PR target/61737.
* config/cris/cris.c (TARGET_LEGITIMATE_CONSTANT_P)
(TARGET_CANNOT_FORCE_CONST_MEM): Define.
(cris_cannot_force_const_mem, cris_legitimate_constant_p): New
functions.
(cris_print_index, cris_print_operand, cris_constant_index_p)
(cris_side_effect_mode_ok): Replace CONSTANT_P with CRIS_CONSTANT_P.
(cris_address_cost): Ditto last CONSTANT_P.
(cris_symbol_type_of): Rename from cris_pic_symbol_type_of. All
callers changed. Yield cris_offsettable_symbol for non-PIC
constant symbolic expressions including labels. Yield cris_unspec
for all unspecs.
(cris_expand_pic_call_address): New parameter MARKERP. Set its
target to pic_offset_table_rtx for calls that will likely go
through PLT, const0_rtx when they can't. All callers changed.
Assert flag_pic. Use CONSTANT_P, not CONSTANT_ADDRESS_P, for
symbolic expressions to be PICified. Remove second, redundant,
assert on can_create_pseudo_p returning non-zero. Use
replace_equiv_address_nv, not replace_equiv_address, for final
operand update.
* config/cris/cris.md ("movsi"): Move variable t to pattern
toplevel. Adjust assert for new cris_symbol_type member. Use
CONSTANT_P instead of CONSTANT_ADDRESS_P.
("*movsi_internal") <case 9>: Make check for valid unspec operands
for lapc stricter.
<case CRIS_UNSPEC_PCREL, CRIS_UNSPEC_PLT_PCREL>: Clear condition
codes.
("call", "call_value"): Use second incoming operand as a marker
for pic-offset-table-register being used.
("*expanded_call_non_v32", "*expanded_call_v32")
("*expanded_call_value_non_v32", "*expanded_call_value_v32"): For
second incoming operand to CALL, match cris_call_type_marker.
("*expanded_call_value_side"): Ditto. Disable before
reload_completed.
("*expanded_call_side"): Ditto. Fix typo in comment.
(moverside, movemside peepholes): Check for CRIS_CONSTANT_P, not
CONSTANT_P.
* config/cris/predicates.md ("cris_call_type_marker"): New predicate.
* config/cris/cris.h (CRIS_CONSTANT_P): New macro.
(enum cris_symbol_type): Rename from cris_pic_symbol_type. All
users changed. Add members cris_offsettable_symbol and
cris_unspec.
(cris_symbol_type): Rename from cris_pic_symbol_type.
* config/cris/constraints.md ("T"): Use CRIS_CONSTANT_P, not
just CONSTANT_P.
* config/cris/cris-protos.h (cris_symbol_type_of,
cris_expand_pic_call_address): Adjust prototypes.
(cris_legitimate_constant_p): New prototype.
From-SVN: r212708
2014-07-16 Vincent Celier <celier@adacore.com>
* gnatls.adb: Get the target parameters only if -nostdinc was
not specified.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* checks.adb (Insert_Valid_Check): If the expression is a packed
component of a modular type of the right size the data is always
valid. This os particularly useful if the component is part of
a volatile variable.
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi, sinfo.ads, freeze.adb, exp_aggr.adb: Minor reformatting
2014-07-16 Thomas Quinot <quinot@adacore.com>
* exp_ch7.ads: Minor documentation fix.
From-SVN: r212663
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Find_Type_Name): Diagnose properly
a private extension completion that is an interface definition
with an interface list.
2014-07-16 Arnaud Charlet <charlet@adacore.com>
* gnatls.adb (Gnatls): Code clean ups.
2014-07-16 Thomas Quinot <quinot@adacore.com>
* sinfo.ads, sinfo.adb (N_Compound_Statement): New node kind.
* sem.adb (Analyze): Handle N_Compound_Statement.
* sprint.adb (Sprint_Node_Actual): Ditto.
* sem_ch5.ads, sem_ch5.adb (Analyze_Compound_Statement): New
procedure to handle N_Compound_Statement.
* exp_aggr.adb (Collect_Initialization_Statements):
Use a proper compound statement node, instead of a bogus
expression-with-actions with a NULL statement as its expression,
to wrap collected initialization statements.
* freeze.ads, freeze.adb
(Explode_Initialization_Compound_Statement): New public procedure,
lifted from Freeze_Entity.
(Freeze_Entity): When freezing
an object with captured initialization statements and without
delayed freezing, explode compount statement.
* sem_ch4.adb (Analyze_Expression_With_Actions): Remove special
case that used to handle bogus EWAs with NULL statement as
the expression.
* exp_ch13.adb (Expand_N_Freeze_Entity): For an object with
delayed freezing and captured initialization statements, explode
compound statement.
From-SVN: r212662
2014-07-16 Vincent Celier <celier@adacore.com>
* make.adb: Do not read gnat.adc when gnatmake is invoked
with -gnatA.
2014-07-16 Pascal Obry <obry@adacore.com>
* gnat_rm.texi, impunit.adb, g-rewdat.adb, g-rewdat.ads: Initial
implementation of GNAT.Rewrite_Data.
From-SVN: r212659
2014-07-16 Vincent Celier <celier@adacore.com>
* gnatls.adb (Normalize): New function.
(Gnatls): Get the target parameters. On targets other than VMS,
normalize the path names in the source search path, the object search
path and the project search path.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* sem_case.adb: Avoid self-checking of case expressions in
dynamic predicates.
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi: Document effect of Suppress_Initialization on tags and
discriminants.
From-SVN: r212658
2014-07-16 Eric Botcazou <ebotcazou@adacore.com>
* switch-b.adb (Scan_Binder_Switches): Add missing guard.
2014-07-16 Ben Brosgol <brosgol@adacore.com>
* gnat_ugn.texi: Fix typo.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* exp_ch4.adb (Expand_N_Case_Expression): Do not expand case
expression if it is the specification of a subtype predicate:
it will be expanded when the return statement is analyzed, or
when a static predicate is transformed into a static expression
for evaluation by the front-end.
* sem_ch13.adb (Get_RList): If the expression for a static
predicate is a case expression, extract the alternatives of the
branches with a True value to create the required statically
evaluable expression.
2014-07-16 Thomas Quinot <quinot@adacore.com>
* exp_prag.adb (Expand_Pragma_Check): Use the location of the
expression, not the location of the aspect, for all generated
code, so that in particular the call to raise_assert_failure
gets the sloc of the associated condition.
* exp_ch6.adb
(Expand_Subprogram_Contract.Build_Postconditions_Procedure):
Set an explicit End_Label on the handled sequence of statements
for the _Postconditions procedure so that the implicit return
statement does not erroneously get associated with code generated
for the last condition in postconditions.
2014-07-16 Thomas Quinot <quinot@adacore.com>
* ug_words: Fix name of VMS synonym for -gnatw.z (SIZE_ALIGNMENT,
not SIZE_ALIGN) and -gnatw.Z (NOSIZE_ALIGNMENT, not NOSIZE_ALIGN).
* vms_data.ads: Add missing spaces in VMS synonyms for -gnatw.z /
-gnatw.Z.
From-SVN: r212657
2014-07-16 Robert Dewar <dewar@adacore.com>
* sem_ch3.adb, sem_prag.adb, sem_util.adb, sem_res.adb, sem_ch13.adb:
Minor code reorganization (use Is_Access_Type, not in Access_Kind).
* exp_ch3.adb: Minor code reorganization, use Is_Access_Type,
not in Access_Kind.
* par-ch4.adb (At_Start_Of_Attribute): New function
(P_Simple_Expression): Better msg for bad attribute prefix.
* scans.ads: Minor reformatting.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* sem_attr.adb (Resolve_Attribute, case 'Update): If choice is a
static constant, check that in belongs to the corresponding index
subtype, to produce the proer warning when expansion is disabled.
2014-07-16 Robert Dewar <dewar@adacore.com>
* freeze.adb (Freeze_Entity): Warn on incompatible size/alignment.
* gnat_ugn.texi: Document -gnatw.z and -gnatw.Z.
* ug_words: VMS synonyms (WARNINGS=[NO]SIZE_ALIGN) for -gnatw.z/-gnatw.Z
* usage.adb: Add lines for -gnatw.z/-gnatw.Z.
* vms_data.ads: VMS synonyms (WARNINGS=[NO]SIZE_ALIGN) for
-gnatw.z/-gnatw.Z
* warnsw.adb: Set Warn_On_Size_Alignment appropriately.
* warnsw.ads (Warn_On_Size_Alignment): New flag Minor
reformatting.
From-SVN: r212656
2014-07-16 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch7.adb (Process_Declarations): Reinstate the check on
a hook object to ensure that the related transient declaration
is finalizable.
* exp_util.adb (Is_Aliased): Do not consider expresison with
actions as a special context.
(Requires_Cleanup_Actions): Reinstate the check on a hook object to
ensure that the related transient declaration is finalizable.
2014-07-16 Robert Dewar <dewar@adacore.com>
* checks.ads, checks.adb (Allocation_Checks_Suppressed): New function.
* snames.ads-tmpl: Add Allocation_Check to list of check names.
* types.ads: Add Allocation_Check to list of check names.
2014-07-16 Thomas Quinot <quinot@adacore.com>
* sem_util.adb (Enter_Name): replace bogus test for presence of
Corresponding_Remote_Type with correct test on Ekind.
* sem_res.adb (Valid_Conversion): ditto; also clarify validity
of calls to Corresponding_ Remote_Type (documentation fix).
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi: Document illegal case of Unrestricted_Access.
* sem_attr.adb (Analyze_Access_Attribute): Set_Non_Aliased_Prefix
where it applies.
(Resolve_Attribute, case Access): Flag illegal Unrestricted_Access use.
* sinfo.ads, sinfo.adb (Non_Aliased_Prefix): New flag.
From-SVN: r212655
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_ugn.texi: Document binder switch -Ra.
* gnatbind.adb (List_Closure_Display): Implement -Ra switch
(List_Closure_All).
* opt.ads (List_Closure_All): New switch.
* switch-b.adb (Scan_Binder_Switches): Recognize -Ra to set
List_Closure_All.
2014-07-16 Ben Brosgol <brosgol@adacore.com>
* gnat_rm.texi: Minor edits, to make case consistent in names
of types and fields.
From-SVN: r212654
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi: Document erroneous mixing of thin pointers and
unrestricted access
* gnat_ugn.texi: Add note on size of access types about thin
pointers and the use of attribute Unrestricted_Access.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* a-cbdlli.ads, a-cbdlli.adb, a-cbhama.ads, a-cbhama.adb,
* a-cbhase.ads, a-cbhase.adb, a-cborma.ads, a-cborma.adb,
* a-cborse.ads, a-cborse.adb, a-cobove.ads a-cobove.adb: Add Control
machinery to detect tampering on bounded vectors.
From-SVN: r212653
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi: Document that leading/trailing asterisks are
now implied for the pattern match string for pragma Warnings
and Warning_As_Error.
* sem_prag.adb (Acquire_Warning_Match_String): New procedure.
(Analyze_Pragma, case Warning_As_Error): Call
Acquire_Warning_Match_String.
(Analyze_Pragma, case Warnings): Call Acquire_Warning_Match_String.
From-SVN: r212651
2014-07-16 Bob Duff <duff@adacore.com>
* gnat_ugn.texi: Document need for project file
for --incremental switch for gnat2xml.
2014-07-16 Robert Dewar <dewar@adacore.com>
* gnat_rm.texi: Fix example of non-packable components in packed
records section.
2014-07-16 Robert Dewar <dewar@adacore.com>
* s-tpoben.adb, s-tasren.adb, s-interr.adb, s-interr-hwint.adb,
s-shasto.adb, s-interr-vms.adb, s-interr-sigaction.adb: Avoid use of
upper case in exception messages.
From-SVN: r212650
2014-07-16 Yannick Moy <moy@adacore.com>
* sem_attr.adb (Analyze_Attribute/cases Pred
and Succ): Always enable range check on fixed-point types if
not suppressed for the type.
2014-07-16 Hristian Kirtchev <kirtchev@adacore.com>
* sem_util.adb (State_Has_Enabled_Property):
Account for an external property denoted by an others choice.
From-SVN: r212647
2014-07-16 Hristian Kirtchev <kirtchev@adacore.com>
* exp_ch4.ads, exp_ch4.adb (Find_Hook_Context): Relocated to Exp_Util.
* exp_ch7.adb (Process_Declarations): There is no need to check
that a transient object being hooked is controlled as it would
not have been hooked in the first place.
* exp_ch9.adb Remove with and use clause for Exp_Ch4.
* exp_util.adb (Find_Hook_Context): Relocated from Exp_Ch4.
(Is_Aliased): A renaming of a transient controlled object is
not considered aliasing when it occurs within an expression
with actions.
(Requires_Cleanup_Actions): There is no need to
check that a transient object being hooked is controlled as it
would not have been hooked in the first place.
* exp_util.ads (Find_Hook_Context): Relocated from Exp_Ch4.
2014-07-16 Hristian Kirtchev <kirtchev@adacore.com>
* sem_ch13.adb (Insert_After_SPARK_Mode): Moved to
the outer level of routine Analyze_Aspect_Specifications. Ensure
that the corresponding pragmas of aspects Initial_Condition and
Initializes are inserted after pragma SPARK_Mode.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* sem_attr.adb (Analyze_Attribute, case 'Update): Handle
properly a choice list with more than one choice, where each
is an aggregate denoting a sequence of array indices for a
multidimentional array. For SPARK use.
From-SVN: r212646
2014-07-16 Vadim Godunko <godunko@adacore.com>
* a-coinho-shared.adb (Adjust): Create
copy of internal shared object and element when source container
is locked.
(Copy): Likewise.
(Query_Element): Likewise.
(Update_Element): Likewise.
(Constant_Reference): Likewise. Raise Constraint_Error on attempt
to get reference for empty holder.
(Reference): Likewise.
2014-07-16 Thomas Quinot <quinot@adacore.com>
* exp_ch4.adb (Find_Hook_Context): New subprogram, extracted
from Process_Transient_Oject.
* exp_ch4.ads: Ditto.
* exp_ch9.adb (Build_Class_Wide_Master): Insert the _master
declaration as an action on the topmost enclosing expression,
not on a possibly conditional subexpreession.
From-SVN: r212645
2014-07-16 Robert Dewar <dewar@adacore.com>
* a-coinho.adb, a-coinho-shared.adb, a-coinho-shared.ads: Minor
reformatting.
2014-07-16 Ed Schonberg <schonberg@adacore.com>
* a-cohase.ads: Type Iterator must be controlled, so that the
tampering bit is properly set through an iteration.
* a-cohase.adb: Add Finalize operation for type Iterator.
From-SVN: r212643
When a system macro is expanded in a non-system file during
out-of-line preprocessing, it can happen that the preprocessor forgets
to emit line markers to express the system-ness status of tokens that
come after the expansion of the macro.
That can lead to situations where the entire non-system file can be
considered as being a system file and thus have its warnings be
discarded during the compilation of the resulting preprocessed file.
My understanding is that this is due to the preprocessor not
systematically detecting (and reporting) the change in system-ness of
tokens.
And this is what this patch does. Each time the system-ness of a
given token is different from the previous token that was emitted by
the preprocessor, it emits a line marker for the sole purpose of
marking the new system-ness of the subsequent tokens to come.
Bootstrapped and tested on x86_64-unknown-linux-gnu against trunk.
gcc/c-family/ChangeLog:
* c-ppoutput.c (struct print::prev_was_system_token): New data
member.
(init_pp_output): Initialize it.
(maybe_print_line_1, maybe_print_line, print_line_1, print_line)
(do_line_change): Return a flag saying if a line marker was
emitted or not.
(scan_translation_unit): Detect if the system-ness of the token we
are about to emit is different from the one of the previously
emitted token. If so, emit a line marker. Avoid emitting useless
adjacent line markers. Avoid emitting line markers for tokens
originating from the expansion of built-in macros.
(scan_translation_unit_directives_only): Adjust.
gcc/testsuite/ChangeLog:
* gcc.dg/cpp/syshdr{4,5}.{c,h}: New test files.
Signed-off-by: Dodji Seketeli <dodji@redhat.com>
Signed-off-by: Dodji Seketeli <dodji@redhat.com>
From-SVN: r212638
When a built-in macro is expanded, the location of the token in the
epansion list is the location of the expansion point of the built-in
macro.
This patch creates a virtual location for that token instead,
effectively tracking locations of tokens resulting from built-in macro
tokens.
libcpp/
* include/line-map.h (line_maps::builtin_location): New data
member.
(line_map_init): Add a new parameter to initialize the new
line_maps::builtin_location data member.
* line-map.c (linemap_init): Initialize the
line_maps::builtin_location data member.
* macro.c (builtin_macro): Create a macro map and track the token
resulting from the expansion of a built-in macro.
gcc/
* input.h (is_location_from_builtin_token): New function
declaration.
* input.c (is_location_from_builtin_token): New function
definition.
* toplev.c (general_init): Tell libcpp what the pre-defined
spelling location for built-in tokens is.
Signed-off-by: Dodji Seketeli <dodji@redhat.com>
From-SVN: r212637
* asan.c (asan_finish_file): Use varpool_finalize_decl instead of
varpool_assemble_decl.
* varpool.c (varpool_assemble_decl): Assert that node->definition is
true.
From-SVN: r212564
PR rtl-optimization/61772
* ifcvt.c (dead_or_predicable): Check jump to be free of side
effects.
testsuite/
* gcc.dg/torture/pr61772.c: New test.
From-SVN: r212563
2014-07-15 Paolo Carlini <paolo.carlini@oracle.com>
* call.c (convert_like_real): Call print_z_candidate and inform only
if permerror returns true.
From-SVN: r212560
Hans-Peter Nilsson
Hans-Peter Nilsson
Jason Merrill
GCC Administrator
David Wohlferd
Denis Chertykov
Arnaud Charlet
Paolo Carlini
Arnaud Charlet
Adhemerval Zanella
Robert Dewar
Vadim Godunko
Ed Schonberg
Dodji Seketeli
Paolo Carlini
Jakub Jelinek
Richard Biener
Eric Christopher
Uros Bizjak
Uros Bizjak
Jan Hubicka
Bernd Schmidt
Michael Matz