sanitizer: do not inline no-sanitize into sanitizer fn

gcc/ChangeLog:

* cif-code.def (ATTRIBUTE_MISMATCH): Rename to...
(SANITIZE_ATTRIBUTE_MISMATCH): ...this.
* ipa-inline.c (sanitize_attrs_match_for_inline_p):
Handle all sanitizer options.
(can_inline_edge_p): Use renamed CIF_* enum value.

gcc/testsuite/ChangeLog:

* c-c++-common/asan/inline.c: New test.
* c-c++-common/asan/inline-kernel.c: New test.
* c-c++-common/tsan/inline.c: New test.
* c-c++-common/ubsan/inline.c: New test.

AArch64+SVE: Add support for unpacked unary ops and BIC

MD patterns extended for unary ops ABS, CLS, CLZ, CNT, NEG and NOT
to support unpacked vectors. Also extended patterns for BIC to
support unpacked vectors where input elements are of the same width.

gcc/ChangeLog:

2020-06-09  Joe Ramsay  <joe.ramsay@arm.com>

* config/aarch64/aarch64-sve.md (<optab><mode>2): Add support for
unpacked vectors.
(@aarch64_pred_<optab><mode>): Add support for unpacked vectors.
(@aarch64_bic<mode>): Enable unpacked BIC.
(*bic<mode>3): Enable unpacked BIC.

gcc/testsuite/ChangeLog:

2020-06-09  Joe Ramsay  <joe.ramsay@arm.com>

* gcc.target/aarch64/sve/logical_unpacked_abs.c: New test.
* gcc.target/aarch64/sve/logical_unpacked_bic_1.c: New test.
* gcc.target/aarch64/sve/logical_unpacked_bic_2.c: New test.
* gcc.target/aarch64/sve/logical_unpacked_bic_3.c: New test.
* gcc.target/aarch64/sve/logical_unpacked_bic_4.c: New test.
* gcc.target/aarch64/sve/logical_unpacked_neg.c: New test.
* gcc.target/aarch64/sve/logical_unpacked_not.c: New test.

gcov: improve --coverage small example

Pushed to master.

gcc/ChangeLog:

PR gcov-profile/95365
* doc/gcov.texi: Compile and link one example in 2 steps.

libgcov: fix TOPN type casting

The patch fixes tree-prof.exp tests on solaris11 and i686-linux-gnu,
problem was that sizeof of a pointer is different from sizeof gcov_type.

I'm going to install it if there are no objections.
Thanks,
Martin

libgcc/ChangeLog:

PR gcov-profile/95494
* libgcov-driver.c (write_top_counters): Cast first to
intptr_t as sizeof(*) != sizeof(gcov_type).
* libgcov.h (gcov_counter_set_if_null): Remove.
(gcov_topn_add_value): Cast first to intptr_t and update
linked list directly.

[Ada] Missing check on private overriding of dispatching primitive

2020-06-09  Javier Miranda  <miranda@adacore.com>

gcc/ada/

* sem_ch6.adb (New_Overloaded_Entity): Add missing call to check
subtype conformance of overriding dispatching primitive.
* sem_eval.adb (Subtypes_Statically_Match): Handle derivations
of private subtypes.
* libgnat/g-exptty.adb, libgnat/g-exptty.ads
(Set_Up_Communications): Fix the profile since null-exclusion is
missing in the access type formals.
* sem_disp.ads (Check_Operation_From_Private_View): Adding
documentation.

[Ada] Small enhancement in XEinfo utility

2020-06-09  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* einfo.ads (XEINFO section): Update format description.
(Is_Subprogram_Or_Entry): Move pragma to regular section.
(Is_Subprogram_Or_Generic_Subprogram): Likewise.
* xeinfo.adb (Get_B4): Rename to...
(Get_B0): ...this.
(Translate_Expr): New procedure extracted from...
(XEinfo): ...here.  Try to apply Get_B0 first and then
call Translate_Expr to translate supported constructs.

[Ada] Implement AI12-0028: Import of variadic C functions

2020-06-09  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* exp_ch6.adb (Freeze_Subprogram): Deal with convention C_Family.
* freeze.adb (Freeze_Profile): Likewise.  Add missing guard.
* sem_mech.adb (Set_Mechanisms): Likewise.
* lib-xref.adb (Output_Import_Export_Info): Ditto for C_Variadic.
* repinfo.adb (List_Subprogram_Info): Likewise.
* sem_prag.adb (Set_Convention_From_Pragma): Move main checks for
Stdcall to...
(Process_Convention): ...here.  Add checks for C_Variadic.
* snames.ads-tmpl: Add Name_C_Variadic_0 .. Name_C_Variadic_16.
Use consistent format for subtype declarations.
(Convention_Id): Add Convention_C_Variadic_0 .. C_Variadic_16
and move Convention_CPP up.
(Convention_C_Family): New subtype of Convention_Id.
(Convention_C_Variadic): Likewise.
(Foreign_Convention): Use explicit upper bound.
Add pragma Inline for Is_Configuration_Pragma_Name,
Is_Function_Attribute_Name, Is_Internal_Attribute_Name
and Is_Procedure_Attribute_Name.
* snames.adb-tmpl (Get_Convention_Id): Deal with Name_Variadic_n.
(Get_Convention_Name): Deal with Convention_Variadic_n.
* types.h (Convention_Id): New typedef.
* xsnamest.adb (Name2): New variable.
(Is_Conv): New pattern.
(Get_Subt1): Likewise.
(Get_Subt2): Likewise.
Output subtypes of Convention_Id into the C header file.

[Ada] Code clean ups and comments updates

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* exp_ch3.adb, exp_ch4.adb, exp_ch6.adb, exp_ch9.adb,
exp_disp.adb, exp_util.adb: Add comments related to errors that
should be moved to semantic analysis. Also replace "?" with "??"
in warning messages.

[Ada] Improve handling of null unbounded strings

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* libgnat/a-strunb__shared.ads, libgnat/a-strunb__shared.adb
(Reference, Unreference): No-op for Empty_Shared_String.
Remove unneeded calls to Reference.

[Ada] Remove kludge for AI05-0087

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* exp_ch5.adb (Expand_N_Assignment): Remove kludge for
AI05-0087.
* sem_ch12.adb (Validate_Derived_Type_Instance): Implement
AI05-0087 retroactively since it's a binding interpretation.

[Ada] Small cleanup in Einfo unit

2020-06-09  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* einfo.ads (Has_Foreign_Convention): Fix description.
(Component_Alignment): Move around.
(Has_DIC): Likewise.
(Has_Interrupt_Handler): Likewise.
(Has_Invariants): Likewise.
(Is_Atomic_Or_VFA): Likewise.
(Next_Index): Likewise.
(Scope_Depth): Likewise.
(Init_Component_Size): Likewise.
(Init_Component_Location): Likewise.
(Init_Size): Likewise.
(Inline Pragmas for functions): Add Corresponding_Function,
Corresponding_Procedure, Entry_Max_Queue_Lengths_Array,
Finalize_Storage_Only, Has_DIC, Has_Invariants,
Initialization_Statements, Is_Anonymous_Access_Type,
Next_Stored_Discriminant, Address_Clause, Alignment_Clause,
Float_Rep, Has_Foreign_Convention, Has_Non_Limited_View,
Is_Constant_Object, Is_Discriminal, Is_Finalizer, Is_Null_State,
Is_Prival, Is_Protected_Component, Is_Protected_Record_Type,
Is_Subprogram_Or_Entry, Is_Task_Record_Type, Size_Clause,
Stream_Size_Clause, Type_High_Bound, Type_Low_Bound, Known_*,
Unknown_*.
(Inline Pragmas for procedures): Add Set_Corresponding_Function,
Set_Corresponding_Procedure, Set_Finalize_Storage_Only,
Set_Float_Rep, Set_Initialization_Statements,
Init_Normalized_First_Bit, Init_Normalized_Position,
Init_Normalized_Position_Max.
* einfo.adb (Was_Hidden): Move around.
(Is_Packed_Array): Likewise.
(Model_Emin_Value): Likewise.
(Model_Epsilon_Value): Likewise.
(Model_Mantissa_Value): Likewise.
(Model_Small_Value): Likewise.

[Ada] Ada2020 AI12-0282: Shared variable control aspects in generics

2020-06-09  Ed Schonberg  <schonberg@adacore.com>

gcc/ada/

* sem_ch12.adb (Check_Shared_Variable_Control_Aspects): Require
exact match between formal and actual for aspects Atomic and
Volatile only for formal derived types.

[Ada] Improve handling of aggregates in Side_Effect_Free

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* exp_util.adb (Side_Effect_Free): Improve handling of
N_Aggregate.

[Ada] Fix wrong type being used for range check generation

2020-06-09  Ghjuvan Lacambre  <lacambre@adacore.com>

gcc/ada/

* sem_res.adb (Resolve_Qualified_Expression): Use Subtype_Mark
type.

[Ada] Expand more others aggregates statically

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* exp_aggr.adb (Max_Aggregate_Size): New function to factorize
code.
(Convert_To_Positional, Aggr_Size_OK): Use Max_Aggregate_Size.

[Ada] Code refactoring on calls to Set_Debug_Info_Needed

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_util.ads, sem_util.adb (Set_Debug_Info_Defining_Id): New.
* exp_ch3.adb, exp_ch8.adb: Call Set_Debug_Info_Defining_Id when
relevant.

[Ada] Missing documentation for Returns_By_Ref

2020-06-09  Justin Squirek  <squirek@adacore.com>

gcc/ada/

* einfo.ads (Returns_By_Ref): Modify documentation to reflect
that Returns_By_Ref can be applied to E_Subprogram_Type
entities.

[Ada] Spurious overlap error on zero-sized arrays with -gnateV

2020-06-09  Justin Squirek  <squirek@adacore.com>

gcc/ada/

* exp_attr.adb (Expand_N_Attribute_Reference): Modify expansion
of 'Overlaps_Storage to take into account zero sized arrays.

[Ada] Write_Invocation_Graph_Vertex: include lib item name

2020-06-09  Bob Duff  <duff@adacore.com>

gcc/ada/

* bindo-graphs.adb, bindo-graphs.ads: For each invocation graph,
record the corresponding library graph.
* bindo-writers.adb (Write_Invocation_Graph_Vertex): Print the
lib item name. Remove library graph parameters.
* bindo-augmentors.adb, bindo-augmentors.ads,
bindo-builders.adb, bindo-diagnostics.adb,
bindo-diagnostics.ads, bindo-elaborators.adb: Remove library
graph parameters.

[Ada] Propagate DIC, Invariant and Predicate attributes to views

2020-06-09  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* checks.adb (Apply_Predicate_Check): Extend trick used for
aggregates to qualified aggregates and object declarations
* einfo.ads (Has_Own_DIC): Mention the underlying full view.
(Has_Own_Invariants): Likewise.
(Has_Predicates): Likewise.
* exp_util.adb (Build_DIC_Procedure_Declaration): Do not deal
with base types explicitly but with underlying full views.
(Build_Invariant_Procedure_Declaration): Likewise.
* sem_ch13.adb (Build_Predicate_Functions): Do not deal with
the full view manually but call Propagate_Predicate_Attributes
to propagate attributes to views.
(Build_Predicate_Function_Declaration): Likewise.
* sem_ch3.adb (Build_Assertion_Bodies_For_Type): Build bodies
for private full views with an underlying full view.
(Build_Derived_Private_Type): Small comment tweak.
(Complete_Private_Subtype): Call Propagate_Predicate_Attributes.
(Process_Full_View): Do not deal with base types explicitly for
DIC and Invariant attributes.  Deal with underlying full views
for them.  Call Propagate_Predicate_Attributes and deal with
underlying full views for them.
* sem_ch7.adb (Preserve_Full_Attributes): Do not cross propagate
DIC and Invariant attributes between full type and its base type.
Propagate Predicate attributes from the full to the private view.
* sem_ch9.adb (Analyze_Protected_Type_Declaration): Likewise.
(Analyze_Task_Type_Declaration): Likewise.
* sem_util.ads (Get_Views): Remove Full_Base parameter and add
UFull_Typ parameter.
(Propagate_Predicate_Attributes): New procedure.
* sem_util.adb (Get_Views): Remove Full_Base parameter and add
UFull_Typ parameter.  Retrieve the Corresponding_Record_Type
from the underlying full view, if any.
(Propagate_DIC_Attributes): Remove useless tests.
(Propagate_Invariant_Attributes): Likewise.
(Propagate_Predicate_Attributes): New procedure.

[Ada] Crash on exit statement within predicated loop

2020-06-09  Justin Squirek  <squirek@adacore.com>

gcc/ada/

* exp_ch5.adb (Expand_Predicated_Loop): Perserve the original
loop identifier within the expansion.

[Ada] gnatbind: Correct assertions in Add_Edge_Kind_Check

2020-06-09  Bob Duff  <duff@adacore.com>

gcc/ada/

* bindo-graphs.ads (Library_Graph_Edge_Kind): Reorder enumerals
to reflect the order of adding edges. Clarify comments.
* bindo-graphs.adb (Add_Edge_Kind_Check): Correct the
assertions.  Reorder the "when"s to match the order of adding
edges, and therefore the order of enumerals in type
Library_Graph_Edge_Kind.  Change names to "Old_" and "New_" to
clarify what's what.  Combine Invocation_Edge into the "<="
test.  Fix the "raise Program_Error" message, which was
backwards.

[Ada] Add debugging message

2020-06-09  Bob Duff  <duff@adacore.com>

gcc/ada/

* bindo-graphs.adb (Add_Edge_Kind_Check): Add the Image of the
old and new Kinds to the raise Program_Error message.

[Ada] Remove bypass for instance bodies from Is_Visible_Component

2020-06-09  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* sem_ch3.adb (Is_Visible_Component): Do not special-case
bodies of instances.

[Ada] Membership test against a non-excluding subtype

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* exp_ch4.adb (Expand_N_In): Fix handling of null exclusion.

[Ada] Ada2020 AI12-0282: Shared variable control aspects in generics

2020-06-09  Ed Schonberg  <schonberg@adacore.com>

gcc/ada/

* sem_ch12.adb (Check_Shared_Variable_Control_Aspects): Require
exact match between formal and actual for aspects Atomic,
Atomic_Component, Volatile, and Volatile_Components.

[Ada] gnatpp: documentation for --no-separate-return switch

2020-06-09  Bob Duff  <duff@adacore.com>

gcc/ada/

* doc/gnat_ugn/gnat_utility_programs.rst: Add documentation for
the --no-separate-return switch of gnatpp.

[Ada] Add missing tag with -gnatw.d

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_warn.adb (Warn_On_Constant_Valid_Condition): Add proper
warning tag.

[Ada] Refine implementation of AI05-0149 missing conversion checks

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_type.adb (Covers): Fix implementation of AI05-0149.
* sem_res.adb: Fix typo.

[Ada] Spurious error on instantiations with Taft_Amendment types and tasks

2020-06-09  Ed Schonberg  <schonberg@adacore.com>

gcc/ada/

* exp_ch9.adb (Build_Master_Renaming): Make name in renaming
declaration unique by adding a numeric suffix, to prevent
accidental name conflict when several instantiations of a
package containing an access_to_incomplete type that designate
tasks appear in the same scope.

[Ada] Annotate Ada.Synchronous_Barriers with SPARK_Mode => Off

2020-06-09  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* libgnarl/a-synbar.ads, libgnarl/a-synbar.adb,
libgnarl/a-synbar__posix.ads, libgnarl/a-synbar__posix.adb
(Ada.Synchronous_Barriers): Annotate with SPARK_Mode => Off.

[Ada] AI1-0201 Relational operators of static string types are now static

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_eval.adb (Eval_Relational_Op, Eval_String_Literal,
Eval_Type_Conversion): Relax rules on relational operators and
type conversions of static string types.

[Ada] Ada2020: AI12-0301 Predicates and Default_Value

2020-06-09  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_util.ads, sem_util.adb (Is_Partially_Initialized_Type):
Take Default_Value and Default_Component_Value into account.
* sem_ch3.adb (Analyze_Object_Declaration): Update comment.

[Ada] Disable assertion regarding Body_Before_Spec_Edge

2020-06-09  Bob Duff  <duff@adacore.com>

gcc/ada/

* bindo-graphs.adb (Add_Edge_Kind_Check): Disable failing part
of the assertion.

libsanitizer: use gnu++14

libsanitizer/ChangeLog:

* asan/Makefile.am: Replace gnu++11 with gnu++14.
* interception/Makefile.am: Likewise.
* libbacktrace/Makefile.am: Likewise.
* lsan/Makefile.am: Likewise.
* sanitizer_common/Makefile.am: Likewise.
* tsan/Makefile.am: Likewise.
* ubsan/Makefile.am: Likewise.
* asan/Makefile.in: Regenerate.
* interception/Makefile.in: Likewise.
* libbacktrace/Makefile.in: Likewise.
* lsan/Makefile.in: Likewise.
* sanitizer_common/Makefile.in: Likewise.
* tsan/Makefile.in: Likewise.
* ubsan/Makefile.in: Likewise.

gcc-changelog: fix deduction for root ChangeLog

contrib/ChangeLog:

* gcc-changelog/git_commit.py: Fix ChangeLog regex in order to
match the top-level ChangeLog.
* gcc-changelog/test_email.py: Add test.
* gcc-changelog/test_patches.txt: Likewise.

c-family: Fix up MEM_REF printing [PR95580]

The C FE in the MEM_REF printing ICEs if the type of the first argument
(which due to useless pointer conversions can be an arbitrary type) is a
pointer to an incomplete type.  The code just wants to avoid printing a cast
if it is a pointer to single byte elements.

2020-06-09  Jakub Jelinek  <jakub@redhat.com>

PR c/95580
* c-pretty-print.c (c_pretty_printer::unary_expression): Handle the
case when MEM_REF's first argument has type pointer to incomplete type.

* gcc.dg/pr95580.c: New test.

match.pd: Optimize ffs comparisons against constants [PR95527]

The following patch implements various optimizations of __builtin_ffs*
against constants.

2020-06-09  Jakub Jelinek  <jakub@redhat.com>

PR tree-optimization/95527
* match.pd (__builtin_ffs (X) cmp CST): New optimizations.

* gcc.dg/tree-ssa/pr95527.c: New test.

testsuite: Fix g++.dg/torture/pr95493.C FAIL on i686-linux with -O0.

2020-06-09  Jakub Jelinek  <jakub@redhat.com>

* g++.dg/torture/pr95493.C: Add -Wno-psabi -w to dg-additional-options.

xtensa: libgcc: fix PR target/95571

Rewrite uw_install_context without function calls to avoid register
spilling in _Unwind_RaiseException during return context installation.

2020-06-08  Max Filippov  <jcmvbkbc@gmail.com>
gcc/testsuite/
* g++.target/xtensa/pr95571.C: New test.
* g++.target/xtensa/xtensa.exp: New testsuite.

libgcc/
* config/xtensa/unwind-dw2-xtensa.c (uw_install_context): Merge
with uw_install_context_1.

rs6000/testsuite: Allow xxperm* instead of only vperm*

Some testcases failed (esp. with --with-cpu=power9) after my change to
prefer xxperm over vperm when all else is equal.  Fix that.  (This also
tightens the relevant REs somewhat).

2020-06-09  Segher Boessenkool  <segher@kernel.crashing.org>

gcc/testsuite/
* gcc.target/powerpc/fold-vec-perm-char.c: Allow both vperm/vpermr and
xxperm/xxpermr.
* gcc.target/powerpc/fold-vec-perm-double.c: Ditto.
* gcc.target/powerpc/fold-vec-perm-float.c: Ditto.
* gcc.target/powerpc/fold-vec-perm-int.c: Ditto.
* gcc.target/powerpc/fold-vec-perm-longlong.c: Ditto.
* gcc.target/powerpc/fold-vec-perm-pixel.c: Ditto.
* gcc.target/powerpc/fold-vec-perm-short.c: Ditto.
* gcc.target/powerpc/lvsl-lvsr.c: Ditto.
* gcc.target/powerpc/vec-mult-char-2.c: Ditto.
* gcc.target/powerpc/vsx-vector-6.p9.c: Also allow xxpermr.

Daily bump.

openmp: ensure variables in offload table are streamed out (PRs 94848 + 95551)

gcc/ChangeLog:

PR lto/94848
PR middle-end/95551
* omp-offload.c (add_decls_addresses_to_decl_constructor,
omp_finish_file): Skip removed items.
* lto-cgraph.c (output_offload_tables): Likewise; set force_output
to this node for variables and functions.

libgomp/ChangeLog:

PR lto/94848
PR middle-end/95551
* testsuite/libgomp.fortran/target-var.f90: New test.

bootstrap: Fix --disable-bootstrap with older g++.

Previously I had AX_CXX_COMPILE_STDCXX in the gcc directory configure, which
added -std=c++11 to CXX if needed, but then CXX is overridden from the
toplevel directory, so it didn't have the desired effect.  Fixed by moving
the check to the toplevel.  Currently it is only used when building GCC
without bootstrapping; other packages that share the toplevel directory
can adjust the condition if they also want to require C++11 support.

/ChangeLog:

* configure.ac: Check AX_CXX_COMPILE_STDCXX if not bootstrapping.
* configure: Regenerate.

gcc/ChangeLog:

* aclocal.m4: Remove ax_cxx_compile_stdcxx.m4.
* configure.ac: Remove AX_CXX_COMPILE_STDCXX.
* configure: Regenerate.

libstdc++: Fix failing tests

These started failing with the previous commit, because I forgot to add
the tests after adjusting them.

* testsuite/20_util/default_delete/48631_neg.cc: Adjust dg-error
line number.
* testsuite/20_util/default_delete/void_neg.cc: Likewise.

Add missing ChangeLog entries

libstdc++: Implement operator<< for std::unique_ptr (LWG 2948)

libstdc++-v3/ChangeLog:

* include/bits/unique_ptr.h (operator<<): Define for C++20.
* testsuite/20_util/unique_ptr/io/lwg2948.cc: New test.

d: Fix regression caused by recent refactoring

gcc/d/ChangeLog:

PR d/95573
* dmd/MERGE: Merge upstream dmd 5041e56f1.

d: Merge upstream dmd 955b8b36f.

Merges AndAndExp and OrOrExp into a LogicalExp AST node.

gcc/d/ChangeLog:

* dmd/MERGE: Merge upstream dmd 955b8b36f.
* expr.cc (ExprVisitor::visit (AndAndExp *)): Rename type to ...
(ExprVisitor::visit (LogicalExp *)): ... this.  Handle both 'and if'
and 'or if' expression nodes.
(ExprVisitor::visit (OrOrExp *)): Remove.

PR fortran/95195 - Fortran testcase should clean up afterwards

Change testcase to check error message (iomsg) at runtime, rather than to crash.

libgfortran/
PR fortran/95091
* io/transfer.c (finalize_transfer): Fix type in error message.

gcc/testsuite/
PR fortran/95195
* gfortran.dg/namelist_97.f90: Adjust testcase.

PR bootstrap/95555 - powepc64 bootstrap failure due to -Wmaybe-uninitialized in reload_cse_simplify_operands

gcc/ChangeLog:

* postreload.c (reload_cse_simplify_operands): Clear first array element
before using it.  Assert a precondition.

Fortran  : ICE in maybe_canonicalize_comparison_1 PR92993

This issue has been fixed by PR94090.  Add test case to ensure that
this does not re-occur.

2020-06-08  Mark Eggleston  <markeggleston@gcc.gnu.org>

gcc/testsuite/

PR fortran/92993
* gfortran.dg/pr92993.f90: New test.

forwprop: Ignore scalar mode vectors in simplify_vector_constructor [PR95528]

As mentioned in the PR, the problem is that at least the x86 backend asumes
that the vec_unpack* and vec_pack* optabs with integral modes are for the
AVX512-ish vector masks rather than for very small vectors done in GPRs.
The only other target that seems to have a scalar mode vec_{,un}pack* optab
is aarch64 as discussed in the PR, so there is also a condition for that.
All other targets have just vector mode optabs.

2020-06-08  Jakub Jelinek  <jakub@redhat.com>

PR target/95528
* tree-ssa-forwprop.c (simplify_vector_constructor): Don't use
VEC_UNPACK*_EXPR or VEC_PACK_TRUNC_EXPR with scalar modes unless the
type is vector boolean.

* g++.dg/opt/pr95528.C: New test.

testsuite: Fix up pr95548.C testcase.

2020-06-08  Jakub Jelinek  <jakub@redhat.com>

PR lto/95548
* g++.dg/torture/pr95548.C: Change from dg-do compile to dg-do link,
add return type for main, for __SIZEOF_INT128__ test with __uint128_t
enumerator constants and add a test with unsigned long long
enumerators for all targets.

AArch64: Expand on comment of stack-clash and implicit probing through LR.

This expands the comment on an assert we have in aarch64_layout_frame
and points to an existing comment somewhere else that has a much longer
explanation of what's going on.

Committed under the GCC Obvious rule.

gcc/ChangeLog:

* config/aarch64/aarch64.c (aarch64_layout_frame): Expand comments.

[arm] Fix vfp_operand_register for VFP HI regs

While looking at PR target/94743 I noticed an ICE when I tried to save
all the FP registers: this was because all HI registers wouldn't match
vfp_register_operand.

gcc/ChangeLog:

* config/arm/predicates.md (vfp_register_operand): Use VFP_HI_REGS
instead of VFP_REGS.

rs6000: Replace FAIL with gcc_unreachable

gcc/ChangeLog:

* config/rs6000/vector.md: Replace FAIL with gcc_unreachable
in all vcond* patterns.

[arm] (header usage fix) include c++ algorithm header via system.h

After the recent commit that forces uses of c++11, the arm part failed
to build because it does not include <algorithm> via system.h as
should be done.

This results in:
from /gcc/common/config/arm/arm-common.c:34:
/usr/lib/gcc/x86_64-linux-gnu/5/include/mm_malloc.h:42:12: error:
attempt to use poisoned "malloc"
     return malloc (size);

This patch fixes the problem by defining INCLUDE_ALGORITHM before
including system.h and no longer includes <algorithm> directly.

gcc/ChangeLog:

* common/config/arm/arm-common.c (INCLUDE_ALGORITHM):
Define. No longer include <algorithm>.

[Ada] Implement AI12-0073 (Ravenscar disallows Synchronous_Barriers)

2020-06-08  Steve Baird  <baird@adacore.com>

gcc/ada/

* sem_prag.adb (Analyze_Pragma.Set_Ravenscar_Profile): Add
appropriate call to Set_Restriction_No_Dependence if Ada_Version
>= Ada2012 and Profile is either Ravenscar or a GNAT-defined
Ravenscar variant (i.e., not Jorvik).

[Ada] AI12-0204 Renaming of a prefixed view

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_ch5.adb: Fix typo.
* sem_ch8.adb (Analyze_Renamed_Primitive_Operation): Check that
the prefix of a prefixed view must be renamable as an object.

[Ada] AI12-0085 Missing aspect cases for Remote_Types

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_cat.ads: Fix typo.
* sem_cat.adb (Validate_Remote_Access_To_Class_Wide_Type): Add
handling of N_Attribute_Definition_Clause.
* sem_ch13.adb (Analyze_Attribute_Definition_Clause): Call
Validate_Remote_Access_To_Class_Wide_Type for Storage_Size and
Storage_Pool.
* sem_attr.adb, exp_ch4.adb: Update comments.

[Ada] Spurious error on call to controlled primitive

2020-06-08  Justin Squirek  <squirek@adacore.com>

gcc/ada/

* sem_ch4.adb (Analyze_One_Call): Add extra condition to the
predicate for deciding when a given controlled call is visible.

[Ada] Port a modified expansion of Enum_Rep from GNAT to GNATprove

2020-06-08  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* exp_spark.adb (Expand_SPARK_N_Attribute_Reference): Port
changes in frontend expander.

[Ada] Style cleanups in new code for Pure_Barriers

2020-06-08  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* sem_ch13.adb (All_Membership_Choices_Static): Fix style.

[Ada] Implement AI12-0291 (Jorvik profile)

2020-06-08  Steve Baird  <baird@adacore.com>

gcc/ada/

* libgnat/s-rident.ads: Add Jorvik to the Profile_Name
enumeration type.  Add an element for Jorvik to the array
aggregate that is the initial value of the constant
Profile_Info.
* targparm.adb (Get_Target_Parameters): Handle "pragma Profile
(Jorvik);" similarly to "pragma Profile (Ravenscar);".
* snames.ads-tmpl: Declare Name_Jorvik Name_Id. Unlike
Ravenscar, Jorvik is not a pragma name and has no corresponding
element in the Pragma_Id enumeration type; this means that its
declaration must not occur between those of First_Pragma_Name
and Last_Pragma_Name.
* sem_prag.adb (Analyze_Pragma): Add call to
Set_Ravenscar_Profile for Jorvik, similar to the existing calls
for Ravenscar and the GNAT Ravenscar variants.

[Ada] AI12-0228 Properties of qualified expressions used as names

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_ch8.adb (Analyze_Object_Renaming): Update Get_Object_Name
to go through N_Qualified_Expression and N_Type_Conversion.  Fix
another case of wrong usage of E_Anonymous_Access_Type instead
of Anonymous_Access_Kind.
* sem_util.adb (Is_Dependent_Component_Of_Mutable_Object): Work
on the original node.
(Is_Aliased_View): Take into account N_Qualified_Expression.

[Ada] Fix handling of 'Enum_Rep and renamings

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_eval.adb (Eval_Type_Conversion): Fix handling of
enumeration to integer conversions.
* exp_attr.adb (Expand_N_Attribute_Reference
[Attribute_Enum_Rep]): Remove special casing for first-level
renaming, best left to the general folding mechanism via
Eval_Type_Conversion.

[Ada] Remove the Has_Dynamic_Range_Check flag

2020-06-08  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* atree.adb (New_Copy): Do not clear Has_Dynamic_Range_Check.
* checks.ads (Append_Range_Checks): Remove Flag_Node parameter.
(Insert_Range_Checks): Likewise and remove default value of
Static_Loc parameter.
* checks.adb (Append_Range_Checks): Remove Flag_Node parameter.
Do not test and set Has_Dynamic_Range_Check.
(Insert_Range_Checks): Likewise and remove default value of
Static_Loc parameter.
* csinfo.adb (CSinfo): Remove 'L' from [NEUB]_Fields pattern and
do not handle Has_Dynamic_Range_Check.
* exp_ch5.adb (Expand_N_Assignment_Statement): Remove argument
in call to Insert_Range_Checks.
* sem_ch3.adb (Analyze_Subtype_Declaration): Do not fiddle
with Has_Dynamic_Range_Check.
(Process_Range_Expr_In_Decl): Remove argument in calls to
Insert_Range_Checks and Append_Range_Checks.
* sinfo.ads (Has_Dynamic_Range_Check): Delete.
(Set_Has_Dynamic_Range_Check): Likewise.
* sinfo.adb (Has_Dynamic_Range_Check): Delete.
(Set_Has_Dynamic_Range_Check): Likewise.
* treepr.adb (Print_Node): Do not print Has_Dynamic_Range_Check.

[Ada] Implement AI12-0290 (Simple_Barriers restriction)

2020-06-08  Steve Baird  <baird@adacore.com>

gcc/ada/

* sem_ch13.ads: Export new function
All_Membership_Choices_Static.
* sem_ch13.adb: Implement new function
All_Membership_Choices_Static.  This involves moving the
functions Is_Static_Choice and All_Membership_Choices_Static,
which were previously declared within the function
Is_Predicate_Static, out to library level so that they can be
called by the new function. The already-exisiting code in
Is_Predicate_Static which became the body of
All_Membership_Choices_Static is replaced with a call to the new
function in order to avoid duplication.
* exp_ch9.adb (Is_Pure_Barrier): Several changes needed to
implement rules of AI12-0290 and RM D.7's definition of
"pure-barrier-eligible". These changes include adding a call to
the new function Sem_13.All_Membership_Choices_Static, as per
the "see 4.9" in RM D.7(1.6/5).

[Ada] Issue with unnesting of 'First/Last and renaming

2020-06-08  Richard Kenner  <kenner@adacore.com>

gcc/ada/

* exp_unst.adb (Visit_Node): When visiting array attribute
nodes, in addition to checking the type of Get_Referenced_Object
of the prefix, also check the actual type of the prefix.

[Ada] Implement predicate checks on qualified expressions (AI12-0100)

2020-06-08  Gary Dismukes  <dismukes@adacore.com>

gcc/ada/

* checks.adb (Apply_Predicate_Check): Refine test for being in a
subprogram body to account for no Corresponding_Body case,
avoiding blowups arising due to other changes here.
* exp_ch4.adb (Expand_N_Qualified_Expression): Apply predicate
checks, if any, after constraint checks are applied.
* sem_eval.ads (Check_Expression_Against_Static_Predicate): Add
Check_Failure_Is_Error formal for conditionalizing warning vs.
error messages.
* sem_eval.adb (Check_Expression_Against_Static_Predicate):
Issue an error message rather than a warning when the new
Check_Failure_Is_Error formal is True. In the nonstatic or
Dynamic_Predicate case where the predicate is known to fail,
emit the check to ensure that folded cases get checks applied.
* sem_res.adb (Resolve_Qualified_Expression): Call
Check_Expression_Against_Static_Predicate, passing True for
Check_Failure_Is_Error, to ensure we reject static predicate
violations. Remove code that was conditionally calling
Apply_Predicate_Check, which is no longer needed, and that check
procedure shouldn't be called from a resolution routine in any
case. Also remove associated comment about preventing infinite
recursion and consistency with Resolve_Type_Conversion, since
that handling was already similarly removed from
Resolve_Type_Convesion at some point.
(Resolve_Type_Conversion): Add passing of True for
Check_Failure_Is_Error parameter on call to
Check_Expression_Against_Static_Predicate, to ensure that static
conversion cases that violate a predicate are rejected as
errors.

[Ada] Restore Snames.Name_SPARK as it used in the GNATprove backend

2020-06-08  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* snames.ads-tmpl (Name_SPARK): Restore after being deleted.

[Ada] Remove processing of SPARK_05 restriction

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* exp_aggr.adb, exp_ch6.adb, par-ch11.adb, par-ch6.adb,
par-ch7.adb, par-prag.adb, restrict.adb, restrict.ads,
scans.ads, scng.adb, sem_aggr.adb, sem_attr.adb, sem_ch11.adb,
sem_ch12.adb, sem_ch3.adb, sem_ch3.ads, sem_ch4.adb,
sem_ch5.adb, sem_ch6.adb, sem_ch7.adb, sem_ch8.adb, sem_ch9.adb,
sem_res.adb, sem_util.adb, sem_util.ads, snames.ads-tmpl,
gnatbind.adb, libgnat/s-rident.ads,
doc/gnat_rm/standard_and_implementation_defined_restrictions.rst:
Remove processing of SPARK_05 restriction.
* gnat_rm.texi: Regenerate.
* opt.ads: Remove processing of old checksum which is now
handled by gprbuild directly.

[Ada] AI12-0287 Legality Rules for null exclusions in renaming are too fierce

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_ch12.adb (Instantiate_Object): Relax rules related to null
exclusions and generic objects. Handle all anonymous types
consistently and not just E_Anonymous_Access_Type.
* sem_ch8.adb (Analyze_Object_Renaming): Change wording so that
it applies to both renamings and instantiations to avoid
confusion.

[Ada] Ada.Text_IO: fix typo

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* libgnat/a-textio.ads (File_Mode): Fix typo in comment.

[Ada] Do not warn on partial access to atomic object with address clause

2020-06-08  Eric Botcazou  <ebotcazou@adacore.com>

gcc/ada/

* sem_res.adb (Resolve_Indexed_Component): Do not give a warning
for a nonatomic component of an atomic array which is subject to
an address clause in Ada 2020 mode.
(Resolve_Selected_Component): Likewise for an atomic record.

[Ada] Add gnatname use for multiple units files support

2020-06-08  Philippe Gil  <gil@adacore.com>

gcc/ada/

* doc/gnat_ugn/the_gnat_compilation_model.rst: in "Handling
Files with Multiple Units" part documents gnatname use for
unmodified files handling and gnatchop use for files
refactoring.
* gnat_ugn.texi: Regenerate.

[Ada] Update doc on Enum_Rep/Enum_Val

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* doc/gnat_rm/implementation_defined_attributes.rst:
Enum_Rep/Enum_Val are standard Ada 202x attributes.
* gnat_rm.texi: Regenerate.

[Ada] Better code generation for nested aggregates

2020-06-08  Javier Miranda  <miranda@adacore.com>

gcc/ada/

* exp_aggr.adb (Safe_Component): Remove code that considers as
unsafe components that are aggregates; such removal allows the
frontend to proceed and evaluate if they are safe by means of
invoking Safe_Aggregate.

[Ada] Reuse standard expansion of 'First and 'Last in GNATprove mode

2020-06-08  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* exp_spark.adb (Expand_SPARK_N_Attribute_Reference): Apply
standard expansion to attributes First and Last.

[Ada] Reuse Get_Index_Subtype in the special expander for GNATprove

2020-06-08  Piotr Trojanek  <trojanek@adacore.com>

gcc/ada/

* exp_attr.adb, exp_util.ads, exp_util.adb (Get_Index_Subtype):
Move from the body of Exp_Attr to Exp_Util and expose from the
spec.
* exp_spark.adb (Expand_SPARK_N_Attribute_Reference): Replace
duplicated code with a call to Get_Index_Subtype.

[Ada] AI12-0226 Make objects more consistent

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* sem_ch8.adb (Analyze_Object_Renaming): Simplify code by moving
many special cases to Is_Object_Reference and removing others by
only checking renamings coming from sources.
* sem_util.adb (Is_Object_Reference): Update for AI12-0226 and
add more regular handling of 'Priority. Remove special cases no
longer needed now that we are only checking renamings coming
from sources.

[Ada] Silence spurious warning on instances of formal vectors

2020-06-08  Claire Dross  <dross@adacore.com>

gcc/ada/

* libgnat/a-cofove.adb (Insert_Space): The computation of Index
generates a spurious compiler warning about a value not being in
range for a statically dead branch.  Silence it using pragma
Warnings.

[Ada] gnatbind: Deterministic No_Entry_Calls_In_Elaboration_Code messages

2020-06-08  Bob Duff  <duff@adacore.com>

gcc/ada/

* bindo-graphs.adb (function Add_Edge): Rename
Add_Edge_With_Return to Add_Edge; we can tell it returns because
it's a function, and overloading seems appropriate in this case.
If Activates_Task=True, and we're not going to add a new edge
because an existing Pred-->Succ edge already exists, then set
Activates_Task to True on the preexisting edge.  This ensures
that the message:

info: use pragma Restrictions (No_Entry_Calls_In_Elaboration_Code)

appears when appropriate, no matter in what order the edges
happened to be processed.
(procedure Add_Edge): Remove redundant assertions.
(Activates_Task): Other kinds of edges can have
Activates_Task=True.  For example, if we had a With_Edge and
then an Invocation_Edge with Activates_Task=True, then the
With_Edge has Activates_Task set to True.
(Add_Edge_Kind_Check): New procedure to prevent other bugs of
this nature. For example, if we were to sometimes call Add_Edge
for a Spec_Before_Body_Edge followed by Add_Edge for a
With_Edge, and sometimes in the other order, that would cause a
similar bug to what we're fixing here.
(Set_Is_Recorded_Edge): Val parameter is not used. Get rid of
it.
(Set_Activates_Task): New procedure to set the Activates_Task flag.
* bindo-graphs.ads (Library_Graph_Edge_Kind): Reorder the
enumeration literals to facilitate Add_Edge_Kind_Check.
* ali.adb (Known_ALI_Lines): The comment about "still available"
was wrong. Fix that by erasing the comment, and encoding the
relevant information in real code. Take advantage of Ada's full
coverage rules by removing "others =>".  Also DRY.

[Ada] Ada_2020: shared variable control aspects on formal derived types

2020-06-08  Ed Schonberg  <schonberg@adacore.com>

gcc/ada/

* par-ch12.adb (P_Formal_Derived_Type_Definition): Handle
properly formal derived types that include aspect
specifications, so that the "with" keyword appears twice in the
formal type declaration.
* sem_ch13.adb (Has_Generic_Parent): Return true if the type
itself is a generic formal.

[Ada] Fix socket timeout correction for Windows Server 2019 case

2020-06-08  Dmitriy Anisimkov  <anisimko@adacore.com>

gcc/ada/

* socket.c (__gnat_minus_500ms): Remove
IsWindowsVersionOrGreater from condition.

[Ada] Add Depends contracts to Delete procedures of formal containers

2020-06-08  Claire Dross  <dross@adacore.com>

gcc/ada/

* libgnat/a-cfdlli.ads, libgnat/a-cfhama.ads,
libgnat/a-cfhase.ads, libgnat/a-cforma.ads, libgnat/a-cforse.ads
(Delete): Add Depends contract.

[Ada] AI12-0309 Missing checks for pragma Suppress

2020-06-08  Arnaud Charlet  <charlet@adacore.com>

gcc/ada/

* snames.ads-tmpl (Name_Program_Error_Check,
Name_Tasking_Check): New constants.
* types.ads (Program_Error_Check, Tasking_Check): New constants.
(All_Checks): Update accordingly.

Daily bump.

d: Merge upstream dmd 73d8e2fec.

Renames the enum PROTKIND to Prot::Kind, updates all uses of the
original enum accordingly.

gcc/d/ChangeLog:

* dmd/MERGE: Merge upstream dmd 73d8e2fec.
* decl.cc (get_symbol_decl): Use new Prot::Kind enum.
* modules.cc (get_internal_fn): Likewise.

i386: Improve expansion of __builtin_parity

GCC currently hides the shift and xor reduction inside a backend
specific UNSPEC PARITY, making it invisible to the RTL optimizers until
very late during compilation.  It is normally reasonable for the
middle-end to maintain wider mode representations for as long as possible
and split them later, but this only helps if the semantics are visible
at the RTL-level (to combine and other passes), but UNSPECs are black
boxes, so in this case splitting early (during RTL expansion) is a
better strategy.

It turns out that that popcount instruction on modern x86_64 processors
has (almost) made the integer parity flag in the x86 ALU completely
obsolete, especially as POPCOUNT's integer semantics are a much better
fit to RTL.  The one remaining case where these transistors are useful
is where __builtin_parity is immediately tested by a conditional branch,
and therefore the result is wanted in a flags register rather than as
an integer.  This case is captured by two peephole2 optimizations in
the attached patch.

2020-06-07  Roger Sayle  <roger@nextmovesoftware.com>

gcc/ChangeLog:

* config/i386/i386.md (paritydi2, paritysi2): Expand reduction
via shift and xor to an USPEC PARITY matching a parityhi2_cmp.
(paritydi2_cmp, paritysi2_cmp): Delete these define_insn_and_split.
(parityhi2, parityqi2): New expanders.
(parityhi2_cmp): Implement set parity flag with xorb insn.
(parityqi2_cmp): Implement set parity flag with testb insn.
New peephole2s to use these insns (UNSPEC PARITY) when appropriate.

gcc/testsuite/ChangeLog:

* gcc.target/i386/parity-3.c: New test.
* gcc.target/i386/parity-4.c: Likewise.
* gcc.target/i386/parity-5.c: Likewise.
* gcc.target/i386/parity-6.c: Likewise.
* gcc.target/i386/parity-7.c: Likewise.
* gcc.target/i386/parity-8.c: Likewise.
* gcc.target/i386/parity-9.c: Likewise.

d: Merge upstream dmd 108ca1bcd.

Renames OutBuffer::peekString to OutBuffer::peekChars, and
OutBuffer::extractString to OutBuffer::extractChars.  All callers have
been updated as appropriate.

Reviewed-on: https://github.com/dlang/dmd/pull/11247

gcc/d/ChangeLog:

* dmd/MERGE: Merge upstream dmd 108ca1bcd.
* d-diagnostic.cc (expand_d_format): Adjust to use extractChars().
* d-frontend.cc (Loc::toChars): Likewise.
* d-lang.cc (deps_write): Likewise.
(d_parse_file): Likewise.
* decl.cc (d_mangle_decl): Likewise.
* intrinsics.cc (maybe_set_intrinsic): Likewise.

d: Merge upstream dmd b0df0e982

Adds a struct ParameterList to encapulate parameter and vararg
information in the front-end.

Reviewed-on: https://github.com/dlang/dmd/pull/11226

gcc/d/ChangeLog:

* dmd/MERGE: Merge upstream dmd b0df0e982.
* d-builtins.cc (build_frontend_type): Use VarArg for varargs_p.
* d-codegen.cc (declaration_type): Call TypeFunction::create with
argument VARARGnone.
(parameter_type): Likewise.
(d_build_call): Use new field names and member functions.
* d-target.cc (Target::cppParameterType): Call TypeFunction::create
with argument VARARGnone.
* types.cc (TypeVisitor::visit (TypeFunction *): Use new field names
and member functions.

PR fortran/95091 - Buffer overflows with submodules and long symbols

Add cast to fix bootstrap error with -Werror=sign-compare.

gcc/fortran/
PR fortran/95091
* class.c (gfc_hash_value): Add cast.

d: Merge upstream dmd 1831b24ff.

Converts some global and param fields from pointers to value types.

Reviewed-on: https://github.com/dlang/dmd/pull/11245

gcc/d/ChangeLog:

* dmd/MERGE: Merge upstream dmd 1831b24ff.
* d-lang.cc (d_init_options): Remove initialization of updated fields.
(d_handle_option): Adjust for new field types.

PR fortran/95091 - Buffer overflows with submodules and long symbols

With submodules, name mangling results in long internal symbols.  This
requires adjustment of the sizes of temporaries to avoid buffer overflows.

2020-06-07  Harald Anlauf  <anlauf@gmx.de>

gcc/fortran/
PR fortran/95091
* class.c (get_unique_type_string, gfc_hash_value): Enlarge
buffers, and check whether the strings returned by
get_unique_type_string() fit.

d: Merge upstream dmd cef1e7991.

Adds a DString type, a struct that has a compatible layout with D
strings.  Many parameters in the Global struct have been switched over
to this type, and users of these params have been adjust to use the
length or ptr field as appropriate.

gcc/d/ChangeLog:

* dmd/MERGE: Merge upstream dmd cef1e7991.
* d-lang.cc (d_parse_file): Adjust for new field types.

rs6000: allow cunroll to grow size according to -funroll-loop or -fpeel-loops

Previously, flag_unroll_loops was turned on at -O2 implicitly.  This
also turned on cunroll with allowance size increasing, and cunroll
will unroll/peel the loop even the loop is complex like code in PR95018.
With this patch, size growth for cunroll is allowed only for if -funroll-loops
or -fpeel-loops or -O3 is specified explicitly.

gcc/ChangeLog
2020-06-07  Jiufu Guo  <guojiufu@linux.ibm.com>

PR target/95018
* config/rs6000/rs6000.c (rs6000_option_override_internal):
Override flag_cunroll_grow_size.

Introduce flag_cunroll_grow_size for cunroll

Currently GIMPLE complete unroller(cunroll) is checking
flag_unroll_loops and flag_peel_loops to see if allow size growth.
Beside affects curnoll, flag_unroll_loops also controls RTL unroler.
To have more freedom to control cunroll and RTL unroller, this patch
introduces flag_cunroll_grow_size.  With this patch, we can control
cunroll and RTL unroller indepently.

gcc/ChangeLog
2020-06-07  Jiufu Guo  <guojiufu@linux.ibm.com>

* common.opt (flag_cunroll_grow_size): New flag.
* toplev.c (process_options): Set flag_cunroll_grow_size.
* tree-ssa-loop-ivcanon.c (pass_complete_unroll::execute):
Use flag_cunroll_grow_size.

Added test case for a PR which has been fixed in the meantime.

gcc/testsuite/ChangeLog:

PR tree-optimization/50439
* gfortran.dg/loop_interchange_2.f: New test.