The fold code also expanded cabs(x+yi) to fsqrt(x*x+y*y) when optimising
for speed. tree-ssa-math-opts.c has this transformation too, but unlike
the fold code, it first checks whether the target implements the sqrt
optab. The patch simply removes the fold code and keeps the
tree-ssa-math-opts.c logic the same.
gcc.dg/lto/20110201-1_0.c was relying on us replacing cabs
with fsqrt even on targets where fsqrt is itself a library call.
The discussion leading up to that patch suggested that we only
want to test the fold on targets with a square root instruction,
so it would be OK to skip the test on other targets:
The patch does that using the sqrt_insn effective target.
It's possible that removing the tree folds renders the LTO trick
unnecessary, but since the test was originally for an ICE, it seems
better to leave it as-is.
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi. 20110201-1_0.c passes on all three.
This patch adds a pass that collects information that is common to
all uses of an SSA name X and back-propagates that information up
the statements that generate X. The general idea is to use the
information to simplify instructions (rather than a pure DCE) so
I've simply called it gimple-ssa-backprop.c, to go with
tree-ssa-forwprop.c.
At the moment the only use of the pass is to remove unnecessary
sign operations, so that it's effectively a global version of
fold_strip_sign_ops. I'm hoping it could be extended in future
to record which bits of an integer are significant. There are
probably other potential uses too.
A later patch gets rid of fold_strip_sign_ops.
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
Aditya Kumar [Wed, 21 Oct 2015 18:53:35 +0000 (18:53 +0000)]
avoid generation of empty guards in trivial cases
We used to insert a trivial integer_onep condition around loops we know have
more than an iteration. This patch avoids generating those conditions.
* graphite-isl-ast-to-gimple.c (graphite_create_new_loop_guard): Do not call
create_empty_if_region_on_edge when cond_expr is true.
(translate_isl_ast_node_for): Check whether a guard has been generated.
Co-Authored-By: Sebastian Pop <s.pop@samsung.com>
From-SVN: r229137
Now that scop contains a list of all the basic blocks inside, it makes sense to
iterate over only those basic blocks in
graphite-sese-to-poly.c:rewrite_reductions_out_of_ssa,rewrite_cross_bb_scalar_deps_out_of_ssa
Passes regtest and bootstrap.
gcc/ChangeLog:
2015-10-20 Aditya Kumar <hiraditya@msn.com>
* graphite-poly.h (struct dr_info): Added invalid_alias_set number.
(operator=): Removed.
(dr_info): Make alias_set number the last argument with default value of invalid_alias_set.
* graphite-sese-to-poly.c (build_scop_drs): Update constructor of dr_info.
(rewrite_reductions_out_of_ssa): Iterate only through the basic blocks which are inside region.
(rewrite_cross_bb_scalar_deps_out_of_ssa): Same.
* sese.h (struct sese_l): Removed assignment operator.
(split_region_for_bb): Removed dead code.
Jan Hubicka [Wed, 21 Oct 2015 18:00:30 +0000 (20:00 +0200)]
re PR middle-end/67966 (ICE in convert_move, at expr.c:282)
PR middle-end/67966
* tree.c (verify_type): Verify that TYPE_MODE match
between TYPE_CANONICAL and type.
* expr.c (store_expr_with_bounds): Revert my previous change.
* expmed.c (store_bit_field_1): Revert prevoius change.
* gimple-expr.c (useless_type_conversion_p): Require TYPE_MODE
to match for all types.
Ilya Enkovich [Wed, 21 Oct 2015 16:01:43 +0000 (16:01 +0000)]
tm.texi: Regenerated.
gcc/
* doc/tm.texi: Regenerated.
* doc/tm.texi.in (TARGET_VECTORIZE_GET_MASK_MODE): New.
* stor-layout.c (layout_type): Use mode to get vector mask size.
* target.def (get_mask_mode): New.
* targhooks.c (default_get_mask_mode): New.
* targhooks.h (default_get_mask_mode): New.
* gcc/tree-vect-stmts.c (get_same_sized_vectype): Add special case
for boolean vector.
* tree.c (MAX_BOOL_CACHED_PREC): New.
(nonstandard_boolean_type_cache): New.
(build_nonstandard_boolean_type): New.
(make_vector_type): Vector mask has no canonical type.
(build_truth_vector_type): New.
(build_same_sized_truth_vector_type): New.
(truth_type_for): Support vector masks.
* tree.h (VECTOR_BOOLEAN_TYPE_P): New.
(build_truth_vector_type): New.
(build_same_sized_truth_vector_type): New.
(build_nonstandard_boolean_type): New.
* tree-cfg.c (verify_gimple_comparison) Require boolean
vector type for vector comparison.
(verify_gimple_assign_ternary): Likewise.
* optabs.c (expand_vec_cond_expr): Accept boolean vector as
condition operand.
* tree-vect-stmts.c (vectorizable_condition): Use boolean
vector type for vector comparison.
* tree-vect-generic.c (elem_op_func): Add new operand to hold
vector type.
(do_unop): Adjust to modified function type.
(do_binop): Likewise.
(do_plus_minus): Likewise.
(do_negate); Likewise.
(expand_vector_piecewise): Likewise.
(do_cond): Likewise.
(do_compare): Use comparison instead of condition.
(expand_vector_divmod): Use boolean vector type for comparison.
(expand_vector_operations_1): Skip scalar mask operations.
gcc/c
* c-typeck.c (build_conditional_expr): Use boolean vector
type for vector comparison.
(build_vec_cmp): New.
(build_binary_op): Use build_vec_cmp for comparison.
gcc/cp
* call.c (build_conditional_expr_1): Use boolean vector
type for vector comparison.
* typeck.c (build_vec_cmp): New.
(cp_build_binary_op): Use build_vec_cmp for comparison.
This patch makes sure that, for every simplification that uses
fold_strip_sign_ops, there are associated match.pd rules for the
leaf sign ops, i.e. abs, negate and copysign. A follow-on patch
will add a pass to handle more complex cases.
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
gcc/
* match.pd: Add rules to simplify ccos, ccosh, hypot, copysign
and x*x in cases where the operands are sign ops. Extend these
rules to handle copysign as a sign op (including for cos, cosh
and pow, which already treated negate and abs as sign ops).
Uros Bizjak [Wed, 21 Oct 2015 08:43:44 +0000 (10:43 +0200)]
re PR target/68018 (ICE: in ix86_compute_frame_layout, at config/i386/i386.c:11308 with -mstackrealign)
PR target/68018
* config/i386/i386.c (ix86_compute_frame_layout): Realign the stack
for 64-bit MS_ABI targets also when default incoming stack boundary
is overriden.
testsuite/ChangeLog:
PR target/68018
* gcc.target/i386/pr68018.c: New test.
Richard Biener [Wed, 21 Oct 2015 08:08:05 +0000 (08:08 +0000)]
re PR tree-optimization/68031 (cc1 crashes when compiling newlib / mktm_r.c)
2015-10-21 Richard Biener <rguenther@suse.de>
PR middle-end/68031
* fold-const.c: Include tree-ssa-operands.h and tree-into-ssa.h.
(tree_ssa_name_nonnegative_warnv_p): Fold into ...
(tree_single_nonnegative_warnv_p): ... here. For SSA names
make sure they are not registered for update.
Max Ostapenko [Wed, 21 Oct 2015 07:51:03 +0000 (10:51 +0300)]
sanitizer_stacktrace.cc (GetCanonicFrame): Assume we compiled code with GCC when extracting the caller PC for ARM if...
libsanitizer/
* sanitizer_common/sanitizer_stacktrace.cc (GetCanonicFrame): Assume we
compiled code with GCC when extracting the caller PC for ARM if no
valid frame pointer is available.
Max Ostapenko [Wed, 21 Oct 2015 07:44:35 +0000 (10:44 +0300)]
re PR sanitizer/63958 (bootstrap failure in the sanitizer libs on sparc-linux-gnu)
libsanitizer/
PR sanitizer/63958
Reapply:
2014-10-14 David S. Miller <davem@davemloft.net>
* sanitizer_common/sanitizer_platform_limits_linux.cc (time_t):
Define at __kernel_time_t, as needed for sparc.
(struct __old_kernel_stat): Don't check if __sparc__ is defined.
* libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.h
(__sanitizer): Define struct___old_kernel_stat_sz,
struct_kernel_stat_sz, and struct_kernel_stat64_sz for sparc.
(__sanitizer_ipc_perm): Adjust for sparc targets.
(__sanitizer_shmid_ds): Likewsie.
(__sanitizer_sigaction): Likewise.
(IOC_SIZE): Likewsie.
Max Ostapenko [Wed, 21 Oct 2015 07:40:54 +0000 (10:40 +0300)]
libsanitizer merge from upstream r250806, compiler part.
gcc/
* asan.c (asan_emit_stack_protection): Don't pass local stack to
asan_stack_malloc_[n] anymore. Check if asan_stack_malloc_[n] returned
NULL and use local stack than.
(asan_finish_file): Insert __asan_version_mismatch_check_v[n] call
in addition to __asan_init.
* sanitizer.def (BUILT_IN_ASAN_INIT): Rename to __asan_init.
(BUILT_IN_ASAN_VERSION_MISMATCH_CHECK): Add new builtin call.
* asan.h (asan_intercepted_p): Handle new string builtins.
* ubsan.c (ubsan_use_new_style_p): New function.
(ubsan_instrument_float_cast): If location is unknown, assign
input_location to loc. Propagate loc to ubsan_create_data if
ubsan_use_new_style_p returned true.
config/
* bootstrap-asan.mk: Replace ASAN_OPTIONS=detect_leaks with
LSAN_OPTIONS=detect_leaks.
Jeff Law [Wed, 21 Oct 2015 05:56:37 +0000 (23:56 -0600)]
[PATCH] Move RTL printing code from sched-vis.c into print-rtl.c
* Makefile.in (OBJS): Remove sched-vis.c
* sched-vis.c: Removed. Code moved into...
* print-rtl.c: Here. Include cfg.h, pretty-print.h and print-rtl.h.
* rtl.h: Remove prototypes for functions now living in print-rtl.c
* print-rtl.h Add prototypes for new functions in print-rtl.c.
* auto-inc-dec.c: Include print-rtl.h
* cfgrtl.c, combine.c, final.c haifa-sched.c: Likewise.
* ira.c, lra-constraints.c, lra.c, sel-sched-dump.c: Likewise.
Richard Biener [Wed, 21 Oct 2015 02:22:04 +0000 (02:22 +0000)]
fold-const.c (fold_binary_loc): Move (-A) * (-B) -> A * B to match.pd.
2015-10-20 Richard Biener <rguenther@suse.de>
Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
* fold-const.c (fold_binary_loc) : Move (-A) * (-B) -> A * B
to match.pd.
Move (a * (1 << b)) is (a << b) to match.pd.
Move convert (C1/X)*C2 into (C1*C2)/X to match.pd.
Move ~X & X, (X == 0) & X, and !X & X are zero to match.pd.
Move X & ~X , X & (X == 0), and X & !X are zero to match.pd.
* match.pd (mult:c @0 (convert? (lshift integer_onep@1 @2))):
New simplifier.
(mult (rdiv:s REAL_CST@0 @1) REAL_CST@2): New simplifier.
(bit_and:c (convert? @0) (convert? (bit_not @0))): New simplifier.
(bit_ior (bit_and:s @0 (bit_not:s @1)) (bit_and:s (bit_not:s @0) @1))
: New simplifier.
(mult:c (convert1? (negate @0)) (convert2? negate_expr_p@1)):
New simplifier.
(match (logical_inverted_value @0) (truth_not @0)) : New Predicate.
Ian Lance Taylor [Tue, 20 Oct 2015 20:44:23 +0000 (20:44 +0000)]
compiler: Report errors after evaluating invalid constant.
The fix for golang/go#11541 made the assertion that getting an invalid
Numeric_constant after evaluating an invalid constant must result in
an error in a binary expression. However, a Numeric_constant can be
invalid if it is unset, which occurs when either operand cannot be
converted to the right type. These errors are issued after the
Numeric_constant is lowered.
Kyrylo Tkachov [Tue, 20 Oct 2015 17:18:24 +0000 (17:18 +0000)]
[AArch64] Add support for 64-bit vector-mode ldp/stp
* config/aarch64/aarch64.c (aarch64_mode_valid_for_sched_fusion_p):
New function.
(fusion_load_store): Use it.
* config/aarch64/aarch64-ldpstp.md: Add new peephole2s for
ldp and stp in VD modes.
* config/aarch64/aarch64-simd.md (load_pair<mode>, VD): New pattern.
(store_pair<mode>, VD): Likewise.
* gcc.target/aarch64/stp_vec_64_1.c: New test.
* gcc.target/aarch64/ldp_vec_64_1.c: Likewise.
to enable compressed debug sections for gas, gold or ld by default. At
the moment, this configure option is ignored by gold and ld. For x86
Linux targets, default to compressing debug sections in gas.
H.J. Lu [Tue, 20 Oct 2015 15:12:37 +0000 (15:12 +0000)]
Properly handle -miamcu and -march=lakemont
-miamcu specifies how parameters are passed to functions and how value
is returned from a function. Inside function, we can use instructions
supported by -march=XXX. -miamcu -march=haswell can generate x87, SSE
and AVX instructions as long as the IA MCU psABI is followed. But since
Lakemont processor doesn't support x87 instructions, we shouldn't
generate x87 instructions with -march=lakemont. This patch separates
code generation from -miamcu and makes -march=lakemont not to generate
x87 instructions.
gcc/
PR target/67963
PR target/67985
* common/config/i386/i386-common.c (ix86_handle_option): Remove
OPT_miamcu handling.
* config/i386/i386.c (PTA_NO_80387): New macro.
(processor_alias_table): Add PTA_NO_80387 to lakemont.
(ix86_option_override_internal): Update MASK_80387 from
PTA_NO_80387. Don't warn x87/MMX/SSE/AVX for -miamcu. Warn
SSE math only if 80387 is supported. Don't change
MASK_FLOAT_RETURNS.
(ix86_valid_target_attribute_tree): Enable FPMATH_387 only if
80387 is supported.
* config/i386/i386.h (TARGET_FLOAT_RETURNS_IN_80387): True only
if TARGET_80387 is true and TARGET_IAMCU is false.
(TARGET_FLOAT_RETURNS_IN_80387_P): True only if TARGET_80387_P
is true and TARGET_IAMCU_P is false.
Yannick Moy [Tue, 20 Oct 2015 12:42:53 +0000 (12:42 +0000)]
fmap.adb, [...]: Fix coding style for marking start of processing of subprograms.
2015-10-20 Yannick Moy <moy@adacore.com>
* fmap.adb, a-cihama.adb, sem_ch5.adb, make.adb, inline.adb,
a-cfhase.adb, scng.adb, sem_ch12.adb, freeze.adb, tempdir.adb,
sem_util.adb, sem_res.adb, s-regexp.adb, a-clrefi.adb: Fix coding
style for marking start of processing of subprograms.
2015-10-20 Yannick Moy <moy@adacore.com>
* lib-xref-spark_specific.adb (Add_SPARK_File): Start traversal
by requesting info from stubs. (Traverse_All_Compilation_Units):
Remove unused procedure.
(Traverse_Declarations_Or_Statements): Handle protected and task units.
* lib-xref.ads (Traverse_All_Compilation_Units): Remove unused
procedure.
* restrict.adb (Check_Restriction): Do not ignore
restrictions in GNATprove_Mode.
Arnaud Charlet [Tue, 20 Oct 2015 12:32:21 +0000 (14:32 +0200)]
[multiple changes]
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch11.adb (Analyze_Handled_Statements): If the scope is a
postcondition subprogram, do not check for useless assignments
because there are no source references in such a body, and the
call will lose deferred references from the enclosing subprogram.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch13.adb: nalyze_Attribute_Definition_Clause, case
'Address): If either object is controlled the overlay is
erroneous, but analysis must be completed so that back-end sees
address clause and completes code generation. Improve text
of warning.
Arnaud Charlet [Tue, 20 Oct 2015 12:24:52 +0000 (14:24 +0200)]
[multiple changes]
2015-10-20 Yannick Moy <moy@adacore.com>
* a-sytaco.ads (Ada.Synchronous_Task_Control): Package
now withs System.Task_Identification. The visible part
of the spec has SPARK_Mode. The private part has pragma
SPARK_Mode (Off).
(Set_True): Added Global and Depends aspects
(Set_False): Added Global and Depends aspects (Current_State):
Added Volatile_Function aspect and added external state
Ada.Task_Identification.Tasking_State as a Global input.
(Suspend_Until_True): Added Global and Depends aspects
* a-sytaco.adb (Ada.Synchronous_Task_Control):
Package body has SPARK_Mode => Off
* a-extiin.ads (Ada.Execution_Time.Interrupts):
Package now withs Ada.Real_Time and has SPARK_Mode.
(Clock): Added Volatile_Function aspect and added external state
Ada.Real_Time.Clock_Time as a Global input.
* a-reatim.ads (Ada.Real_Time): The visible part of the spec has
SPARK_Mode. The private part has pragma SPARK_Mode (Off). The package
declares external state Clock_Time with properties Async_Readers and
Async_Writers.
(Clock): Added Volatile_Function aspect and
added external state Clock_Time as a Global input.
* a-reatim.adb (Ada.Real_Time): Package body has SPARK_Mode => Off
* a-exetim-default.ads, a-exetim-mingw.ads (Ada.Execution_Time):
The visible part of the spec has SPARK_Mode. The private part
has pragma SPARK_Mode (Off).
(Clock): Added Volatile_Function
aspect and added external state Clock_Time as a Global input.
(Clock_For_Interrupts): Added Volatile_Function aspect and added
external state Ada.Real_Time.Clock_Time as a Global input.
* a-exetim-mingw.adb (Ada.Execution_Time): Package body has
SPARK_Mode => Off
* a-interr.ads (Ada.Interrupts): Package now
withs Ada.Task_Identification (Is_Reserved): Added
SPARK_Mode, Volatile_Function and external state
Ada.Task_Identification.Tasking_State as a Global input.
(Is_Attached): Added SPARK_Mode, Volatile_Function and external
state Ada.Task_Identification.Tasking_State as a Global input.
(Attach_Handler): Added SPARK_Mode => Off (Exchange_Handler):
Added SPARK_Mode => Off (Detach_Handler): Added SPARK_Mode
and external state Ada.Task_Identification.Tasking_State as a
Global In_Out. (Reference): Added SPARK_Mode => Off
* a-disedf.ads (Get_Deadline): Added SPARK_Mode, Volatile_Function
and external state Ada.Task_Identification.Tasking_State as a
Global input.
* a-taside.ads (Ada.Task_Identification): The visible part of
the spec has SPARK_Mode. The private part has pragma SPARK_Mode
(Off). The package declares external state Tasking_State with
properties Async_Readers and Async_Writers.
(Current_Task): Added
Volatile_Function aspect and added external state Tasking_State
as a Global input.
(Environment_Task): Added SPARK_Mode => Off
(Is_Terminated): Added Volatile_Function aspect and added external
state Tasking_State as a Global input. (Is_Callable): Added
Volatile_Function aspect and added external state Tasking_State as
a Global input.
(Activation_Is_Complete): Added Volatile_Function
aspect and added external state Tasking_State as a Global input.
* a-taside.adb (Ada.Task_Identification): Package body has
SPARK_Mode => Off.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* atree.ads, atree.adb: Enable List38 and List39 on entities.
* einfo.ads, einfo.adb (Class_Wide_Preconds) new attribute defined
on subprograms. Holds the list of class-wide precondition
functions inherited from ancestors. Each such function is an
instantiation of the generic function generated from an explicit
aspect specification for a class-wide precondition. A type is
an ancestor of itself, and therefore a root type has such an
instance on its own list.
(Class_Wide_Postconds): ditto for postconditions.
2015-10-20 Vincent Celier <celier@adacore.com>
* prj-attr.adb: Add packages Prove and GnatTest.
2015-10-20 Steve Baird <baird@adacore.com>
* a-conhel.adb: Add an Annotate pragma to help suppress CodePeer's
analysis of internals of container generic instances. This pragma
has no other effect.
* a-conhel.adb (Generic_Implementation) Add "pragma Annotate
(CodePeer, Skip_Analysis);".
Arnaud Charlet [Tue, 20 Oct 2015 12:21:37 +0000 (14:21 +0200)]
[multiple changes]
2015-10-20 Steve Baird <baird@adacore.com>
* pprint.adb: Code clean up.
2015-10-20 Bob Duff <duff@adacore.com>
* a-cfinve.ads, a-coboho.ads: Improve comments.
* a-coboho.adb (Size_In_Storage_Elements): Improve error message
in case of "Size is too big" exception.
2015-10-20 Bob Duff <duff@adacore.com>
* a-contai.ads: Remove check names (moved to snames.ads-tmpl).
* snames.ads-tmpl: Add check names that were previously in
a-contai.ads, so they are now visible in configuration files.
* types.ads: Add checks corresponding to snames.ads-tmpl.
Arnaud Charlet [Tue, 20 Oct 2015 12:20:20 +0000 (14:20 +0200)]
[multiple changes]
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch5.adb (Analyze_Loop_Statement): Attach generated loop
identifier to the tree, because it may be the root of a tree
traversal in Pop_Scope when freeze actions are pending.
2015-10-20 Steve Baird <baird@adacore.com>
* pprint.ads (Expression_Image) Add new generic formal flag
Hide_Parameter_Blocks.
* pprint.adb (Expression_Image) If new flag is set, then display
dereferences of parameter block components accordingly.
* exp_ch6.adb (Expand_N_Extended_Return_Statement): Code cleanup.
(Make_Build_In_Place_Call_In_Object_Declaration): Update the
parameter profile. Code cleanup. Request debug info for the
object renaming declaration.
(Move_Activation_Chain): Add new formal parameter and update the
comment on usage.
* exp_ch6.ads (Make_Build_In_Place_Call_In_Object_Declaration):
Update the parameter profile and comment on usage.
* sem_util.ads, sem_util.adb (Remove_Overloaded_Entity): New routine,
currently unused.
Ed Schonberg [Tue, 20 Oct 2015 12:16:37 +0000 (12:16 +0000)]
sem_ch13.adb (Analyze_One_Aspect, [...]): If expander is not active...
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch13.adb (Analyze_One_Aspect, case
Aspect_Disable_Controlled): If expander is not active, pre-analyze
expression anyway for ASIS and other tools use.
* sem_prag.adb (Build_Generic_Class_Condition): Handle properly
anonymous access types in parameter specifications. Make the
formal type a formal derived type of the controlling type of
the subprogram.
* init.c: Fix build issue on arm-vx6 when building the RTP
run-time.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Analyze_Object_Declaration): If the expression
is an aggregate and compilation is in -gnatI mode (ignore rep
clauses) do not delay resolution of aggregate, to prevent freeze
actions out of order in the backend.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_prag.ads, sem_prag.adb (Build_Generic_Class_Condition):
New procedure to construct a generic function for a class-wide
precondition, to implement AI12-0113 concerning the new semantics
of class-wide preconditions for overriding uperations.
* exp_ch9.adb (Expand_N_Protected_Type_Declaration):
Check for No_Implicit_Protected_Object_Allocations.
* fe.h (Check_No_Implicit_Task_Alloc,
Check_No_Implicit_Protected_Alloc): Define and declare.
* restrict.ads, restrict.adb (Check_No_Implicit_Task_Alloc,
Check_No_Implicit_Protected_Alloc): New procedures to check the
restrictions.
* s-rident.ads (No_Implicit_Task_Allocations)
(No_Implicit_Protected_Object_Allocations): Declare new
restrictions.
2015-10-20 Yannick Moy <moy@adacore.com>
* sem_res.adb (Resolve_Selected_Component): Only set flag
when component is defined in a variant part.
* sem_util.adb,
* sem_util.ads (Is_Declared_Within_Variant): Promote local query
as publicy visible one for use in Resolve_Selected_Component.
2015-10-20 Philippe Gil <gil@adacore.com>
* g-debpoo.adb: allow instrumented System.Memory to use Debug_Pool
from foreign threads.
* g-debpoo.adb (Print_Traceback): NEW print traceback if available
added to support Stack_Trace_Depth = 0.
(Print_Address): NEW print System.Address without no secondary
stack use (Address_Image uses secondary stack)
* sem_ch5.adb (Analyze_Loop_Statement): Element iterators
over multidimensional arrays create additional loops during
expansion. For such loops we create a label as a scope
name. Attach this label properly to tree, for use in GNATProve
over such element iterators.
* sem_attr.adb (Analyze_Attribute, case Loop_Entry): The loop to
which the attribute applies comes from source, not from expansion
of an element iterator or a quantified expression.
* exp_attr.adb (Expand_N_Attribute_Reference): Ditto.
* sem_res.adb (Is_OK_Volatile_Context): Add a guard when checking a
possible call to an instance of Ada.Unchecked_Conversion to avoid
testing protected function calls. Allow references to protected objects
in prefixed protected calls.
(Is_Protected_Operation_Call): New routine.
2015-10-20 Yannick Moy <moy@adacore.com>
* exp_ch5.adb, exp_ch5.ads (Expand_Iterator_Loop_Over_Array): Make
query public. Remove code handling with iterator loop over array
of the 'in' form, which is not allowed in Ada. * exp_spark.adb
(Expand_SPARK): Expand loop statements that take the form of an
iterator over an array.
* sem_ch5.adb (Analyze_Loop_Statement): Do not analyze loop statements
that take the form of an iterator over an array, so that the rewritten
form gets analyzed instead.
* sem_util.adb, sem_util.ads (Is_Iterator_Over_Array): New query
to recognize iterators over arrays.
Martin Liska [Tue, 20 Oct 2015 11:59:51 +0000 (13:59 +0200)]
Correctly fill up cgraph_node::local.versionable flag.
* cgraphclones.c (cgraph_node::create_virtual_clone):
Verify cgraph_node.local.versionable instead of calling
tree_versionable_function_p.
* ipa-cp.c (determine_versionability): Save the information
to ipa_node_params summary.
(ipcp_versionable_function_p): Use it.
(ipcp_propagate_stage): Pass IPA_NODE_REF to a called function.
(ipcp_generate_summary): Do not compute cgraph_node
versionability.
* ipa-inline-analysis.c (inline_generate_summary): Compute
versionability for all cgraph nodes.
* ipa-prop.c (ipa_node_params_t::duplicate): Duplicate
ipa_node_params::versionability.
* ipa-prop.h (struct ipa_node_params): Declare it.
Arnaud Charlet [Tue, 20 Oct 2015 11:59:22 +0000 (13:59 +0200)]
[multiple changes]
2015-10-20 Yannick Moy <moy@adacore.com>
* sem_warn.adb (Is_OK_Fully_Initialized): Consider types with DIC as
fully default initialized.
* sem_ch6.adb: minor style fix in comment
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* s-diflio.ads, s-diflio.adb (Image): New function for dimensioned
quantities, to produce a string that includes the dimension
synbol for the quantity, or the vector of dimensions in standard
notation.
* sem_dim.adb (Expand_Put_Call_With_Symbol): Process new function
Image, to include dimension information in the generated string,
identical to the string produced by the Put procedure on a string
for a dimensioned quantity.
* exp_ch7.adb (Process_Declarations): A loop
parameter does not require finalization actions.
2015-10-20 Eric Botcazou <ebotcazou@adacore.com>
* exp_ch6.adb (Expand_Simple_Function_Return): Do not create an
actual subtype for a mutable record return type if the expression
is itself a function call.
* s-atocou.adb, s-atocou-builtin.adb: Fix implementation description
related to new type support.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_res.adb (Resolve_Explicit_Dererence): Call Analyze_Dimension
to propagate dimension information from prefix.
* sem_dim.adb (Analyze_Dimension): Handle Explicit_Dereference.
* inline.ads: minor whitespace fix in comment
* sem_ch6.adb: minor gramar fix in comment
Arnaud Charlet [Tue, 20 Oct 2015 11:45:37 +0000 (13:45 +0200)]
[multiple changes]
2015-10-20 Bob Duff <duff@adacore.com>
* a-cobove.adb (Set_Length): Restore previous logic, but with "Checks
and then" on the check.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* par-ch3.adb (P_Known_Discriminant_Part_Opt): Handle properly
a misplaced "constant" keyword in a discriminant specification.
2015-10-20 Steve Baird <baird@adacore.com>
* einfo.ads (Is_Param_Block_Component_Type): New function decl
for querying the flag.
(Set_Is_Param_Block_Component_Type): New procedure decl for setting
the flag.
* einfo.adb (Is_Param_Block_Component_Type): New function body.
(Set_Is_Param_Block_Component_Type): New procedure body.
(Write_Entity_Flags): Display the new flag.
* exp_ch9.adb (Build_Parameter_Block): Set flag on parameter
block component types.
(Expand_N_Entry_Declaration): Set flag on parameter block component
types.
2015-10-20 Steve Baird <baird@adacore.com>
* sem_elab.adb: Do not pass an N_Attribute_Reference node to
Sinfo.Set_No_Elaboration_Check.
* sem_elab.adb (Check_Elab_Call) Do not pass a non-call
node (specifically, an N_Attribute_Reference node) to
Sinfo.Set_No_Elaboration_Check.
2015-10-20 Tristan Gingold <gingold@adacore.com>
* adaint.c: File names are case sensitive on aarch64-ios.
* rtsfind.ads Remove the entries for Ada.Synchronous_Task_Control
and Suspension_Object from tables RE_Id, RE_Unit_Table and RTU_Id.
* sem_util.adb (Is_Descendant_Of_Suspension_Object): Update
the comment on usage. Use routine Is_Suspension_Object to detect
whether a type matches Suspension_Object.
(Is_Suspension_Object): New routine.
* snames.ads-tmpl: Add predefined names for Suspension_Object
and Synchronous_Task_Control.
* aspects.adb Add aspect Volatile_Function to table
Canonical_Aspect.
* aspect.ads Add aspect Volatile_Function to tables
Aspect_Argument, Aspect_Delay, Aspect_Id, Aspect_Names
and Implementation_Defined_Aspect. Aspects Async_Readers,
Async_Writers, Effective_Reads and Effective_Writes are no
longer Boolean.
* einfo.adb (Get_Pragma): Add an entry for pragma
Volatile_Function.
* par-prag.adb (Prag): Pragma Volatile_Function does not need
special processing by the parser.
* rtsfind.ads Add an entry for Ada.Synchronous_Task_Control in
table RTU_Id. Add an entry for Suspension_Object in table RE_Id.
* sem_ch3.adb Fix SPARK RM references.
(Analyze_Object_Contract): Update the error guard.
* sem_ch5.adb Fix SPARK RM references.
* sem_ch6.adb (Analyze_Subprogram_Body_Contract): Ensure
that a non-volatile function does not contain an effectively
volatile parameter.
(Analyze_Subprogram_Contract): Ensure
that a non-volatile function does not contain an effectively
volatile parameter.
* sem_ch12.adb (Instantiate_Object): Remove the reference to
the SPARK RM from the error message.
* sem_ch13.adb (Analyze_Aspect_Specifications): Add
processing for aspects Async_Readers, Async_Writers,
Effective_Reads, Effective_Writes and Volatile_Function.
(Check_Aspect_At_Freeze_Point): Aspects Async_Readers,
Async_Writers, Effective_Reads, Effective_Writes and
Volatile_Function do not need special processing at the freeze
point.
* sem_prag.adb Add an entry for pragma Volatile_Function in
table Sig_Flags.
(Analyze_External_Property_In_Decl_Part):
Reimplemented as Async_Readers, Async_Writers, Effective_Reads
and Effective_Writes are no longer Boolean pragmas.
(Analyze_Global_Item): An external state or effectively
volatile object cannot appear as an item in pragma
[Refined_]Global.
(Analyze_Pragma): Change the implementation
of Async_Readers, Async_Writers, Effective_Reads and
Effective_Writes as these are no longer Boolean pragmas.
Use routine Check_Static_Boolean_Expression to verify the
optional Boolean expression of Async_Readers, Async_Writers,
Constant_After_Elaboration, Effective_Reads, Effective_Writes,
Extensions_Visible and Volatile_Function. Add processing for
pragma Volatile_Function.
(Check_Static_Boolean_Expression): New routine.
(Find_Related_Context): Update the comment on usage.
(Is_Enabled_Pragma): New routine.
* sem_prag.ads (Is_Enabled_Pragma): New routine.
* sem_res.adb Fix SPARK RM references.
(Is_OK_Volatile_Context): Add detection for return statements.
(Resolve_Actuals): Remove the check concerning an effectively volatile
OUT actual parameter as this is now done by the SPARK flow analyzer.
(Resolve_Entity_Name): Remove the check concerning an effectively
volatile OUT formal parameter as this is now done by the SPARK
flow analyzer. (Within_Volatile_Function): New routine.
* sem_util.adb (Add_Contract_Item): Add processing for pragma
Volatile_Function.
(Check_Nonvolatile_Function_Profile): New routine.
(Is_Descendant_Of_Suspension_Object): New routine.
(Is_Effectively_Volatile): Protected types and descendants of
Suspension_Object are now treated as effectively volatile.
(Is_Enabled): The optional Boolean expression of pragmas
Async_Readers, Async_Writers, Effective_Reads and Effective_Writes
now appears as the first argument.
(Is_Volatile_Function): New routine.
* sem_util.ads Add SPARK RM references.
(Add_Contract_Item): Update the comment on usage.
(Check_Nonvolatile_Function_Profile): New routine.
(Is_Effectively_Volatile): Update the comment on usage.
(Is_Volatile_Function): New routine.
* snames.ads-tmpl Add a predefined name and pragma id for
Volatile_Function.
Bob Duff [Tue, 20 Oct 2015 10:23:46 +0000 (10:23 +0000)]
sem_ch13.adb (Analyze_One_Aspect): Avoid analyzing the expression in a 'Disable_Controlled attribute when...
2015-10-20 Bob Duff <duff@adacore.com>
* sem_ch13.adb (Analyze_One_Aspect): Avoid
analyzing the expression in a 'Disable_Controlled attribute when
Expander_Active is False, because otherwise, we get errors about
nonstatic expressions in pragma-Preelaborate generic packages.
* restrict.ads: minor whitespace cleanup in comment
2015-10-20 Bob Duff <duff@adacore.com>
* a-conhel.adb: Remove "use SAC;", because otherwise the compiler
complains about use clauses in run-time units. Use "use type"
instead.
* a-btgbso.adb, a-btgbso.ads, a-cbdlli.adb, a-cbdlli.ads,
* a-cbhama.adb, a-cbhama.ads, a-cbhase.adb, a-cbhase.ads,
* a-cbmutr.adb, a-cbmutr.ads, a-cborma.adb, a-cborma.ads,
* a-cborse.adb, a-cborse.ads, a-cdlili.adb, a-cdlili.ads,
* a-chtgbk.adb, a-chtgbk.ads, a-chtgbo.adb, a-chtgbo.ads,
* a-chtgke.adb, a-chtgke.ads, a-chtgop.adb, a-chtgop.ads,
* a-cidlli.adb, a-cidlli.ads, a-cihama.adb, a-cihama.ads,
* a-cihase.adb, a-cihase.ads, a-cimutr.adb, a-cimutr.ads,
* a-ciorma.adb, a-ciorma.ads, a-ciormu.adb, a-ciormu.ads,
* a-ciorse.adb, a-ciorse.ads, a-cobove.adb, a-cobove.ads,
* a-cohama.adb, a-cohama.ads, a-cohase.adb, a-cohase.ads,
* a-cohata.ads, a-coinve.adb, a-comutr.adb, a-comutr.ads,
* a-convec.adb, a-coorma.adb, a-coorma.ads, a-coormu.adb,
* a-coormu.ads, a-coorse.adb, a-coorse.ads, a-crbltr.ads,
* a-crbtgk.adb, a-crbtgk.ads, a-crbtgo.adb, a-crbtgo.ads,
* a-rbtgbk.adb, a-rbtgbk.ads, a-rbtgbo.adb, a-rbtgbo.ads,
* a-rbtgso.adb, a-rbtgso.ads: Change all the predefined containers
to share the tampering machinery in Ada.Containers.Helpers. This
reduces the amount of duplicated code, and takes advantage of
efficiency improvements in Helpers.
Protect all run-time checks and supporting machinery with "if
Checks" or "if T_Check", so this code can be suppressed with
pragma Suppress or -gnatp.
Add Pseudo_Reference and Get_Element_Access to remaining
containers, so that the compiler can optimize "for ... of" loops.
Arnaud Charlet [Tue, 20 Oct 2015 10:20:37 +0000 (12:20 +0200)]
[multiple changes]
2015-10-20 Bob Duff <duff@adacore.com>
* a-contai.adb, a-coinve.ads, a-contai.ads, a-conhel.adb, a-conhel.ads,
Makefile.rtl, a-convec.ads: Move helper code from Ada.Containers to a
new package Ada.Containers.Helpers, because otherwise it's not
visible everywhere it needs to be (e.g. in the package
Ada.Containers.Red_Black_Trees, Generic_Tree_Types wants to have
a component of type Tamper_Counts).
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_type.adb (Intersect_Types): Specialize error message when
one operand is a limited view which is a priori incompatible
with all other named types.
* sem_prag.adb: minor fix in comment
* sem_ch13.adb: Code clean up.
2015-10-20 Eric Botcazou <ebotcazou@adacore.com>
* sem_ch12.adb (Need_Subprogram_Instance_Body): Also return true
for a subprogram nested in an inlined subprogram.
Arnaud Charlet [Tue, 20 Oct 2015 10:07:09 +0000 (12:07 +0200)]
[multiple changes]
2015-10-20 Philippe Gil <gil@adacore.com>
* g-debpoo.ads (Dump): NEW print Debug_Pool statistics & main
contributors.
(Dump_Stdout): NEW print to stdout Debug_Pool statistics &
main contributors.
(Reset): NEW reset counters to 0.
(Get_Size): NEW return size allocated at parameter.
(High_Water_Mark): NEW.
(Current_Water_Mark): NEW.
(System_Memory_Debug_Pool): NEW tell Debug_Pools that
System.Memory uses it.
* g-debpoo.adb (Traceback_Htable_Elem): add Frees, Total_Frees
components.
(Find_Or_Create_Traceback): don't manage in System.Memory
Debug_Pool Deallocate Traceback's.
(Validity): add optional Handled table when System.Memory asked
for Allow_Unhandled_Memory.
(Allocate): handle Allocate reentrancy occuring when System.Memory
uses Debug_Pools.
(Deallocate): handle when Allow_Unhandled_Memory
is set deallocation of unhandled memory. Dont't check
Size_In_Storage_Elements if equal to Storage_Count'Last. update
Frees, Total_Frees new components.
Arnaud Charlet [Tue, 20 Oct 2015 10:04:44 +0000 (12:04 +0200)]
[multiple changes]
2015-10-20 Vincent Celier <celier@adacore.com>
* sem_cat.adb (Check_Categorization_Dependencies): Do nothing
when -gnatdu is specified.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch8.adb (analyze_Subprogram_Renaming): The actual for a
formal abstract subprogram must have a controlling type.
* stand.ads: Minor whitespace cleanup.
Arnaud Charlet [Tue, 20 Oct 2015 10:00:15 +0000 (12:00 +0200)]
[multiple changes]
2015-10-20 Gary Dismukes <dismukes@adacore.com>
* sem_ch13.adb: Minor reference change (RM => AARM).
2015-10-20 Eric Botcazou <ebotcazou@adacore.com>
* make.adb (Check): Skip multilib switches reinstated by the
compiler only when counting the number of switches, since it is
what really matters in the regular operating mode.
2015-10-20 Arnaud Charlet <charlet@adacore.com>
* einfo.adb: Add extra assertion for small clause.
* cstand.adb: Minor style fix in comment.
* debug.adb: Minor reformatting.
* exp_util.adb: Fix minor typo.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch12.adb (Same_Instantiated_Function): New predicate in
Check_Formal_Package_Instance, used to verify that the formal
and the actual of an actual package match when both are functions
given as attribute references.
Arnaud Charlet [Tue, 20 Oct 2015 09:56:56 +0000 (11:56 +0200)]
[multiple changes]
2015-10-20 Bob Duff <duff@adacore.com>
* a-coinve.ads, a-coinve.adb: Do the same efficiency
improvements that were already done in the definite case
(Ada.Containers.Vectors, i.e. a-convec). This includes the
ability to suppress checks, the fast path for Append, inlining
as appropriate, and special-casing of "for ... of" loops. Reuse
the tampering machinery that is now in Ada.Containers. Simplify
many operations.
* a-convec.ads, a-convec.adb: Change the code to be more similar
to a-coinve.
* a-finali.ads, a-finali.adb: Expose the "null"-ness of the
operations. This may enable optimizations in the future, and
seems cleaner anyway.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch13.adb (Is_Operational_Item): Attributes related to
Ada 2012 iterators are operational items, and can be specified
on partial views.
* sem_prag.adb (Check_Usage): Update the calls to Usage_Error.
(Usage_Error): Remove formal parameter Item. Emit a clearer message
concerning a missing dependency item and place it on the related pragma.
2015-10-20 Bob Duff <duff@adacore.com>
* debug.adb, expander.adb: Implement -gnatd.B switch, which
triggers a bug box when an abort_statement is seen. This is
useful for testing Comperr.Compiler_Abort.
* gnat1drv.adb: Trigger bug box on all exceptions other than
Unrecoverable_Error.
Arnaud Charlet [Tue, 20 Oct 2015 09:53:11 +0000 (11:53 +0200)]
[multiple changes]
2015-10-20 Thomas Quinot <quinot@adacore.com>
* Makefile.rtl: add the following...
* g-binenv.ads, g-binenv.adb: New unit providing runtime access
to bind time captured values ("bind environment")
* init.c: declare new global variable __gl_bind_env_addr.
* bindgen.ads, bindgen.adb (Set_Bind_Env): record a bind
environment key=value pair.
(Gen_Bind_Env_String): helper to produce the bind environment data
called in the binder generated file.
(Gen_Output_File_Ada): Call the above (Gen_Adainit): Set
__gl_bind_env_addr accordingly.
* switch-b.adb: Support for command line switch -V (user interface
to set a build environment key=value pair)
* bindusg.adb: Document the above
2015-10-20 Vincent Celier <celier@adacore.com>
* sem_prag.adb (Analyse_Pragma: Pragma Pure): Do not set the
entity as Pure if Debug_Flag_U is set.
Arnaud Charlet [Tue, 20 Oct 2015 09:48:00 +0000 (11:48 +0200)]
[multiple changes]
2015-10-20 Bob Duff <duff@adacore.com>
* output.adb (Write_Int): Work with negative numbers in order to avoid
negating Int'First and thereby causing overflow.
* sem_util.adb: Minor comment fix.
2015-10-20 Eric Botcazou <ebotcazou@adacore.com>
* make.adb (Check): Skip multilib switches reinstated by the
compiler when doing the comparison with switches passed to
gnatmake.
* einfo.adb (Get_Pragma): Minor reformatting. Rename local constant
Is_CDG to Is_CLS. Add pragma Constant_After_Elaboration to the list of
classification pragmas.
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Analyze_Declarations); At the of the visible part,
perform name resolution on the expressions in aspects of visible
entities.
* sem_ch13.ads, sem_ch13.adb (Resolve_Aspect_Expressions): Resolve
expressions in aspects independently of whether corresponding
entity is frozen. Used to complete name resolution of aspect
expressions for entities declared in the visible part of a
package or generic package declaration.
Ed Schonberg [Tue, 20 Oct 2015 09:40:24 +0000 (09:40 +0000)]
sem_ch3.adb (Check_Nonoverridable_Aspects): New procedure within Analyze_Full_Type_ Declaration...
2015-10-20 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Check_Nonoverridable_Aspects): New procedure within
Analyze_Full_Type_ Declaration, used to apply legality rules in
13,1,1 (18.3.3) concerning aspects that cannot be overridden in
a type extension.
(Check_Duplicate_Aspects): It is not legal to specify the
Implicit_Dereference aspect on a full view if partial view has
known discriminants.
* sem_ch13.adb (Analyze_Aspect_Implicit_Dereference): Verify that
the specification of the aspect on a derived type confirms the
value of the inherited one.
* sem_util.adb (Reference_Discriminant): Return empty if none
specified.
Jason Merrill [Tue, 20 Oct 2015 02:48:43 +0000 (22:48 -0400)]
Expand PTRMEM_CST only when necessary.
* constexpr.c (cxx_eval_constant_expression): Expand PTRMEM_CST
only when necessary.
(cxx_eval_component_reference): Like here.
* decl2.c (lower_var_init): And here.
(c_parse_final_cleanups): Call it.
* typeck2.c (digest_init_r): Not here.
* decl.c (complete_vars): Or here.
(cp_finish_decl): Add local statics to symbol table.
projects / thirdparty / gcc.git / log
