* einfo.adb Anonymous_Master now uses Node35.
(Anonymous_Master): Update the assertion and node reference.
(Set_Anonymous_Master): Update the assertion and node reference.
(Write_Field35_Name): Add output for Anonymous_Master.
(Write_Field36_Name): The output is now undefined.
* einfo.ads Update the node and description of attribute
Anonymous_Master. Remove prior occurrences in entities as this
is now a type attribute.
* exp_ch3.adb (Expand_Freeze_Array_Type): Remove local variable
Ins_Node. Anonymous access- to-controlled component types no
longer need finalization masters. The master is now built when
a related allocator is expanded.
(Expand_Freeze_Record_Type): Remove local variable Has_AACC. Do not
detect whether the record type has at least one component of anonymous
access-to- controlled type. These types no longer need finalization
masters. The master is now built when a related allocator is expanded.
* exp_ch4.adb Remove with and use clauses for Lib and Sem_Ch8.
(Current_Anonymous_Master): Removed.
(Expand_N_Allocator): Call Build_Anonymous_Master to create a
finalization master for an anonymous access-to-controlled type.
* exp_ch6.adb (Add_Finalization_Master_Actual_To_Build_In_Place_Call):
Call routine Build_Anonymous_Master to create a finalization master
for an anonymous access-to-controlled type.
* exp_ch7.adb (Allows_Finalization_Master): New routine.
(Build_Anonymous_Master): New routine.
(Build_Finalization_Master): Remove formal parameter
For_Anonymous. Use Allows_Finalization_Master to determine whether
circumstances warrant a finalization master. This routine no
longer creates masters for anonymous access-to-controlled types.
(In_Deallocation_Instance): Removed.
* exp_ch7.ads (Build_Anonymous_Master): New routine.
(Build_Finalization_Master): Remove formal parameter For_Anonymous
and update the comment on usage.
* sem_util.adb (Get_Qualified_Name): New routines.
(Output_Name): Reimplemented.
(Output_Scope): Removed.
* sem_util.ads (Get_Qualified_Name): New routines.
* debug.adb: Document the use of switch -gnatd.H.
* gnat1drv.adb (Adjust_Global_Switches): Set ASIS_GNSA mode when
-gnatd.H is present.
(Gnat1drv): Suppress the call to gigi when ASIS_GNSA mode is active.
* opt.ads: Add new option ASIS_GNSA_Mode.
* sem_ch13.adb (Alignment_Error): New routine.
(Analyze_Attribute_Definition_Clause): Suppress certain errors in
ASIS mode for attribute clause Alignment, Machine_Radix, Size, and
Stream_Size.
(Check_Size): Use routine Size_Too_Small_Error to
suppress certain errors in ASIS mode.
(Get_Alignment_Value): Use routine Alignment_Error to suppress certain
errors in ASIS mode.
(Size_Too_Small_Error): New routine.
Arnaud Charlet [Mon, 2 May 2016 10:00:00 +0000 (10:00 +0000)]
2016-05-02 Arnaud Charlet <charlet@adacore.com>
* spark_xrefs.ads Description of the spark cross-references
clarified; small style fixes.
* lib-xref-spark_specific.adb (Add_SPARK_Scope,
Detect_And_Add_SPARK_Scope): consider protected types and bodies
as yet another scopes.
(Enclosing_Subprogram_Or_Library_Package): refactored using
Hristian's suggestions; added support for scopes of protected
types and bodies; fix for entries to return the scope of the
enclosing concurrent type, which is consistent with what is
returned for protected subprograms.
* sem_intr.adb: Minor style fix in comment.
* lib-xref.ads, lib-xref-spark_specific.adb, get_spark_xrefs.adb,
put_spark_xrefs.adb: Minor reformatting.
2016-05-02 Doug Rupp <rupp@adacore.com>
* g-traceb.ads: Document traceback for ARM.
2016-05-02 Javier Miranda <miranda@adacore.com>
* exp_disp.adb (Make_Tags): Do not generate the
external name of interface tags adding the suffix counter since
it causes problems at link time when the IP routines are inlined
across units with optimization.
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* einfo.ads, einfo.adb (Predicates_Ignared): new flag to indicate
that predicate checking is disabled for predicated subtypes in
the context of an Assertion_Policy pragma.
* checks.adb (Apply_Predicate_Check): Do nothing if
Predicates_Ignored is true.
* exp_ch3.adb (Expand_Freeze_Enumeration_Type): If
Predicates_Ignores is true, the function Rep_To_Pos does raise
an exception for invalid data.
* exp_ch4.adb (Expand_N_Type_Conversion): IF target is a predicated
type do not apply check if Predicates_Ignored is true.
* exp_ch5.adb (Expand_N_Case_Statement): If Predicates_Ignored
is true, sem_prag.adb:
* sem_ch3.adb (Analyze_Object_Declaration): If Predicates_Ignored
is true do not emit predicate check on initializing expression.
Arnaud Charlet [Mon, 2 May 2016 09:53:16 +0000 (09:53 +0000)]
get_spark_xrefs.adb (Get_Nat, Get_Name): Initialize variables when they are declared...
2016-05-02 Arnaud Charlet <charlet@adacore.com>
* get_spark_xrefs.adb (Get_Nat, Get_Name): Initialize variables when
they are declared; refine type of a counter from Integer to Natural.
* sem_ch5.adb, gnatcmd.adb, s-intman-posix.adb, eval_fat.adb,
prj.adb, sem_util.adb, s-intman-android.adb, prj-nmsc.adb, sem_ch8.adb,
exp_ch3.adb: Minor editing.
Yannick Moy [Mon, 2 May 2016 09:50:45 +0000 (09:50 +0000)]
a-tigeli.adb (Get_Line): Always set Last prior to returning.
2016-05-02 Yannick Moy <moy@adacore.com>
* a-tigeli.adb (Get_Line): Always set Last prior to returning.
2016-05-02 Yannick Moy <moy@adacore.com>
* lib-xref.adb: Minor style fix in whitespace of declarations.
* put_spark_xrefs.adb (Put_SPARK_Xrefs): printing of strings
refactored without loops.
* put_spark_xrefs.ads (Write_Info_Str): new formal argument of
generic procedure.
* spark_xrefs.adb (Write_Info_Str): new actual in instantiation
of generic procedure.
Uros Bizjak [Mon, 2 May 2016 09:49:39 +0000 (11:49 +0200)]
predicates.md (nonimm_ssenomem_operand): New predicate.
* config/i386/predicates.md (nonimm_ssenomem_operand): New predicate.
(register_mixssei387nonimm_operand): Remove predicate.
* config/i386/i386.md (*fop_<mode>_comm): Merge from
*fop_<mode>_comm_mixed and *fop_<mode>_comm_i387. Disable unsupported
alternatives using "enabled" attribute. Also check X87_ENABLE_ARITH
for TARGET_MIX_SSE_I387 alternatives.
(*fop_<mode>_1): Merge from *fop_<mode>_1_mixed and *fop_<mode>_1_i387.
Disable unsupported alternatives using "enabled" attribute. Use
nonimm_ssenomem_operand as operand 1 predicate. Also check
X87_ENABLE_ARITH for TARGET_MIX_SSE_I387 alternatives.
* config/i386/predicates.md (nonimm_ssenomem_operand): New predicate.
(register_mixssei387nonimm_operand): Remove predicate.
Arnaud Charlet [Mon, 2 May 2016 09:48:55 +0000 (09:48 +0000)]
lib-xref-spark_specific.adb (Add_SPARK_Scope): add task type scope.
2016-05-02 Arnaud Charlet <charlet@adacore.com>
* lib-xref-spark_specific.adb (Add_SPARK_Scope): add task type scope.
(Detect_And_Add_SPARK_Scope): detect and add task type scope.
(Enclosing_Subprogram_Or_Package): Respect boundaries of task
and entry declarations.
* spark_xrefs.ads: minor typo in comment.
Arnaud Charlet [Mon, 2 May 2016 09:47:42 +0000 (09:47 +0000)]
make.adb: Minor...
2016-05-02 Arnaud Charlet <charlet@adacore.com>
* make.adb: Minor: avoid an exception when calling gnatmake with
no argument and gnatmake is built with checks on.
* lib-xref-spark_specific.adb: Minor code cleanup.
Ed Schonberg [Mon, 2 May 2016 09:44:54 +0000 (09:44 +0000)]
sem_util.adb (Normalize_Actuals): Take into account extra actuals that may have been introduced previously.
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* sem_util.adb (Normalize_Actuals): Take into account extra
actuals that may have been introduced previously. Normally extra
actuals are introduced when a call is expanded, but a validity
check may copy and reanalyze a call that carries an extra actual
(e.g. an accessibility parameter) before the call itself is
marked Analzyed, and the analysis of the copy has to be able to
cope with the added actual.
While looking at the use of cst_and_fits_in_hwi in tree-ssa-loop-ivopts.c,
I had difficulty working out what the function actually tests. The
final NUNITS check seems redundant, since it asks about the number of
HWIs in the _unextended_ constant. We've already checked that the
unextended constant has no more than HOST_BITS_PER_WIDE_INT bits, so the
length must be 1.
As Richard says, we ought to have a convenient way of converting
an INTEGER_CST to a wide_int of a particular precision without
having to extract the sign of the INTEGER_CST's type each time.
This patch adds a wi::to_wide helper for that, alongside the
existing wi::to_offset and wi::to_widest.
Tested on x86_64-linux-gnu and aarch64-linux-gnu.
gcc/
* tree.h (wi::to_wide): New function.
* expr.c (expand_expr_real_1): Use wi::to_wide.
* fold-const.c (int_const_binop_1): Likewise.
(extract_muldiv_1): Likewise.
gcc/c-family/
* c-common.c (shorten_compare): Use wi::to_wide.
Following on from the comparison patch, I think it makes sense to
support << and >> for offset_int (int128_t) and widest_int (intNNN_t),
with >> being arithmetic shift. It doesn't make sense to use
logical right shift on a potentially negative offset_int, since
the precision of 128 bits has no meaning on the target.
Tested on x86_64-linux-gnu and aarch64-linux-gnu.
gcc/
* wide-int.h: Update offset_int and widest_int documentation.
(WI_SIGNED_SHIFT_RESULT): New macro.
(wi::binary_shift): Define signed_shift_result_type for
shifts on offset_int- and widest_int-like types.
(generic_wide_int): Support <<= and >>= if << and >> are supported.
* tree.h (int_bit_position): Use shift operators instead of wi::
shifts.
* alias.c (adjust_offset_for_component_ref): Likewise.
* expr.c (get_inner_reference): Likewise.
* fold-const.c (fold_comparison): Likewise.
* gimple-fold.c (fold_nonarray_ctor_reference): Likewise.
* gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
* tree-dfa.c (get_ref_base_and_extent): Likewise.
* tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
(stmt_kills_ref_p): Likewise.
* tree-ssa-ccp.c (bit_value_binop_1): Likewise.
* tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
* tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
(ao_ref_init_from_vn_reference): Likewise.
gcc/cp/
* init.c (build_new_1): Use shift operators instead of wi:: shifts.
Support <, <=, > and >= for offset_int and widest_int
offset_int and widest_int are supposed to be at least one bit wider
than all the values they need to represent, with the extra bits
being signs. Thus offset_int is effectively int128_t and widest_int
is effectively intNNN_t, for target-dependent NNN.
Because the types are signed, there's not really any need to specify
a sign for operations like comparison. I think things would be clearer
if we supported <, <=, > and >= for them (but not for wide_int, which
doesn't have a sign).
Tested on x86_64-linux-gnu and aarch64-linux-gnu.
gcc/
* wide-int.h: Update offset_int and widest_int documentation.
(WI_SIGNED_BINARY_PREDICATE_RESULT): New macro.
(wi::binary_traits): Allow ordered comparisons between offset_int and
offset_int, between widest_int and widest_int, and between either
of these types and basic C types.
(operator <, <=, >, >=): Define for the same combinations.
* tree.h (tree_int_cst_lt): Use comparison operators instead
of wi:: comparisons.
(tree_int_cst_le): Likewise.
* gimple-fold.c (fold_array_ctor_reference): Likewise.
(fold_nonarray_ctor_reference): Likewise.
* gimple-ssa-strength-reduction.c (record_increment): Likewise.
* tree-affine.c (aff_comb_cannot_overlap_p): Likewise.
* tree-parloops.c (try_transform_to_exit_first_loop_alt): Likewise.
* tree-sra.c (completely_scalarize): Likewise.
* tree-ssa-alias.c (stmt_kills_ref_p): Likewise.
* tree-ssa-reassoc.c (extract_bit_test_mask): Likewise.
* tree-vrp.c (extract_range_from_binary_expr_1): Likewise.
(check_for_binary_op_overflow): Likewise.
(search_for_addr_array): Likewise.
* ubsan.c (ubsan_expand_objsize_ifn): Likewise.
Arnaud Charlet [Mon, 2 May 2016 09:30:59 +0000 (11:30 +0200)]
[multiple changes]
2016-05-02 Bob Duff <duff@adacore.com>
* sem_ch10.adb (Analyze_Compilation_Unit): Preserve
treeishness. Previous version had Context_Items shared between
the spec and body.
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* sem_aggr.adb (Resolve_Aggr_Expression): For both array and
record cases, apply predicate check on component for expression
only if expression has been analyzed already. For expressions
that need to be duplicated when they cover multiple components,
resolution and predicate checking take place later.
2016-05-02 Olivier Hainque <hainque@adacore.com>
* a-direct.adb (Delete_Tree): Use full names to designate subdirs
and files therein, instead of local names after a change of
current directory.
* sem_ch4.adb (Find_Indexing_Operations): Use the underlying type
of the container base type in case the container is a subtype.
* sem_ch5.adb (Analyze_Iterator_Specification): Ensure that
the selector has an entity when checking for a component of a
mutable object.
2016-05-02 Arnaud Charlet <charlet@adacore.com>
Remove dead code.
* opt.ads (Latest_Ada_Only): New flag.
* sem_prag.adb, par-prag.adb: Ignore pragma Ada_xx under this flag.
* usage.adb, switch-c.adb: Disable support for -gnatxx under this flag.
* einfo.ads (Has_Predicates, Predicate_Function):
Clarify that Has_Predicates does not imply that Predicate_Function
will return a non-empty entity.
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* sem_res.adb (Resolve_Qualified_Expression): Generate a predicate
check if type requires it.
* checks.adb (Apply_Predicate_Check): Disable checks in the
object declaration created for an expression with side-effects
that requires a predicate check to prevent infinite recursion
during expansion.
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* sem_ch6.adb (Process_Formals): Check properly the type of a
formal to determine whether a given convention applies to it.
2016-05-02 Doug Rupp <rupp@adacore.com>
* tracebak.c: Add incantations for arm-vxworks[67] traceback.
2016-05-02 Thomas Quinot <quinot@adacore.com>
* freeze.adb (Check_Component_Storage_Order): Make it a warning, not an
error, to have a component with implicit SSO within a composite type
that has explicit SSO.
Arnaud Charlet [Mon, 2 May 2016 09:08:44 +0000 (11:08 +0200)]
[multiple changes]
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* sem_ch4.adb (Analyze_Allocator): If the expression does not
have a subtype indication and the type is an unconstrained tagged
type with defaulted discriminants, create an explicit constraint
for it during analysis to prevent out-of-order freezing actions
on generated classwide types.
2016-05-02 Javier Miranda <miranda@adacore.com>
* exp_ch5.adb (Expand_N_Assignment_Statement):
In the runtime check that ensures that the tags of source an
target match, add missing displacement of the pointer to the
objects if they cover interface types.
2016-05-02 Ed Schonberg <schonberg@adacore.com>
* sem_attr.adb (Analyze_Attribute, case 'Old): Do not use
base type for attribute when type is discrete: transformation
is not needed for such types, and leads to spurious errors if
the context is a case construct.
Eric Botcazou [Mon, 2 May 2016 07:46:05 +0000 (07:46 +0000)]
decl.c (elaborate_reference_1): Do not bother about operand #2 for COMPONENT_REF.
* gcc-interface/decl.c (elaborate_reference_1): Do not bother about
operand #2 for COMPONENT_REF.
* gcc-interface/utils2.c (gnat_save_expr): Likewise.
(gnat_protect_expr): Likewise.
(gnat_stabilize_reference_1): Likewise.
(gnat_rewrite_reference): Do not bother about operand #3 for ARRAY_REF.
(get_inner_constant_reference): Likewise.
(gnat_invariant_expr): Likewise.
* gcc-interface/trans.c (fold_constant_decl_in_expr): Likewise.
Eric Botcazou [Mon, 2 May 2016 07:38:37 +0000 (07:38 +0000)]
trans.c (Range_to_gnu): New static function.
* gcc-interface/trans.c (Range_to_gnu): New static function.
(Raise_Error_to_gnu) <N_In>: Call it to translate the range.
(gnat_to_gnu) <N_In>: Likewise.
Oleg Endo [Mon, 2 May 2016 05:25:46 +0000 (05:25 +0000)]
re PR target/52898 (SH Target: Inefficient DImode comparisons)
gcc/
PR target/52898
* config/sh/sh.c (sh_option_override): Remove TARGET_CBRANCHDI4,
TARGET_CMPEQDI_T.
(prepare_cbranch_operands): Don't use scratch register. Assume that
function is used when pseudos can be created.
(expand_cbranchdi4): Likewise. Remove unused TARGET_CMPEQDI_T paths.
* config/sh/sh.md (cbranchsi4): Allow only when pseudos can be created.
(cbranchdi4, cbranchdi4_i): Simplify to single cbranchdi4
define_expand. Allow it only when pseudos can be created.
* config/sh/sh.opt (mcbranchdi, mcmpeqdi): Delete.
Eric Botcazou [Sun, 1 May 2016 07:32:01 +0000 (07:32 +0000)]
rs6000.c (altivec_expand_lv_builtin): Do not use switch statement on instruction code.
* config/rs6000/rs6000.c (altivec_expand_lv_builtin): Do not use switch
statement on instruction code. Remove trailing spaces.
(altivec_expand_stv_builtin): Likewise.
Eric Botcazou [Sat, 30 Apr 2016 10:42:14 +0000 (10:42 +0000)]
Make-lang.in (ACATSCMD): New variable.
ada/
* gcc-interface/Make-lang.in (ACATSCMD): New variable.
(check-acats): Use it.
(check_acats_targets): Likewise.
testsuite/
* ada/acats/run_acats: Rename into...
* ada/acats/run_acats.sh: ...this. Only export BASE variable.
* ada/acats/run_all.sh: Remove redundant test.
(target_run): Move around.
(target_gnatchop): Use newly built executable.
(target_gnatmake): Likewise.
Check that the compilation of impbit succeeds.
Alan Modra [Sat, 30 Apr 2016 00:35:39 +0000 (10:05 +0930)]
[RS6000] Allow saving of fixed regs.
As I noted a long time ago in the comment on fixed_reg_p, the real
problem with saving fixed/global regs is that exception frame
unwinding might restore them. So don't emit eh_frame info for any
such reg, and the unwinder won't restore them.
Also, tidy rs6000_savres_strategy. Delaying some checks means we
won't iterate over regs quite so often.
* config/rs6000/rs6000.c (rs6000_savres_strategy): Force inline
restoring when fixed_reg_p, but allow out-of-line or stmw save.
Check for user regs later to avoid unnecessary looping over regs.
Merge user reg check with non-saved reg check. Don't force
inline VR restore when static chain used.
(rs6000_frame_related): Omit eh_frame info for user regs when
saving.
(fixed_regs_p): Delete.
Alan Modra [Sat, 30 Apr 2016 00:01:07 +0000 (09:31 +0930)]
Goodbye REG_LIVE_LENGTH
* regs.h (struct reg_info_t): Delete live_length.
(REG_LIVE_LENGTH): Delete macro.
* regstat.c (regstat_bb_compute_ri): Delete artificial_uses,
local_live, local_processed and local_live_last_luid params.
Replace bb_index param with bb. Don't set REG_LIVE_LENGTH.
Formatting fixes.
(regstat_compute_ri): Adjust for above. Don't set
REG_LIVE_LENGTH.
(dump_reg_info): Don't print live length.
* ira.c (update_equiv_regs): Replace test of REG_LIVE_LENGTH
with test of setjmp_crosses. Don't set REG_LIVE_LENGTH.
Localize loop_depth var.
Paolo Carlini [Sat, 30 Apr 2016 00:00:51 +0000 (00:00 +0000)]
re PR c++/66644 (Rejects C++11 in-class anonymous union members initialization)
/cp
2016-04-29 Paolo Carlini <paolo.carlini@oracle.com>
PR c++/66644
* class.c (check_field_decl): Remove final int* parameter, change
the return type to bool; fix logic in order not to reject multiple
initialized fields in anonymous struct.
(check_field_decls): Adjust call.
/testsuite
2016-04-29 Paolo Carlini <paolo.carlini@oracle.com>
Alan Modra [Sat, 30 Apr 2016 00:00:22 +0000 (09:30 +0930)]
ira.c validate_equiv_mem
This function is used to validate REG_EQUIV notes generated by ira,
and to validate potential insn combines performed by ira. The two
conditions are not exactly the same, with reload being more
restrictive. Separate them so more combines/moves can occur.
For example, this sequence from cfgexpand.c:expand_gimple_cond
callq _Z18update_bb_for_insnP15basic_block_def
mov 0x10(%rbx),%rdi
mov 0x0(%rip),%rbp # x_rtl+0x34
callq _Z9safe_as_aIP8rtx_insn7rtx_defET_PT0_
mov %r13,%rdx
mov %rbp,%rsi
mov %rax,%rdi
callq _Z18create_basic_blockP7rtx_defS0_P15basic_block_def
Alan Modra [Fri, 29 Apr 2016 23:59:22 +0000 (09:29 +0930)]
ira.c use DF infrastructure for combine_and_move_insns
This patch actually improves generated code, because REG_DEAD notes
used by the old insn scan are not always present. On x86_64, see
gcc/wide-int-print.o:print_hex for an example of a function that is
smaller and uses one less callee saved reg.
* ira.c (combine_and_move_insns): Rather than scanning insns,
use DF infrastucture to find use and def insns.
Alan Modra [Fri, 29 Apr 2016 23:58:17 +0000 (09:28 +0930)]
ira.c combine_and_move_insns, and ordering of functions
Notes added by add_store_equivs are not used directly or indirectly by
combine_and_move_insns. add_store_equivs can therefore run later
without affecting the output of combine_and_move_insns, and thus
add_store_equivs need not take into account potentially moved insns.
Since not all potentially combined/moved insns are in fact combined or
moved, this may allow add_store_equivs to add more REG_EQUIV notes.
grow_reg_equivs isn't needed until the reload reg_equivs array is
changed.
ira.c (combine_and_move_insns): Move invariant conditions..
(ira.c): ..to here. Call combine_and_move_insns before
add_store_equivs. Call grow_reg_equivs later. Allocate
req_equiv later using max_reg_num() rather than global max_regno.
(contains_replace_regs): Delete.
(add_store_equivs): Remove contains_replace_regs test.
Alan Modra [Fri, 29 Apr 2016 23:57:00 +0000 (09:27 +0930)]
ira.c tidies: split update_reg_equivs
* ira.c (add_store_equivs, combine_and_move_insns): New functions,
split out from..
(update_reg_equivs): ..here. Move allocation and freeing of
reg_equiv, and calls to grow_reg_equivs, init_alias_analysis,
end_alias_analysis to..
(ira): ..here.
Patrick Palka [Fri, 29 Apr 2016 19:15:25 +0000 (19:15 +0000)]
tree-ssa-threadedge.c (simplify_control_stmt_condition): Split out into ...
2016-04-29 Patrick Palka <ppalka@gcc.gnu.org>
* tree-ssa-threadedge.c (simplify_control_stmt_condition): Split
out into ...
(simplify_control_stmt_condition_1): ... here. Recurse into
BIT_AND_EXPRs and BIT_IOR_EXPRs.
* gcc.dg/tree-ssa/ssa-thread-14.c: New test.
* gcc.dg/tree-ssa/ssa-thread-11.c: Update expected output.
re PR middle-end/70626 (bogus results in 'acc parallel loop' reductions)
gcc/c-family/
PR middle-end/70626
* c-common.h (c_oacc_split_loop_clauses): Add boolean argument.
* c-omp.c (c_oacc_split_loop_clauses): Use it to duplicate
reduction clauses in acc parallel loops.
gcc/c/
PR middle-end/70626
* c-parser.c (c_parser_oacc_loop): Don't augment mask with
OACC_LOOP_CLAUSE_MASK.
(c_parser_oacc_kernels_parallel): Update call to
c_oacc_split_loop_clauses.
gcc/cp/
PR middle-end/70626
* parser.c (cp_parser_oacc_loop): Don't augment mask with
OACC_LOOP_CLAUSE_MASK.
(cp_parser_oacc_kernels_parallel): Update call to
c_oacc_split_loop_clauses.
gcc/fortran/
PR middle-end/70626
* trans-openmp.c (gfc_trans_oacc_combined_directive): Duplicate
the reduction clause in both parallel and loop directives.
gcc/testsuite/
PR middle-end/70626
* c-c++-common/goacc/combined-reduction.c: New test.
* gfortran.dg/goacc/reduction-2.f95: Add check for kernels reductions.
libgomp/
PR middle-end/70626
* testsuite/libgomp.oacc-c++/template-reduction.C: Adjust test.
* testsuite/libgomp.oacc-c-c++-common/combined-reduction.c: New test.
* testsuite/libgomp.oacc-fortran/combined-reduction.f90: New test.
tree-vect-loop.c (vect_transform_loop): Fix nb_iterations_upper_bound computation for vectorized loop.
gcc/
* tree-vect-loop.c (vect_transform_loop): Fix
nb_iterations_upper_bound computation for vectorized loop.
gcc/testsuite/
* gcc.target/i386/vect-unpack-2.c (avx512bw_test): Avoid
optimization of vector loop.
* gcc.target/i386/vect-unpack-3.c: New test.
* gcc.dg/vect/vect-nb-iter-ub-1.c: New test.
* gcc.dg/vect/vect-nb-iter-ub-2.c: New test.
* gcc.dg/vect/vect-nb-iter-ub-3.c: New test.
Andreas Krebbel [Fri, 29 Apr 2016 09:17:35 +0000 (09:17 +0000)]
S/390: Replace LDER with LDR.
For performance reasons it is important to write the full 64 bits of
an FPR target reg even when dealing with 32 bit values. So we chose
lder over ler for 32 bit float register moves. lder zero-extends the
32 bit value from the source reg to 64 bit in the target. However,
since it actually doesn't matter whether we write the upper 32 bits
with zeros or with any other garbage we can also use ldr instead. It
is bit shorter and therefore will do good for I-Cache usage.
gcc/ChangeLog:
2016-04-29 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
This fixes an issue with the long displacement memory address
constraints S and T. These were defined to only accept long
displacement addresses. This is wrong since a memory constraint must
not reject an address with a 0 displacement. Reload relies on being
able to turn an invalid memory address into a valid one by reloading
the address into a base register. The S and T constraints would
reject such an address.
This isn't really a problem for the backend since we used the
constraints with that knowledge there but it is a problem for people
writing inline assemblies.
gcc/ChangeLog:
2016-04-29 Ulrich Weigand <uweigand@de.ibm.com>
* config/s390/constraints.md ("U", "W"): Invoke
s390_mem_constraint with "ZR" and "ZT".
* config/s390/s390.c (s390_check_qrst_address): Reject invalid
addresses when using LRA. Accept also short displacements for S
and T constraints. Do not check for long displacement target for
S and T constraints.
(s390_mem_constraint): Remove handling of U and W constraints.
* config/s390/s390.md (various patterns): Remove the short
displacement constraints (Q and R) if a long displacement
constraint is present. Add longdisp as required CPU capability.
* config/s390/vector.md: Likewise.
* config/s390/vx-builtins.md: Likewise.
[ARC] Fix unwanted match for sign extend 16-bit constant.
The combine pass may conclude umulhisi3_imm pattern can accept also sign
extended 16-bit constants. This patch prohibits the combine in considering
this pattern as suitable.
Richard Biener [Fri, 29 Apr 2016 08:36:49 +0000 (08:36 +0000)]
re PR tree-optimization/13962 ([tree-ssa] make "fold" use alias information to optimize pointer comparisons)
2016-04-29 Richard Biener <rguenther@suse.de>
PR tree-optimization/13962
PR tree-optimization/65686
* tree-ssa-alias.h (ptrs_compare_unequal): Declare.
* tree-ssa-alias.c (ptrs_compare_unequal): New function
using PTA to compare pointers.
* match.pd: Add pattern for pointer equality compare simplification
using ptrs_compare_unequal.
i386.md (Load+RegOp to Mov+MemOp peephole2): Use SWI mode iterator.
* config/i386/i386.md (Load+RegOp to Mov+MemOp peephole2):
Use SWI mode iterator. Use general_reg_operand predicate.
(Load+RegOp to Mov+MemOp peephole2 with vector regs): Split
peephole to MMX and SSE part. Use mmx_reg_operand and sse_reg_operand
predicates.
projects / thirdparty / gcc.git / log
