Tom Tromey [Tue, 17 Sep 2019 08:02:41 +0000 (08:02 +0000)]
[Ada] Update the character type comment in exp_dbug.ads
The character type comment in exp_dbug.ads documented the QU and QW
encodings, but did not document the plain "Qc" encoding, where 'c' is
the character itself. This patch updates the comment to follow the
implementation.
[Ada] Minor fixes mostly in comments of runtime arithmetic unit
Multiple comments in functions Double_Divide and Scaled_Divide were
incorrect. Now fixed.
Also change the expression (if Zhi /= 0 then Ylo * Zhi else 0) to the
simpler equivalent (Ylo * Zhi) in Double_Divide.
Also add a comment explaining why the implementation of Algorithm D for
multiple-precision division from the 2nd Edition of The Art of Computer
Programming does not suffer from two bugs discovered on that version.
There is no impact on execution, hence no test.
2019-09-17 Yannick Moy <moy@adacore.com>
gcc/ada/
* libgnat/s-arit64.adb (Double_Divide): Simplify needlessly
complex computation. Fix comments.
(Scaled_Divide): Fix comments. Explain why implementation does
not suffer from bugs in Algorithm D from 2nd Edition of TAOCP.
[Ada] Raise Constraint_Error in overflow case involving rounding
Function Scaled_Divide in s-arith runtime unit computes the combined
multiplication and division of its arguments ((X*Y) / Z). In a very
special case where the quotient computed before rounding is 2**64-1,
then rounding may lead to undesirable wrap-around, leading to a computed
quotient of 0 instead of raising Constraint_Error as expected.
This function is only called in the runtime for arithmetic operations
involving fixed-point types. Rounding is performed only when the target
type is of a decimal fixed-point type, and the attribute 'Round of the
type is used to round the result of the arithmetic operation.
This fix correctly raises Constraint_Error in this special case.
2019-09-17 Yannick Moy <moy@adacore.com>
gcc/ada/
* libgnat/s-arit64.adb (Scaled_Divide): Add protection against
undesirable wrap-around.
[Ada] Fix possible suppressed overflows in arithmetic run-time
Function Double_Divide computes the division of its parameters
(X / (Y*Z)) in a way that avoids overflows on signed integers, except in
two specific cases, when X = -2**63, abs(Y) = abs(Z) = 1 (leading to an
overflow in -To_Int(Qu)) and when X = -2**63 and Y*Z is large enough
that Qu=0 and so the remainder Ru=2**63 (leading to an overflow in
-To_Int(Ru)), for example with Y = Z = 2**32-1.
This fix avoids the overflow by applying "-" on the unsigned value
before the conversion to signed integer.
The issue cannot manifest as an overflow check failure in our runtime,
as overflow checks are suppressed by using pragma Suppress at the start
of the file. Assuming a machine implements wraparound semantics here,
the result was correct even with the suppressed overflow.
As a result, there can be no test showing the difference.
2019-09-17 Yannick Moy <moy@adacore.com>
gcc/ada/
* libgnat/s-arit64.adb (Double_Divide): Fix two possible
overflows.
Old versions forced even file timestamp in Windows in
GNAT.OS_Lib.GM_Split implementation. We removed this time distortion in
GNAT.OS_Lib.GM_Split a few commits. But gprbuild became unnecessary
recompiling the sources in 50% cases, when the source has odd file
timestamp. This commit is restoring the compatibility between new
gprbuild and old GNAT compilers.
* make_util.ads (On_Windows): Move...
* osint.ads (On_Windows): There.
* osint.adb (OS_Time_To_GNAT_Time): If odd incremented on
Windows before conversion to Time_Stamp_Type.
[Ada] GNATprove: avoid crash on illegal borrow during package elaboration
In GNATprove, a borrow during package elaboration is illegal, but a
crash could be issued when analyzing the corresponding declaration. Now
avoid analyzing the declaration in that case.
There is no test as this does not impact compilation.
2019-09-17 Yannick Moy <moy@adacore.com>
gcc/ada/
* sem_spark.adb (Check_Declaration): Do not check the assignment
from an illegal declaration.
Sem_Util.Wrong_Type usually emits an error message, but in some cases it
does not. The code which prevents emitting an error message was going
too far in some cases, causing illegal constructs to be accepted. For
example, a qualified expression such as Integer'(null) might be passed
in as an actual parameter in an instantiation of a generic and generate
no error message.
Running this command:
gcc -c inst.ads
On the following sources:
package Inst is
type Ptr is new Integer;
generic
type TElement is private;
NonDefini : TElement;
package ArbMgr is
end ArbMgr;
package Pack is new ArbMgr (Ptr, Ptr'(null));
procedure Dummy;
end Inst;
Should produce the following output:
inst.ads:10:42: expected type "Ptr" defined at line 2
inst.ads:10:42: found an access type
compilation abandoned due to previous error
2019-09-17 Steve Baird <baird@adacore.com>
gcc/ada/
* sem_util.adb (Wrong_Type): In deciding to suppress a message,
it is not enough for In_Instance to be True; in addition,
In_Generic_Actual (Expr) must be False.
* sem_type.adb (In_Generic_Actual): Fix bug where traversal of
parents skips every other node.
[Ada] Support chained calls to traversal functions in SPARK
This change only affects the SPARK toolset. In the part of semantic
analysis enforcing ownership rules for SPARK, it corrects a crash in
analysis of a declaration of a local borrower whose definition is a
chain of several calls to traversal functions.
2019-09-17 Claire Dross <dross@adacore.com>
gcc/ada/
* sem_spark.adb (Get_Observed_Or_Borrowed_Expr): If the
definition of a local borrower contains calls to traversal
functions, the borrowed expression is the first parameter of the
first traversal function call in the definition.
[Ada] Close file descriptors allocated for tty only once
2019-09-17 Vadim Godunko <godunko@adacore.com>
gcc/ada/
* libgnat/g-exptty.ads (Close_Input): New subprogram.
* libgnat/g-exptty.adb (Close_Input): New subprogram.
(Close): Move close of TTY to Close_Input.
* terminals.c (__gnat_close_tty): Set file descriptors to
invalid value after close.
'Close' subprogram of GNAT.Expect can close irrelevant file descriptors
when 'Expect' was terminated by Process_Died exception and any file open
operations was done before call to 'Close'.
2019-09-17 Vadim Godunko <godunko@adacore.com>
gcc/ada/
* libgnat/g-expect.ads, libgnat/g-expect.adb (Close_Input): New
subprogram.
(Get_Command_Output): Call Close_Input to close input stream.
(Expect_Internal): Likewise.
(Close): Likewise.
* libgnat/g-exptty.adb (Close): Likewise.
[Ada] Do not inline dispatching operations in GNATprove mode
In GNATprove, local subprograms without contracts are candidates for
inlining, so that they are only analyzed in the context of their calls.
This does not apply to dispatching operations, which may be called
through dispatching, in an unknown calling context. Hence such
operations should not be considered as candidates for inlining.
There is no test as this has no effect on compilation.
2019-09-17 Yannick Moy <moy@adacore.com>
gcc/ada/
* inline.adb (Can_Be_Inlined_In_GNATprove_Mode): Add handling
for dispatching operations.
Ed Schonberg [Tue, 17 Sep 2019 08:00:07 +0000 (08:00 +0000)]
[Ada] In a generic use Presanalyze_Spec_Expression on Predicates
When verifying that the meaning of an aspect has not changed between the
freeze point of the entity and the end of the declarations, we analkyze
a copy of the expression to verify its conformance to previous analysis.
If the expression contains overloaded references, these have to be
resolved, which is not done if the expression is only preanalyzed. This
applies in particular to expressions in predicates.
2019-09-17 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_ch13.adb (Check_Aspect_At_End_Of_Declarations): In a
generic context, for a Predicate aspect, use
Preanalyze_Spec_Expression to verify conformance.
gcc/testsuite/
* gnat.dg/predicate13.adb, gnat.dg/predicate13.ads: New
testcase.
Javier Miranda [Tue, 17 Sep 2019 07:59:58 +0000 (07:59 +0000)]
[Ada] Missing tagged type decoration in corresponding record subtypes
The frontend silently skips propagating attribute Is_Tagged_Type to the
constrained corresponding record subtype associated with a protected
type or task type with discriminants. This change does not affect the
functionality of the compiler; it leaves more clean the decoration of
internal entities.
Javier Miranda [Tue, 17 Sep 2019 07:59:53 +0000 (07:59 +0000)]
[Ada] Ada 2020: Raise expressions in limited contexts (AI12-0172)
This patch adds support for the use of raise expressions in more
limited contexts (as described in the Ada Isssue AI12-0172).
2019-09-17 Javier Miranda <miranda@adacore.com>
gcc/ada/
* exp_ch3.adb (Build_Record_Init_Proc): Do not generate code to
adjust the tag component when the record is initialized with a
raise expression.
* sem_aggr.adb (Valid_Limited_Ancestor): Return True for
N_Raise_Expression nodes.
(Valid_Ancestor_Type): Return True for raise expressions.
* sem_ch3.adb (Analyze_Component_Declaration): Do not report an
error when a component is initialized with a raise expression.
* sem_ch4.adb (Analyze_Qualified_Expression): Do not report an
error when the aggregate has a raise expression.
Piotr Trojanek [Tue, 17 Sep 2019 07:59:48 +0000 (07:59 +0000)]
[Ada] Ignore missing ALI files in GNATprove mode
This change only affects GNATprove backend, where it avoids spurious
errors on missing ALI files for units indirectly withed from the current
unit and processed as part of a different .gpr project.
No test provided, because only GNATprove is affected.
2019-09-17 Piotr Trojanek <trojanek@adacore.com>
gcc/ada/
* ali.ads: Fix casing in comment.
* ali-util.ads, ali-util.adb (Read_Withed_ALIs): Remove
Ignore_Errors parameter; it was only set to non-default value of
True when running in GNATprove_Mode and wrongly reset to False
when calling this routine recursively. Now in GNATprove mode we
want it to be always True, so in fact it is equivalent to
GNATProve_Mode flag itself (which was already used in this
routine).
We use Long_Long_Integer'Size / 2 instead of 4 * 8 to support
specifying a target configuration file where the largest integer is
32 bits instead of 64.
2019-09-17 Arnaud Charlet <charlet@adacore.com>
gcc/ada/
* libgnat/s-bitfie.ads (Val_Bits, Val_Bytes): Define from
Long_Long_Integer'Size.
Javier Miranda [Tue, 17 Sep 2019 07:59:33 +0000 (07:59 +0000)]
[Ada] Adding assertions on extra formals for BIP function calls
This patch adds assertions to ensure that the frontend passes to the
backend the right number of extra parameters required for build in place
function calls. No functional change.
2019-09-17 Javier Miranda <miranda@adacore.com>
gcc/ada/
* exp_ch6.ads (Needs_BIP_Task_Actuals): New subprogram.
* exp_ch6.adb (Add_Task_Actuals_To_Build_In_Place_Call): Code
cleanup.
(Check_Number_Of_Actuals): New subprogram.
(Make_Build_In_Place_Call_In_Allocator): Adding assertion.
(Make_Build_In_Place_Call_In_Anonymous_Context): Adding
assertion.
(Make_Build_In_Place_Call_In_Assignment): Adding assertion.
(Make_Build_In_Place_Call_In_Object_Declaration): Code cleanup
plus assertion addition.
(Needs_BIP_Task_Actuals): New subprogram.
* sem_ch6.adb (Create_Extra_Formals): Rely on
Needs_BIP_Task_Actuals() to check if the master of the tasks to
be created, and the caller's activation chain formals are
needed.
...in cases where the Val_2 might cross a page boundary, and the second
page is now known to exist.
Copy_Bitfield is still disabled in the compiler: no test possible.
2019-09-17 Bob Duff <duff@adacore.com>
gcc/ada/
* libgnat/s-bituti.adb (Get_Val_2, Set_Val_2): Use new routines
for getting and setting a Val_2, avoiding touching the second
half when that half might not exist.
* exp_ch5.adb (Expand_Assign_Array_Loop_Or_Bitfield): Correct
tests for potential volatile or independent components. In
particular, do not call Prefix unless we know it's a slice.
[Ada] Support for local unix sockets in GNAT.Sockets API
Sock_Addr_Type has Family_Unix variant now. This variant can be created
with function Unix_Local_Addr call. And this variant is working in
GNAT.Socket routines where it is appropriate.
* gsocket.h: Include sys/un.h.
* s-oscons-tmplt.c (AF_UNIX): New constant generation.
(SIZEOF_sockaddr_un): Idem.
* libgnat/g-socket.ads (Family_Type): New value Family_Unix
added.
(Family_Inet_4_6): New subtype only for network families.
(Sock_Addr_Type): Add Unbounded_String field for Family_Unix
variant.
(Unix_Socket_Address): Create Sock_Addr_Type from socket
pathname.
(Network_Socket_Address): Create Sock_Addr_Type from
Inet_Addr_Type and Port_Type parameters.
* libgnat/g-socket.adb: Support local unix address in socket
routines.
(Get_Address_Info): Disable warning about Result may be
referenced before it has a value. Remove duplicated code to exit
from Look_For_Supported.
* libgnat/g-sothco.ads (Unix_Name_Length): New constant defining
maximum number of characters in local socket address path.
(Sockaddr): Add variant for Family_Unix address family. Move
Sin_Port and Sin_Family to Family_Inet section. Add Sin6_Port
and Sin6_Family to Family_Inet6 section.
(Set_Address): Add out parameter Length to return valuable
Sockaddr data length.
(Get_Address): Add input parameter Length to set valuable
Sockaddr data length.
* libgnat/g-sothco.adb: Support local unix address in socket
routines.
Eric Botcazou [Tue, 17 Sep 2019 07:59:16 +0000 (07:59 +0000)]
[Ada] Fix wrong value of 'Size for slices of bit-packed arrays
This fixes a long-standing issue in the compiler which would return
a wrong value for the Size attribute applied to slices of bit-packed
arrays whose size is not a multiple of the storage unit.
The problem is that the computation was done in the code generator
after the bit-packed array had been internally rewritten into an
array of bytes, so the Size was always rounded up to the next byte.
The computation is now rewritten into the product of the Length and
Compnent_Size attribute of the slices before being sent to the code
generator.
2019-09-17 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* exp_attr.adb (Expand_Size_Attribute): Chain the special cases
on the back-end path and rewrite the attribute appled to slices
of bit-packed arrays into the product of the Length and the
Compoent_Size attributes of the slices.
* exp_ch5.adb (Expand_Assign_Array_Bitfield): Use Size attribute
directly to compute the bitfield's size.
Paolo Carlini [Mon, 16 Sep 2019 23:04:10 +0000 (23:04 +0000)]
decl.c (grokdeclarator): Use declspecs->locations and declarator->id_loc in a few error messages.
/cp
2019-09-16 Paolo Carlini <paolo.carlini@oracle.com>
* decl.c (grokdeclarator): Use declspecs->locations and
declarator->id_loc in a few error messages.
* pt.c (finish_member_template_decl): Use DECL_SOURCE_LOCATION.
(push_template_decl_real): Likewise.
/testsuite
2019-09-16 Paolo Carlini <paolo.carlini@oracle.com>
Martin Liska [Mon, 16 Sep 2019 14:22:36 +0000 (16:22 +0200)]
Rewrite part of and_comparisons_1 into match.pd.
2019-09-16 Martin Liska <mliska@suse.cz>
* genmatch.c (dt_node::append_simplify): Do not print
warning when we have duplicate patterns belonging
to a same simplify rule.
* gimple-fold.c (and_comparisons_1): Remove matching moved to match.pd.
(maybe_fold_comparisons_from_match_pd): Handle
tcc_comparison as a results.
* match.pd: Handle (X == CST1) && (X OP2 CST2) conditions.
Li Jia He [Mon, 16 Sep 2019 14:22:16 +0000 (14:22 +0000)]
Fix PR88784, middle end is missing some optimizations about unsigned
2019-09-16 Li Jia He <helijia@linux.ibm.com>
Qi Feng <ffengqi@linux.ibm.com>
PR middle-end/88784
* match.pd (x > y && x != XXX_MIN): Optimize into 'x > y'.
(x > y && x == XXX_MIN): Optimize into 'false'.
(x <= y && x == XXX_MIN): Optimize into 'x == XXX_MIN'.
(x < y && x != XXX_MAX): Optimize into 'x < y'.
(x < y && x == XXX_MAX): Optimize into 'false'.
(x >= y && x == XXX_MAX): Optimize into 'x == XXX_MAX'.
(x > y || x != XXX_MIN): Optimize into 'x != XXX_MIN'.
(x <= y || x != XXX_MIN): Optimize into 'true'.
(x <= y || x == XXX_MIN): Optimize into 'x <= y'.
(x < y || x != XXX_MAX): Optimize into 'x != XXX_MAX'.
(x >= y || x != XXX_MAX): Optimize into 'true'.
(x >= y || x == XXX_MAX): Optimize into 'x >= y'.
2019-09-16 Li Jia He <helijia@linux.ibm.com>
Qi Feng <ffengqi@linux.ibm.com>
PR middle-end/88784
* gcc.dg/pr88784-1.c: New testcase.
* gcc.dg/pr88784-2.c: New testcase.
* gcc.dg/pr88784-3.c: New testcase.
* gcc.dg/pr88784-4.c: New testcase.
* gcc.dg/pr88784-5.c: New testcase.
* gcc.dg/pr88784-6.c: New testcase.
* gcc.dg/pr88784-7.c: New testcase.
* gcc.dg/pr88784-8.c: New testcase.
* gcc.dg/pr88784-9.c: New testcase.
* gcc.dg/pr88784-10.c: New testcase.
* gcc.dg/pr88784-11.c: New testcase.
* gcc.dg/pr88784-12.c: New testcase.
Jason Merrill [Mon, 16 Sep 2019 04:34:28 +0000 (00:34 -0400)]
PR c++/30277 - int-width bit-field promotion.
Here, if cp_perform_integral_promotions saw that the TREE_TYPE of a
bit-field reference was the same as the type it promotes to, it didn't do
anything. But then decay_conversion saw that the bit-field reference was
unchanged, and converted it to its declared type. So I needed to add
something to make it clear that promotion has been done. But then the 33819
change caused trouble by looking through the NOP_EXPR I just added. This
was the wrong fix for that bug; I've now fixed that better by recognizing in
cp_perform_integral_promotions that we won't promote a bit-field larger than
32 bits, so we should use the declared type.
PR c++/33819 - long bit-field promotion.
* typeck.c (cp_perform_integral_promotions): Handle large bit-fields
properly. Handle 32-bit non-int bit-fields properly.
(is_bitfield_expr_with_lowered_type): Don't look through NOP_EXPR.
Jason Merrill [Mon, 16 Sep 2019 04:34:23 +0000 (00:34 -0400)]
PR c++/82165 - enum bitfields and operator overloading.
In this testcase, !f.b0 was failing to call the overloaded operator because
TREE_TYPE is the magic bitfield integer type, and we weren't using
unlowered_expr_type the way we do in other places. It would be nice if we
could give bit-field COMPONENT_REFs their declared type until genericization
time...
* call.c (build_new_op_1): Use unlowered_expr_type.
Jason Merrill [Mon, 16 Sep 2019 04:34:18 +0000 (00:34 -0400)]
Fix conversions for built-in operator overloading candidates.
While working on C++20 operator<=>, I noticed that build_new_op_1 was doing
too much conversion when a built-in candidate was selected; the standard
says it should only perform user-defined conversions, and then leave the
normal operator semantics to handle any standard conversions. This is
important for operator<=> because a comparison of two different unscoped
enums is ill-formed; if we promote the enums to int here, cp_build_binary_op
never gets to see the original operand types, so we can't give the error.
* call.c (build_new_op_1): Don't apply any standard conversions to
the operands of a built-in operator. Don't suppress conversions in
cp_build_unary_op.
* typeck.c (cp_build_unary_op): Do integral promotions for enums.
Jason Merrill [Mon, 16 Sep 2019 04:34:12 +0000 (00:34 -0400)]
Handle location wrappers better in warn_logical_operator.
When we introduced location wrappers, we added fold_for_warn to warnings
that are interested in a constant value, or wrapper-stripping to warnings
that are interested in literal constants. This particular warning is
looking for a literal constant, but was wrongly changed to use
fold_for_warn; this patch makes it strip instead.
gcc/testsuite/
* lib/target-supports.exp
(check_effective_target_arm_neon_fp16_hw)
(check_effective_target_arm_fp16_hw): Use check_runtime
instead of check_runtime_nocache.
Steven G. Kargl [Sun, 15 Sep 2019 17:49:44 +0000 (17:49 +0000)]
re PR fortran/91727 (ICE in conformable_arrays, at fortran/resolve.c:7490)
2019-09-15 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/91727
* resolve.c (conformable_arrays): If array-spec is NULL, then
allocate-object is a scalar. a conformability check only occurs
for an array source-expr.
2019-09-15 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/91727
* gfortran.dg/pr91727.f90: New test.
Thomas Koenig [Sun, 15 Sep 2019 14:57:48 +0000 (14:57 +0000)]
re PR fortran/91550 (ICE in do_subscript, at fortran/frontend-passes.c:2652)
2019-09-15 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/91550
* frontend-passes.c (do_subscript): If step equals
zero, a previuos error has been reported; do nothing
in this case.
* resolve.c (gfc_resolve_iterator): Move error checking
after type conversion.
2019-09-15 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/91550
* gfortran.dg/do_subscript_6.f90: New test.
Thomas Koenig [Sat, 14 Sep 2019 20:40:55 +0000 (20:40 +0000)]
re PR fortran/91557 (Bogus warning about unused dummy argument _formal_*)
2019-09-14 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/91557
PR fortran/91556
* frontend-passes.c (check_externals_procedure): Reformat argument
list. Use gfc_compare_actual_formal instead of gfc_procedure_use.
* gfortran.h (gfc_symbol): Add flag error.
* interface.c (gfc_compare_interfaces): Reformat.
(argument_rank_mismatch): Add where_formal argument. If it is
present, note that the error is between different calls.
(compare_parameter): Change warnings that previously dependended
on -Wargument-mismatch to unconditional. Issue an error / warning
on type mismatch only once. Pass where_formal to
argument_rank_mismatch for artificial variables.
(compare_actual_formal): Change warnings that previously
dependeded on -Wargument-mismatch to unconditional.
(gfc_check_typebound_override): Likewise.
(gfc_get_formal_from_actual_arglist): Set declared_at for
artificial symbol.
* invoke.texi: Extend description of -fallow-argument-mismatch.
Delete -Wargument-mismatch.
* lang.opt: Change -Wargument-mismatch to do-nothing option.
* resolve.c (resolve_structure_cons): Change warnings that
previously depended on -Wargument-mismatch to unconditional.
* trans-decl.c (generate_local_decl): Do not warn if the symbol is
artificial.
2019-09-14 Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/91557
PR fortran/91556
* gfortran.dg/argument_checking_20.f90: New test.
* gfortran.dg/argument_checking_21.f90: New test.
* gfortran.dg/argument_checking_22.f90: New test.
* gfortran.dg/argument_checking_23.f90: New test.
* gfortran.dg/warn_unused_dummy_argument_5.f90: New test.
* gfortran.dg/bessel_3.f90: Add pattern for type mismatch.
* gfortran.dg/g77/20010519-1.f: Adjust dg-warning messages to new
handling.
* gfortran.dg/pr24823.f: Likewise.
* gfortran.dg/pr39937.f: Likewise.
Kewen Lin [Sat, 14 Sep 2019 09:01:21 +0000 (09:01 +0000)]
Consider doloop IV in IVOPTs
For the targets supporting low-overhead loops, IVOPTs has to take care of
the loops which will probably be transformed in RTL doloop optimization,
to try to make selected IV candidate set optimal. The process of doloop
support includes:
1) Analyze the current loop will be transformed to doloop or not, find and
mark its compare type IV use as doloop use (iv_group field doloop_p), and
set flag doloop_use_p of ivopts_data to notify subsequent processings on
doloop. See analyze_and_mark_doloop_use and its callees for the details.
The target hook predict_doloop_p can be used for target specific checks.
2) Add one doloop dedicated IV cand {(may_be_zero ? 1 : (niter + 1)), +, -1},
set flag doloop_p of iv_cand, step cost is set as zero and no extra cost
like biv. For cost determination between doloop IV cand and IV use, the
target hooks doloop_cost_for_generic and doloop_cost_for_address are
provided to add on extra costs for generic type and address type IV use.
Zero cost is assigned to the pair between doloop IV cand and doloop IV
use, and bound zero is set for IV elimination.
3) With the cost setting in step 2), the current cost model based IV
selection algorithm will process as usual, pick up doloop dedicated IV if
profitable.
gcc/ChangeLog
2019-09-14 Kewen Lin <linkw@gcc.gnu.org>
PR middle-end/80791
* config/rs6000/rs6000.c (TARGET_HAVE_COUNT_REG_DECR_P): New macro.
(TARGET_DOLOOP_COST_FOR_GENERIC): Likewise.
(TARGET_DOLOOP_COST_FOR_ADDRESS): Likewise.
* target.def (have_count_reg_decr_p): New hook.
(doloop_cost_for_generic): Likewise.
(doloop_cost_for_address): Likewise.
* doc/tm.texi.in (TARGET_HAVE_COUNT_REG_DECR_P): Likewise.
(TARGET_DOLOOP_COST_FOR_GENERIC): Likewise.
(TARGET_DOLOOP_COST_FOR_ADDRESS): Likewise.
* doc/tm.texi: Regenerate.
* tree-ssa-loop-ivopts.c (comp_cost::operator+=): Consider infinite cost
addend.
(record_group): Init doloop_p.
(add_candidate_1): Add optional argument doloop, change the handlings
accordingly.
(add_candidate): Likewise.
(generic_predict_doloop_p): Update attribute.
(force_expr_to_var_cost): Add costing for expressions COND_EXPR/LT_EXPR/
LE_EXPR/GT_EXPR/GE_EXPR/EQ_EXPR/NE_EXPR/UNORDERED_EXPR/ORDERED_EXPR/
UNLT_EXPR/UNLE_EXPR/UNGT_EXPR/UNGE_EXPR/UNEQ_EXPR/LTGT_EXPR/MAX_EXPR/
MIN_EXPR.
(get_computation_cost): Update for doloop IV cand extra cost.
(determine_group_iv_cost_cond): Update for doloop IV cand.
(determine_iv_cost): Likewise.
(ivopts_estimate_reg_pressure): Likewise.
(may_eliminate_iv): Update handlings for doloop IV cand.
(add_iv_candidate_for_doloop): New function.
(find_iv_candidates): Call function add_iv_candidate_for_doloop.
(iv_ca_set_no_cp): Update for doloop IV cand.
(iv_ca_set_cp): Likewise.
(iv_ca_dump): Dump register cost.
(find_doloop_use): New function.
(analyze_and_mark_doloop_use): Likewise.
(tree_ssa_iv_optimize_loop): Call function analyze_and_mark_doloop_use.
gnatmake: Accept the `--sysroot=' GCC driver option
According to `gnatmake' documentation:
"Any uppercase or multi-character switch that is not a 'gnatmake' switch
is passed to 'gcc' (e.g., '-O', '-gnato,' etc.)"
however the `--sysroot=' switch is actually rejected:
gnatmake: invalid switch: --sysroot=...
likely because it is one of the very few GCC driver options that have a
leading double dash and therefore we don't have a blanket fall-through
for such switches that would satisfy what our documentation claims.
The option is actually shared between the compiler and the linker, so
pass the switch to both build stages if requested, removing GNAT
testsuite issues like:
gnatmake: invalid switch: --sysroot=.../sysroot
compiler exited with status 1
Executing on host: .../gcc/gnatclean -c -q -n ./abstract_with_anonymous_result (timeout = 300)
spawn -ignore SIGHUP .../gcc/gnatclean -c -q -n ./abstract_with_anonymous_result
PASS: gnat.dg/abstract_with_anonymous_result.adb (test for excess errors)
UNRESOLVED: gnat.dg/abstract_with_anonymous_result.adb compilation failed to produce executable
in a test environment where `--with-build-sysroot=.../sysroot' has been
used to build a cross-compiler. Passing to the compilation stage only
would lead to errors like:
.../bin/riscv64-linux-gnu-ld: cannot find crt1.o: No such file or directory
.../bin/riscv64-linux-gnu-ld: cannot find -lc
collect2: error: ld returned 1 exit status
gnatlink: error when calling .../gcc/xgcc
gnatmake: *** link failed.
compiler exited with status 1
Executing on host: .../gcc/gnatclean -c -q -n ./abstract_with_anonymous_result (timeout = 300)
spawn -ignore SIGHUP .../gcc/gnatclean -c -q -n ./abstract_with_anonymous_result
./abstract_with_anonymous_result.ali
./abstract_with_anonymous_result.o
FAIL: gnat.dg/abstract_with_anonymous_result.adb (test for excess errors)
Excess errors:
.../bin/riscv64-linux-gnu-ld: cannot find crt1.o: No such file or directory
.../bin/riscv64-linux-gnu-ld: cannot find -lc
gnatlink: error when calling .../gcc/xgcc
UNRESOLVED: gnat.dg/abstract_with_anonymous_result.adb compilation failed to produce executable
instead.
gcc/ada/
* make.adb (Scan_Make_Arg): Also accept `--sysroot=' for the
compiler and the linker.
Jonathan Wakely [Thu, 12 Sep 2019 10:51:39 +0000 (11:51 +0100)]
PR libstdc++/91748 fix std::for_each_n for random access iterators
PR libstdc++/91748
* include/bits/stl_algo.h (for_each_n): Fix random access iterator
case.
* testsuite/25_algorithms/for_each/for_each_n.cc: Test with random
access iterators.
Ian Lance Taylor [Wed, 11 Sep 2019 14:25:52 +0000 (14:25 +0000)]
compiler: suppress type descriptor generation for aliases
Change Named_object::get_backend to ignore aliases when creating type
descriptors for types, to be consistent with
Type::needs_specific_type_functions and the Specific_type_functions
traversal class. For example, when compiling a package that creates an
alias to an an externally defined type, e.g.
import "foo"
type MyFoo = foo.Foo
it makes sense to skip the alias (not try to generate type specific
functions for it) and let the normal mechanisms take care of the alias
target, depending on whether the target is defined locally or defined
elsewhere.
Testcase for this problen can be found in CL 193261.
Ian Lance Taylor [Wed, 11 Sep 2019 12:44:12 +0000 (12:44 +0000)]
gotest: force test package to be imported first
When compiling the x_test package, force the test package to be
imported first. That ensures that we will see the types defined in
the test package before the types defined in the non-test version of
the package. This matters if the types differ in some way, such as by
adding a new method.
This avoids a failure in internal/poll on Solaris, in which the test
package adds a method to a type (FD.EOFError). I think it was Solaris-
specific because files are sorted in a different order by default.
The go tool handles this kind of thing correctly, by rebuilding
dependent packages. This is just a hack sufficient to run the libgo
testsuite without using the go tool.
Jonathan Wakely [Wed, 11 Sep 2019 11:38:15 +0000 (12:38 +0100)]
Fix libstdc++ tests for -Wvolatile warnings in C++2a mode
* testsuite/20_util/result_of/sfinae_friendly_1.cc: Add -Wno-volatile
for C++2a and up. Define HAS_52748_FIXED and fix incorrect tests.
* testsuite/tr1/3_function_objects/result_of.cc: Add -Wno-volatile
for C++2a and up.
Eric Botcazou [Wed, 11 Sep 2019 10:41:27 +0000 (10:41 +0000)]
re PR rtl-optimization/89795 (wrong code with -O2 -fno-dce -fno-forward-propagate -fno-sched-pressure)
PR rtl-optimization/89795
* rtlanal.c (nonzero_bits1) <SUBREG>: Do not propagate results from
inner REGs to paradoxical SUBREGs if WORD_REGISTER_OPERATIONS is set.
Jakub Jelinek [Wed, 11 Sep 2019 08:34:41 +0000 (10:34 +0200)]
re PR tree-optimization/91723 (builtin fma is not optimized or vectorized as *+)
PR tree-optimization/91723
* tree-vect-stmts.c (vectorizable_call): Use types_compatible_p check
instead of pointer equality when checking if argument vectypes are
the same.
Marek Polacek [Tue, 10 Sep 2019 23:22:37 +0000 (23:22 +0000)]
PR c++/91673 - ICE with noexcept in alias-declaration.
* parser.c (CP_PARSER_FLAGS_DELAY_NOEXCEPT): New parser flag.
(cp_parser_lambda_declarator_opt): Pass CP_PARSER_FLAGS_NONE to
cp_parser_exception_specification_opt.
(cp_parser_direct_declarator): Adjust a call to
cp_parser_exception_specification_opt.
(cp_parser_member_declaration): Pass CP_PARSER_FLAGS_DELAY_NOEXCEPT
to cp_parser_declarator if not processing a friend or typedef
declaration.
(cp_parser_late_noexcept_specifier): Adjust a call to
cp_parser_noexcept_specification_opt.
(cp_parser_noexcept_specification_opt): New parameter for parser flags,
drop the FRIEND_P parameter. Use the new parameter.
(cp_parser_exception_specification_opt): Likewise.
(cp_parser_transaction): Adjust a call to
cp_parser_noexcept_specification_opt.
(cp_parser_transaction_expression): Likewise.
* g++.dg/cpp1z/using7.C: New test.
* g++.dg/cpp1z/using8.C: New test.
Ian Lance Taylor [Tue, 10 Sep 2019 20:25:32 +0000 (20:25 +0000)]
go/internal/gccgoimporter: support embedded field in pointer loop
Backport of https://golang.org/cl/194440. Original description:
If an embedded field refers to a type via a pointer, the parser needs
to know the name of the embedded field. It is possible that the
pointer type is not yet resolved. This CL fixes the parser to handle
that case by setting the pointer element type to the unresolved named
type while the pointer is being resolved.
After previous patches, it's now possible for call_used_regs to be
the "real" set of call-clobbered registers, without any special
handling for fixed registers. This patch therefore removes the
separate call_really_used_regs and updates the targets that define
CALL_REALLY_USED_REGISTERS so that they handle call_used_regs in
the same way that they used to handle call_really_used_regs.
With this change, it's no longer necessary for targets that define
CALL_REALLY_USED_REGISTERS to define CALL_USED_REGISTER as well.
2019-09-10 Richard Sandiford <richard.sandiford@arm.com>
gcc/
* doc/tm.texi.in: Document that exactly one of CALL_USED_REGISTERS
and CALL_REALLY_USED_REGISTERS must be defined, and that
CALL_REALLY_USED_REGISTERS is preferred.
* doc/tm.texi: Regenerate.
* hard-reg-set.h (target_hard_regs::x_call_really_used_regs): Delete.
(call_really_used_regs): Likewise.
* reginfo.c: Raise an #error if both CALL_USED_REGISTERS and
CALL_REALLY_USED_REGISTERS are defined.
(initial_call_used_regs): Use CALL_REALLY_USED_REGISTERS as the
initial value if defined.
(initial_call_really_used_regs): Delete.
(saved_call_really_used_regs): Likewise.
(CALL_REALLY_USED_REGNO_P): Likewise.
(init_reg_sets): Remove handling of call_really_used_regs.
(save_register_info, restore_register_info, globalize_reg): Likewise.
(init_reg_sets_1): Likewise. Use call_used_regs instead of
CALL_REALLY_USED_REGNO_P. Don't set call_used_regs for registers
outside operand_reg_set.
(fix_register): Don't change call_used_regs if
CALL_REALLY_USED_REGISTERS is defined.
* config/csky/csky.h (CALL_USED_REGISTERS): Delete.
* config/csky/csky.c (get_csky_live_regs): Use call_used_regs
instead of call_really_used_regs.
(csky_conditional_register_usage): Remove the old handling of
call_used_regs and change the handling of call_really_used_regs
to use call_used_regs instead.
* config/ia64/ia64.h (CALL_USED_REGISTERS): Delete.
* config/ia64/ia64.c (fix_range): Don't set call_used_regs when
making a register fixed.
* config/m32r/m32r.h (CALL_USED_REGISTERS): Delete.
* config/m32r/m32r.c (MUST_SAVE_REGISTER): Use call_used_regs
instead of call_really_used_regs.
(m32r_conditional_register_usage): Don't set call_used_regs when
making a register fixed.
* config/mips/mips.h (CALL_USED_REGISTERS): Delete.
* config/mips/mips.c (mips_global_pointer): Use call_used_regs
instead of call_really_used_regs.
(mips_interrupt_extra_call_saved_reg_p): Likewise.
(mips_cfun_call_saved_reg_p): Likewise.
(mips_swap_registers): Remove the old handling of call_used_regs
and change the handling of call_really_used_regs to use call_used_regs
instead.
(mips_conditional_register_usage): Likewise.
* config/mn10300/mn10300.h (CALL_USED_REGISTERS): Delete.
* config/mn10300/mn10300.c (fp_regs_to_save): Use call_used_regs
instead of call_really_used_regs.
(mn10300_get_live_callee_saved_regs): Likewise.
(mn10300_expand_prologue, mn10300_expand_epilogue): Likewise.
(mn10300_conditional_register_usage): Don't set call_used_regs when
making a register fixed.
* config/rs6000/rs6000.h (CALL_USED_REGISTERS): Delete.
* config/rs6000/rs6000.c (rs6000_conditional_register_usage):
Remove the old handling of call_used_regs and change the handling
of call_really_used_regs to use call_used_regs instead.
* config/s390/s390.h (CALL_USED_REGISTERS): Delete.
* config/s390/s390.c (s390_regs_ever_clobbered): Use call_used_regs
instead of call_really_used_regs.
(s390_register_info_gprtofpr, s390_register_info): Likewise.
(s390_hard_regno_rename_ok, s390_hard_regno_scratch_ok): Likewise.
(s390_emit_prologue, s300_set_up_by_prologue): Likewise.
(s390_can_use_return_insn, s390_optimize_prologue): Likewise.
(s390_conditional_register_usage): Remove the old handling of
call_used_regs and change the handling of call_really_used_regs
to use call_used_regs instead.
* config/sh/sh.h (CALL_USED_REGISTERS): Delete.
* config/sh/sh.c (output_stack_adjust, calc_live_regs): Likewise.
(sh_fix_range, reg_unused_after): Likewise.
(sh_conditional_register_usage): Remove the old handling of
call_used_regs and change the handling of call_really_used_regs
to use call_used_regs instead.
* config/sparc/sparc.h (CALL_USED_REGISTERS): Delete.
* config/sparc/sparc.c (sparc_conditional_register_usage): Don't set
call_used_regs when making a register fixed.
* config/tilegx/tilegx.h (CALL_USED_REGISTERS): Delete.
* config/tilegx/tilegx.c (tilegx_conditional_register_usage): Don't set
call_used_regs when making a register fixed.
* config/tilepro/tilepro.h (CALL_USED_REGISTERS): Delete.
* config/tilepro/tilepro.c (tilepro_conditional_register_usage): Don't
set call_used_regs when making a register fixed.
* config/visium/visium.h (CALL_USED_REGISTERS): Delete.
* config/visium/visium.c (visium_conditional_register_usage): Remove
the old handling of call_used_regs and change the handling of
call_really_used_regs to use call_used_regs instead.
Now that tests of call_used_regs go through call_used_or_fixed_reg_p,
we can hide call_used_regs from target-independent code. (It still
needs to be available to targets for the conditional register usage
hooks.)
2019-09-10 Richard Sandiford <richard.sandiford@arm.com>
gcc/
* hard-reg-set.h (call_used_regs): Only define if IN_TARGET_CODE.
(call_used_or_fixed_reg_p): Expand definition of call_used_regs.
* reginfo.c (call_used_regs): New macro.
projects / thirdparty / gcc.git / log
