Steven G. Kargl [Tue, 23 Jul 2019 21:43:21 +0000 (21:43 +0000)]
arith.c (gfc_convert_integer, [...]): Move to ...
2019-07-23 Steven G. Kargl <kargl@gcc.gnu.org>
* arith.c (gfc_convert_integer, gfc_convert_real, gfc_convert_complex):
Move to ...
* primary.c (convert_integer, convert_real, convert_complex): ... here.
Rename and make static functions.
(match_integer_constant): Use convert_integer
(match_real_constant): Use convert_real.
(match_complex_constant: Use convert_complex.
* arith.h (gfc_convert_integer, gfc_convert_real, gfc_convert_complex):
Remove prototypes.
* array.c (match_array_cons_element): A BOZ cannot be a data
statement value. Jump to a common exit point.
* check.c (gfc_invalid_boz): New function. Emit error or warning
for a BOZ in an invalid context.
(boz_args_check): Move to top of file to prevent need of forward
declaration.
(is_boz_constant): New function. Check that BOZ expr is constant.
(gfc_b z2real): New function. In-place conversion of BOZ literal
constant to REAL in accordance to F2018.
(gfc_boz2int): New function. In-place conversion of BOZ literal
onstant to INTEGER in accordance to F2018.
(gfc_check_achar, gfc_check_char, gfc_check_float): Use gfc_invalid_boz.
Convert BOZ as needed.
(gfc_check_bge_bgt_ble_blt): Enforce F2018 requirements on BGE,
BGT, BLE, and BLT intrinsic functions.
(gfc_check_cmplx): Re-organize to check kind, if present, first.
Convert BOZ real and/or imaginary parts as needed in accordance to
F2018.
(gfc_check_complex): Use gfc_invalid_boz. Convert BOZ as needed.
(gfc_check_dcmplx, gfc_check_dble ): Convert BOZ as needed.
(gfc_check_dshift): Make dshift[lr] conform to F2018 standard.
gfc_check_float (gfc_expr *a)
(gfc_check_iand_ieor_ior): Make IAND, IEOR, and IOR conform to
F2018 standard.
(gfc_check_int): Conform to F2018 standard.
(gfc_check_intconv): Deprecate SHORT and LONG aliases for INT2 and
INT. Simply return for a BOZ argument. See gfc_simplify_intconv.
(gfc_check_merge_bits): Make MERGE_BITS conform to Fortran 2018
standard.
(gfc_check_real): Remove incorrect comment. Check kind, if present,
first. Simply return for a BOZ argument. See gfc_simplify_real.
(gfc_check_and): Re-do error handling for BOZ arguments. Remove
special casing ts.type != BT_INTEGER or BT_LOGICAL.
* decl.c (match_old_style_init): Check for BOZ in old-style
initialization. Issue error or warning depending on
-fallow-invalid-boz option. Issue error if variable is not an
INTEGER or REAL and the value is BOZ.
* expr.c (gfc_copy_expr): Copy a BT_BOZ gfc_expr.
(gfc_check_assign): Re-do error handling for a BOZ in an assignment
statement. Do in-place conversion of RHS based on LHS type of
INTEGER or REAL.
* gfortran.h (gfc_expr): Add a boz component. Remove is_boz component.
(gfc_boz2int, gfc_boz2real, gfc_invalid_boz): New prototypes.
* interface.c (gfc_extend_assign): Guard against replacing an
intrinsic involving a BOZ literal constant on RHS.
* invoke.texi: Doument -fallow-invalid-boz.
* lang.opt: New option. -fallow-invalid-boz.
* libgfortran.h (bt): Elevate BOZ to a basic type.
* misc.c (gfc_basic_typename, gfc_typename): Translate BT_BOZ to BOZ.
* primary.c (convert_integer, convert_real, convert_complex): to here.
Rename and make static functions.
* primary.c(match_boz_constant): Rewrite parsing of a BOZ. Re-do
error handling. Deprecate 'X' for hexidecimal and postfix notation.
Use -fallow-invalid-boz and gfc_invalid_boz to accept deprecated code.
* resolve.c (resolve_ordinary_assign): Rework a RHS that is a
BOZ literal constant. Use gfc_invalid_boz to allow previous
nonstandard behavior. Remove range checking of BOZ conversion.
* simplify.c (convert_boz): Remove function.
(simplify_cmplx): Remove conversion of BOZ constants, because
conversion is done in gfc_check_cmplx.
(gfc_simplify_float): Remove conversion of BOZ constant, because
conversion is done in gfc_check_float.
(simplify_intconv): Use gfc_boz2int to convert BOZ to INTEGER.
Remove range checking for BOZ conversion.
(gfc_simplify_real): Use k, if present, to determine kind. Convert
BOZ to REAL. Remove range checking for BOZ conversion.
target-memory.c (gfc_convert_boz): Rewrite to deal with convert of
a BOZ to a REAL value.
This is about 32/64b host and multilib support across the range of Darwin
systems.
Prior to Darwin8 (OS X 10.4), the toolchains support only PowerPC and only 32b.
On Darwin8 it is possible to target a 64b multilib, but with support limited
to a few of the main libraries on the system (not a recommended configuration).
From Darwin9 to Darwin17 (OSX 10.5 to 10.13) it is possible to have either
32 or 64b hosted toolchains, with support for a 64 or 32b multilib respectively.
On Darwin9 the kernel is 32b, but with support for 64b executables, so it's
conventional to build a 32b host toolchain supporting a 64b multilib. However
this is not enforced (merely a convention).
There is also some platform hardware supporting Darwin10/11 which is only 32b
and for which the same situation applies. However, from Darwin10 to Darwin17,
the majority of platform hardware supports a 64b kernel and it's conventional
to build a 64b host toolchain with support for a 32b multilib.
On/from Darwin18 (OS X 10.14), the development headers (in the SDK) no longer
expose the interfaces for the 32b multilib support (although sufficient runtime
support remains installed that the testsuite can be run for a 32b multilib).
The PR is raised against this latter situation since the absence of exposed
interfaces causes a 'default' bootstrap fail regardless of the availability of
the runtimes. Given the number of permutations, I felt it warranted a general
solution, especially since the current scheme of target headers and t-make
fragments has become somewhat messy.
The changes here enforce the single 32b PowerPC multilib for Darwin < 8 and the
single X86 64b multilib for Darwin >= 18. This means that there is no longer
any need to configure Darwin18+ '--disable-multilib', but also that if you want
to use the ability to continue to test the compiler's 32b multilib there, you
need to make a configuration targeting an earlier OS version (and using the
SDK from that).
gcc/
PR bootstrap/87030
* config.gcc (*-*-darwin*): Don't include CPU t-darwin here.
(i[34567]86-*-darwin*): Adjust to use biarch files. Produce
an error message if i686-darwin configuration is attempted for
Darwin >= 18.
(x86_64-*-darwin*): Switch to single multilib for Darwin >= 18.
(powerpc-*-darwin*): Use biarch files where needed.
(powerpc64-*-darwin*): Likewise.
* config/i386/darwin.h (REAL_LIBGCC_SPEC): Move to new biarch file.
(DARWIN_ARCH_SPEC, DARWIN_SUBARCH_SPEC): Revise for default single
arch case.
* config/i386/darwin32-biarch.h: New.
* config/i386/darwin64.h: Rename.
* gcc/config/i386/darwin64-biarch.h: To this.
* config/i386/t-darwin: Rename.
* gcc/config/i386/t-darwin32-biarch: To this.
* config/i386/t-darwin64: Rename.
* gcc/config/i386/t-darwin64-biarch: To this.
* config/rs6000/darwin32-biarch.h: New.
* config/rs6000/darwin64.h: Rename.
* config/rs6000/darwin64-biarch.h: To this.
(DARWIN_ARCH_SPEC, DARWIN_SUBARCH_SPEC): Revise for default single
arch case.
* config/rs6000/t-darwin8: Rename.
* config/rs6000/t-darwin32-biarch: To this.
* config/rs6000/t-darwin64 Rename.
* config/rs6000/t-darwin64-biarch: To this.
Ian Lance Taylor [Tue, 23 Jul 2019 17:20:36 +0000 (17:20 +0000)]
compiler: use correct value type in 2-case select send
In the channel-send case, the value to be sent may needs an
(implicit) type conversion to the channel element type. This CL
ensures that we use the correct value type for the send.
Richard Biener [Tue, 23 Jul 2019 10:00:24 +0000 (10:00 +0000)]
re PR tree-optimization/83518 (Missing optimization: useless instructions should be dropped)
2019-07-23 Richard Biener <rguenther@suse.de>
PR tree-optimization/83518
* tree-ssa-sccvn.c (vn_reference_lookup_3): Handle aggregate
init from a constant even when partial defs are already recorded.
c/
* gimple-parser.c (c_parser_parse_gimple_body): When we have
a CFG also rebuild cgraph edges.
Jan Hubicka [Tue, 23 Jul 2019 09:38:49 +0000 (11:38 +0200)]
i386-common.c: Use PROCESSOR_ZNVER2 scheduler for znver2.
* i386-common.c: Use PROCESSOR_ZNVER2 scheduler for znver2.
* config/i386/znver1.md: Enable patterns for znver2 and add store
variants which use extra AGU unit.
Ed Schonberg [Tue, 23 Jul 2019 08:13:41 +0000 (08:13 +0000)]
[Ada] Aspect CPU may depend on a discriminant of a task type
2019-07-23 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_ch13.adb (Check_Aspect_At_End_Of_Declarations,
Freeze_Entity_Checks): Include Aspect_CPU with other aspects
whose expresssion may depend on a discriminant, and thus require
that components of the type be made visible.
Eric Botcazou [Tue, 23 Jul 2019 08:13:32 +0000 (08:13 +0000)]
[Ada] Plug small loophole in Generate_Range_Check
The Generate_Range_Check routine is responsible for generating range
checks in the scalar case. It automatically deals with possible
overflow in the process when the source and the target base types are
different.
However there is one case where overflow is not dealt with correctly,
namely when the target base type is narrower than the source base type
and both are floating-point types. In this case, the routine will
convert the source type to the target base type without checking for
overflow. In practice this does not matter much because the conversion
would yield an infinity on overflow, which would then fail the
subsequent range check. However it's more correct to have a proper
overflow check with -gnateF than relying on the infinity.
2019-07-23 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* checks.adb (Convert_And_Check_Range): Add Suppress parameter
and pass it in the call to Insert_Actions. Rename local
variable.
(Generate_Range_Check): Minor comment fixes. Pass Range_Check
in the first call to Convert_And_Check_Range and All_Checks in
the second call.
* exp_ch4.adb (Expand_N_Type_Conversion): Reset the
Do_Overflow_Check flag in the float-to-float case too if there
is also a range check.
Eric Botcazou [Tue, 23 Jul 2019 08:13:26 +0000 (08:13 +0000)]
[Ada] Eliminate redundant overflow checks for conversions from fixed-point
This eliminates redundant overflow checks that are generated for
conversions from fixed-point to integer types when range checks are also
enabled (which is the default), as the former checks are subsumed into
the latter checks.
No functional changes.
2019-07-23 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* checks.adb (Activate_Overflow_Check): Remove redundant
argument.
* exp_ch4.adb (Discrete_Range_Check): Reset the overflow flag.
(Expand_N_Type_Conversion): Do not reset it here.
Eric Botcazou [Tue, 23 Jul 2019 08:13:21 +0000 (08:13 +0000)]
[Ada] Minor tweak to -gnatR output
This makes sure that the numbers present in the -gnatR output are
printed in decimal format in all cases, since the hexadecimal format is
not compatible with the JSON syntax.
2019-07-23 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* repinfo.adb (List_Component_Layout): Pass Decimal to UI_Write.
(Write_Val): Likewise.
Ed Schonberg [Tue, 23 Jul 2019 08:13:15 +0000 (08:13 +0000)]
[Ada] Iterators are view-specific
Operational aspects, such as Default_Iterator, are view-specific, and if
such an aspect appears on the full view of a private type, an object of
the type cannot be iterated upon if it is not in the scope of the full
view, This patch diagnoses properly an attempt to iterate over such an
object.
2019-07-23 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* aspects.ads: New table Operational_Aspect, used to distinguish
between aspects that are view-specific, such as those related to
iterators, and representation aspects that apply to all views of
a type.
* aspects.adb (Find_Aspect): If the aspect being sought is
operational, do not ecamine the full view of a private type to
retrieve it.
* sem_ch5.adb (Analyze_Iterator_Specification): Improve error
message when the intended domain of iteration does not implement
the required iterator aspects.
gcc/testsuite/
* gnat.dg/iter5.adb: Add an expected error.
* gnat.dg/iter6.adb: New testcase.
[Ada] Issue error on SPARK ownership rule violation
A modified rule in SPARK RM specifies that object declarations of
anonymous access type should only occur immediately in subprogram, entry
or block. Now checked.
There is no impact on compilation.
2019-07-23 Yannick Moy <moy@adacore.com>
gcc/ada/
* sem_spark.ads (Is_Local_Context): New function.
* sem_spark.adb (Check_Declaration): Issue errors on violations
of SPARK RM 3.10(4)
(Process_Path): Do not issue error on borrow/observe during
elaboration, as these are caught by the new rule.
Linking of an enabled ghost unit which requires a finalizer lead to an
error, as the name generated by the binder for calling the finalizer was
not the same as the name chosen by the compiler. Now fixed.
2019-07-23 Yannick Moy <moy@adacore.com>
gcc/ada/
* exp_ch7.adb (Create_Finalizer): Force finalizer not to be
Ghost enabled.
* exp_dbug.adb (Get_External_Name): Explain special case of
Ghost finalizer.
gcc/testsuite/
* gnat.dg/ghost6.adb, gnat.dg/ghost6_pkg.ads: New testcase.
Jan Beulich [Tue, 23 Jul 2019 08:07:49 +0000 (08:07 +0000)]
x86/AVX512: improve generated code for mask-to-vector-register conversions
Conversion of comparison results to full vectors does, when VPMOVM2* are
unavailable, not require any intermediate VMOVDQ{A,U}*: Simply use
embedded masking on VPTERNLOG* right away, which is available with
AVX512F (while VPMOVM2{D,Q} are available only with AVX512DQ).
Note that the chosen immediate is only one of many possible ones; I was
trying to make the insn here distinguishable from the pre-existing uses
of vpternlog.
gcc/
2019-07-23 Jan Beulich <jbeulich@suse.com>
* config/i386/sse.md (<avx512>_cvtmask2<ssemodesuffix><mode>):
Require only AVX512F.
(*<avx512>_cvtmask2<ssemodesuffix><mode>): Likewise. Add
alternative expanding to vpternlog.
Richard Biener [Tue, 23 Jul 2019 06:58:39 +0000 (06:58 +0000)]
tree-cfg.c (label_for_bb): Remove global var.
2019-07-23 Richard Biener <rguenther@suse.de>
* tree-cfg.c (label_for_bb): Remove global var.
(main_block_label): Take label_for_bb as argument.
(cleanup_dead_labels_eh): Likewise, adjust.
(cleanup_dead_labels): Adjust.
Ian Lance Taylor [Tue, 23 Jul 2019 00:34:38 +0000 (00:34 +0000)]
compiler: follow-on fix for finalizing imported methods
This patch is a revision to CL 185518, which added code to perform
finalization of methods on types created by the importer and not
directly reachable until inlining is done.
The original fix invoked the finalization code at the end of
Import::read_types(), but it turns out this doesn't handle the case
where a type with methods is read in due to a reference from something
later in the export data (a function or variable). The fix is to move
the import finalization call to the end of Import::import().
Jonathan Wakely [Mon, 22 Jul 2019 16:53:36 +0000 (17:53 +0100)]
Adjust std::rotl, std::rotr etc to match final P0553R4 proposal
This proposal has now been accepted for C++20, with a few changes. This
patch adjusts std::rotl and std::rotr to match the final specification
and declares the additions for C++2a mode even when __STRICT_ANSI__ is
defined.
* include/std/bit (__rotl, __rotr): Change second parameter from
unsigned int to int and handle negative values.
(rotl, rotr): Remove check for __STRICT_ANSI__. Change second
parameter from unsigned int to int. Add nodiscard attribute.
* testsuite/26_numerics/bit/bitops.rot/rotl.cc: Rename to ...
* testsuite/26_numerics/bit/bit.rotate/rotl.cc: Here. Test negative
shifts.
* testsuite/26_numerics/bit/bitops.rot/rotr.cc: Rename to ...
* testsuite/26_numerics/bit/bit.rotate/rotr.cc: Here. Test negative
shifts.
Jonathan Wakely [Mon, 22 Jul 2019 16:53:27 +0000 (17:53 +0100)]
Change std::ceil2 to be undefined if the result can't be represented
* include/std/bit (__ceil2): Make unrepresentable results undefined,
as per P1355R2. Add debug assertion. Perform one left shift, not two,
so that out of range values cause undefined behaviour. Ensure that
shift will still be undefined if left operand is promoted.
* testsuite/26_numerics/bit/bit.pow.two/ceil2.cc: Replace checks for
unrepresentable values with checks that they are not core constant
expressions.
* testsuite/26_numerics/bit/bit.pow.two/ceil2_neg.cc: New test.
Eric Botcazou [Mon, 22 Jul 2019 13:58:32 +0000 (13:58 +0000)]
[Ada] Fix missing check for no-op conversion to fixed-point type
This plugs a small loophole in the compiler for the case of a
multiplication or a division in a fixed-point type wrapped in a no-op
conversion, e.g. to the same fixed-point type.
The front-end fails to generate a range check for the operation. This
used to be caught by the back-end, which would generate the range check,
but this is no longer the case because we now make sure to reset the
Do_Range_Check flag in all cases before invoking the back-end.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* exp_ch4.adb (Expand_N_Type_Conversion): Beef up comment.
(Fixup_Universal_Fixed_Operation): Set the base type instead of
the type of the enclosing type conversion on the operation.
Ed Schonberg [Mon, 22 Jul 2019 13:58:27 +0000 (13:58 +0000)]
[Ada] Remove misleading warning/suggestion in membership test
This patch removes a warning on a membership test whose right operand is
given by a range. In many cases the check can be replaced by the use of
attribute 'Valid, but if the bounds of range are type conversion this
replacement would be invorrect and the warning and suggestion are
misleading.
2019-07-22 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* exp_ch4.adb (Expand_N_In): Do not suggest the use of attribute
'Valid as a replacement for a range check on a discrete type
when the bounds of the range are given by type conversions,
because in such a case there are distinct types involved and the
subbested attribute replacement would be misplaced.
[Ada] Adapt ownership checking in SPARK to traversal functions
A traversal function, especially when implemented as an expression
function, may need to return an if-expression or case-expression, while
still respecting Legality Rule SPARK RM 3.10(5). This case is now
allowed in GNATprove.
There is no impact on compilation.
2019-07-22 Yannick Moy <moy@adacore.com>
gcc/ada/
* sem_spark.adb (Get_Root_Object, Is_Path_Expression,
Is_Subpath_Expression): Add parameter Is_Traversal to adapt
these functions to the case of paths returned from a traversal
function.
(Read_Indexes): Handle the case of an if-expression or
case-expression.
(Check_Statement): Check Emit_Messages only when issuing an
error message. This is important as Emit_Messages may store the
information that an error was detected.
This ovehauls the code implementing conversions involving fixed-point
types in the front-end because it leaks the Do_Range_Check flag in
several places to the back-end, which is a violation of the documented
interface between front-end and back-end.
This also does a bit of housekeeping work throughout it in the process.
There should be essentially no functional changes.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* checks.adb (Apply_Type_Conversion_Checks): Do not set
Do_Range_Check flag on conversions from fixed-point types
either.
* exp_attr.adb: Add use and with clause for Expander.
(Expand_N_Attribute_Reference) <Fixed_Value, Integer_Value>: Set
the Conversion_OK flag and do not generate overflow/range checks
manually.
* exp_ch4.adb (Expand_N_Qualified_Expression): Remove
superfluous clearing of Do_Range_Check flag.
(Discrete_Range_Check): New procedure to generate a range check
for discrete types.
(Real_Range_Check): Remove redundant local variable and adjust.
Remove useless shortcut. Clear Do_Range_Check flag on all
paths.
(Expand_N_Type_Conversion): Remove redundant test on
Conversion_OK. Call Discrete_Range_Check to generate range
checks on discrete types. Remove obsolete code for
float-to-integer conversions. Add code to generate range checks
for conversions involving fixed-point types.
Javier Miranda [Mon, 22 Jul 2019 13:58:09 +0000 (13:58 +0000)]
[Ada] Spurious error passing access to class-wide interface type
The compiler reports an spurious error when the formal parameter of a
subprogram is an access to a class wide interface type and the actual
parameter is an allocator of an object covering such interface type.
2019-07-22 Javier Miranda <miranda@adacore.com>
gcc/ada/
* sem_res.adb (Resolve_Actuals): Replace code that displaces the
pointer to an allocated object to reference its secondary
dispatch table by a type conversion (which takes care of
handling all cases).
Eric Botcazou [Mon, 22 Jul 2019 13:58:04 +0000 (13:58 +0000)]
[Ada] Small enhancement to the -gnatD/-gnatG output for fixed-point types
This is a small enhancement to the -gnatD/-gnatG output: the base type
of fixed-point types, which is usually an itype, used to be printed as
??? in this case. It is now printed in a similar fashion as the first
subtype.
For the following package:
package P is
type D is delta 128.0 / (2 ** 15) range 0.0 .. 256.0;
end P;
the -gnatD/-gnatG must now be:
Source recreated from tree for P (spec)
---------------------------------------
p_E : short_integer := 0;
package p is
type p__d is delta [1.0/256.0] range 0.0 .. 256.0;
[type p__TdB is delta [1.0/256.0] range -[2147483648.0*2**(-8)] ..
[2147483647.0*2**(-8)]]
freeze p__TdB []
end p;
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* sprint.adb (Sprint_Node_Actual)
<N_Decimal_Fixed_Point_Definition>: Swap a couple of spaces.
(Write_Itype): Minor consistency fixes throughout. Add support
for printing ordinary and decimal fixed-point types and
subtypes.
Ed Schonberg [Mon, 22 Jul 2019 13:57:55 +0000 (13:57 +0000)]
[Ada] Optimization loses exception in improper use of 'Value
This patch prevents an improper removal of an evaluation of attribute
'Value on an illegal input that will raise Constraint_Error, when a
subsequent use of this evaluation might be optimized away by the
back-end.
2019-07-22 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* libgnat/s-valboo.ads, libgnat/s-valcha.ads,
libgnat/s-valdec.ads, libgnat/s-valenu.ads,
libgnat/s-valint.ads, libgnat/s-vallld.ads,
libgnat/s-vallli.ads, libgnat/s-valllu.ads,
libgnat/s-valrea.ads, libgnat/s-valuns.ads,
libgnat/s-valwch.ads: Change categorization of packages that
implement attribute 'Value from Pure to Preelaborate, to prevent
undesirable optimizations when the evaluation of the attribute
raises Constraint_Error, but subsequent use of the result of
this evsaluation is removed by a subsequent optimization.
Ed Schonberg [Mon, 22 Jul 2019 13:57:51 +0000 (13:57 +0000)]
[Ada] Misleading warning on variable not assigned
This patch removes a warning on a referenced entity with no explicit
prior assignment, if the type of the entity has
Preelaborable_Initialixation, such as Exception_Occurrence.
2019-07-22 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_warn.adb (Check_References): Do not emit s warning on a
referenced entity with no explicit assignment if the type of the
entity has Preelaborable_Initialixation, such as
Exception_Occurrence.
Javier Miranda [Mon, 22 Jul 2019 13:57:46 +0000 (13:57 +0000)]
[Ada] Usage of signed type in array bounds in CCG
2019-07-22 Javier Miranda <miranda@adacore.com>
gcc/ada/
* exp_ch4.adb (Size_In_Storage_Elements): Improve the expansion
to handle array indexes that are modular type.
(Expand_N_Allocator): For 32-bit targets improve the generation
of the runtime check associated with large arrays supporting
arrays initialized with a qualified expression.
* libgnat/s-imenne.adb (Image_Enumeration_8,
Image_Enumeration_16, Image_Enumeration_32): Define the index of
Index_Table with range Natural'First .. Names'Length since in
the worst case all the literals of the enumeration type would be
single letter literals and the Table built by the frontend would
have as many components as the length of the names string. As a
result of this enhancement, the internal tables declared using
Index_Table have a length closer to the real needs, thus
avoiding the declaration of large arrays on 32-bit CCG targets.
[Ada] Issue warning or error message on ignored typing constraint
GNAT ignores the discriminant constraint on a component when it applies
to the type of the record being analyzed. Now issue a warning on Ada
code when ignoring this constraint, or an error on SPARK code.
2019-07-22 Yannick Moy <moy@adacore.com>
gcc/ada/
* sem_ch3.adb (Constrain_Access): Issue a message about ignored
constraint.
Eric Botcazou [Mon, 22 Jul 2019 13:57:37 +0000 (13:57 +0000)]
[Ada] Fix spurious visibility error for tagged type with inlining
This fixes a spurious visibility error for the very peculiar case where
an operator that operates on the class-wide type of a tagged type is
declared in a package, the operator is renamed in another package where
a subtype of the tagged type is declared, and both packages end up in
the transititive closure of a unit compiled with optimization and
inter-inlining (-gnatn).
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* sem_ch8.adb (End_Use_Type): Reset the In_Use flag on the
class-wide type if the type is tagged.
(Use_One_Type): Add commentary on the handling of the class-wide
type.
gcc/testsuite/
* gnat.dg/inline17.adb, gnat.dg/inline17_pkg1.adb,
gnat.dg/inline17_pkg1.ads, gnat.dg/inline17_pkg2.ads,
gnat.dg/inline17_pkg3.adb, gnat.dg/inline17_pkg3.ads: New
testcase.
This change removes the Is_For_Access_Subtype machinery from the
compiler. This machinery was devised a long time ago to deal with a
peculiarity of the freezing for access-to-record subtypes but has been
degenerate for quite some time now and does not seem to serve any useful
purpose at this point.
Morever it has an annoying side effect whereby it causes Underlying_Type
to return the (unconstrained) base record type when invoked on the
designated record subtype, which is very problematic for GNATprove.
There should be no functional changes.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* einfo.ads (Is_For_Access_Subtype): Delete.
(Set_Is_For_Access_Subtype): Likewise.
* einfo.adb (Is_For_Access_Subtype): Likewise.
(Set_Is_For_Access_Subtype): Likewise.
(Write_Entity_Flags): Do not write Is_For_Access_Subtype.
* exp_ch4.adb (Expand_N_Selected_Component): Do not deal with
it.
* exp_spark.adb (Expand_SPARK_N_Selected_Component): Likewise.
* sem_ch4.adb (Analyze_Explicit_Dereference): Likewise.
* sem_ch3.adb (Build_Discriminated_Subtype): Do not build a
special private subtype for access-to-record subtypes.
Eric Botcazou [Mon, 22 Jul 2019 13:57:26 +0000 (13:57 +0000)]
[Ada] Spurious error on private subtype of derived access type
This patch fixes a spurious type error on a dynamic predicate on a
subtype of a private type whose full view is a derived access type.
Prior to it, the base type of the subtype would appear to be the parent
type of the derived type instead of the derived type itself, leading to
problems downstream.
The following package must now compile quietly:
with S;
package T is
type B_Pointer is private;
Null_B_Pointer : constant B_Pointer;
function OK (B : B_Pointer) return Boolean is (B /= Null_B_Pointer);
subtype Valid_B_Pointer is B_Pointer
with Dynamic_Predicate => OK (Valid_B_Pointer);
private
type B_Pointer is new S.A_Pointer;
Null_B_Pointer : constant B_Pointer := B_Pointer (S.Null_A_Pointer);
end;
package S is
type A_Type is new Integer;
type A_Pointer is access A_Type;
Null_A_Pointer : constant A_Pointer := null;
end;
Moreover, it also plugs a loophole in the compiler whereby an
instantiation of a generic with a formal subprogram declaration nested
in an enclosing generic package would be done even if there was a
mismatch between an original and a derived types involved in the
instantiation.
The compiler must now gives the following error:
p.adb:11:43: no visible subprogram matches the specification for "Action"
on
with Q;
with R;
with G;
procedure P is
package My_G is new G (Q.T);
procedure Proc (Value : R.T) is null;
procedure Iter is new My_G.Iteration_G (Proc);
begin
null;
end;
with R;
package Q is
type T is new R.T;
end Q;
package R is
type T is private;
private
type T is access Integer;
end R;
generic
type Value_T is private;
package G is
generic
with procedure Action (Value : Value_T);
procedure Iteration_G;
end G;
package body G is
procedure Iteration_G is null;
end G;
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* sem_ch3.adb (Complete_Private_Subtype): Rework the setting of
the Etype of the full view for full base types that cannot
contain any discriminant. Remove code and comment about it in
the main path.
Ed Schonberg [Mon, 22 Jul 2019 13:57:22 +0000 (13:57 +0000)]
[Ada] Type inconsistency in floating_point type declarations
This patch fixes an inconsistency in the typing of the bounds of a
floting point type declaration, when some bound is given by a dtatic
constant of an explicit type, instead of a real literal, Previous to
this patch the bound of the type retained the given type, leading to
spurious errors in Codepeer.
2019-07-22 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_ch3.adb (Convert_Bound): Subsidiary of
Floating_Point_Type_Declaration, to handle properly range
specifications with bounds that may include static constants of
a given type rather than real literals.
Eric Botcazou [Mon, 22 Jul 2019 13:57:18 +0000 (13:57 +0000)]
[Ada] Further fix non-stored discriminant in aggregate for GNATprove
GNATprove expects discriminants appearing in aggregates and their types
to be resolved to stored discriminants. This extends the machinery that
makes sure this is the case for default initialization expressions so as
to also handle component associations in these expressions.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* sem_aggr.adb (Rewrite_Bound): Be prepared for discriminals
too.
(Rewrite_Range;): Minor tweak.
(Resolve_Record_Aggregate): For a component with default
initialization whose expression is an array aggregate, also
rewrite the bounds of the component associations, if any.
Gary Dismukes [Mon, 22 Jul 2019 13:57:13 +0000 (13:57 +0000)]
[Ada] Premature finalization of controlled temporaries in case expressions
The compiler was generating finalization of temporary objects used in
evaluating case expressions for controlled types in cases where the case
statement created by Expand_N_Expression_With_Actions is rewritten as an
if statement. This is fixed by inheriting the From_Condition_Expression
flag from the rewritten case statement.
The test below must generate the following output when executed:
$ main
Xs(1): 1
----
package Test is
type E is (E1, E2);
procedure Test (A : in E);
end Test;
----
with Ada.Text_IO;
with Ada.Finalization;
package body Test is
type T is new Ada.Finalization.Controlled with
record
N : Natural := 0;
end record;
overriding procedure Finalize (X : in out T) is
begin
X.N := 42;
end Finalize;
type T_Array is array (Positive range <>) of T;
function Make_T (N : Natural) return T is
begin
return (Ada.Finalization.Controlled with N => N);
end Make_T;
X1 : constant T := Make_T (1);
X2 : constant T := Make_T (2);
procedure Test (A : in E)
is
Xs : constant T_Array := (case A is
when E1 => (1 => X1),
when E2 => (1 => X2));
begin
Ada.Text_IO.Put_Line ("Xs(1):" & Natural'Image (Xs (1).N));
end Test;
end Test;
----
with Test;
procedure Main is
begin
Test.Test (Test.E1);
end Main;
2019-07-22 Gary Dismukes <dismukes@adacore.com>
gcc/ada/
* exp_ch5.adb (Expand_N_Case_Statement): In the case where a
case statement is rewritten as an equivalent if statement,
inherit the From_Condition_Expression flag from the case
statement.
Eric Botcazou [Mon, 22 Jul 2019 13:57:09 +0000 (13:57 +0000)]
[Ada] Internal error on iterator for limited private discriminated type
This patch further extends the short-circuit, aka optimization, present
in the Check_Constrained_Object procedure used for renaming declarations
to all limited types, so as to prevent type mismatches downstream in
more cases.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* sem_ch8.adb (Check_Constrained_Object): Further extend the
special optimization to all limited types.
gcc/testsuite/
* gnat.dg/iter5.adb, gnat.dg/iter5_pkg.ads: New testcase.
Eric Botcazou [Mon, 22 Jul 2019 13:57:04 +0000 (13:57 +0000)]
[Ada] Fix missing Constraint_Error for Enum_Val attribute
This fixes an old issue involving the Enum_Val attribute: it does not
always raise a Constraint_Error exception when the specified value is
not valid for the enumeration type (instead a modulo computation is
applied to the value).
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* exp_attr.adb (Expand_N_Attribute_Reference)
<Attribute_Enum_Val>: Set No_Truncation on the
N_Unchecked_Type_Conversion built around the argument passed to
the attribute.
Eric Botcazou [Mon, 22 Jul 2019 13:56:50 +0000 (13:56 +0000)]
[Ada] Fix spurious loop warning for function with Out parameter
The compiler gives a spurious warning about a possible infinite while
loop whose condition contains a call to a function that takes an Out or
In/Out parameter and whose actual is a variable that is not modified in
the loop, because it still thinks that functions can only have In
parameters.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* sem_warn.adb (Find_Var): Bail out for a function call with an
Out or In/Out parameter.
Nicolas Roche [Mon, 22 Jul 2019 13:56:45 +0000 (13:56 +0000)]
[Ada] Ensure Ctrl-C is not emited on terminated processes
Due to the reuse policy of PID on Windows. Sending a Ctrl-C to a dead
process might result in a Ctrl-C sent to the wrong process. The check is
also implemented on Unix platforms and avoid unecessary waits.
2019-07-22 Nicolas Roche <roche@adacore.com>
gcc/ada/
* terminals.c (__gnat_tty_waitpid): Support both blocking and
not blocking mode.
* libgnat/g-exptty.ads (Is_Process_Running): New function.
* libgnat/g-exptty.adb (Close): Don't try to interrupt/terminate
a process if it is already dead.
----
with Text_IO; use Text_IO;
procedure Conv is
V_P : constant := 10.0 ** (-6);
M_V : constant := 9000.0;
N_V : constant := -9000.0;
type V_T is delta V_P range N_V .. M_V with Small => V_P;
subtype S_T is V_T range 0.0 .. M_V;
function Convert (Input : in S_T) return Long_Float is
begin
Put_Line (Input'Img);
Put_Line (Long_Float'Image (Long_Float (Input)));
return Long_Float (Input);
end Convert;
* freeze.adb (Freeze_Fixed_Point_Type): When freezing a
fixed-point subtype, check whether the parent type declarastion
includes an aspect specification for the 'Small type attribute,
and inherit the specified value.
Javier Miranda [Mon, 22 Jul 2019 13:56:36 +0000 (13:56 +0000)]
[Ada] Crash in C++ constructor without external and link name
The compiler blows up processing the declaration of a tagged type
variable that has a C++ constructor without external or link name. After
this patch the frontend reports an error.
2019-07-22 Javier Miranda <miranda@adacore.com>
gcc/ada/
* freeze.adb (Freeze_Subprogram): Check that C++ constructors
must have external or link name.
Ed Schonberg [Mon, 22 Jul 2019 13:56:31 +0000 (13:56 +0000)]
[Ada] Spurious warning about a useless assignment
This patch removes a spurious warning about a useless assignment, when a
composite object is the target of an assignment and is an actual for an
out parameter in a subsewuent call, and there is an intervening use of
the object as the prefix of a selected component in an intervening
operation.
2019-07-22 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_res.adb (Resolve_Selected_Component): If the prefix has a
deferred reference, generate the correct reference now, to
indicate that the previous assignment is used. This prevents
spurious warnings on useless assignments when compiling with all
warnings enabled. when there is a subsequent call in the same
stqtement list, in which the prefix of the selected component is
the actual for an out parameter.
Eric Botcazou [Mon, 22 Jul 2019 13:56:26 +0000 (13:56 +0000)]
[Ada] Fix internal error on array slice in loop and Loop_Invariant
This fixes an internal error caused by the presence of an Itype in a
wrong scope. This Itype is created for an array slice present in the
condition of a while loop whose body also contains a pragma
Loop_Invariant, initially in the correct scope but then relocated into a
function created for the pragma.
2019-07-22 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* exp_attr.adb (Expand_Loop_Entry_Attribute): Copy the condition
of a while loop instead of simply relocating it.
gcc/testsuite/
* gnat.dg/loop_invariant1.adb, gnat.dg/loop_invariant1.ads: New
testcase.
Richard Biener [Mon, 22 Jul 2019 11:18:55 +0000 (11:18 +0000)]
re PR tree-optimization/91221 (ICE in get_int_cst_ext_nunits, at tree.c:1299 since r273548)
2019-07-22 Richard Biener <rguenther@suse.de>
PR tree-optimization/91221
* tree-ssa-sccvn.c (vn_reference_lookup_3): Appropriately
restrict partial-def handling of empty constructors and
memset to refs with known offset.
Jan Beulich [Mon, 22 Jul 2019 08:07:29 +0000 (08:07 +0000)]
x86/AVX512: improve generated code for bit-wise negation of vectors of integers
NOT on vectors of integers does not require loading a constant vector of
all ones into a register - VPTERNLOG can be used here (and could/should
be further used to carry out other binary and ternary logical operations
which don't have a special purpose instruction).
gcc/
2019-07-22 Jan Beulich <jbeulich@suse.com>
* config/i386/sse.md (ternlogsuffix): New.
(one_cmpl<mode>2): Don't force CONSTM1_RTX into a register when
AVX512F is in use.
(<mask_codefor>one_cmpl<mode>2<mask_name>): New.
Martin Liska [Mon, 22 Jul 2019 07:34:32 +0000 (09:34 +0200)]
Simplify LTO section format.
2019-07-22 Martin Liska <mliska@suse.cz>
* lto-section-in.c (lto_get_section_data):
Use new function get_compression.
* lto-streamer-out.c (produce_lto_section): Use
set_compression to encode compression algorithm.
* lto-streamer.h (struct lto_section): Do not
use bitfields in the format.
Martin Liska [Mon, 22 Jul 2019 07:34:10 +0000 (09:34 +0200)]
Make a warning for -Werror=wrong-language (PR driver/91172).
2019-07-22 Martin Liska <mliska@suse.cz>
PR driver/91172
* opts-common.c (decode_cmdline_option): Decode
argument of -Werror and check it for a wrong language.
* opts-global.c (complain_wrong_lang): Remove such case.
2019-07-22 Martin Liska <mliska@suse.cz>
The force_reg in or1k_expand_compare is hard coded for SImode, which is fine as
this used to only be used on SI expands. However, with FP support this will
cause issues. In general we should only force the right hand operand to a
register if its an immediate. This patch adds an condition to check for that.
gcc/ChangeLog:
* config/or1k/or1k.c (or1k_expand_compare): Check for int before
force_reg.
This adds support for OpenRISC hardware floating point instructions.
This is enabled with the -mhard-float option.
Double-prevision floating point operations work using register pairing as
specified in: https://openrisc.io/proposals/orfpx64a32. This has just been
added in the OpenRISC architecture specification 1.3.
This is enabled with the -mdouble-float option.
Not all architectures support unordered comparisons so an option,
-munordered-float is added.
Currently OpenRISC does not support sf/df or df/sf conversions, but this has
also just been added in architecture specification 1.3.
gcc/ChangeLog:
* config.gcc (or1k*-*-*): Add mhard-float, mdouble-float, msoft-float
and munordered-float validations.
* config/or1k/constraints.md (d): New register constraint.
* config/or1k/predicates.md (fp_comparison_operator): New.
* config/or1k/or1k.c (or1k_print_operand): Add support for printing 'd'
operands.
(or1k_expand_compare): Normalize unordered comparisons.
* config/or1k/or1k.h (reg_class): Define DOUBLE_REGS.
(REG_CLASS_NAMES): Add "DOUBLE_REGS".
(REG_CLASS_CONTENTS): Add contents for DOUBLE_REGS.
* config/or1k/or1k.md (type): Add fpu.
(fpu): New instruction reservation.
(F, f, fr, fi, FI, FOP, fop): New.
(<fop><F:mode>3): New ALU instruction definition.
(float<fi><F:mode>2): New conversion instruction definition.
(fix_trunc<F:mode><fi>2): New conversion instruction definition.
(fpcmpcc): New code iterator.
(*sf_fp_insn): New instruction definition.
(cstore<F:mode>4): New expand definition.
(cbranch<F:mode>4): New expand definition.
* config/or1k/or1k.opt (msoft-float, mhard-float, mdouble-float,
munordered-float): New options.
* doc/invoke.texi: Document msoft-float, mhard-float, mdouble-float and
munordered-float.
* config.gcc (or1k*-*-*): Add mrori and mror to validation.
* doc/invoke.texi (OpenRISC Options): Add mrori option, rewrite all
documenation to be more clear.
* config/or1k/elf.opt (mboard=, mnewlib): Rewrite documentation to be
more clear.
* config/or1k/or1k.opt (mrori): New option.
(mhard-div, msoft-div, mhard-mul, msoft-mul, mcmov, mror, msext,
msfimm, mshftimm): Rewrite documentation to be more clear.
* config/or1k/or1k.md (insn_support): Add ror and rori.
(enabled): Add conditions for ror and rori.
(rotrsi3): Replace condition for shftimm with ror and rori.
gcc/testsuite/ChangeLog:
* gcc.target/or1k/ror-4.c: New file.
* gcc.target/or1k/shftimm-1.c: Update test from rotate to shift
as the shftimm option no longer controls rotate.
Fixes bad assembly logic with software divide as reported by Richard Selvaggi.
Also, add a basic test to verify the soft math works when enabled.
gcc/testsuite/ChangeLog:
PR target/90362
* gcc.target/or1k/div-mul-3.c: New test.
libgcc/ChangeLog:
PR target/90362
* config/or1k/lib1funcs.S (__udivsi3): Change l.sfeqi
to l.sfeq and l.sfltsi to l.sflts equivalents as the immediate
instructions are not available on every processor. Change a
l.bnf to l.bf to fix logic issue.
Volatile memory does not match the memory_operand predicate. This
causes extra extend/mask instructions instructions when reading
from volatile memory. On OpenRISC loading volatile memory can be
treated the same as regular memory loads which supports combined
sign/zero extends. Fixing this eliminates the need for extra
extend/mask instructions.
This also adds a test provided by Richard Selvaggi which uncovered the
issue while we were looking into another issue.
The recent change to move code into the new rs6000-call.c file is missing a
default value for the TARGET_NO_PROTOTYPE value (which only affects targets
that don’t include svr4.h). Fixed by moving the fallback setting from
rs6000.c (which has no uses now) to rs6000-call.c.
2019-07-21 Iain Sandoe <iain@sandoe.co.uk>
* config/rs6000/rs6000.c (TARGET_NO_PROTOTYPE): Move from here...
* config/rs6000/rs6000-call.c: ... to here.
PR libfortran/91030
* io/unix.c (BUFFER_SIZE): Delete.
(BUFFER_FORMATTED_SIZE_DEFAULT): New variable.
(BUFFER_UNFORMATTED_SIZE_DEFAULT): New variable.
(unix_stream): Add buffer_size.
(buf_read): Use s->buffer_size instead of BUFFER_SIZE.
(buf_write): Likewise.
(buf_init): Add argument unformatted. Handle block sizes
for unformatted vs. formatted, using defaults if provided.
(fd_to_stream): Add argument unformatted in call to buf_init.
* libgfortran.h (options_t): Add buffer_size_formatted and
buffer_size_unformatted.
* runtime/environ.c (variable_table): Add
GFORTRAN_UNFORMATTED_BUFFER_SIZE and
GFORTRAN_FORMATTED_BUFFER_SIZE.
Jason Merrill [Sat, 20 Jul 2019 14:43:49 +0000 (10:43 -0400)]
Fix ICE on class template argument deduction with inherited ctor.
In general, when we see a dependent using-declaration we don't know whether
it names a function or not, so it doesn't get an OVERLOAD unless we see
overloads of the same name in the current class. In the case of an
inherited constructor we could figure that out from the name, but it's
simpler to handle USING_DECL properly.
* cp-tree.h (ovl_iterator::using_p): A USING_DECL by itself was also
introduced by a using-declaration.
Jason Merrill [Sat, 20 Jul 2019 13:48:38 +0000 (09:48 -0400)]
Reduce memory consumption for push/pop_access_scope.
I was seeing memory consumption issues on the concepts-cxx2a
branch. push_scope was, surprisingly, at the top of -fmem-report, and
push_access_scope was pretty high. Fixing them was pretty simple.
* name-lookup.c (leave_scope): Do add class levels other than
previous_class_level to free_binding_level.
(invalidate_class_lookup_cache): Move from class.c, add to
free_binding_level.
* pt.c (saved_access_scope): Change from list to vec.
Ian Lance Taylor [Fri, 19 Jul 2019 23:10:55 +0000 (23:10 +0000)]
compiler: don't export bodies for functions marked "go:noinline"
The current Mark_inline_candidates helper looks only at budget when
deciding to mark a function or method as inline (with the proviso that
IR constructs not yet supported by the inliner are given artificially
high cost). This patch changes the helper to also look at whether a
function has the "go:noinline" pragma; if it does have the pragma
there is no point putting it into the export data (it will just make
the export data bigger).
This strips out a few driver specs that are only specifying a default state.
Also warn on an option now ignored, and add some comments to the driver specs
section.
2019-07-19 Iain Sandoe <iain@sandoe.co.uk>
* config/darwin.h (DRIVER_SELF_SPECS): Ignore X and Mach specs which
refer to default conditions. Warn for the 'y' spec which is ignored
by current linkers.
Bill Seurer [Fri, 19 Jul 2019 18:33:59 +0000 (18:33 +0000)]
[PATCH, rs6000] Split up rs6000.c.
The source file rs6000.c has grown to unreasonable size and is being
split up into several smaller source files. This should improve
compilation speed for building gcc.
This is the second of several patches to do this and moves most of the
function call and builtin code to a new source file.
Bootstrapped and tested on powerpc64le-unknown-linux-gnu and
powerpc64-unknown-linux-gnu with no regressions. Is this ok for trunk?
2019-07-17 Bill Seurer <seurer@linux.vnet.ibm.com>
projects / thirdparty / gcc.git / log
