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.
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.
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.
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.
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>.
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).
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.
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.
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.
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.
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.
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.
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).
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.