.
(ix86_register_move_cost): Be more sensible about mismatch stall;
model AVX moves correctly; make difference between sse->integer and
integer->sse.
(ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
moves; make difference between SSE and AVX.
* i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
and zmm_move. Increase size of sse load and store tables;
add unaligned load and store tables; add ssemmx_to_integer.
* x86-tune-costs.h: Update all entries according to real
move latencies from Agner Fog's manual and chip documentation.

From-SVN: r254012

re PR rtl-optimization/82628 (wrong code at -Os on x86_64-linux-gnu in the 32-bit mode)

PR target/82628
* config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
* config/i386/constraints.md (Wf): New constraint.
* config/i386/i386.md (UNSPEC_SBB): New unspec.
(cmp<dwi>_doubleword): Removed.
(sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
(sub<mode>3_carry_ccgz): Use unspec instead of compare.
* config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
expand with cmp<dwi>_doubleword.  For LTU and GEU use
sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.

From-SVN: r254011

common.opt (gcolumn-info): Enable by default.

* common.opt (gcolumn-info): Enable by default.
* doc/invoke.texi (gcolumn-info): Document new default.

* lib/scanasm.exp (dg-function-on-line): Accept optional column info.
* gcc.dg/debug/dwarf2/pr53948.c: Likewise.
* g++.dg/debug/dwarf2/pr77363.C: Likewise.
* gcc.dg/debug/dwarf2/asm-line1.c: Add -gno-column-info to dg-options.
* gcc.dg/debug/dwarf2/discriminator.c: Likewise.
* g++.dg/debug/dwarf2/typedef6.C: Likewise.

From-SVN: r254010

re PR tree-optimization/82672 ([GRAPHITE] ICE in verify_gimple_in_cfg)

2017-10-23  Richard Biener  <rguenther@suse.de>

PR tree-optimization/82672
* graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
Fold the stmt if we propagated into it.

* gfortran.dg/graphite/pr82672.f90: New testcase.

From-SVN: r254009

Implement C++17 Filesystem library

Based on Filesystem TS implementation, with the changes applied by:

- P0219R1 Relative Paths for Filesystem
- P0317R1 Directory Entry Caching for Filesystem
- P0492R2 Resolution of C++17 National Body Comments

Where appropriate code is shared between the TS and C++17
implementations.

* include/Makefile.am: Add new headers for C++17 filesystem library.
* include/Makefile.in: Regenerate.
* include/bits/fs_dir.h: New header, based on Filesystem TS code in
include/experimental/bits directory.
* include/bits/fs_fwd.h: Likewise.
* include/bits/fs_ops.h: Likewise.
* include/bits/fs_path.h: Likewise.
* include/experimental/bits/fs_dir.h: Rename Doxygen group.
* include/experimental/bits/fs_fwd.h: Likewise.
* include/experimental/bits/fs_ops.h: Likewise.
* include/experimental/bits/fs_path.h: Likewise.
* include/experimental/filesystem (filesystem_error::_M_gen_what):
Remove inline definition.
* include/precompiled/stdc++.h: Add <filesystem> to precompiled
header.
* include/std/filesystem: New header.
* python/libstdcxx/v6/printers.py: Enable printer for std::filesystem
paths.
* src/filesystem/Makefile.am: Add new files. Compile as C++17.
* src/filesystem/Makefile.in: Regenerate.
* src/filesystem/cow-dir.cc: Update comment.
* src/filesystem/cow-ops.cc: Likewise.
* src/filesystem/cow-path.cc: Likewise.
* src/filesystem/cow-std-dir.cc: New file.
* src/filesystem/cow-std-ops.cc: New file.
* src/filesystem/cow-std-path.cc: New file.
* src/filesystem/dir-common.h (_Dir_base, get_file_type): New header
for common code.
* src/filesystem/dir.cc (_Dir): Derive from _Dir_base.
(open_dir): Move to _Dir_base constructor.
(get_file_type): Move to dir-common.h.
(recurse): Move to _Dir_base::should_recurse.
* src/filesystem/ops-common.h: New header for common code.
* src/filesystem/ops.cc (is_set, make_file_type, make_file_status)
(is_not_found_errno, file_time, do_copy_file): Move to ops-common.h.
* src/filesystem/path.cc (filesystem_error::_M_gen_what): Define.
* src/filesystem/std-dir.cc: New file, based on Filesystem TS code.
* src/filesystem/std-ops.cc: Likewise.
* src/filesystem/std-dir.cc: Likewise.
* testsuite/27_io/filesystem/iterators/directory_iterator.cc: New
test.
* testsuite/27_io/filesystem/iterators/pop.cc: New test.
* testsuite/27_io/filesystem/iterators/recursive_directory_iterator.cc:
New test.
* testsuite/27_io/filesystem/operations/absolute.cc: New test.
* testsuite/27_io/filesystem/operations/canonical.cc: New test.
* testsuite/27_io/filesystem/operations/copy.cc: New test.
* testsuite/27_io/filesystem/operations/copy_file.cc: New test.
* testsuite/27_io/filesystem/operations/create_directories.cc: New
test.
* testsuite/27_io/filesystem/operations/create_directory.cc: New test.
* testsuite/27_io/filesystem/operations/create_symlink.cc: New test.
* testsuite/27_io/filesystem/operations/current_path.cc: New test.
* testsuite/27_io/filesystem/operations/equivalent.cc: New test.
* testsuite/27_io/filesystem/operations/exists.cc: New test.
* testsuite/27_io/filesystem/operations/file_size.cc: New test.
* testsuite/27_io/filesystem/operations/is_empty.cc: New test.
* testsuite/27_io/filesystem/operations/last_write_time.cc: New test.
* testsuite/27_io/filesystem/operations/permissions.cc: New test.
* testsuite/27_io/filesystem/operations/proximate.cc: New test.
* testsuite/27_io/filesystem/operations/read_symlink.cc: New test.
* testsuite/27_io/filesystem/operations/relative.cc: New test.
* testsuite/27_io/filesystem/operations/remove_all.cc: New test.
* testsuite/27_io/filesystem/operations/space.cc: New test.
* testsuite/27_io/filesystem/operations/status.cc: New test.
* testsuite/27_io/filesystem/operations/symlink_status.cc: New test.
* testsuite/27_io/filesystem/operations/temp_directory_path.cc: New
test.
* testsuite/27_io/filesystem/operations/weakly_canonical.cc: New test.
* testsuite/27_io/filesystem/path/append/path.cc: New test.
* testsuite/27_io/filesystem/path/assign/assign.cc: New test.
* testsuite/27_io/filesystem/path/assign/copy.cc: New test.
* testsuite/27_io/filesystem/path/compare/compare.cc: New test.
* testsuite/27_io/filesystem/path/compare/path.cc: New test.
* testsuite/27_io/filesystem/path/compare/strings.cc: New test.
* testsuite/27_io/filesystem/path/concat/path.cc: New test.
* testsuite/27_io/filesystem/path/concat/strings.cc: New test.
* testsuite/27_io/filesystem/path/construct/copy.cc: New test.
* testsuite/27_io/filesystem/path/construct/default.cc: New test.
* testsuite/27_io/filesystem/path/construct/locale.cc: New test.
* testsuite/27_io/filesystem/path/construct/range.cc: New test.
* testsuite/27_io/filesystem/path/construct/string_view.cc: New test.
* testsuite/27_io/filesystem/path/decompose/extension.cc: New test.
* testsuite/27_io/filesystem/path/decompose/filename.cc: New test.
* testsuite/27_io/filesystem/path/decompose/parent_path.cc: New test.
* testsuite/27_io/filesystem/path/decompose/relative_path.cc: New
test.
* testsuite/27_io/filesystem/path/decompose/root_directory.cc: New
test.
* testsuite/27_io/filesystem/path/decompose/root_name.cc: New test.
* testsuite/27_io/filesystem/path/decompose/root_path.cc: New test.
* testsuite/27_io/filesystem/path/decompose/stem.cc: New test.
* testsuite/27_io/filesystem/path/generation/normal.cc: New test.
* testsuite/27_io/filesystem/path/generation/proximate.cc: New test.
* testsuite/27_io/filesystem/path/generation/relative.cc: New test.
* testsuite/27_io/filesystem/path/generic/generic_string.cc: New test.
* testsuite/27_io/filesystem/path/itr/traversal.cc: New test.
* testsuite/27_io/filesystem/path/modifiers/clear.cc: New test.
* testsuite/27_io/filesystem/path/modifiers/make_preferred.cc: New
test.
* testsuite/27_io/filesystem/path/modifiers/remove_filename.cc: New
test.
* testsuite/27_io/filesystem/path/modifiers/replace_extension.cc: New
test.
* testsuite/27_io/filesystem/path/modifiers/replace_filename.cc: New
test.
* testsuite/27_io/filesystem/path/modifiers/swap.cc: New test.
* testsuite/27_io/filesystem/path/native/string.cc: New test.
* testsuite/27_io/filesystem/path/nonmember/hash_value.cc: New test.
* testsuite/27_io/filesystem/path/query/empty.cc: New test.
* testsuite/27_io/filesystem/path/query/has_extension.cc: New test.
* testsuite/27_io/filesystem/path/query/has_filename.cc: New test.
* testsuite/27_io/filesystem/path/query/has_parent_path.cc: New test.
* testsuite/27_io/filesystem/path/query/has_relative_path.cc: New
test.
* testsuite/27_io/filesystem/path/query/has_root_directory.cc: New
test.
* testsuite/27_io/filesystem/path/query/has_root_name.cc: New test.
* testsuite/27_io/filesystem/path/query/has_root_path.cc: New test.
* testsuite/27_io/filesystem/path/query/has_stem.cc: New test.
* testsuite/27_io/filesystem/path/query/is_relative.cc: New test.
* testsuite/experimental/filesystem/path/construct/string_view.cc:
Define USE_FILESYSTEM_TS.
* testsuite/util/testsuite_fs.h: Allow use with C++17 paths as well
as Filesystem TS.

From-SVN: r254008

tree-ssa-pre.c (bitmap_remove_from_set): Rename to...

2017-10-23  Richard Biener  <rguenther@suse.de>

* tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
(bitmap_remove_expr_from_set): ... this.  All callers call this
for non-constant values.
(bitmap_set_subtract): Rename to...
(bitmap_set_subtract_expressions): ... this.  Adjust and
optimize.
(bitmap_set_contains_value): Remove superfluous check.
(bitmap_set_replace_value): Inline into single caller ...
(bitmap_value_replace_in_set): ... here and simplify.
(dependent_clean): Merge into ...
(clean): ... this using an overload.  Adjust.
(prune_clobbered_mems): Adjust.
(compute_antic_aux): Likewise.
(compute_partial_antic_aux): Likewise.

From-SVN: r254007

re PR c++/77555 (unused inline function in-function static variable accessed from outside leads to linker error)

2017-10-23  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/77555
* g++.dg/torture/pr77555.C: New.

From-SVN: r254006

re PR tree-optimization/82129 (ICE in compute_antic, at tree-ssa-pre.c:2447)

2017-10-23  Richard Biener  <rguenther@suse.de>

PR tree-optimization/82129
Revert
2017-08-01  Richard Biener  <rguenther@suse.de>

PR tree-optimization/81181
* tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
(compute_antic): ... end of iteration here.

From-SVN: r254005

PR libstdc++/82644 document IS 29124 support

Also fix declarations of special functions in C++17, to import them into
the global namespace in <math.h>, and to prevent defining the
non-standard hypergeometric functions in strict mode.

PR libstdc++/82644
* doc/xml/manual/intro.xml: Include new section.
* doc/xml/manual/status_cxxis29124.xml: New section on IS 29124
status.
* include/bits/specfun.h [__STRICT_ANSI__] (hyperg, hypergf, hypergl)
(conf_hyperg, conf_hypergf, conf_hypergl): Don't declare.
* include/c_compatibility/math.h: Import special functions into
global namespace for C++17.
* testsuite/26_numerics/headers/cmath/82644.cc: New test.
* testsuite/26_numerics/headers/cmath/functions_global_c++17.cc: New
test.

From-SVN: r254004

Convert STARTING_FRAME_OFFSET to a hook

I took the documentation of the FRAME_GROWS_DOWNWARD behaviour from the
version that was in most header files, since the one in the manual seemed
less clear.

The patch deliberately keeps FIRST_PARM_OFFSET(FNDECL) in
microblaze_starting_frame_offset; this seems to be a port-local
convention and takes advantage of the fact that FIRST_PARM_OFFSET
doesn't read FNDECL.

2017-10-23  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* target.def (starting_frame_offset): New hook.
* doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
(TARGET_STARTING_FRAME_OFFSET): ...this new hook.
* doc/tm.texi.in: Regenerate.
* hooks.h (hook_hwi_void_0): Declare.
* hooks.c (hook_hwi_void_0): New function.
* doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
STARTING_FRAME_OFFSET.
* builtins.c (expand_builtin_setjmp_receiver): Likewise.
* reload1.c (reload): Likewise.
* cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
instead of STARTING_FRAME_OFFSET.
* function.c (try_fit_stack_local): Likewise.
(assign_stack_local_1): Likewise
(instantiate_virtual_regs): Likewise.
* rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
* config/avr/avr.md (nonlocal_goto_receiver): Likewise.
* config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
* config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
* config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
* config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
* config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
* config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
* config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
* config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
* config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
* config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
* config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
* config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
* config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
* config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
* config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
* config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
* config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
* config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
* config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
* config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
* config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
* config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
* config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
* config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
* config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
* config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
* config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
* config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
* config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
* config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
* config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
* config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
* config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
* config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
* config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
* config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
* config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
* config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
* config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
* config/avr/avr.c (avr_starting_frame_offset): Make static and
return a HOST_WIDE_INT.
(avr_builtin_setjmp_frame_value): Use it instead of
STARTING_FRAME_OFFSET.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
* config/epiphany/epiphany.c (epiphany_starting_frame_offset):
New function.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
* config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
(TARGET_CONSTANT_ALIGNMENT): Redefine.
* config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
* config/lm32/lm32.c (lm32_starting_frame_offset): New function.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
* config/m32r/m32r.c (m32r_starting_frame_offset): New function.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
* config/microblaze/microblaze.c (microblaze_starting_frame_offset):
New function.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
* config/mips/mips.c (mips_compute_frame_info): Refer to
TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
(mips_starting_frame_offset): New function.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
* config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
* config/mmix/mmix.c (mmix_starting_frame_offset): Make static
and return a HOST_WIDE_INT.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
(mmix_initial_elimination_offset): Refer to
TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
* config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
* config/pa/pa.c (pa_starting_frame_offset): New function.
(pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
(pa_expand_prologue): Likewise.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
!FRAME_GROWS_DOWNWARD handling to...
(RS6000_STARTING_FRAME_OFFSET): ...this new macro.
* config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
!FRAME_GROWS_DOWNWARD handling to...
(RS6000_STARTING_FRAME_OFFSET): ...this new macro.
* config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
!FRAME_GROWS_DOWNWARD handling to...
(RS6000_STARTING_FRAME_OFFSET): ...this new macro.
* config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
Redefine.
(rs6000_starting_frame_offset): New function.
* config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
!FRAME_GROWS_DOWNWARD handling to...
(RS6000_STARTING_FRAME_OFFSET): ...this new macro.
* config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
!FRAME_GROWS_DOWNWARD handling to...
(RS6000_STARTING_FRAME_OFFSET): ...this new macro.
* config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
!FRAME_GROWS_DOWNWARD handling to...
(RS6000_STARTING_FRAME_OFFSET): ...this new macro.
* config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
(rs6000_starting_frame_offset): New function.
* config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
* config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
* config/vax/vax.c (vax_starting_frame_offset): New function.
(vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
* config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
(TARGET_STARTING_FRAME_OFFSET): Redefine.
* system.h (STARTING_FRAME_OFFSET): Poison.

From-SVN: r254003

Use SCALAR_TYPE_MODE in vect_create_epilog_for_reduction

This follows on from similar changes a couple of months ago and
is needed when general modes have variable size.

2017-10-23  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* tree-vect-loop.c (vect_create_epilog_for_reduction): Use
SCALAR_TYPE_MODE instead of TYPE_MODE.

From-SVN: r254002

Use SCALAR_INT_TYPE_MODE in loc_list_from_tree_1

This follows on from similar changes a couple of months ago and
is needed when general modes have variable size.

2017-10-23  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r254001

Use scalar_int/float_mode in brig_langhook_type_for_mode

This follows on from similar changes a couple of months ago and
is needed when general modes have variable size.

2017-10-23  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/brig/
* brig-lang.c (brig_langhook_type_for_mode): Use scalar_int_mode
and scalar_float_mode.

From-SVN: r254000

Use scalar_mode in expand_shift_1

Since this function handles scalar and vector shifts:

  machine_mode scalar_mode = mode;
  if (VECTOR_MODE_P (mode))
    scalar_mode = GET_MODE_INNER (mode);

is equivalent to:

  scalar_mode = GET_MODE_INNER (mode);

2017-10-23  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r253999

re PR tree-optimization/82129 (ICE in compute_antic, at tree-ssa-pre.c:2447)

2017-10-23  Richard Biener  <rguenther@suse.de>

PR tree-optimization/82129
* tree-ssa-pre.c (bitmap_set_and): Remove.
(compute_antic_aux): Compute ANTIC_OUT intersection in a way
canonicalizing expressions in the set to those with lowest
ID rather than taking that from the first edge.

* gcc.dg/torture/pr82129.c: New testcase.

From-SVN: r253998

Fix HWI + -unsigned in combine.c

rtx_equal_for_field_assignment_p had:

  x = adjust_address_nv (x, GET_MODE (y),
       -subreg_lowpart_offset (GET_MODE (x),
       GET_MODE (y)));

But subreg_lowpart_offset returns an unsigned int and
adjust_address_nv takes a HWI, so a subreg offset of 4 would
give a memory offset of 0x00000000fffffffffc.

2017-10-23  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* combine.c (rtx_equal_for_field_assignment_p): Use
byte_lowpart_offset.

From-SVN: r253997

Daily bump.

From-SVN: r253996

i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost latencies instead of having separate table...

* i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
latencies instead of having separate table; make difference between
integer and float costs.
* i386.h (processor_costs): Remove scalar_stmt_cost,
scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
vec_store_cost.
* x86-tune-costs.h: Remove entries which has been removed in
procesor_costs from all tables; make cond_taken_branch_cost
and cond_not_taken_branch_cost COST_N_INSNS based.

From-SVN: r253993

SUBREG_PROMOTED_VAR_P handling in expand_direct_optab_fn

This is needed by the later SVE LAST reductions, where an 8-bit
or 16-bit result is zero- rather than sign-extended to 32 bits.
I think it could occur in other situations too.

2017-09-19  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* internal-fn.c (expand_direct_optab_fn): Don't assign directly
to a SUBREG_PROMOTED_VAR.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r253992

Make more use of GET_MODE_UNIT_PRECISION

This patch is like the earlier GET_MODE_UNIT_SIZE one,
but for precisions rather than sizes.  There is one behavioural
change in expand_debug_expr: we shouldn't use lowpart subregs
for non-scalar truncations, since that would just reinterpret
some of the scalars and drop the rest.  (This probably doesn't
trigger in practice.)  Using TRUNCATE is fine for scalars,
since simplify_gen_unary knows when a subreg can be used.

2017-10-22  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
(expand_debug_source_expr): Likewise.
* combine.c (combine_simplify_rtx): Likewise.
* cse.c (fold_rtx): Likewise.
* optabs.c (expand_float): Likewise.
* simplify-rtx.c (simplify_unary_operation_1): Likewise.
(simplify_binary_operation_1): Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r253991

Make more use of HWI_COMPUTABLE_MODE_P

This patch uses HWI_COMPUTABLE_MODE_P (X) instead of
GET_MODE_PRECISION (X) <= HOST_BITS_PER_WIDE_INT in cases
where X also needs to be a scalar integer.

2017-10-22  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
(record_promoted_value): Likewise.
* expr.c (expand_expr_real_2): Likewise.
* ree.c (update_reg_equal_equiv_notes): Likewise.
(combine_set_extension): Likewise.
* rtlanal.c (low_bitmask_len): Likewise.
* simplify-rtx.c (neg_const_int): Likewise.
(simplify_binary_operation_1): Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r253990

Make more use of subreg_size_lowpart_offset

This patch uses subreg_size_lowpart_offset in places that open-coded
the calculation.  The reload use (and the LRA one that was based on it)
seemed to ignore the BYTES_BIG_ENDIAN != WORDS_BIG_ENDIAN case; it's not
obvious whether that was deliberate or an oversight.

2017-10-22  Richard Sandiford  <richard.sandiford@linaro.org>
    Alan Hayward  <alan.hayward@arm.com>
    David Sherwood  <david.sherwood@arm.com>

gcc/
* lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
* regcprop.c (maybe_mode_change): Likewise.
* reload1.c (alter_reg): Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r253989

Add wide_int version of inchash::hash::add_wide_int

This patch adds an inchash hasher for wide_int-based types.
It means that hash_tree no longer hashes TREE_INT_CST_EXT_NUNITS,
but that was redundant with hashing the type.

2017-10-22  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* inchash.h (inchash::hash::add_wide_int): New function.
* lto-streamer-out.c (hash_tree): Use it.

From-SVN: r253988

Rename inchash::hash::add_wide_int

The name inchash::add_wide_int is a bit misleading, since it sounds
like it's hashing a wide_int.  This patch renames it to add_hwi instead.

2017-10-22  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* inchash.h (inchash::hash::add_wide_int): Rename to...
(inchash::hash::add_hwi): ...this.
* ipa-devirt.c (hash_odr_vtable): Update accordingly.
(polymorphic_call_target_hasher::hash): Likewise.
* ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
(sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
(sem_item_optimizer::update_hash_by_addr_refs): Likewise.
* lto-streamer-out.c (hash_tree): Likewise.
* optc-save-gen.awk: Likewise.
* tree.c (add_expr): Likewise.

From-SVN: r253987