Jakub Jelinek [Tue, 7 Jun 2022 08:05:08 +0000 (10:05 +0200)]
openmp: Add support for OpenMP 5.2 linear clause syntax for C/C++
The syntax for linear clause changed in 5.2, the original syntax
which is still valid is:
linear (var1, var2)
linear (var3, var4 : step1)
The 4.5 syntax with modifiers like:
linear (val (var5, var6))
linear (val (var7, var8) : step2)
is still supported in 5.2, but is deprecated there.
Instead, one can use a new syntax:
linear (var9, var10 : val)
linear (var11, var12 : step (step3), val)
As val, ref, uval or step (someexpr) can be valid expressions (and especially
in C++ can be const / constexpr / consteval), the spec says that
when the whole step expression is val (or ref or uval) or step ( ... )
then it is the new modifier syntax, one can use + 0 or 0 + or 1 * or * 1
or ()s to say it is the old step expression.
Also, 5.2 now allows val modifier to be specified even outside of declare simd
(but not the other modifiers). I've implemented this for the new modifier
syntax only, the old one keeps the old restriction (which is why
OMP_CLAUSE_LINEAR_OLD_LINEAR_MODIFIER flag has been introduced).
2022-06-07 Jakub Jelinek <jakub@redhat.com>
gcc/
* tree.h (OMP_CLAUSE_LINEAR_OLD_LINEAR_MODIFIER): Define.
* tree-pretty-print.cc (dump_omp_clause) <case OMP_CLAUSE_LINEAR>:
Adjust clause printing style depending on
OMP_CLAUSE_LINEAR_OLD_LINEAR_MODIFIER.
gcc/c/
* c-parser.cc (c_parser_omp_clause_linear): Parse OpenMP 5.2
style linear clause modifiers. Set
OMP_CLAUSE_LINEAR_OLD_LINEAR_MODIFIER flag on the clauses when
old style modifiers are used.
* c-typeck.cc (c_finish_omp_clauses): Only reject linear clause
with val modifier on simd or for if the old style modifiers are
used.
gcc/cp/
* parser.cc (cp_parser_omp_clause_linear): Parse OpenMP 5.2
style linear clause modifiers. Set
OMP_CLAUSE_LINEAR_OLD_LINEAR_MODIFIER flag on the clauses when
old style modifiers are used.
* semantics.cc (finish_omp_clauses): Only reject linear clause
with val modifier on simd or for if the old style modifiers are
used.
gcc/fortran/
* trans-openmp.cc (gfc_trans_omp_clauses): Set
OMP_CLAUSE_LINEAR_OLD_LINEAR_MODIFIER on OMP_CLAUSE_LINEAR
clauses unconditionally for now.
gcc/testsuite/
* c-c++-common/gomp/linear-2.c: New test.
* c-c++-common/gomp/linear-3.c: New test.
* g++.dg/gomp/linear-3.C: New test.
* g++.dg/gomp/linear-4.C: New test.
* g++.dg/gomp/linear-5.C: New test.
Jan Beulich [Tue, 7 Jun 2022 07:18:28 +0000 (09:18 +0200)]
x86: harmonize __builtin_ia32_psadbw*() types
The 64-bit, 128-bit, and 512-bit variants have V<n>DI return type, in
line with instruction behavior. Make the 256-bit builtin match, thus
also making it match the insn it expands to (using VI8_AVX2_AVX512BW).
gcc/
* config/i386/i386-builtin.def (__builtin_ia32_psadbw256):
Change type.
* config/i386/i386-builtin-types.def: New function type
(V4DI, V32QI, V32QI).
* config/i386/i386-expand.cc (ix86_expand_args_builtin): Handle
V4DI_FTYPE_V32QI_V32QI.
Jan Beulich [Tue, 7 Jun 2022 07:17:25 +0000 (09:17 +0200)]
x86-64: make "length_vex" also account for VEX.B use by register operand
The length attribute ought to be "the (bounding maximum) length of an
instruction" according to the comment next to its definition. A register
operand encoded using the ModR/M.rm field will additionally use VEX.B
for encoding the highest bit of the register number. Hence for the high
8 GPR registers as well as the [xy]mm{8..15} ones 3-byte VEX encoding
may be needed. Since it isn't known to the function calculating the
length which register goes where in the insn encoding, be conservative
and assume a 3-byte VEX prefix whenever any such register operand is
present and there's no memory operand.
gcc/
* config/i386/i386.cc (ix86_attr_length_vex_default): Take REX.B
into account for reg-only insns.
Roger Sayle [Tue, 7 Jun 2022 06:54:13 +0000 (07:54 +0100)]
PR c++/96442: Improved error recovery in enumerations.
This patch is a revised fix for PR c++/96442 providing a cleaner
solution, setting ENUM_UNDERLYING_TYPE to integer_type_node when
issuing an error, so that this invariant holds during the parser's
error recovery.
2022-06-07 Roger Sayle <roger@nextmovesoftware.com>
gcc/cp/ChangeLog
PR c++/96442
* decl.cc (start_enum): When emitting a "must be integral" error,
set ENUM_UNDERLYING_TYPE to integer_type_node, to avoid an ICE
downstream in build_enumeration.
gcc/testsuite/ChangeLog
PR c++/96442
* g++.dg/parse/pr96442.C: New test case.
Roger Sayle [Tue, 7 Jun 2022 06:49:40 +0000 (07:49 +0100)]
Recognize vpcmov in combine with -mxop on x86.
By way of an apology for causing PR target/105791, where I'd overlooked
the need to support V1TImode in TARGET_XOP's vpcmov instruction, this
patch further improves support for TARGET_XOP's vpcmov instruction, by
recognizing it in combine.
on x86_64 with -O2 -mxop generates:
vpxor %xmm2, %xmm1, %xmm1
vpand %xmm0, %xmm1, %xmm1
vpxor %xmm2, %xmm1, %xmm0
ret
but with this patch now generates:
vpcmov %xmm0, %xmm2, %xmm1, %xmm0
ret
On its own, the new combine splitter works fine on TARGET_64BIT, but
alas with -m32 combine incorrectly thinks the replacement instruction
is more expensive, as IF_THEN_ELSE isn't currently/correctly handled
in ix86_rtx_costs. So to avoid the need for a target selector in the
new tescase, I've updated ix86_rtx_costs to report that AMD's vpcmov
has a latency of two cycles [it's now an obsolete instruction set
extension and there's unlikely to ever be a processor where this
instruction has a different timing], and while there I also added
rtx_costs for x86_64's integer conditional move instructions (which
have single cycle latency).
2022-06-07 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
* config/i386/i386.cc (ix86_rtx_costs): Add a new case for
IF_THEN_ELSE, and provide costs for TARGET_XOP's vpcmov and
TARGET_CMOVE's (scalar integer) conditional moves.
* config/i386/sse.md (define_split): Recognize XOP's vpcmov
from its equivalent (canonical) pxor;pand;pxor sequence.
gcc/testsuite/ChangeLog
* gcc.target/i386/xop-pcmov3.c: New test case.
Kewen Lin [Tue, 7 Jun 2022 03:08:23 +0000 (22:08 -0500)]
Update document for VECTOR_MODES_WITH_PREFIX
r10-3912 updated the format of VECTOR_MODES_WITH_PREFIX by
adding one more parameter ORDER, the related document is out
of date. So update the document for ORDER.
gcc/ChangeLog:
* machmode.def (VECTOR_MODES_WITH_PREFIX): Update document for
parameter ORDER.
Patrick Palka [Mon, 6 Jun 2022 18:29:12 +0000 (14:29 -0400)]
c++: function NTTP argument considered unused [PR53164, PR105848]
Here at parse time the template argument f (an OVERLOAD) in A<f> gets
resolved ahead of time to the FUNCTION_DECL f<int>, and we defer marking
f<int> as used until instantiation (of g) as usual.
Later when instantiating g the type A<f> (where f has already been
resolved) is non-dependent, so tsubst_aggr_type avoids re-processing its
template arguments, and we end up never actually marking f<int> as used
(which means we never instantiate it) even though A<f>::h() later calls
it, leading to a link error.
This patch works around this issue by looking through ADDR_EXPR when
calling mark_used on the substituted callee of a CALL_EXPR.
PR c++/53164
PR c++/105848
gcc/cp/ChangeLog:
* pt.cc (tsubst_copy_and_build) <case CALL_EXPR>: Look through an
ADDR_EXPR callee when calling mark_used.
cp_parser_attributes_opt doesn't accept GNU attributes followed by
[[]] attributes and vice versa; only a sequence of attributes of the
same kind. That causes grief for code like:
struct __attribute__ ((may_alias)) alignas (2) struct S { };
or
#define EXPORT __attribute__((visibility("default")))
struct [[nodiscard]] EXPORT F { };
It doesn't seem to a documented restriction, so this patch fixes the
problem.
However, the patch does not touch the C FE. The C FE doesn't have
a counterpart to C++'s cp_parser_attributes_opt -- it only has
c_parser_transaction_attributes (which parses both GNU and [[]]
attributes), but that's TM-specific. The C FE seems to use either
c_parser_gnu_attributes or c_parser_std_attribute_specifier_sequence.
As a consequence, this works:
Roger Sayle [Sat, 4 Jun 2022 11:21:51 +0000 (12:21 +0100)]
PR middle-end/95126: Expand small const structs as immediate constants.
This patch resolves PR middle-end/95126 which is a code quality regression,
by teaching the RTL expander to emit small const structs/unions as integer
immediate constants.
The motivating example from the bugzilla PR is:
struct small{ short a,b; signed char c; };
extern int func(struct small X);
void call_func(void)
{
static struct small const s = { 1, 2, 0 };
func(s);
}
2022-06-04 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
PR middle-end/95126
* calls.cc (load_register_parameters): When loading a suitable
immediate_const_ctor_p VAR_DECL into a single word_mode register,
construct it directly in a pseudo rather than read it (by parts)
from memory.
* expr.cc (int_expr_size): Make tree argument a const_tree.
(immediate_const_ctor_p): Helper predicate. Return true for
simple constructors that may be materialized in a register.
(expand_expr_real_1) [VAR_DECL]: When expanding a constant
VAR_DECL with a suitable immediate_const_ctor_p constructor
use store_constructor to materialize it directly in a pseudo.
* expr.h (immediate_const_ctor_p): Prototype here.
* varasm.cc (initializer_constant_valid_for_bitfield_p): Change
VALUE argument from tree to const_tree.
* varasm.h (initializer_constant_valid_for_bitfield_p): Update
prototype.
gcc/testsuite/ChangeLog
PR middle-end/95126
* gcc.target/i386/pr95126-m32-1.c: New test case.
* gcc.target/i386/pr95126-m32-2.c: New test case.
* gcc.target/i386/pr95126-m32-3.c: New test case.
* gcc.target/i386/pr95126-m32-4.c: New test case.
* gcc.target/i386/pr95126-m64-1.c: New test case.
* gcc.target/i386/pr95126-m64-2.c: New test case.
* gcc.target/i386/pr95126-m64-3.c: New test case.
* gcc.target/i386/pr95126-m64-4.c: New test case.
Jakub Jelinek [Sat, 4 Jun 2022 08:36:24 +0000 (10:36 +0200)]
i386: Fix up *_doubleword_mask [PR105825]
My PR105778 patch apparently broke the following testcase.
If the mask has the top relevant bit clear (i.e. we know we are shifting
by 0 to wordsize bits - 1) but doesn't have all the bits below it set,
we emit andsi3 before the shift sequence. When the pattern had :SI
for that operand, that was just fine, but now that it can be also HImode
or for -m64 DImode, we either can use a lowpart or paradoxical subreg to
SImode as the following patch, or we use a HImode or DImode AND.
This patch does the latter.
2022-06-04 Jakub Jelinek <jakub@redhat.com>
PR target/105825
* config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
*<insn><dwi>3_doubleword_mask): If top bit of mask is clear, but lower
bits of mask aren't all set, use operands[2] mode for the AND
operation instead of always SImode.
Jason Merrill [Fri, 3 Jun 2022 18:53:59 +0000 (14:53 -0400)]
c++: more-specialized test
I noticed the need for this testcase while working on PR102629; since there
is no information about the target type, we don't want to choose the most
specialized overload.
Patrick Palka [Fri, 3 Jun 2022 18:58:22 +0000 (14:58 -0400)]
c++: value-dep but not type-dep decltype expr [PR105756]
Here during ahead of time instantiation of the value-dependent but not
type-dependent decltype expression (5 % N) == 0, cp_build_binary_op folds
the operands of the == via cp_fully_fold, which performs speculative
constexpr evaluation, and from which we crash for (5 % N) due to the
value-dependence.
Since the operand folding performed by cp_build_binary_op appears to
be solely for sake of diagnosing overflow, and since these diagnostics
are suppressed when in an unevaluated context, this patch avoids this
crash by suppressing cp_build_binary_op's operand folding accordingly.
PR c++/105756
gcc/cp/ChangeLog:
* typeck.cc (cp_build_binary_op): Don't fold operands
when c_inhibit_evaluation_warnings.
Jason Merrill [Fri, 3 Jun 2022 16:35:12 +0000 (12:35 -0400)]
c++: redeclared hidden friend [PR105761]
Here, when we see the second declaration of f we match it with the first
one, copy over DECL_TEMPLATE_INFO, and then try to use it when parsing the
definition, leading to confusion.
PR c++/105761
gcc/cp/ChangeLog:
* decl.cc (duplicate_decls): Don't copy DECL_TEMPLATE_INFO
from a hidden friend.
Patrick Palka [Fri, 3 Jun 2022 16:06:59 +0000 (12:06 -0400)]
c++: cv-quals of dummy obj for non-dep memfn call [PR105637]
In non-dependent23.C below we expect the Base::foo calls to
resolve to the second, third and fourth overloads respectively in light
of the cv-qualifiers of 'this' in each case. But ever since r12-6075-g2decd2cabe5a4f, the calls incorrectly resolve to the first
overload at instantiation time.
This happens because the calls to Base::foo are all deemed
non-dependent (ever since r7-755-g23cb72663051cd made us ignore 'this'
dependence when considering the dependence of a non-static memfn call),
hence we end up checking the call ahead of time, using as the object
argument a dummy object of type Base. Since this object argument is
cv-unqualified, the calls in turn resolve to the unqualified overload
of baseDevice. Before r12-6075 this incorrect result would just get
silently discarded and we'd end up redoing OR at instantiation time
using 'this' as the object argument. But after r12-6075 we now reuse
this incorrect result at instantiation time.
This patch fixes this by making maybe_dummy_object respect the cv-quals
of (the non-lambda) 'this' when returning a dummy object. Thus, ahead
of time OR using a dummy object will give us the right answer that's
consistent with the instantiation time answer.
An earlier version of this patch didn't handle 'this'-capturing lambdas
correctly, which broke lambda-this22.C below.
PR c++/105637
gcc/cp/ChangeLog:
* tree.cc (maybe_dummy_object): When returning a dummy
object, respect the cv-quals of 'this' if available.
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/lambda/lambda-this22.C: New test.
* g++.dg/template/non-dependent23.C: New test.
This patch makes us avoid substituting into the TEMPLATE_PARM_CONSTRAINTS
of each template parameter except as necessary for declaration matching,
like we already do for the other constituent constraints of a declaration.
This patch also improves the CA104 implementation of explicit
specialization matching of a constrained function template inside a
class template, by considering the function's combined constraints
instead of just its trailing constraints. This allows us to correctly
handle the first three explicit specializations in concepts-spec2.C
below, but because we compare the constraints as a whole, it means we
incorrectly accept the fourth explicit specialization which writes #3's
constraints in a different way. For complete correctness here,
determine_specialization should use tsubst_each_template_parm_constraints
and template_parameter_heads_equivalent_p.
PR c++/100374
gcc/cp/ChangeLog:
* pt.cc (determine_specialization): Compare overall constraints
not just the trailing constraints.
(tsubst_each_template_parm_constraints): Define.
(tsubst_friend_function): Use it.
(tsubst_friend_class): Use it.
(tsubst_template_parm): Don't substitute TEMPLATE_PARM_CONSTRAINTS.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/concepts-spec2.C: New test.
* g++.dg/cpp2a/concepts-template-parm11.C: New test.
Patrick Palka [Fri, 3 Jun 2022 13:08:41 +0000 (09:08 -0400)]
c++: find_template_parameters and PARM_DECLs [PR105797]
As explained in r11-4959-gde6f64f9556ae3, the atom cache assumes two
equivalent expressions (according to cp_tree_equal) must use the same
template parameters (according to find_template_parameters). This
assumption turned out to not hold for TARGET_EXPR, which was addressed
by that commit.
But this assumption apparently doesn't hold for PARM_DECL either:
find_template_parameters walks its DECL_CONTEXT but cp_tree_equal by
default doesn't consider DECL_CONTEXT unless comparing_specializations
is set. Thus in the first testcase below, the atomic constraints of #1
and #2 are equivalent according to cp_tree_equal, but according to
find_template_parameters the former uses T and the latter uses both T
and U (surprisingly).
We could fix this assumption violation by setting comparing_specializations
in the atom_hasher, which would make cp_tree_equal return false for the
two atoms, but that seems overly pessimistic here. Ideally the atoms
should continue being considered equivalent and we instead fix
find_template_paremeters to return just T for #2's atom.
To that end this patch makes for_each_template_parm_r stop walking the
DECL_CONTEXT of a PARM_DECL. This should be safe to do because
tsubst_copy / tsubst_decl only substitutes the TREE_TYPE of a PARM_DECL
and doesn't bother substituting the DECL_CONTEXT, thus the only relevant
template parameters are those used in its type. any_template_parm_r is
currently responsible for walking its TREE_TYPE, but I suppose it now makes
sense for for_each_template_parm_r to do so instead.
In passing this patch also makes for_each_template_parm_r stop walking
the DECL_CONTEXT of a VAR_/FUNCTION_DECL since doing so after walking
DECL_TI_ARGS is redundant, I think.
I experimented with not walking DECL_CONTEXT for CONST_DECL, but the
second testcase below demonstrates it's necessary to walk it.
PR c++/105797
gcc/cp/ChangeLog:
* pt.cc (for_each_template_parm_r) <case FUNCTION_DECL, VAR_DECL>:
Don't walk DECL_CONTEXT.
<case PARM_DECL>: Likewise. Walk TREE_TYPE.
<case CONST_DECL>: Simplify.
(any_template_parm_r) <case PARM_DECL>: Don't walk TREE_TYPE.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/concepts-decltype4.C: New test.
* g++.dg/cpp2a/concepts-memfun3.C: New test.
The following patch is an incremental change to the PR30314 enhancement,
this one handles signed types.
For signed types (but still, the same for 1st and result element type
and non-zero constant that fits into that type), we actually need to
watch for overflow in direction to positive and negative infinity
and it also depends on whether the cst operand is positive or negative.
For __builtin_mul_overflow_p (x, cst, (stype) 0):
For cst > 0, we can simplify it to:
x > INT_MAX / cst || x < INT_MIN / cst
aka:
x + (unsigned) (INT_MIN / cst) > (unsigned) (INT_MAX / cst) - (unsigned) (INT_MIN / cst)
and for cst < 0 to:
x < INT_MAX / cst || x > INT_MIN / cst
aka:
x + (unsigned) (INT_MAX / cst) > (unsigned) (INT_MIN / cst) - (unsigned) (INT_MAX / cst)
Additionally, I've added executable testcases, so we don't just check for
the optimization to be performed, but also that it is correct (done that
even for the other PR's testcase).
2022-06-03 Jakub Jelinek <jakub@redhat.com>
PR middle-end/30314
PR middle-end/105777
* match.pd (__builtin_mul_overflow_p (x, cst, (stype) 0) ->
x > stype_max / cst || x < stype_min / cst): New simplification.
* gcc.dg/tree-ssa/pr30314.c: Add noipa attribute to all functions.
* gcc.dg/tree-ssa/pr105777.c: New test.
* gcc.c-torture/execute/pr30314.c: New test.
* gcc.c-torture/execute/pr105777.c: New test.
Aldy Hernandez [Wed, 1 Jun 2022 14:58:36 +0000 (16:58 +0200)]
Flesh out unsupported_range.
It's cleaner to have the unsupported_range fully fleshed out, instead
of trapping on every operation. It can also serve as the basis for
the default vrange methods that frange and prange will inherit.
This patch implements most methods, including union and intersect, to
handle an UNDEFINED and a VARYING range.
Since this can serve as the basis for other classes, I have moved
everything into the vrange class, making the unsupported_range
trivial.
Note that vrange is still an abstract class, as I have purposely left
the dump() method abstract.
Also, I have made the unsupported range in the temporary class
(Value_Range) a method field, instead of a static member. This way
the temporary can set UNDEFINED and VARYING as needed.
Tested on x86-64 Linux.
gcc/ChangeLog:
* value-range.cc (vrange::contains_p): Implement.
(vrange::type): Return void.
(vrange::supports_type_p): Implement.
(irange::fits_p): Same.
(vrange::set_undefined): Same.
(irange::set_nonnegative): Same.
(vrange::set_varying): Same.
(vrange::union_): Same.
(unsupported_range::set): Move to vrange.
(unsupported_range::type): Move to vrange.
(vrange::intersect): Implement for varying and undefined.
(vrange::zero_p): Implement.
(unsupported_range::supports_type_p): Move to vrange.
(vrange::nonzero_p): Implement.
(unsupported_range::set_undefined): Move to vrange.
(unsupported_range::set_varying): Same.
(unsupported_range::dump): Same.
(unsupported_range::union_): Same. Implement for varying and
undefined.
(unsupported_range::intersect): Move to vrange.
(unsupported_range::zero_p): Same.
(unsupported_range::nonzero_p): Same.
(unsupported_range::set_nonzero): Same.
(unsupported_range::set_zero): Same.
(unsupported_range::set_nonnegative): Same.
(unsupported_range::fits_p): Same.
* value-range.h (class vrange): Remove abstract markers for most
methods.
(class unsupported_range): Remove most methods as they will now be
inherited from vrange.
Aldy Hernandez [Tue, 31 May 2022 12:04:51 +0000 (14:04 +0200)]
Implement vrange::supports_type_p.
[I have conservatively assumed that both the loop-ch and loop-unswitch
passes, which also use the ranger, only support integers and pointers.
If the goal is to handle other types as well, irange::supports_p()
should be Value_Range::supports_type_p(), and any uses of
int_range_max should be converted to Value_Range. I can help in the
conversion if you'd like.]
As discussed, this patch disambiguates the use of supports_type_p
throughout, as what ranger supports is a totally different question
than what a given range variant (irange, frange, etc) supports.
Unfortunately we need both a static method and a virtual method, and
they can't be named the same. The uses are documented in the vrange
class:
+// To query what types ranger and the entire ecosystem can support,
+// use Value_Range::supports_type_p(tree type). This is a static
+// method available independently of any vrange object.
+//
+// To query what a given vrange variant can support, use:
+// irange::supports_p ()
+// frange::supports_p ()
+// etc
+//
+// To query what a range object can support, use:
+// void foo (vrange &v, irange &i, frange &f)
+// {
+// if (v.supports_type_p (type)) ...
+// if (i.supports_type_p (type)) ...
+// if (f.supports_type_p (type)) ...
+// }
The value_range_equiv::supports_p() method can be use to determine
what legacy VRP supports, as irange::supports_p() will no longer be
applicable in the evrp analyzer code base once irange and prange are
split.
Jan Beulich [Fri, 3 Jun 2022 08:22:25 +0000 (10:22 +0200)]
configure: arrange to use appropriate objcopy
Using the system objcopy is wrong when other configure checks have
probed a different set of binutils (I've noticed the problem on a system
where the base objcopy can't deal with compressed debug sections).
Arrange for the matching one to be picked up, first and foremost if an
"in tree" one is available, by mirroring respective logic already
present for nm.
Jan Beulich [Fri, 3 Jun 2022 08:21:21 +0000 (10:21 +0200)]
x86: {,v}psadbw have commutative source operands
Like noticed for gas as well (binutils-gdb commit c8cad9d389b7), the
"absolute difference" aspect of the insns makes their source operands
commutative.
gcc/
* config/i386/mmx.md (mmx_psadbw): Convert to expander.
(*mmx_psadbw): New. Mark as commutative.
* config/i386/sse.md (<sse2_avx2>_psadbw): Convert to expander.
(*<sse2_avx2>_psadbw): New. Mark as commutative.
Alexandre Oliva [Fri, 3 Jun 2022 06:59:03 +0000 (03:59 -0300)]
[PR105665] ivopts: check defs of names in base for undefs
The patch for PR 100810 tested for undefined SSA_NAMEs appearing
directly in the base expression of the potential IV candidate, but
that's not enough. The testcase for PR105665 shows an undefined
SSA_NAME has the same ill effect if it's referenced as an PHI_NODE arg
in the referenced SSA_NAME. The variant of that test shows it can be
further removed from the referenced SSA_NAME.
To avoid deep recursion, precompute maybe-undefined SSA_NAMEs: start
from known-undefined nonvirtual default defs, and propagate them to
any PHI nodes reached by a maybe-undefined arg, as long as there
aren't intervening non-PHI uses, that would imply the maybe-undefined
name must be defined at that point, otherwise it would invoke
undefined behavior. Also test for intervening non-PHI uses of DEFs in
the base expr.
The test for intervening uses implemented herein relies on dominance;
this could be further extended, regarding conditional uses in every
path leading to a point as an unconditional use dominating that point,
but I haven't implemented that.
Alexandre Oliva [Fri, 3 Jun 2022 06:59:01 +0000 (03:59 -0300)]
libcody: fix nonportable shell code in revision.stamp build rule
Two non-portable shell constructs have been long present in libcody's
build rule for revision.stamp: $() instead of ``, and += to append to
a shell variable. The former seems to work even when bash is
operating as /bin/sh, but += doesn't, and it ends up trying to run
revision+=M as a command name, and issuing an error as that command is
(hopefully) not found.
This patch replaces both constructs with more portable ones.
for libcody/ChangeLog
* Makefile.in (revision.stamp): Replace $() and += with more
portable shell constructs.
Marek Polacek [Wed, 1 Jun 2022 22:21:36 +0000 (18:21 -0400)]
c++: ICE with template NEW_EXPR [PR105803]
Here we ICE because value_dependent_expression_p gets a NEW_EXPR
whose operand is a type, and we go to the default case which just
calls v_d_e_p on each operand of the NEW_EXPR. Since one of them
is a type, we crash on the new assert in t_d_e_p.
t_d_e_p has code to handle {,VEC_}NEW_EXPR, which at this point
was already performed, so I think we can handle these two codes
specifically and skip the second operand, which is always going
to be a type.
PR c++/105803
gcc/cp/ChangeLog:
* pt.cc (value_dependent_expression_p): Handle {,VEC_}NEW_EXPR
in the switch.
David Malcolm [Thu, 2 Jun 2022 19:40:22 +0000 (15:40 -0400)]
diagnostics: add SARIF output format
This patch adds support to gcc's diagnostic subsystem for emitting
diagnostics in SARIF, aka the Static Analysis Results Interchange Format:
https://sarifweb.azurewebsites.net/
by extending -fdiagnostics-format= to add two new options:
-fdiagnostics-format=sarif-stderr
and:
-fdiagnostics-format=sarif-file
The patch targets SARIF v2.1.0
This is a JSON-based format suited for capturing the results of static
analysis tools (like GCC's -fanalyzer), but it can also be used for plain
GCC warnings and errors.
SARIF supports per-event metadata in diagnostic paths such as
["acquire", "resource"] and ["release", "lock"] (specifically, the
threadFlowLocation "kinds" property: SARIF v2.1.0 section 3.38.8), so
the patch extends GCC"s diagnostic_event subclass with a "struct meaning"
with similar purpose. The patch implements this for -fanalyzer so that
the various state-machine-based warnings set these in the SARIF output.
The heart of the implementation is in the new file
diagnostic-format-sarif.cc. Much of the rest of the patch is interface
classes, isolating the diagnostic subsystem (which has no knowledge of
e.g. tree or langhook) from the "client" code in the compiler proper
cc1 etc).
The patch adds a langhook for specifying the SARIF v2.1.0
"artifact.sourceLanguage" property, based on the list in
SARIF v2.1.0 Appendix J.
The patch adds automated DejaGnu tests to our testsuite via new
scan-sarif-file and scan-sarif-file-not directives (although these
merely use regexps, rather than attempting to use a proper JSON parser).
I've tested the patch by hand using the validator at:
https://sarifweb.azurewebsites.net/Validation
and the react-based viewer at:
https://microsoft.github.io/sarif-web-component/
which successfully shows most of the information (although not paths,
and not CWE IDs), and I've fixed all validation errors I've seen (though
bugs no doubt remain).
I've also tested the generated SARIF using the VS Code extension linked
to from the SARIF website; I'm a novice with VS Code, but it seems to be
able to handle my generated SARIF files (e.g. showing the data in the
SARIF tab, and showing squiggly underlines under issues, and when I
click on them, it visualizes the events in the path inline within the
source window).
Has anyone written an Emacs mode for SARIF files? (pretty please)
gcc/ChangeLog:
* Makefile.in (OBJS): Add tree-diagnostic-client-data-hooks.o and
tree-logical-location.o.
(OBJS-libcommon): Add diagnostic-format-sarif.o; reorder.
(CFLAGS-tree-diagnostic-client-data-hooks.o): Add TARGET_NAME.
* common.opt (fdiagnostics-format=): Add sarif-stderr and sarif-file.
(sarif-stderr, sarif-file): New enum values.
* diagnostic-client-data-hooks.h: New file.
* diagnostic-format-sarif.cc: New file.
* diagnostic-path.h (enum diagnostic_event::verb): New enum.
(enum diagnostic_event::noun): New enum.
(enum diagnostic_event::property): New enum.
(struct diagnostic_event::meaning): New struct.
(diagnostic_event::get_logical_location): New vfunc.
(diagnostic_event::get_meaning): New vfunc.
(simple_diagnostic_event::get_logical_location): New vfunc impl.
(simple_diagnostic_event::get_meaning): New vfunc impl.
* diagnostic.cc: Include "diagnostic-client-data-hooks.h".
(diagnostic_initialize): Initialize m_client_data_hooks.
(diagnostic_finish): Clean up m_client_data_hooks.
(diagnostic_event::meaning::dump_to_pp): New.
(diagnostic_event::meaning::maybe_get_verb_str): New.
(diagnostic_event::meaning::maybe_get_noun_str): New.
(diagnostic_event::meaning::maybe_get_property_str): New.
(get_cwe_url): Make non-static.
(diagnostic_output_format_init): Handle
DIAGNOSTICS_OUTPUT_FORMAT_SARIF_STDERR and
DIAGNOSTICS_OUTPUT_FORMAT_SARIF_FILE.
* diagnostic.h (enum diagnostics_output_format): Add
DIAGNOSTICS_OUTPUT_FORMAT_SARIF_STDERR and
DIAGNOSTICS_OUTPUT_FORMAT_SARIF_FILE.
(class diagnostic_client_data_hooks): New forward decl.
(class logical_location): New forward decl.
(diagnostic_context::m_client_data_hooks): New field.
(diagnostic_output_format_init_sarif_stderr): New decl.
(diagnostic_output_format_init_sarif_file): New decl.
(get_cwe_url): New decl.
* doc/invoke.texi (-fdiagnostics-format=): Add sarif-stderr and
sarif-file.
* doc/sourcebuild.texi (Scan a particular file): Add
scan-sarif-file and scan-sarif-file-not.
* langhooks-def.h (lhd_get_sarif_source_language): New decl.
(LANG_HOOKS_GET_SARIF_SOURCE_LANGUAGE): New macro.
(LANG_HOOKS_INITIALIZER): Add
LANG_HOOKS_GET_SARIF_SOURCE_LANGUAGE.
* langhooks.cc (lhd_get_sarif_source_language): New.
* langhooks.h (lang_hooks::get_sarif_source_language): New field.
* logical-location.h: New file.
* plugin.cc (struct for_each_plugin_closure): New.
(for_each_plugin_cb): New.
(for_each_plugin): New.
* plugin.h (for_each_plugin): New decl.
* tree-diagnostic-client-data-hooks.cc: New file.
* tree-diagnostic.cc: Include "diagnostic-client-data-hooks.h".
(tree_diagnostics_defaults): Populate m_client_data_hooks.
* tree-logical-location.cc: New file.
* tree-logical-location.h: New file.
gcc/testsuite/ChangeLog:
* c-c++-common/diagnostic-format-sarif-file-1.c: New test.
* c-c++-common/diagnostic-format-sarif-file-2.c: New test.
* c-c++-common/diagnostic-format-sarif-file-3.c: New test.
* c-c++-common/diagnostic-format-sarif-file-4.c: New test.
* gcc.dg/analyzer/file-meaning-1.c: New test.
* gcc.dg/analyzer/malloc-meaning-1.c: New test.
* gcc.dg/analyzer/malloc-sarif-1.c: New test.
* gcc.dg/plugin/analyzer_gil_plugin.c
(gil_diagnostic::get_meaning_for_state_change): New vfunc impl.
* gcc.dg/plugin/diagnostic-test-paths-5.c: New test.
* gcc.dg/plugin/plugin.exp (plugin_test_list): Add
diagnostic-test-paths-5.c to tests for
diagnostic_plugin_test_paths.c.
* lib/gcc-dg.exp: Load scansarif.exp.
* lib/scansarif.exp: New test.
libatomic/ChangeLog:
* testsuite/lib/libatomic.exp: Add load_gcc_lib of scansarif.exp.
libgomp/ChangeLog:
* testsuite/lib/libgomp.exp: Add load_gcc_lib of scansarif.exp.
libitm/ChangeLog:
* testsuite/lib/libitm.exp: Add load_gcc_lib of scansarif.exp.
libphobos/ChangeLog:
* testsuite/lib/libphobos-dg.exp: Add load_gcc_lib of scansarif.exp.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
David Malcolm [Thu, 2 Jun 2022 19:38:38 +0000 (15:38 -0400)]
Add -fdiagnostics-format={json-stderr|json-file}
This commit adds -fdiagnostics-format=json-file, writing to
DUMP_BASE_NAME.gcc.json, and adds -fdiagnostics-format=json-stderr,
a synonym for the existing -fdiagnostics-format=json.
gcc/ChangeLog:
* common.opt (fdiagnostics-format=): Add json-stderr and json-file
to description.
(DIAGNOSTICS_OUTPUT_FORMAT_JSON): Rename to...
(DIAGNOSTICS_OUTPUT_FORMAT_JSON_STDERR): ...this.
(diagnostics_output_format): Add json-stderr and json-file.
* diagnostic-format-json.cc (json_flush_to_file): New.
(json_final_cb): Convert to...
(json_flush_to_file): ...this, ...
(json_stderr_final_cb): ...this, and...
(json_file_final_cb): ...this.
(diagnostic_output_format_init): Move to diagnostic.cc.
(json_output_base_file_name): New.
(diagnostic_output_format_init_json): New.
(diagnostic_output_format_init_json_stderr): New.
(diagnostic_output_format_init_json_file): New.
* diagnostic.cc (diagnostic_output_format_init): Move here from
diagnostic-format-json.cc; update for changes to enum.
* diagnostic.h (enum diagnostics_output_format): Rename
DIAGNOSTICS_OUTPUT_FORMAT_JSON to
DIAGNOSTICS_OUTPUT_FORMAT_JSON_STDERR, and add
DIAGNOSTICS_OUTPUT_FORMAT_JSON_FILE.
(diagnostic_output_format_init): Add base_file_name param.
(diagnostic_output_format_init_json_stderr): New decl.
(diagnostic_output_format_init_json_file): New dec.
* doc/invoke.texi (-fdiagnostics-format=): Add "json-stderr" and
"json-file". Rewrite so that the existing "json" is a synonym of
"json-stderr".
* gcc.cc (driver_handle_option): Pass dump_base_name to
diagnostic_output_format_init.
* opts.cc (common_handle_option): Likewise.
gcc/testsuite/ChangeLog:
* c-c++-common/diagnostic-format-json-file-1.c: New test.
* c-c++-common/diagnostic-format-json-stderr-1.c: New test.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
Philipp Tomsich [Mon, 29 Jun 2020 13:15:10 +0000 (15:15 +0200)]
RISC-V: bitmanip: improve constant-loading for (1ULL << 31) in DImode
The SINGLE_BIT_MASK_OPERAND() is overly restrictive, triggering for
bits above 31 only (to side-step any issues with the negative SImode
value 0x80000000/(-1ull << 31)/(1 << 31)). This moves the special
handling of this SImode value (i.e. the check for (-1ull << 31) to
riscv.cc and relaxes the SINGLE_BIT_MASK_OPERAND() test.
With this, the code-generation for loading (1ULL << 31) from:
li a0,1
slli a0,a0,31
to:
bseti a0,zero,31
gcc/ChangeLog:
* config/riscv/riscv.cc (riscv_build_integer_1): Rewrite value as
(-1 << 31) for the single-bit case, when operating on (1 << 31)
in SImode.
* config/riscv/riscv.h (SINGLE_BIT_MASK_OPERAND): Allow for
any single-bit value, moving the special case for (1 << 31) to
riscv_build_integer_1 (in riscv.c).
Signed-off-by: Philipp Tomsich <philipp.tomsich@vrull.eu>
Jason Merrill [Wed, 1 Jun 2022 20:13:48 +0000 (16:13 -0400)]
c++: constexpr empty aggr [PR105795]
In this testcase, leaving ctx->ctor pointing to the enclosing object meant
that evaluating the initializer for the subobject clobbered previous
initializers for the enclosing object. So do update ctx->ctor, just don't
add it to the enclosing object ctor.
Roger Sayle [Thu, 2 Jun 2022 17:46:37 +0000 (18:46 +0100)]
PR target/105791: Add V1TI to V_128_256 for xop_pcmov_v1ti on x86_64.
This patch resolves PR target/105791 which is a regression that was
accidentally introduced for my workaround to PR tree-optimization/10566.
(a deeper problem in GCC's vectorizer creating VEC_COND_EXPR when it
shouldn't). The latest issues is that by providing a vcond_mask_v1tiv1ti
pattern in sse.md, the backend now calls ix86_expand_sse_movcc with
V1TImode operands, which has a special case for TARGET_XOP to generate
a vpcmov instruction. Unfortunately, there wasn't previously a V1TImode
variant, xop_pcmov_v1ti, so we'd ICE.
This is easily fixed by adding V1TImode (and V2TImode) to V_128_256
which is only used for defining XOP's vpcmov instruction. This in turn
requires V1TI (and V2TI) to be supported by <avxsizesuffix> (though
the use if <avxsizesuffix> in the names xop_pcmov_<mode><avxsizesuffix>
seems unnecessary; the mode makes the name unique).
2022-06-02 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
PR target/105791
* config/i386/sse.md (V_128_256):Add V1TI and V2TI.
(define_mode_attr avxsizesuffix): Add support for V1TI and V2TI.
gcc/testsuite/ChangeLog
PR target/105791
* gcc.target/i386/pr105791.c: New test case.
Eric Botcazou [Tue, 19 Apr 2022 19:53:35 +0000 (21:53 +0200)]
[Ada] Couple of small preparatory adjustments
No functional changes.
gcc/ada/
* gcc-interface/decl.cc (gnat_to_gnu_entity): Do not check the
scope of anonymous access Itypes.
* gcc-interface/trans.cc (Identifier_to_gnu): Do not translate
the return type of a subprogram here.
Eric Botcazou [Sat, 21 May 2022 19:54:45 +0000 (21:54 +0200)]
[Ada] Restore full generation of static dispatch tables with -gnatzr
The -gnatzr switch triggers the creation of distribution stubs for use
by the implementation of PolyORB. Now these stubs declare tagged types
and are generated at the very end of the analysis of compilation units,
after the static dispatch tables have been built, so these tables are
missing for the tagged types of the stubs.
Therefore this change defers the generation of static dispatch tables
for compilation units, which is the common case, until after the stubs
are (potentially) generated. For the other cases, in particular the
generic instances that are not compilation units, nothing is changed.
gcc/ada/
* exp_ch7.adb (Expand_N_Package_Body): Build static dispatch
tables only for units that are not compilation units, unless
they are generic instances. Do not push a scope for this.
(Expand_N_Package_Declaration): Build static dispatch tables
only for units that are both not compilation units and generic
instances.
* exp_disp.adb (Build_Static_Dispatch_Tables): Remove redundant
early return. Push a scope for package bodies.
* sem_ch10.adb: Add with and use clauses for Exp_Disp.
(Analyze_Compilation_Unit): Build static dispatch tables here.
Piotr Trojanek [Fri, 20 May 2022 07:41:30 +0000 (09:41 +0200)]
[Ada] Remove redundant checks for missing lists
When iterating over list elements with First/Next there is no need to
check if the list is present, because First intentionally returns Empty
if list is not present and the condition of subsequent loop will not be
satisfied.
Code cleanup; semantics is unaffected.
Occurrences of the redundant pattern were found with:
$ grep First -B 3 | less
and examining the output for the calls to Present.
gcc/ada/
* exp_ch13.adb, exp_ch5.adb, exp_ch9.adb, exp_strm.adb,
sem_ch10.adb, sem_ch13.adb, sem_ch5.adb, sem_ch6.adb,
sem_ch8.adb, sem_elab.adb, sem_eval.adb, sem_prag.adb,
sem_util.adb: Remove checks for the missing list before
iterating with First/Next; reindent code and refill comments.
Eric Botcazou [Wed, 18 May 2022 10:17:27 +0000 (12:17 +0200)]
[Ada] Get rid of secondary stack for most calls returning tagged types
This eliminates the use of the secondary stack to return specific tagged
types from functions in calls that are not dispatching on result, which
comprises returning controlled types, by introducing thunks whose only
purpose is to move the result from the primary to the secondary stack
for primitive functions that are controlling on result, and referencing
them in the dispatch table in lieu of the primitive functions.
The implementation reuses the existing machinery of interface thunks and
thus creates another kind of thunks, secondary stack thunks, which only
perform a call to the primitive function and return the result.
gcc/ada/
* einfo.ads (Has_Controlling_Result): Document new usage.
(Is_Thunk): Document secondary stack thunks.
(Returns_By_Ref): Adjust.
* exp_ch6.adb (Caller_Known_Size): Return true for tagged types.
(Expand_N_Extended_Return_Statement): Do not call Set_By_Ref.
(Expand_Simple_Function_Return): For a BIP return with an Alloc_Form
parameter, mark the node as returning on the secondary stack.
Replace call to Is_Limited_Interface with Is_Limited_View. Deal wit
secondary stack thunks. Do not call Set_By_Ref. Optimize the case
of a call to a function whose type also needs finalization.
(Needs_BIP_Task_Actuals): Replace Thunk_Entity with Thunk_Target.
(Needs_BIP_Finalization_Master): Cosmetic fixes.
(Needs_BIP_Alloc_Form): Check No_Secondary_Stack restriction and
return true for tagged types.
* exp_ch7.adb (Transient Scope Management): Update description.
* exp_disp.adb (Expand_Dispatching_Call): Always set Returns_By_Ref
on designated type if the call is dispatching on result. Tidy up.
(Expand_Interface_Thunk): Change type of Thunk_Code from Node_Id to
List_Id. Change type of local variables from Node_Id to Entity_Id.
Propagate Aliased_Present flag to create the formals and explicitly
set Has_Controlling_Result to False. Build a secondary stack thunk
if necessary in the function case.
(Expand_Secondary_Stack_Thunk): New function.
(Make_Secondary_DT): Build secondary stack thunks if necessary.
(Make_DT): Likewise.
(Register_Predefined_Primitive): Likewise.
(Register_Primitive): Likewise.
* exp_util.ads (Is_Secondary_Stack_Thunk): Declare.
(Thunk_Target): Likewise.
* exp_util.adb (Is_Secondary_Stack_Thunk): New function.
(Thunk_Target): Likewise.
* fe.h (Is_Secondary_Stack_Thunk): Declare.
(Thunk_Target): Likewise.
* gen_il-fields.ads (Opt_Field_Enum): Remove By_Ref.
* gen_il-gen-gen_nodes.adb (N_Simple_Return_Statement): Likewise.
(N_Extended_Return_Statement): Likewise.
* sem_ch6.adb (Analyze_Subprogram_Specification): Skip check for
abstract return type in the thunk case.
(Create_Extra_Formals): Replace Thunk_Entity with Thunk_Target.
* sem_disp.adb (Check_Controlling_Formals): Skip in the thunk case.
* sem_util.adb: Add use and with clauses for Exp_Ch6.
(Compute_Returns_By_Ref): Do not process procedures and only set
the flag for direct return by reference.
(Needs_Secondary_Stack): Do not return true for specific tagged
types and adjust comments accordingly.
* sinfo.ads (By_Ref): Delete.
(N_Simple_Return_Statement): Remove By_Ref.
(N_Extended_Return_Statement): Likewise.
* gcc-interface/ada-tree.h (TYPE_RETURN_UNCONSTRAINED_P): Delete.
* gcc-interface/decl.cc (gnat_to_gnu_subprog_type): Do not use it.
Return by direct reference if the return type needs the secondary
stack as well as for secondary stack thunks.
* gcc-interface/gigi.h (fntype_same_flags_p): Remove parameter.
* gcc-interface/misc.cc (gnat_type_hash_eq): Adjust to above change.
* gcc-interface/trans.cc (finalize_nrv): Replace test on
TYPE_RETURN_UNCONSTRAINED_P with TYPE_RETURN_BY_DIRECT_REF_P.
(Subprogram_Body_to_gnu): Do not call maybe_make_gnu_thunk for
secondary stack thunks.
(Call_to_gnu): Do not test TYPE_RETURN_UNCONSTRAINED_P.
(gnat_to_gnu) <N_Simple_Return_Statement>: In the return by direct
reference case, test for the presence of Storage_Pool on the node
to build an allocator.
(maybe_make_gnu_thunk): Deal with Thunk_Entity and Thunk_Target.
* gcc-interface/utils.cc (fntype_same_flags_p): Remove parameter.
Marc Poulhiès [Mon, 16 May 2022 12:52:41 +0000 (14:52 +0200)]
[Ada] Fix references to old variables that have been refactored
Local_Entity_Suppress and Global_Entity_Suppress variables referencing
tables were refactored to Local_Suppress_Stack_Top and
Global_Suppress_Stack_Top stacks back in 2007. Fix remaining references
to these variables.
gcc/ada/
* einfo.ads: Fix reference to Global_Entity_Suppress and
Local_Entity_Suppress variable in the comments.
* sem.ads: Likewise.
* sem_prag.adb: Likewise.
Claire Dross [Wed, 18 May 2022 13:22:39 +0000 (15:22 +0200)]
[Ada] Rename GNATprove annotate pragma for termination to Always_Return
GNATprove changed the name of the pragma Annotate used to verify that
a subprogram always returns normally. It is now called Always_Return
instead of Terminating.
gcc/ada/
* libgnat/s-aridou.adb: Use Always_Return instead of Terminating
to annotate termination for GNATprove.
* libgnat/s-arit32.adb: Idem.
* libgnat/s-spcuop.ads: Idem.
Before this patch, the Functional Sets ans Maps were bounded both from
the user and the implementation points of view. To make them closer to
mathematical Sets ans Maps, this patch removes the bounds from the
contracts. Note that, in practice, they are still bounded by
Count_Type'Last, even if the user is not aware of it anymore.
This patch removed constraints on length of sets and maps from the
preconditions of functions. The function Length and Num_Overlaps now
return a Big_Natural.
gcc/ada/
* libgnat/a-cofuse.ads, libgnat/a-cofuse.adb,
libgnat/a-cofuma.ads, libgnat/a-cofuma.adb: Make Length and
Num_Overlaps return Big_Natural.
* libgnat/a-cforse.ads, libgnat/a-cforse.adb,
libgnat/a-cforma.adb, libgnat/a-cfhase.ads,
libgnat/a-cfhase.adb, libgnat/a-cfhama.adb,
libgnat/a-cfdlli.adb: Adapt code to handle Big_Integers instead
of Count_Type.
Ghjuvan Lacambre [Tue, 17 May 2022 18:03:44 +0000 (20:03 +0200)]
[Ada] Add precise subprogram identification to -fdiagnostics-format=json
This commit adds information allowing identification of the subprogram
surrounding the message emitted by gnat when using -gnatdJ along with
-fdiagnostics-format=json.
gcc/ada/
* errout.adb (Write_JSON_Span): Add subprogram name to emitted
JSON.
Arnaud Charlet [Tue, 17 May 2022 12:22:20 +0000 (12:22 +0000)]
[Ada] Refine previous changes on making symbols public/private
Inline_Always procedures should be kept public for proper inter unit
inlining.
gcc/ada/
* sem_ch7.adb (Set_Referencer_Of_Non_Subprograms): New local
procedure, used for code refactoring. Also take into account
Inline_Always pragma when deciding to make a symbol public for
C generation.
Yannick Moy [Mon, 16 May 2022 07:58:18 +0000 (09:58 +0200)]
[Ada] Fix spurious errors on ghost code in generics
After the recent fix for detecting illegal use of ghost entities in
code, spurious errors could be raised on generic code with ghost, due to
wrong setting of the ghost flags on copied entities from the generic to
the instantiation.
gcc/ada/
* atree.adb (New_Copy): Reset flags related to ghost entities
before marking the new node.
Eric Botcazou [Mon, 16 May 2022 14:14:46 +0000 (16:14 +0200)]
[Ada] Build static dispatch tables always at the end of declarative part
The static dispatch tables of library-level tagged types are either built
on the first object declaration or at the end of the declarative part of
the package spec or body. There is no real need for the former case, and
the tables are not built for other constructs that freeze (tagged) types.
Therefore this change removes the former case, thus causing the tables to
be always built at the end of the declarative part; that's orthogonal to
freezing and the tagged types are still frozen at the appropriate place.
Moreover, it wraps the code in the Actions list of a freeze node (like
for the nonstatic case) so that it is considered elaboration code by the
processing done in Sem_Elab and does not disturb it.
No functional changes.
gcc/ada/
* exp_ch3.adb (Expand_Freeze_Record_Type): Adjust comment.
(Expand_N_Object_Declaration): Do not build static dispatch tables.
* exp_disp.adb (Make_And_Insert_Dispatch_Table): New procedure.
(Build_Static_Dispatch_Tables): Call it to build the dispatch tables
and wrap them in the Actions list of a freeze node.
Doug Rupp [Mon, 16 May 2022 18:57:13 +0000 (11:57 -0700)]
[Ada] vx7r2-arm/aarch64 - Support Atomic Primitives True
This feature is an architecture feature, not an OS feature, so enable
on vx7r2 for arm and aarch64 to coincide with what is done on similarly
capable targets.
Yannick Moy [Mon, 2 May 2022 15:38:41 +0000 (17:38 +0200)]
[Ada] Issue errors on wrong context for ghost entities
References to ghost entities should only occur in ghost context. This
was not checked systematically on all references.
gcc/ada/
* sem_ch2.adb (Analyze_Identifier): Add checking for ghost
context.
* sem_ch5.adb (Analyze_Implicit_Label_Declaration): Treat
implicit labels like other entities by setting their ghost
status according to context.
* ghost.adb (Check_Ghost_Context): Adapt checking.
Eric Botcazou [Wed, 11 May 2022 08:10:52 +0000 (10:10 +0200)]
[Ada] Propagate Has_Inherit{able,ed}_Invariants to base types
The two flags apply to base types only like Has_Own_Invariants.
gcc/ada/
* sem_util.adb (Propagate_DIC_Attributes): Add ??? comment.
(Propagate_Invariant_Attributes): Likewise. Propagate the
Has_Inheritable_Invariants and Has_Inherited_Invariants to
the base type of the target type.
Justin Squirek [Tue, 10 May 2022 18:19:10 +0000 (18:19 +0000)]
[Ada] Gnatbind crash during checksum calculation
This patch corrects an error in the compiler whereby gnatbind may crash
during calculation of file checksums in certain corner cases due to
uninitialized lookup tables.
Claire Dross [Mon, 9 May 2022 13:44:22 +0000 (15:44 +0200)]
[Ada] Fix iteration on formal vectors
We need to use Extended_Index for the Position parameter of the Element
function in formal vectors so it is compatible with other primitives of
the Iterable aspect.
gcc/ada/
* libgnat/a-cfinve.ads (Element): Change the type of the
Position parameter to Extended_Index.
* libgnat/a-cfinve.adb (Element): Idem.
* libgnat/a-cofove.ads (Element): Idem.
* libgnat/a-cofove.adb (Element): Idem.
Joffrey Huguet [Mon, 9 May 2022 13:25:30 +0000 (15:25 +0200)]
[Ada] Add contracts to System.Address_To_Access_Conversions
This patch adds SPARK annotations to subprograms from
System.Address_To_Access_Conversions. To_Pointer is considered to have
no global items, if the returned value has no aliases. To_Address is
forbidden in SPARK because addresses are not handled.
Joffrey Huguet [Tue, 3 May 2022 12:46:35 +0000 (14:46 +0200)]
[Ada] Add contracts to Interfaces.C.Strings
This patch adds Global contracts and preconditions to subprograms of
Interfaces.C.Strings. Effects on allocated memory are modelled
through an abstract state, C_Memory. The preconditions protect against
Dereference_Error, but not Storage_Error (which is not handled by
SPARK). This patch also disables the use of To_Chars_Ptr, which
creates an alias between an ownership pointer and the abstract state,
and the use of Free, in SPARK code. Thus, memory leaks will happen
if the user creates the Chars_Ptr using New_Char_Array and New_String.
gcc/ada/
* libgnat/i-cstrin.ads (To_Chars_Ptr): Add SPARK_Mode => Off.
(Free): Likewise.
(New_Char_Array): Add global contracts and Volatile attribute.
(New_String): Likewise.
(Value, Strlen, Update): Add global contracts and preconditions.
* libgnat/i-cstrin.adb: Add SPARK_Mode => Off to the package
body.
Jakub Jelinek [Thu, 2 Jun 2022 08:40:12 +0000 (10:40 +0200)]
i386: Optimize away shift count masking of shifts/rotates some more [PR105778]
As the following testcase shows, our x86 backend support for optimizing
out useless masking of shift/rotate counts when using instructions
that naturally modulo the count themselves is insufficient.
The *_mask define_insn_and_split patterns use
(subreg:QI (and:SI (match_operand:SI) (match_operand "const_int_operand")))
for the masking, but that can catch only the case where the masking
is done in SImode, so typically in SImode in the source.
We then have another set of patterns, *_mask_1, which use
(and:QI (match_operand:QI) (match_operand "const_int_operand"))
If the masking is done in DImode or in theory in HImode, we don't match
it.
The following patch does 4 different things to improve this:
1) drops the mode from AND and MATCH_OPERAND inside of the subreg:QI
and replaces that by checking that the register shift count has
SWI48 mode - I think doing it this way is cheaper than adding
another mode iterator to patterns which use already another mode
iterator and sometimes a code iterator as well
2) the doubleword shift patterns were only handling the case where
the shift count is masked with a constant that has the most significant
bit clear, i.e. where we know the shift count is less than half the
number of bits in double-word. If the mask is equal to half the
number of bits in double-word minus 1, the masking was optimized
away, otherwise the AND was kept.
But if the most significant bit isn't clear, e use a word-sized shift
and SHRD instruction, where the former does the modulo and the latter
modulo with 64 / 32 depending on what mode the CPU is in (so 64 for
128-bit doubleword and 32 or 64-bit doubleword). So we can also
optimize away the masking when the mask has all the relevant bits set,
masking with the most significant bit will remain for the cmove
test.
3) as requested, this patch adds a bunch of force_reg calls before
gen_lowpart
4) 1-3 above unfortunately regressed
+FAIL: gcc.target/i386/bt-mask-2.c scan-assembler-not and[lq][ \\t]
+FAIL: gcc.target/i386/pr57819.c scan-assembler-not and[lq][ \\t]
where we during combine match the new pattern we didn't match
before and in the end don't match the pattern we were testing for.
These 2 tests are fixed by the *jcc_bt<mode>_mask_1 pattern
addition and small tweak to target rtx_costs, because even with
the pattern around we'd refuse to match it because it appeared to
have higher instruction cost
2022-06-02 Jakub Jelinek <jakub@redhat.com>
PR target/105778
* config/i386/i386.md (*ashl<dwi>3_doubleword_mask): Remove :SI
from AND and its operands and just verify operands[2] has HImode,
SImode or for TARGET_64BIT DImode. Allow operands[3] to be a mask
with all low 6 (64-bit) or 5 (32-bit) bits set and in that case
just throw away the masking. Use force_reg before calling
gen_lowpart.
(*ashl<dwi>3_doubleword_mask_1): Allow operands[3] to be a mask
with all low 6 (64-bit) or 5 (32-bit) bits set and in that case
just throw away the masking.
(*ashl<mode>3_doubleword): Rename to ...
(ashl<mode>3_doubleword): ... this.
(*ashl<mode>3_mask): Remove :SI from AND and its operands and just
verify operands[2] has HImode, SImode or for TARGET_64BIT DImode.
Use force_reg before calling gen_lowpart.
(*<insn><mode>3_mask): Likewise.
(*<insn><dwi>3_doubleword_mask): Likewise. Allow operands[3] to be
a mask with all low 6 (64-bit) or 5 (32-bit) bits set and in that
case just throw away the masking. Use force_reg before calling
gen_lowpart.
(*<insn><dwi>3_doubleword_mask_1): Allow operands[3] to be a mask
with all low 6 (64-bit) or 5 (32-bit) bits set and in that case just
throw away the masking.
(*<insn><mode>3_doubleword): Rename to ...
(<insn><mode>3_doubleword): ... this.
(*<insn><mode>3_mask): Remove :SI from AND and its operands and just
verify operands[2] has HImode, SImode or for TARGET_64BIT DImode.
Use force_reg before calling gen_lowpart.
(splitter after it): Remove :SI from AND and its operands and just
verify operands[2] has HImode, SImode or for TARGET_64BIT DImode.
(*<btsc><mode>_mask, *<btsc><mode>_mask): Remove :SI from AND and its
operands and just verify operands[1] has HImode, SImode or for
TARGET_64BIT DImode. Use force_reg before calling gen_lowpart.
(*jcc_bt<mode>_mask_1): New define_insn_and_split pattern.
* config/i386/i386.cc (ix86_rtx_costs): For ZERO_EXTRACT with
ZERO_EXTEND QI->SI in last operand ignore the cost of the ZERO_EXTEND.
Richard Biener [Tue, 31 May 2022 07:37:05 +0000 (09:37 +0200)]
tree-optimization/101668 - relax SLP of existing vectors
This relaxes the conditions on SLPing extracts from existing vectors
leveraging the relaxed VEC_PERM conditions on the input vs output
vector type compatibility. It also handles lowpart extracts
and concats without VEC_PERMs now.
2022-05-25 Richard Biener <rguenther@suse.de>
PR tree-optimization/101668
* tree-vect-slp.cc (vect_build_slp_tree_1): Allow BIT_FIELD_REFs
for vector types with compatible lane types.
(vect_build_slp_tree_2): Deal with this.
(vect_add_slp_permutation): Adjust. Emit lowpart/concat
special cases without VEC_PERM.
(vectorizable_slp_permutation): Select the operand vector
type and relax requirements. Handle identity permutes
with mismatching operand types.
* optabs-query.cc (can_vec_perm_const_p): Only allow variable
permutes for op_mode == mode.
* gcc.target/i386/pr101668.c: New testcase.
* gcc.dg/vect/bb-slp-pr101668.c: Likewise.
Richard Biener [Wed, 1 Jun 2022 13:42:06 +0000 (15:42 +0200)]
tree-optimization/105802 - another unswitching type issue
This also fixes the type of the irange used for unswitching of
switch statements.
PR tree-optimization/105802
* tree-ssa-loop-unswitch.cc (find_unswitching_predicates_for_bb):
Make sure to also compute the range in the type of the switch index.
H.J. Lu [Tue, 18 Jan 2022 00:18:56 +0000 (16:18 -0800)]
libgcc: Align __EH_FRAME_BEGIN__ to pointer size
Aligne __EH_FRAME_BEGIN__ to pointer size since gcc/unwind-dw2-fde.h has
/* The first few fields of a CIE. The CIE_id field is 0 for a CIE,
to distinguish it from a valid FDE. FDEs are aligned to an addressing
unit boundary, but the fields within are unaligned. */
struct dwarf_cie
{
uword length;
sword CIE_id;
ubyte version;
unsigned char augmentation[];
} __attribute__ ((packed, aligned (__alignof__ (void *))));
/* The first few fields of an FDE. */
struct dwarf_fde
{
uword length;
sword CIE_delta;
unsigned char pc_begin[];
} __attribute__ ((packed, aligned (__alignof__ (void *))));
which indicates that CIE/FDE should be aligned at the pointer size.
PR libgcc/27576
* crtstuff.c (__EH_FRAME_BEGIN__): Aligned to pointer size.
David Seifert [Wed, 1 Jun 2022 21:21:44 +0000 (17:21 -0400)]
[PATCH] configure: use OBJDUMP determined by libtool [PR95648]
$ac_cv_prog_OBJDUMP contains the --host OBJDUMP that
libtool has inferred. Current config/gcc-plugin.m4 does
not respect the user's choice for OBJDUMP.
PR plugins/95648
config/
* gcc-plugin.m4: Use libtool's $ac_cv_prog_OBJDUMP.
H.J. Lu [Wed, 18 May 2022 20:00:47 +0000 (13:00 -0700)]
DSE: Use the constant store source if possible
RTL DSE tracks redundant constant stores within a basic block. When RTL
loop invariant motion hoists a constant initialization out of the loop
into a separate basic block, the constant store value becomes unknown
within the original basic block. When recording store for RTL DSE, check
if the source register is set only once to a constant by a non-partial
unconditional load. If yes, record the constant as the constant store
source. It eliminates unrolled zero stores after memset 0 in a loop
where a vector register is used as the zero store source.
gcc/
PR rtl-optimization/105638
* df-core.cc (df_find_single_def_src): Moved and renamed from
find_single_def_src in loop-iv.cc. Change the argument to rtx
and use rtx_equal_p. Return null for partial or conditional
defs.
* df.h (df_find_single_def_src): New prototype.
* dse.cc (record_store): Use the constant source if the source
register is set only once.
* loop-iv.cc (find_single_def_src): Moved to df-core.cc.
(replace_single_def_regs): Replace find_single_def_src with
df_find_single_def_src.
gcc/testsuite/
PR rtl-optimization/105638
* g++.target/i386/pr105638.C: New test.
Wilco Dijkstra [Wed, 1 Jun 2022 15:46:36 +0000 (16:46 +0100)]
AArch64: Cleanup option processing code
Further cleanup option processing. Remove the duplication of global
variables for CPU and tune settings so that CPU option processing is
simplified even further. Move global variables that need save and
restore due to target option processing into aarch64.opt. This removes
the need for explicit saving/restoring and unnecessary reparsing of
options.
gcc/
* config/aarch64/aarch64.opt (explicit_tune_core): Rename to
selected_tune.
(explicit_arch): Rename to selected_arch.
(x_aarch64_override_tune_string): Remove.
(aarch64_ra_sign_key): Add as TargetVariable so it gets saved/restored.
(aarch64_override_tune_string): Add Save so it gets saved/restored.
* config/aarch64/aarch64.h (aarch64_architecture_version): Remove.
* config/aarch64/aarch64.cc (aarch64_architecture_version): Remove.
(processor): Remove archtecture_version field.
(selected_arch): Remove global.
(selected_cpu): Remove global.
(selected_tune): Remove global.
(aarch64_ra_sign_key): Move global to aarch64.opt so it is saved.
(aarch64_override_options_internal): Use aarch64_get_tune_cpu.
(aarch64_override_options): Further simplify code to only set
selected_arch and selected_tune globals.
(aarch64_option_save): Remove now that target options are saved.
(aarch64_option_restore): Remove redundant target option restores.
* config/aarch64/aarch64-c.cc (aarch64_update_cpp_builtins): Use
AARCH64_ISA_V9.
* config/aarch64/aarch64-opts.h (aarch64_key_type): Add, moved from...
* config/aarch64/aarch64-protos.h (aarch64_key_type): Remove.
(aarch64_ra_sign_key): Remove.
Jakub Jelinek [Wed, 1 Jun 2022 15:54:39 +0000 (17:54 +0200)]
match.pd: Optimize __builtin_mul_overflow_p (x, cst, (utype)0) to x > ~(utype)0 / cst [PR30314]
A comparison with a constant is most likely always faster than
.MUL_OVERFLOW from which we only check whether it overflowed and not the
multiplication result, and even if not, it is simpler operation on GIMPLE
and even if a target exists where such multiplications with overflow checking
are cheaper than comparisons, because comparisons are so much more common
than overflow checking multiplications, it would be nice if it simply
arranged for comparisons to be emitted like those multiplications on its
own...
2022-06-01 Jakub Jelinek <jakub@redhat.com>
PR middle-end/30314
* match.pd (__builtin_mul_overflow_p (x, cst, (utype) 0) ->
x > ~(utype)0 / cst): New simplification.
Nathan Sidwell [Wed, 1 Jun 2022 11:52:21 +0000 (04:52 -0700)]
c++: Static init guard generation
The guard generation for a static var init was overly verbose. We can
use a bit of RAII and avoid some rechecking. Also in the !cxa_atexit
case, the only difference is whether can become whether to use
post-inc or pre-dec.
gcc/cp/
* decl2.cc (fix_temporary_vars_context_r): Use data argument
for new context.
(one_static_initialization_or_destruction): Adjust tree walk
call. Refactor guard generation.
Nathan Sidwell [Tue, 31 May 2022 20:22:06 +0000 (13:22 -0700)]
c++: Cleanup static init generation
The static init/fini generation is showing some bitrot. This cleans
up several places to use C++, and also take advantage of already
having checked a variable for non-nullness.
gcc/cp/
* decl2.cc (ssdf_decl): Delete global.
(start_static_storage_duration_function): Use some RAII.
(do_static_initialization_or_destruction): Likewise.
(c_parse_final_cleanups): Likewise. Avoid rechecking 'vars'.
Nathan Sidwell [Tue, 31 May 2022 14:56:53 +0000 (07:56 -0700)]
c++: Make static init generation more consistent
The end-of-compilation static init code generation functions are:
* Inconsistent in argument ordering (swapping 'is-init' and 'priority',
wrt each other and other arguments).
* Inconsistent in naming. mostly calling the is-init argument 'initp',
but sometimes calling it 'constructor_p' and in the worst case using
a transcoded 'methody_type' character, and naming the priority
argument 'initp'.
* Inconsistent in typing. Sometimes the priority is unsigned,
sometimes signed. And the initp argument can of course be a bool.
* Several of the function comments have bit-rotted.
This addresses those oddities. Name is-init 'initp', name priority
'priority'. Place initp first, make priority unsigned.
gcc/cp/
* decl2.cc (start_objects): Replace 'method_type' parameter
with 'initp' boolean, rename and retype 'priority' parameter.
(finish_objects): Likewise. Do not expand here.
(one_static_initialization_or_destruction): Move 'initp'
parameter first.
(do_static_initialization_or_destruction): Likewise.
(generate_ctor_or_dtor_function): Rename 'initp' parameter.
Adjust start_objects/finish_obects calls and expand here.
(generate_ctor_and_dtor_functions_for_priority): Adjust calls.
(c_parse_final_cleanups): Likewise.
(vtv_start_verification_constructor_init): Adjust.
(vtv_finish_verification_constructor_init): Use finish_objects.
Richard Biener [Wed, 1 Jun 2022 12:13:01 +0000 (14:13 +0200)]
tree-optimization/105786 - avoid strlen replacement for pointers
This avoids matching strlen to a pointer result, avoiding ICEing
because of an integer adjustment using PLUS_EXPR on pointers.
2022-06-01 Richard Biener <rguenther@suse.de>
PR tree-optimization/105786
* tree-loop-distribution.cc
(loop_distribution::transform_reduction_loop): Only do strlen
replacement for integer type reductions.
Jakub Jelinek [Wed, 1 Jun 2022 12:00:49 +0000 (14:00 +0200)]
unswitch: Fold case label lo/hi values to index type [PR105770]
The following testcase ICEs because we use different types in comparison,
idx has int type, while CASE_LOW has char type.
While I believe all CASE_{LOW,HIGH} in the same switch have to use the same
or compatible type, the index expression can have a promoted type as happens
in this testcase. Other spots that handle switches do such foldings too.
2022-06-01 Jakub Jelinek <jakub@redhat.com>
PR tree-optimization/105770
* tree-ssa-loop-unswitch.cc (find_unswitching_predicates_for_bb): Cast
CASE_LOW and CASE_HIGH to TREE_TYPE (idx) before comparisons with idx.
Aldy Hernandez [Sun, 22 May 2022 18:17:40 +0000 (20:17 +0200)]
Convert ranger and clients to vrange.
Finally, the meat of the work. Convert ranger and associated clients
to vrange.
Everything's relatively mechanical given the previous patches. I did
include a minor cleanup in the edge code. There's no need to check
that the type of the switch is an integer as non-integer switches are
invalid. I verified this with an appropriately coded assert.
Aldy Hernandez [Sun, 22 May 2022 18:17:39 +0000 (20:17 +0200)]
Revamp irange_allocator to handle vranges.
This patch revamps the range allocator to handle generic vrange's.
I've cleaned it up somehow to make it obvious the various things you
can allocate with it. I've also moved away from overloads into
distinct names when appropriate.
The various entry points are now:
// Allocate a range of TYPE.
vrange *alloc_vrange (tree type);
// Allocate a memory block of BYTES.
void *alloc (unsigned bytes);
// Return a clone of SRC.
template <typename T> T *clone (const T &src);
It is now possible to allocate a clone of an irange, or any future
range types:
You can actually do so without the <>, but I find it clearer to
specify the vrange type.
So with it you can allocate a specific range type, or vrange, or a
block of memory.
I have rewritten the C style casts to C++ casts, since casts tend to
be hints of problematic designs. With the C++ casts you can at least
grep for them easier. Speak of which, the next patch, which converts
ranger to vrange, will further clean this space by removing some
unnecessary casts.
Aldy Hernandez [Mon, 14 Mar 2022 12:31:18 +0000 (13:31 +0100)]
Convert range-op.* to vrange.
This patch provides the infrastructure to make range-ops type agnostic.
First, the range_op_handler function has been replaced with an object
of the same name. It's coded in such a way to minimize changes to the
code base, and to encapsulate the dispatch code.
Instead of:
range_operator *op = range_op_handler (code, type);
if (op)
op->fold_range (...);
We now do:
range_op_handler op (code, type);
if (op)
op->fold_range (...);
I've folded gimple_range_handler into the range_op_handler class,
since it's also a query into the range operators.
This all has the added benefit of moving all the dispatch code into an
independent class and avoid polluting range_operator (which we'll
further split later when frange and prange come live).
There's this annoying "using" keyword that's been added to each
operator due to hiding rules in C++. The issue is that we will have
different virtual versions of fold_range() for each combination of
operands. For example:
// Traditional binary op on irange's.
fold_range (irange &lhs, const irange &op1, const irange &op2);
// For POINTER_DIFF_EXPR:
fold_range (irange &lhs, const prange &op1, const prange &op2);
// Cast from irange to prange.
fold_range (prange &lhs, const irange &op1, const irange &op2);
Overloading virtuals when there are multiple same named methods causes
hidden virtuals warnings from -Woverloaded-virtual, thus the using
keyword. An alternative would be to have different names:
fold_range_III, fold_range_IPP, fold_range_PII, but that's uglier
still.
Aldy Hernandez [Mon, 14 Mar 2022 12:27:36 +0000 (13:27 +0100)]
Implement generic range temporaries.
Now that we have generic ranges, we need a way to define generic local
temporaries on the stack for intermediate calculations in the ranger
and elsewhere. We need temporaries analogous to int_range_max, but
for any of the supported types (currently just integers, but soon
integers, pointers, and floats).
The Value_Range object is such a temporary. It is designed to be
transparently used as a vrange. It shares vrange's abstract API, and
implicitly casts itself to a vrange when passed around.
The ultimate name will be value_range, but we need to remove legacy
first for that to happen. Until then, Value_Range will do.
Sample usage is as follows. Instead of:
extern void foo (vrange &);
int_range_max t;
t.set_nonzero (type);
foo (t);
one does:
Value_Range t (type);
t.set_nonzero (type);
foo (t);
You can also delay initialization, for use in loops for example:
Creating an supported range type, will result in an unsupported_range
object being created, which will trap if anything but set_undefined()
and undefined_p() are called on it. There's no size penalty for the
unsupported_range, since its immutable and can be shared across
instances.
Since supports_type_p() is called at construction time for each
temporary, I've removed the non-zero check from this function, which
was mostly unneeded. I fixed the handful of callers that were
passing null types, and in the process sped things up a bit.
As more range types come about, the Value_Range class will be augmented
to support them by adding the relevant bits in the initialization
code, etc.
Tested on x86-64 & ppc64le Linux.
gcc/ChangeLog:
* gimple-range-fold.h (gimple_range_type): Check type before
calling supports_type_p.
* gimple-range-path.cc (path_range_query::range_of_stmt): Same.
* value-query.cc (range_query::get_tree_range): Same.
* value-range.cc (Value_Range::lower_bound): New.
(Value_Range::upper_bound): New.
(Value_Range::dump): New.
* value-range.h (class Value_Range): New.
(irange::supports_type_p): Do not check if type is non-zero.
Aldy Hernandez [Mon, 14 Mar 2022 13:04:56 +0000 (14:04 +0100)]
Implement abstract vrange class.
This is a series of patches making ranger type agnostic in preparation
for contributing support for other types of ranges (pointers and
floats initially).
The first step in this process is to implement vrange, an abstract
class that will be exclusively used by ranger, and from which all
ranges will inherit. Vrange provides the minimum operations for
ranger to work. The current virtual methods are what we've used to
implement frange (floats) and prange (pointers), but we may restrict
the virtual methods further as other ranges come about
(i.e. set_nonnegative() has no meaning for a future string range).
This patchset also provides a mechanism for declaring local type
agnostic ranges that can transparently hold an irange, frange,
prange's, etc, and a dispatch mechanism for range-ops to work with
various range types. More details in the relevant patches.
FUTURE PLAN
===========
The plan after this is to contribute a bare bones implementation for
floats (frange) that will provide relationals, followed by a
separation of integers and pointers (irange and prange). Once this is
in place, we can further enhance both floats and pointers. For
example, pointer tracking, pointer plus optimizations, and keeping
track of NaN's, etc.
Once frange and prange come live, all ranger clients will immediately
benefit from these enhancements. For instance, in our local branch,
the threader is already float aware with regards to relationals.
We expect to wait a few weeks before starting to contribute further
enhancements to give the tree a time to stabilize, and Andrew time to
rebase his upcoming patches :-P.
NOTES
=====
In discussions with Andrew, it has become clear that with vrange
coming about, supports_type_p() is somewhat ambiguous. Prior to
vrange it has been used to (a) determine if a type is supported by
ranger, (b) as a short-cut for checking if a type is pointer or integer,
as well as (c) to see if a given range can hold a type. These things
have had the same meaning in irange, but are slightly different with
vrange. I will address this in a follow-up patch.
Speaking of supported types, we now provide an unsupported_range
for passing around ranges for unsupported types. We've been silently
doing this for a while, in both vr-values by creating VARYING for
unsupported types with error_mark_node end points, and in ranger when
we pass an unsupported range before we realize in range_of_expr that
it's unsupported. This class just formalizes what we've already been
doing in an irange, but making it explicit that you can't do anything
with these ranges except pass them. Any other operation traps.
There is no GTY support for vrange yet, as we don't store it long
term. When we contribute support for global ranges (think
SSA_NAME_RANGE_INFO but for generic ranges), we will include it. There
was just no need to pollute this patchset with it.
TESTING
=======
The patchset has been tested on x86-64 Linux as well as ppc64 Linux.
I have also verified that we fold the same number of conditionals in
evrp as well as thread the same number of paths. There should be no
user visible changes.
We have also benchmarked the work, with the final numbers being an
*improvement* of 1.92% for evrp, and 0.82% for VRP. Overall
compilation has a miniscule improvement. This is despite the extra
indirection level.
The improvements are mostly because of small cleanups required for the
generalization of ranges. As a sanity check, I stuck kcachegrind on a
few sample .ii files to see where the time was being gained. Most of
the gain came from gimple_range_global() being 19% faster. This
function is called a lot, and it was constructing a legacy
value_range, then returning it by value, which the caller then had to
convert to an irange. This is in line with other pending work:
anytime we get rid of legacy, we gain time.
I will wait a few days before committing to welcome any comments.
Yannick Moy [Mon, 9 May 2022 13:26:22 +0000 (15:26 +0200)]
[Ada] Allow confirming volatile properties on No_Caching variables
Volatile variables marked with the No_Caching aspect can now have
confirming aspects for other volatile properties, with a value of
False.
gcc/ada/
* contracts.adb (Check_Type_Or_Object_External_Properties): Check
the validity of combinations only when No_Caching is not used.
* sem_prag.adb (Analyze_External_Property_In_Decl_Part): Check
valid combinations with No_Caching.
git://gcc.gnu.org / gcc.git / log