platform/upstream/gcc.git
11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:42:01 +0000 (12:42 +0200)]
[multiple changes]

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch6.adb: Add with and use clause for Stringt.
(Expand_Contract_Cases): Moved from sem_ch6. Add formal parameters
Decls and Stmts along with comments on their usage.
* exp_ch6.ads (Expand_Contract_Cases): Moved from sem_ch6.
* sem_ch6.adb (Expand_Contract_Cases): Moved to exp_ch6.
(Process_Contract_Cases): Update the call to Expand_Contract_Cases.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* gnat_rm.texi: Minor editing, to clarify use of dimension aspects.
* sem_util.adb (Is_OK_Variable_For_Out_Formal): Reject an
aggregate for a packed type, which may be converted into an
unchecked conversion of an object.

From-SVN: r198292

11 years agosem_prag.adb: Minor code reorganization (correct misspelling Restiction).
Robert Dewar [Thu, 25 Apr 2013 10:39:36 +0000 (10:39 +0000)]
sem_prag.adb: Minor code reorganization (correct misspelling Restiction).

2013-04-25  Robert Dewar  <dewar@adacore.com>

* sem_prag.adb: Minor code reorganization (correct misspelling
Restiction).
* sem_util.adb, aspects.ads, sem_ch6.adb: Minor reformatting.
* gnat_rm.texi: Document impl-defined aspects.
* sem_dim.adb, sem_dim.ads, gnat_ugn.texi, s-dimmks.ads: Minor
reformatting.

From-SVN: r198291

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:35:29 +0000 (12:35 +0200)]
[multiple changes]

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* einfo.adb (Set_Abstract_States): The attribute now applies
to generic packages.
* sem_ch4.adb (Referenced): Moved to sem_util.
* sem_ch7.adb (Unit_Requires_Body): A [generic] package with
a non-null abstract state needs a body.
* sem_prag.adb (Analyze_Depends_In_Decl_Part): Update the calls
to Collect_Subprogram_Inputs_Outputs.
(Analyze_Global_Item): Verify the proper usage of an item with mode
In_Out or Output relative to the enclosing context.
(Analyze_Pragma): Abstract_State can now be applied to a generic
package. Do not reset the Analyzed flag for pragmas Depends and Global
as this is not needed.
(Appears_In): Moved to library level.
(Check_Mode_Restiction_In_Enclosing_Context): New routine.
(Collect_Subprogram_Inputs_Outputs): Moved to library level. Add
formal parameters Subp_Id, Subp_Inputs, Subp_Outputs and Global
seen along with comments on usage.
* sem_util.ads, sem_util.adb (Referenced): New routine.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch6.adb (Expand_Contract_Cases): Generate
detailed error messages only when switch -gnateE is in effect.

2013-04-25  Yannick Moy  <moy@adacore.com>

* sem_attr.adb (Analyze_Attribute): Do not issue
an error for a possibly misplaced 'Result or 'Old attribute when
analyzing the aspect.

From-SVN: r198290

11 years agosem_ch12.adb, [...]: Minor reformatting.
Robert Dewar [Thu, 25 Apr 2013 10:33:23 +0000 (10:33 +0000)]
sem_ch12.adb, [...]: Minor reformatting.

2013-04-25  Robert Dewar  <dewar@adacore.com>

* sem_ch12.adb, sem_util.adb, sem_ch4.adb: Minor reformatting.

From-SVN: r198289

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:31:36 +0000 (12:31 +0200)]
[multiple changes]

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch4.adb (Analyze_Quantified_Expression):
Add local variable Loop_Id. Verify that the loop variable
is used within the condition of the quantified expression.
(Referenced): New routine.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_case.adb (Analyze_Choices): Enhance the error message
given on a bad use of subtype predicate.
* sem_ch5.adb (Analyze_Loop_Parameter_Specification): Enhance
the error message given on a bad use of subtype predicate.
* sem_util.adb (Bad_Predicated_Subtype_Use): Add formal parameter
Suggest_Static. Emit an extra error message advising how to
remedy the bad use of the predicate if the context warrants it.
* sem_util.ads (Bad_Predicated_Subtype_Use): Add formal parameter
Suggest_Static along with a comment explaining its usage.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* sem_disp.adb (Check_Dispatching_Operation): Further refinement
to checks for AI05-0125: the check for a hidden primitive that
may be overridden by the new declaration only applies if the
hidden operation is never declared. This is not the case if the
operation is declared in a parent unit.

From-SVN: r198288

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:28:45 +0000 (12:28 +0200)]
[multiple changes]

2013-04-25  Robert Dewar  <dewar@adacore.com>

* debug.adb: Remove d.X and d.Y entries and documentation.
* exp_ch4.adb (Expand_N_If_Expression): Remove special code used
if expression with actions not available (now always available).
(Expand_Short_Circuit_Operator): Same change.
* gnat1drv.adb (Adjust_Global_Switches) Remove setting
Use_Expression_With_Actions flag, since this is now obsolete.
* opt.ads (Use_Expression_Actions): Removed (always True now).
* sinfo.ads: Minor comment updates.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* sem_ch12.adb (Check_Generic_Actuals): If an actual is an array
subtype whose base type is currently private, install full view
when compiling instance body.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* sem_disp.adb (Check_Dispatching_Operation): Refine checks for
AI05-0125: the check for a hidden primitive that may be overridden
by the new declaration is only performed if the declaration comes
from source, and it must carry an explicit overriding indicator.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* einfo.adb (Abstract_States): The attribute now applies to
generic packages.
* sem_ch3.adb (Analyze_Object_Declaration): Check whether an
object declaration introduces an illegal hidden state.
* sem_prag.adb (Analyze_Abstract_State): Check whether a state
declaration introduces an illegal hidden state.
* sem_util.ads, sem_util.adb (Check_No_Hidden_State): New routine.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* exp_ch6.adb (Is_Build_In_Place_Function_Call): The call may
be to a protected function, in which case the name in the call
is a selected component.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch4.adb (Analyze_Quantified_Expression):
Warn on a suspicious use of quantifier "some" when "all" was meant.
(No_Else_Or_Trivial_True): New routine.

From-SVN: r198287

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:25:54 +0000 (12:25 +0200)]
[multiple changes]

2013-04-25  Robert Dewar  <dewar@adacore.com>

* einfo.ads, einfo.adb: Put back with/use for Namet.
(Get_Pragma): New name (wi new spec) for Find_Pragma.
* sem_ch6.adb: Change name Find_Pragma to Get_Pragma with
different interface.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb (Is_Visible_Component): In an instance all
components are visible.

From-SVN: r198286

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:24:16 +0000 (12:24 +0200)]
[multiple changes]

2013-04-25  Matthew Heaney  <heaney@adacore.com>

* a-rbtgbo.adb, a-crbtgo.adb (Generic_Equal): do not test for
tampering when container empty.
* a-crbtgk.adb (Ceiling, Find, Floor): ditto.
(Generic_Conditional_Insert, Generic_Conditional_Insert_With_Hint):
ditto.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* par-ch12.adb: Move aspects from package specification to
generic package declaration.
* sem_ch12.adb: Analyze aspect specifications before building
and analyzing the generic copy, so that the generated pragmas
are properly taken into account.
* sem_ch13.adb: For compilation unit aspects that apply to a
generic package declaration, insert corresponding pragmas ahead
of visible declarations.
* sprint.adb: Display properly the aspects of a generic type
declaration.

2013-04-25  Robert Dewar  <dewar@adacore.com>

* frontend.adb: Minor reformatting.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* einfo.ads: Extend documentation on use of Is_Private_Ancestor
for untagged types.
* sem_ch3.adb (Is_Visible_Component): Refine predicate for the
case of untagged types derived from private types, to reject
illegal selected components.

From-SVN: r198285

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:19:59 +0000 (12:19 +0200)]
[multiple changes]

2013-04-25  Gary Dismukes  <dismukes@adacore.com>

* sem_util.adb (Is_Dependent_Component_Of_Mutable_Object): Test
for case of selecting from an unexpanded implicit dereference
and do not make a recursive call on such a prefix.

2013-04-25  Doug Rupp  <rupp@adacore.com>

* targparm.adb (VXF{_Str}): New tag for vaxfloat.
(Get_Target_Parameters): Handle VXF tag.
* targparm.ads (VAX_Float_On_Target): New boolean.
* system-vms-ia64.ads (VAX_Float): New boolean.
* frontend.adb (Frontend): Handle VAX float boolean.

From-SVN: r198284

11 years agoeinfo.ads, einfo.adb: Remove with and use clauses for Namet.
Hristian Kirtchev [Thu, 25 Apr 2013 10:13:14 +0000 (10:13 +0000)]
einfo.ads, einfo.adb: Remove with and use clauses for Namet.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* einfo.ads, einfo.adb: Remove with and use clauses for Namet.
(Find_Pragma): New routine.
* sem_util.ads, sem_util.adb (Find_Pragma): Moved to einfo.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch13.adb (Add_Call): Do not capture the nature of the inherited
predicate.
(Add_Predicates): Save the static predicate for diagnostics and error
reporting purposes.
(Process_PPCs): Remove local variables Dynamic_Predicate_Present and
Static_Predicate_Present. Add local variable Static_Pred. Ensure that
the expression of a static predicate is static.

From-SVN: r198283

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 10:10:58 +0000 (12:10 +0200)]
[multiple changes]

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* einfo.adb (Is_Ghost_Subprogram): Remove useless code.

2013-04-25  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Minor addition of index entry.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch6.adb (Check_Access_Invariants): Test whether an
invariant procedure is empty before generating a call to it.
(Has_Enabled_Predicate): New routine.
(Has_Null_Body): New routine.
(Process_PPCs): Test whether an invariant procedure is
empty before generating a call to it. Test whether predicates are
enabled for a particular type before generating a predicate call.
* sem_util.ads, sem_util.adb (Find_Pragma): New routine.

From-SVN: r198282

11 years agosem_ch7.adb, [...]: Minor reformatting.
Robert Dewar [Thu, 25 Apr 2013 10:09:35 +0000 (10:09 +0000)]
sem_ch7.adb, [...]: Minor reformatting.

2013-04-25  Robert Dewar  <dewar@adacore.com>

* sem_ch7.adb, einfo.adb, repinfo.adb, snames.adb-tmpl,
snames.ads-tmpl: Minor reformatting.

From-SVN: r198281

11 years agosem_ch7.adb: Minor reformatting.
Thomas Quinot [Thu, 25 Apr 2013 10:08:30 +0000 (10:08 +0000)]
sem_ch7.adb: Minor reformatting.

2013-04-25  Thomas Quinot  <quinot@adacore.com>

* sem_ch7.adb: Minor reformatting.

From-SVN: r198280

11 years agore PR libstdc++/57065 (incorrect default allocator template for debug and profile...
Paolo Carlini [Thu, 25 Apr 2013 09:42:46 +0000 (09:42 +0000)]
re PR libstdc++/57065 (incorrect default allocator template for debug and profile unordered_map)

2013-04-25  Paolo Carlini  <paolo.carlini@oracle.com>

PR libstdc++/57065
* include/debug/unordered_map (unordered_map, unordered_multimap):
Fix default allocator type.
* include/profile/unordered_map (unordered_map, unordered_multimap):
Likewise.
* include/bits/unordered_map.h: Fix comments.

From-SVN: r198279

11 years ago[AArch64] Fix printf format warning in aarch64_print_operand
James Greenhalgh [Thu, 25 Apr 2013 09:14:27 +0000 (09:14 +0000)]
[AArch64] Fix printf format warning in aarch64_print_operand

gcc/
* config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
format specifier in 'X' case.

From-SVN: r198278

11 years ago[multiple changes]
Arnaud Charlet [Thu, 25 Apr 2013 08:35:16 +0000 (10:35 +0200)]
[multiple changes]

2013-04-25  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Minor fix to Loop_Variant doc (Loop_Entry allowed).
* s-tarest.adb: Minor reformatting.

2013-04-25  Hristian Kirtchev  <kirtchev@adacore.com>

* aspects.ads, aspects.adb: Remove aspect Ghost from all relevant
tables.
* einfo.adb: Remove with and use clause for Aspects.
(Is_Ghost_Function): Removed.
(Is_Ghost_Entity): New routine.
(Is_Ghost_Subprogram): New routine.
* einfo.ads: Remove synthesized attribute Is_Ghost_Function
along with its uses in entities.  Add synthesized attributes
Is_Ghost_Entity and Is_Ghost_Subprogram along with uses in related
entities.
(Is_Ghost_Function): Removed.
(Is_Ghost_Entity): New routine.
(Is_Ghost_Subprogram): New routine.
* par-prag.adb: Remove pragma Ghost from the processing machinery.
* repinfo.adb (List_Mechanisms): Add a value for convention Ghost.
* sem_attr.adb (Analyze_Access_Attribute): Update the check
for ghost subprograms.
* sem_ch4.adb (Analyze_Call): Update the check for calls
to ghost subprograms.
(Check_Ghost_Function_Call): Removed.
(Check_Ghost_Subprogram_Call): New routine.
* sem_ch6.adb (Check_Convention): Rewritten.
(Check_Overriding_Indicator): Remove the check for overriding
ghost functions.
(Convention_Of): New routine.
* sem_ch12.adb (Preanalyze_Actuals): Update the check for ghost
generic actual subprograms.
* sem_mech.adb (Set_Mechanisms): Add an entry for convention Ghost.
* sem_prag.adb: Remove the value for pragma Ghost from
table Sig_Flags.
(Analyze_Pragma): Remove the processing for pragma Ghost.
(Process_Convention): Emit an error when a ghost
subprogram attempts to override.
(Set_Convention_From_Pragma): Emit an error when a ghost subprogram
attempts to override.
* sinfo.ads: Clarify the usage of field Label_Construct.
* snames.adb-tmpl (Get_Convention_Id): Add an entry for
predefined name Ghost.
(Get_Convention_Name): Add an entry for convention Ghost.
* snames.ads-tmpl: Move predefined name Ghost to the sublist
denoting conventions. Add convention id Ghost. Remove pragma
id Ghost.

2013-04-25  Ed Schonberg  <schonberg@adacore.com>

* sem_ch7.adb (Swap_Private_Dependents): Do no recurse on child
units if within a generic hierarchy.

From-SVN: r198275

11 years agore PR target/57052 (missed optimization with rotate and mask)
Alan Modra [Thu, 25 Apr 2013 01:00:13 +0000 (10:30 +0930)]
re PR target/57052 (missed optimization with rotate and mask)

PR target/57052
* config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
(rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
Repeat for many other rotate/shift and mask patterns using subregs.
Name lshiftrt insns.
(ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
on WORDS_BIG_ENDIAN.

From-SVN: r198274

11 years agoconfig.host: Match little-endian powerpc-linux.
Alan Modra [Thu, 25 Apr 2013 00:57:16 +0000 (10:27 +0930)]
config.host: Match little-endian powerpc-linux.

libgcc/
* config.host: Match little-endian powerpc-linux.
gcc/
* config.gcc: Support little-endian powerpc-linux targets.
* config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
(LINK_OS_LINUX_SPEC): Define.
* config/rs6000/linuxspe.h (TARGET_DEFAULT):
Preserve MASK_LITTLE_ENDIAN.
* config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
* config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
* config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
(LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
(LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
* config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
Correct fp word order for little-endian.  Don't shift toc entries
smaller than a word for little-endian.
* config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
(bswapdi2 splits): Correct low-part subreg for little-endian.
Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
low/high where such is correct only for be.
* config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
little-endian for -mcall-aixdesc.

From-SVN: r198273

11 years agors6000.c (rs6000_secondary_reload_inner): Use replace_equiv_address_nv.
Alan Modra [Thu, 25 Apr 2013 00:50:57 +0000 (10:20 +0930)]
rs6000.c (rs6000_secondary_reload_inner): Use replace_equiv_address_nv.

* config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
replace_equiv_address_nv.

From-SVN: r198272

11 years agors6000.c (rs6000_emit_set_long_const): Tidy.
Alan Modra [Thu, 25 Apr 2013 00:49:08 +0000 (10:19 +0930)]
rs6000.c (rs6000_emit_set_long_const): Tidy.

* config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.

From-SVN: r198271

11 years agoDaily bump.
GCC Administrator [Thu, 25 Apr 2013 00:16:24 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r198270

11 years agoFix date of ChangeLog entry.
Jonathan Wakely [Wed, 24 Apr 2013 22:44:01 +0000 (23:44 +0100)]
Fix date of ChangeLog entry.

From-SVN: r198267

11 years agorevert: rtl.h (struct rtx_def): ...
Vladimir Makarov [Wed, 24 Apr 2013 22:33:06 +0000 (22:33 +0000)]
revert: rtl.h (struct rtx_def): ...

2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>

Revert:
2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
* rtl.h (struct rtx_def): ...

From-SVN: r198266

11 years agore PR libstdc++/56905 ([C++11][DR 1130] std::copy_exception should be removed or...
Jonathan Wakely [Wed, 24 Apr 2013 22:00:16 +0000 (22:00 +0000)]
re PR libstdc++/56905 ([C++11][DR 1130] std::copy_exception should be removed or no longer be used)

PR libstdc++/56905
* libsupc++/exception_ptr.h (copy_exception): Deprecate and
move implementation to make_exception_ptr.
* include/std/future (_State_base::_M_break_promise): Replace
copy_exception with make_exception_ptr.
* testsuite/18_support/exception_ptr/move.cc: Likewise.
* testsuite/18_support/exception_ptr/rethrow_exception.cc: Likewise.
* testsuite/30_threads/future/members/get2.cc: Likewise.
* testsuite/30_threads/promise/members/set_exception.cc: Likewise.
* testsuite/30_threads/promise/members/set_exception2.cc: Likewise.
* testsuite/30_threads/promise/members/set_value2.cc: Likewise.
* testsuite/30_threads/shared_future/members/get2.cc: Likewise.

From-SVN: r198265

11 years agore PR rtl-optimization/57046 (wrong code generated by gcc 4.8.0 on i686)
Vladimir Makarov [Wed, 24 Apr 2013 20:27:33 +0000 (20:27 +0000)]
re PR rtl-optimization/57046 (wrong code generated by gcc 4.8.0 on i686)

2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>

PR rtl-optimizations/57046
* lra-constraints (split_reg): Set up lra_risky_transformations_p
for multi-reg splits.

2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>

PR rtl-optimizations/57046
* gcc.target/i386/pr57046.c: New test.

From-SVN: r198263

11 years agoHandle -mx32 in x86-64 ASM_SPEC
H.J. Lu [Wed, 24 Apr 2013 20:07:50 +0000 (20:07 +0000)]
Handle -mx32 in x86-64 ASM_SPEC

* config/i386/x86-64.h (ASM_SPEC): Support -mx32.

From-SVN: r198262

11 years agocpplib.h (enum c_lang): Add CLK_GNUCXX1Y and CLK_CXX1Y.
Paolo Carlini [Wed, 24 Apr 2013 19:33:54 +0000 (19:33 +0000)]
cpplib.h (enum c_lang): Add CLK_GNUCXX1Y and CLK_CXX1Y.

/libcpp
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* include/cpplib.h (enum c_lang): Add CLK_GNUCXX1Y and CLK_CXX1Y.
* init.c (lang_defaults): Add defaults for the latter.
(cpp_init_builtins): Define __cplusplus as 201300L for the latter.
* lex.c (_cpp_lex_direct): Update.

/gcc/c-family
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* c-opts.c (set_std_cxx11): Use CLK_CXX1Y and CLK_GNUCXX1Y.

/gcc/testsuite
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* g++.dg/cpp1y/cplusplus.C: New.

From-SVN: r198261

11 years ago* invoke.texi: Update expected date for C++1y.
Jason Merrill [Wed, 24 Apr 2013 18:23:51 +0000 (14:23 -0400)]
* invoke.texi: Update expected date for C++1y.

From-SVN: r198260

11 years ago2013-04-24 Sterling Augustine <saugustine@google.com>
Sterling Augustine [Wed, 24 Apr 2013 17:19:47 +0000 (17:19 +0000)]
2013-04-24  Sterling Augustine  <saugustine@google.com>

* dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
(comp_dir_string, debug_str_dwo_section): New.
(DEBUG_STR_DWO_SECTION): Rename to ...
(DEBUG_DWO_STR_SECTION): ... this.
(DEBUG_NORM_STR_SECTION): Delete.
(DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
(DEBUG_STR_DWO_SECTION_FLAGS): New.
(find_AT_string): Move most logic to ...
(find_AT_string_in_table): ... here.  New.
(add_top_level_skeleton_die_attrs): Call comp_dir_string and
add_skeleton_AT_string.  Delete logic.
(output_skeleton_debug_sections): Remove call to
add_top_level_skeleton_die_attrs.
(add_comp_dir_attribute): Move logic to comp_dir_string.
(dwarf2out_init): Initialize debug_str_dwo_section.
(output_indirect_string): Call find_string_form.
(output_indirect_strings): Rewrite.
(prune_unused_types): Empty skeleton_debug_str_hash.
Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
(dwarf2out_finish):  Call output_indirect_strings.

From-SVN: r198259

11 years agocpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
Paolo Carlini [Wed, 24 Apr 2013 17:09:22 +0000 (17:09 +0000)]
cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.

2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.

/c-family
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* c-cppbuiltin.c (c_cpp_builtins): Do not define
__GXX_EXPERIMENTAL_CXX1Y__.

/testsuite
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* g++.dg/cpp1y/cxx1y_macro.C: Remove.

From-SVN: r198258

11 years agore PR c++/53721 ([C++11] "this" not allowed in trailing return type)
Jason Merrill [Wed, 24 Apr 2013 16:46:53 +0000 (12:46 -0400)]
re PR c++/53721 ([C++11] "this" not allowed in trailing return type)

PR c++/53721
* parser.c (cp_parser_postfix_dot_deref_expression): Fix thinko.

From-SVN: r198257

11 years agoc.opt ([Wpointer-arith]): Enabled by -Wpedantic, as documented.
Paolo Carlini [Wed, 24 Apr 2013 16:35:17 +0000 (16:35 +0000)]
c.opt ([Wpointer-arith]): Enabled by -Wpedantic, as documented.

/c-family
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
            Manuel Lopez-Ibanez  <manu@gcc.gnu.org>

* c.opt ([Wpointer-arith]): Enabled by -Wpedantic, as documented.
* c-common.c (pointer_int_sum): Change -Wpointer-arith pedwarns
to simply use OPT_Wpointer_arith.
(c_sizeof_or_alignof_type): Likewise.

/cp
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* typeck.c (cxx_sizeof_or_alignof_type): Change -Wpointer-arith
pedwarn to simply use OPT_Wpointer_arith.
(cp_build_unary_op): Likewise.

/c
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* c-typeck.c (pointer_diff): Change -Wpointer-arith pedwarns
to simply use OPT_Wpointer_arith.
(build_unary_op): Likewise.

/testsuite
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* c-c++-common/Wpointer-arith-1.c: New.

Co-Authored-By: Manuel López-Ibáñez <manu@gcc.gnu.org>
From-SVN: r198256

11 years agortl.h (struct rtx_def): Add comment for field jump.
Vladimir Makarov [Wed, 24 Apr 2013 15:45:42 +0000 (15:45 +0000)]
rtl.h (struct rtx_def): Add comment for field jump.

2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>

        * rtl.h (struct rtx_def): Add comment for field jump.
(LRA_SUBREG_P): New macro.
* recog.c (register_operand): Check LRA_SUBREG_P.
* lra.c (lra): Add note at the end of RTL code. Align non-empty
stack frame.
* lra-spills.c (lra_spill): Align stack after spilling pseudos.
(lra_final_code_change): Skip subreg change for operators.
* lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
if there are no operand changes.
* lra-constraints.c (curr_insn_set): New.
(match_reload): Set LRA_SUBREG_P.
(emit_spill_move): Ditto.
(check_and_process_move): Use curr_insn_set. Process only single
set insns.  Don't initialize sec_mem_p and change_p.
(simplify_operand_subreg): Use LRA_SUBREG_P.
(reg_in_class_p): New function.
(process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
of #ifdef.  Add code to remove cycling.
(process_address): Check EXTRA_CONSTRAINT_STR. Process even if
non-null disp.  Reload inner instead of disp when base and index
are null.  Try to put lo_sum into register.
(EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
        (check_and_process_move): Move code for move cost check to
simple_move_p.  Remove equiv_substitution.
        (simple_move_p): New function.
(curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
curr_insn_set.  Call check_and_process_move only for single set
insns.  Use the new function.  Move call of check_and_process_move
after operand equiv substitution and address process.

From-SVN: r198252

11 years ago[AArch64] Fix vld1<q>_* asm constraints in arm_neon.h
James Greenhalgh [Wed, 24 Apr 2013 15:18:38 +0000 (15:18 +0000)]
[AArch64] Fix vld1<q>_* asm constraints in arm_neon.h

gcc/
* config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
(vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
(vld1<q>_<sufp><8, 16, 32, 64>): Likewise.

From-SVN: r198250

11 years agocpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
Paolo Carlini [Wed, 24 Apr 2013 15:12:50 +0000 (15:12 +0000)]
cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.

2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.

/c-family
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* c-cppbuiltin.c (c_cpp_builtins): Define __GXX_EXPERIMENTAL_CXX1Y__.

/testsuite
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

* g++.dg/cpp1y/cxx1y_macro.C: New.

From-SVN: r198249

11 years ago* libsupc++/initializer_list: Fix typos.
Jason Merrill [Wed, 24 Apr 2013 15:04:33 +0000 (11:04 -0400)]
* libsupc++/initializer_list: Fix typos.

From-SVN: r198248

11 years agoN3648: init-captures are named.
Jason Merrill [Wed, 24 Apr 2013 15:03:58 +0000 (11:03 -0400)]
N3648: init-captures are named.

* semantics.c (add_capture): Don't prepend "__" to init-captures.
(build_capture_proxy): Adjust.
* error.c (dump_simple_decl): Check DECL_NORMAL_CAPTURE_P.

From-SVN: r198247

11 years agoN3648: Allow braced and parenthesized initializers.
Jason Merrill [Wed, 24 Apr 2013 15:03:51 +0000 (11:03 -0400)]
N3648: Allow braced and parenthesized initializers.

* parser.c (cp_parser_lambda_introducer): Use cp_parser_initializer.
* pt.c (tsubst) [DECLTYPE_TYPE]: Handle DECLTYPE_FOR_INIT_CAPTURE.
* semantics.c (lambda_capture_field_type): Use do_auto_deduction.
(add_capture): Collapse a parenthesized initializer into a single
expression.
* cp-tree.h (DECLTYPE_FOR_INIT_CAPTURE): New.

From-SVN: r198246

11 years agoUpdate dependencies.
Arnaud Charlet [Wed, 24 Apr 2013 14:58:18 +0000 (16:58 +0200)]
Update dependencies.

From-SVN: r198245

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:56:49 +0000 (16:56 +0200)]
[multiple changes]

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch6.adb (Expand_Actuals): Add a predicate check on an
actual the related type has a predicate function.
* sem_ch3.adb (Constant_Redeclaration): Ensure that the related
type has an invariant procedure before building a call to it.
* sem_ch6.adb (Append_Enabled_Item): New routine.
(Check_Access_Invariants): Use routine
Append_Enabled_Item to chain onto the list of postconditions.
(Contains_Enabled_Pragmas): Removed.
(Expand_Contract_Cases): Use routine Append_Enabled_Item to chain onto
the list of postconditions.
(Invariants_Or_Predicates_Present): Removed.
(Process_PPCs): Partially reimplemented.

2013-04-24  Sergey Rybin  <rybin@adacore.com frybin>

* tree_io.ads: Update ASIS_Version_Number because of changes
in the way how entities are chained in a scope by means of
Next_Entity link.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* exp_ch13.adb (Expand_N_Attribute_Definition_Clause, case
Storage_Size): If the clause is not from an aspect, insert
assignment to size variable of task type at the point of the
clause, not after the task definition, to prevent access before
elaboration in the back-end.

2013-04-24  Yannick Moy  <moy@adacore.com>

* sem_prag.adb (Sig_Flags): Set correct value for Pragma_Assume.

2013-04-24  Yannick Moy  <moy@adacore.com>

* gnat_rm.texi: Document 'Loop_Entry.

2013-04-24  Jose Ruiz  <ruiz@adacore.com>

* s-tassta.adb, s-tarest.adb (Task_Wrapper): Start looking for
fall-back termination handlers from the parents, because they apply
only to dependent tasks.
* s-solita.adb (Task_Termination_Handler_T): Do not look for fall-back
termination handlers because the environment task has no parent,
and if it defines one of these handlers it does not apply to
itself because they apply only to dependent tasks.

From-SVN: r198244

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:49:08 +0000 (16:49 +0200)]
[multiple changes]

2013-04-24  Robert Dewar  <dewar@adacore.com>

* sem_type.adb, exp_attr.adb, exp_ch4.adb: Minor reformatting.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Document 'Update attribute.
* sem_attr.adb (Analyze_Attribute, case Update): Remove call
to S14_Attribute (S14_Attribute): removed.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* interfac.ads: Add size clauses for IEEE_Float_32/64

2013-04-24  Claire Dross  <dross@adacore.com>

* gnat1drv.adb (Adjust_Global_Switches): Remove
special assignment of Use_Expression_With_Actions for SPARK_Mode.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* checks.adb (Apply_Predicate_Check): Check for the presence
of the dynamic predicate aspect when trying to determine if the
predicate of a type is non-static.
* sem_ch5.adb (Analyze_Loop_Parameter_Specification): Check
for the presence of the dynamic predicate aspect when trying to
determine if the predicate of a type is non- static.
* sem_ch13.adb (Add_Call): Capture the nature of the
inherited ancestor predicate.
(Build_Predicate_Functions): Update comments. Rewrite the checks on
static predicate application. Complain about the form of a non-static
expression only when the type is static.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_prag.adb: Add guard to tree traversal.

2013-04-24  Vincent Celier  <celier@adacore.com>

* clean.adb (Clean): Remove local variable Root_Environment,
use Makeutl.Root_Environment instead.
* gnatcmd.adb: Remove local variable Root_Environment, use
Makeutl.Root_Environment instead.
* make.adb (Gnatmake): Remove local variable Root_Environment,
use Makeutl.Root_Environment instead.
* prj-makr.adb: Remove local variable Root_Environment, use
Makeutl.Root_Environment instead.

From-SVN: r198243

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:44:32 +0000 (16:44 +0200)]
[multiple changes]

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_attr.adb (Expand_Loop_Entry_Attribute): Clarify the
extraction of the declarative part of the conditional block. Move
the processing of simple infinite loops to the start of the
expansion logic. Correct the check which determines whether the
proper scope is installed in visibility.
* sem_attr.adb (Analyze_Attribute): Add local variable Attr
to keep track of the attribute in case the enclosing indexed
component has to be rewritten. When searching for the enclosing
loop, start from the proper attribute reference in case of a
rewriting. Do not allow for 'Loop_Entry to appear in pragma
Assert. Replace loop variable J with Index. Set the type of the
proper attribute.
* sem_ch5.adb (Check_Unreachable_Code): Detect a specialized
block that services a loop statement subject to at least one
'Loop_Entry attribute.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_type.adb (Disambiguate): In Ada 2012 mode, when trying to
resolve a fixed point operation, use first subtype to determine
whether type and operator are declared in the same list of
declarations.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* par-ch6.adb (P_Subprogram): Detect an illegal
placement of the aspect specification list in the context of
expression functions.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* exp_ch4.adb (Expand_N_Allocator):  If the designated object
has tasks, and the pointer type is an itype that has no master
id, create a master renaming in the current context, which can
only be an init_proc.

From-SVN: r198242

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:41:04 +0000 (16:41 +0200)]
[multiple changes]

2013-04-24  Robert Dewar  <dewar@adacore.com>

* sem_ch3.adb, sem_ch7.adb: Minor reformatting.
* gnat_rm.texi: Document pragma Loop_Invariant.
* sem_attr.adb (Analyze_Attribute, case Loop_Entry): This is
no longer an S14_Attribute.
* sem_prag.adb (Analyze_Pragma, case Loop_Invariant): Combine
processing with Assert, allow message parameter, remove call
to S14_Pragma.

2013-04-24  Thomas Quinot  <quinot@adacore.com>

* exp_ch4.adb: Minor reformatting.

From-SVN: r198241

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:38:50 +0000 (16:38 +0200)]
[multiple changes]

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch7.adb (Swap_Private_Dependents): New internal routine
to Install_Private_Declarations, to make the installation of
private dependents recursive in the presence of child units.
* sem_ch3.adb (Build_Discriminated_Subtype): Initialize properly
the Private_Dependents of a private subtype.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_attr.adb (Expand_Loop_Entry_Attribute): Update the
retrieval of the block declarations.
* par-ch4.adb (P_Name): Let the name parsing machinery create
a sequence of nested indexed components for attribute Loop_Entry.
* sem_attr.adb (Analyze_Attribute): Add local constant
Context. Reimplement part of the analysis of attribute Loop_Entry.
(Convert_To_Indexed_Component): Removed.
* sem_ch4.adb (Analyze_Indexed_Component_Form): Do not analyze
an indexed component after it has been rewritten into attribute
Loop_Entry.

From-SVN: r198240

11 years agosnames.ads-tmpl: Minor change to list Loop_(In)variant not in configuration pragma.
Yannick Moy [Wed, 24 Apr 2013 14:37:37 +0000 (14:37 +0000)]
snames.ads-tmpl: Minor change to list Loop_(In)variant not in configuration pragma.

2013-04-24  Yannick Moy  <moy@adacore.com>

* snames.ads-tmpl: Minor change to list
Loop_(In)variant not in configuration pragma.
* sem_ch3.adb (Analyze_Declarations): Do not look at the original node
for analyzing the expressions in pre/postconditions.

From-SVN: r198239

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:35:42 +0000 (16:35 +0200)]
[multiple changes]

2013-04-24  Robert Dewar  <dewar@adacore.com>

* gnatcmd.adb, xref_lib.adb, gnatls.adb, sem_ch13.adb: Minor
reformatting.

2013-04-24  Yannick Moy  <moy@adacore.com>

* sem_ch6.adb (Analyze_Generic_Subprogram_Body,
Analyze_Subprogram_Body_Helper): Reset contract node to Empty
before setting entity to E_Subprogram_Body.
* sem_ch8.adb (Analyze_Subprogram_Renaming): Reset contract node to
Empty before setting entity to E_Subprogram_Body.

From-SVN: r198238

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:33:20 +0000 (16:33 +0200)]
[multiple changes]

2013-04-24  Vincent Celier  <celier@adacore.com>

* gnat_ugn.texi: Document new gnatls switch -aPdir.
* gnatcmd.adb: Pass switch -aP<dir> to gnatls.
* gnatls.adb (Scan_Ls_Arg): Process new switch -aP<dir>. Issue
a warning for unknown switches.
(Usage): Add line for new switch -aPdir.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_util.adb, sem_util.ads (Is_Limited_Class_Wide_Type): Return true
if the type comes from a limited view, so that task attributes can be
constructed.

2013-04-24  Yannick Moy  <moy@adacore.com>

* checks.adb (Apply_Float_Conversion_Check): Do not apply checks if
full expansion is not enabled.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch6.adb (Create_Extra_Formals): In Ada 2012, create extra
formals if the type does not yet have a completion, and thus
has no underlying view.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb (Analyze_Aspect_Specifications): Treat an aspect
specification for Address as a reference, to suppress warnings
on entities that may be read by an external device.

From-SVN: r198237

11 years ago2013-04-24 Sergey Rybin <rybin@adacore.com frybin>
Arnaud Charlet [Wed, 24 Apr 2013 14:22:24 +0000 (16:22 +0200)]
2013-04-24  Sergey Rybin  <rybin@adacore.com frybin>

* gnat_ugn.texi: Add description of '--help' and '--version'
options for ASIS tools: gnatelim, gnatmetric, gnatstub, gnatpp.

2013-04-24  Arnaud Charlet  <charlet@adacore.com>

* gnat_rm.texi: Minor syntax fix.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_attr.adb (Expand_Loop_Entry_Attribute): Add extra comments on
what and why is being analyzed. Remove the decoration of renamings as
this simply falls out of the general analysis mechanism.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_res.adb (Explain_Redundancy): New routine.
(Resolve_Equality_Op): Place the error concerning a redundant
comparison to True at the "=". Try to explain the nature of the
redundant True.

2013-04-24  Javier Miranda  <miranda@adacore.com>

* checks.adb, exp_ch6.adb (Install_Null_Excluding_Check): No
check in interface thunks since it is performed at the caller
side.
(Expand_Simple_Function_Return): No accessibility check
needed in thunks since the check is done by the target routine.

2013-04-24  Vincent Celier  <celier@adacore.com>

* xref_lib.adb (Add_Entity): Use the canonical file names
so that source file names with capital letters are found on
platforms where file names are case insensitive.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* par-ch4.adb (P_Name): Continue to parse the name extension when the
construct is attribute Loop_Entry. Do not convert the attribute
reference into an indexed component when there is at least one
expression / range following 'Loop_Entry.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch6.adb (Contains_Enabled_Pragmas): New routine.
(Process_PPCs): Generate procedure _Postconditions
only when the context has invariants or predicates or enabled
aspects/pragmas.

From-SVN: r198236

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 14:18:30 +0000 (16:18 +0200)]
[multiple changes]

2013-04-24  Thomas Quinot  <quinot@adacore.com>

* g-socket.adb (Host_Entry): Introduce intermediate copy of
memory location pointed to by Hostent_H_Addr, as it might not
have sufficient alignment.

2013-04-24  Yannick Moy  <moy@adacore.com>

* repinfo.adb (List_Rep_Info): Set the value of Unit_Casing before
calling subprograms which may read it.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* einfo.adb: Remove Loop_Entry_Attributes from the usage of
nodes. Flag 260 is now used.
(Has_Loop_Entry_Attributes): New routine.
(Loop_Entry_Attributes): Removed.
(Set_Has_Loop_Entry_Attributes): New routine.
(Set_Loop_Entry_Attributes): Removed.
(Write_Entity_Flags): Write out Flag 260.
(Write_Field10_Name): Remove the output for Loop_Entry_Attributes.
* einfo.ads: Remove attribute Loop_Entry_Attributes,
its related comment and uses in nodes. Add new attribute
Has_Loop_Entry_Attributes, related comment and uses in loop nodes.
(Has_Loop_Entry_Attributes): New routine and pragma Inline.
(Loop_Entry_Attributes): Removed along with pragma Inline.
(Set_Has_Loop_Entry_Attributes): New routine and pragma Inline.
(Set_Loop_Entry_Attributes): Removed along with pragma Inline.
* exp_attr.adb (Expand_Loop_Entry_Attribute): New routine.
(Expand_N_Attribute_Reference): Expand attribute 'Loop_Entry.
* exp_ch5.adb: Remove with and use clause for Elists.
(Expand_Loop_Entry_Attributes): Removed.
(Expand_N_Loop_Statement): Add local variable Stmt. Rename local
constant Isc to Scheme. When a loop is subject to attribute
'Loop_Entry, retrieve the nested loop from the conditional
block. Move the processing of controlled object at the end of
loop expansion.
* sem_attr.adb (Analyze_Attribute): Do not chain attribute
'Loop_Entry to its related loop.
* sem_ch5.adb (Analyze_Loop_Statement): Add local variable
Stmt. When the iteration scheme mentions attribute 'Loop_Entry,
the entire loop is rewritten into a block.  Retrieve the nested
loop in such cases to complete the analysis.
* sem_util.ads, sem_util.adb (Find_Loop_In_Conditional_Block): New
routine.
(Subject_To_Loop_Entry_Attributes): New routine.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* exp_prag.adb (Expand_Loop_Variant): Generate pragma Check
(Loop_Variant, xxx) rather than Assert (xxx).
* gnat_rm.texi: Document pragma Loop_Variant.
* sem_prag.adb (Analyze_Pragma, case Loop_Variant): Remove call
to S14_Pragma.

From-SVN: r198235

11 years agoadabkend.adb, [...]: Everything with name 'Alfa' renamed in 'SPARK'.
Yannick Moy [Wed, 24 Apr 2013 14:14:27 +0000 (14:14 +0000)]
adabkend.adb, [...]: Everything with name 'Alfa' renamed in 'SPARK'.

2013-04-24  Yannick Moy  <moy@adacore.com>

* adabkend.adb, ali-util.adb, ali.adb, debug.adb,
errout.adb, errout.ads, erroutc.adb, exp_ch3.adb, exp_ch4.adb,
exp_ch6.adb, exp_ch7.adb, exp_dbug.adb, exp_util.adb,
expander.adb, freeze.adb, gnat1drv.adb, lib-writ.adb,
lib-writ.ads, lib-xref.adb, lib-xref.ads, opt.adb, opt.ads,
restrict.adb, sem_aggr.adb, sem_attr.adb, sem_ch3.adb,
sem_ch4.adb, sem_ch5.adb, sem_ch6.adb, sem_eval.adb, sem_prag.adb,
sem_res.adb, sem_util.adb: Everything with name
'Alfa' renamed in 'SPARK'. Update comments.
Renaming of units with name 'Alfa', renamed with 'SPARK' instead.
* exp_alfa.adb: renamed exp_spark.adb.
* exp_alfa.ads: renamed exp_spark.ads.
* get_alfa.adb: renamed get_spark_xrefs.adb.
* get_alfa.ads: renamed get_spark_xrefs.ads.
* lib-xref-alfa.adb: renamed lib-xref-spark_specific.adb.
* put_alfa.adb: renamed put_spark_xrefs.adb.
* put_alfa.ads: renamed put_spark_xrefs.ads.
* alfa.adb: renamed spark_xrefs.adb.
* alfa.ads: renamed spark_xrefs.ads.
* alfa_test.adb: renamed spark_xrefs_test.adb.

From-SVN: r198234

11 years agoPrune SCEV
Marek Polacek [Wed, 24 Apr 2013 14:04:44 +0000 (14:04 +0000)]
Prune SCEV

From-SVN: r198233

11 years agore PR c++/56970 (SFINAE does not apply correctly to sizeof.)
Paolo Carlini [Wed, 24 Apr 2013 13:58:11 +0000 (13:58 +0000)]
re PR c++/56970 (SFINAE does not apply correctly to sizeof.)

/cp
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56970
* init.c (build_offset_ref): Add tsubst_flags_t parameter.
* semantics.c (finish_qualified_id_expr): Likewise.
(finish_id_expression): Update.
* typeck.c (cp_build_addr_expr_1): Likewise.
* pt.c (tsubst_qualified_id, resolve_nondeduced_context): Likewise.
* cp-tree.h: Update declarations.

/testsuite
2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56970
* g++.dg/cpp0x/sfinae45.C: New.

From-SVN: r198232

11 years agognat_rm.texi: Document pragma Assume.
Robert Dewar [Wed, 24 Apr 2013 13:19:24 +0000 (13:19 +0000)]
gnat_rm.texi: Document pragma Assume.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Document pragma Assume.
* sem_prag.adb (Analyze_Pragma, case Assume): Now processed as
part of Assert, and no longer requires -gnatd.F

From-SVN: r198231

11 years agognat_rm.texi: Document pragma Assert_And_Cut.
Robert Dewar [Wed, 24 Apr 2013 13:18:21 +0000 (13:18 +0000)]
gnat_rm.texi: Document pragma Assert_And_Cut.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Document pragma Assert_And_Cut.
* sem_prag.adb (Analyze_Pragma, case Assert_And_Cut): Remove
S14_Pragma call.

From-SVN: r198230

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 13:17:28 +0000 (15:17 +0200)]
[multiple changes]

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_aux.adb: Add guard in Available_View.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Analyze_Depends_In_Decl_Part): Use
Find_Related_Subprogram to find the associated subprogram.
(Analyze_Global_In_Decl_List): Use Find_Related_Subprogram
to find the associated subprogram.
(Analyze_Pragma): Use Find_Related_Subprogram to find the associated
subprogram.

From-SVN: r198229

11 years agomicromips.md (jraddiusp): New pattern.
Catherine Moore [Wed, 24 Apr 2013 13:16:33 +0000 (09:16 -0400)]
micromips.md (jraddiusp): New pattern.

2013-04-02  Catherine Moore  <clm@codesourcery.com>
    Chao-ying Fu <fu@mips.com>

        * config/mips/micromips.md (jraddiusp): New pattern.
        * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
        instruction if possible.

Co-Authored-By: Chao-ying Fu <fu@mips.com>
From-SVN: r198228

11 years agoexp_ch6.adb: Remove with and use clause for Sem_Prag.
Hristian Kirtchev [Wed, 24 Apr 2013 13:15:27 +0000 (13:15 +0000)]
exp_ch6.adb: Remove with and use clause for Sem_Prag.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch6.adb: Remove with and use clause for Sem_Prag.
(Freeze_Subprogram): Call Analyze_Subprogram_Contract to analyze
the contract of a subprogram.
* sem_ch3.adb: Remove with and use clause for Sem_Prag.
(Analyze_Declarations): Call Analyze_Subprogram_Contract to
analyze the contract of a subprogram.
* sem_ch6.adb (Analyze_Subprogram_Contract): New routine.
(Check_Subprogram_Contract): Removed.
* sem_ch6.ads (Analyze_Subprogram_Contract): New routine.
(Check_Subprogram_Contract): Removed.
(Expand_Contract_Cases): Add a guard against malformed contract cases.
* sem_ch13.adb (Analyze_Aspect_Specifications): Call
Decorate_Delayed_Aspect_And_Pragma to decorate aspects
Contract_Cases, Depends and Global. Reimplement the analysis of
aspect Contract_Cases.
(Decorate_Delayed_Aspect_And_Pragma): New routine.
* sem_prag.adb (Analyze_Contract_Cases_In_Decl_Part): New routine.
(Analyze_CTC_In_Decl_Part): Removed.
(Analyze_Pragma): Reimplement the analysis of pragma Contract_Cases.
(Analyze_Test_Case_In_Decl_Part): New routine.
(Find_Related_Subprogram): New routine.
(Requires_Profile_Installation): Add new formal Prag. Update
the logic to take into account the origin of the pragma.
* sem_prag.ads (Analyze_Contract_Cases_In_Decl_Part): New routine.
(Analyze_CTC_In_Decl_Part): Removed.
(Analyze_Test_Case_In_Decl_Part): New routine.

From-SVN: r198227

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 13:13:04 +0000 (15:13 +0200)]
[multiple changes]

2013-04-24  Robert Dewar  <dewar@adacore.com>

* sem_prag.adb (Process_Convention): Move Stdcall tests to
Set_Convention_From_Pragma so that they are applied to each
entry of a homonym set.
(Process_Convention): Don't try to set convention if already set.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* gnatbind.adb: Minor reformatting.

2013-04-24  Vincent Celier  <celier@adacore.com>

* clean.adb (Gnatclean): Add the default project search
directories in the project search path after scanning the
switches on the command line.
(Initialize): Do not put the default project search directories in the
project search path.
* gnatcmd.adb (GNATcmd): Add the default project search
directories in the project search path after scanning the switches
on the command line.
* make.adb (Initialize): Add the default project search
directories in the project search path after scanning the switches
on the command line.

2013-04-24  Yannick Moy  <moy@adacore.com>

* restrict.ads (Restriction_Warnings): Initialize with all False value.

2013-04-24  Robert Dewar  <dewar@adacore.com>

* checks.ads, checks.adb (Predicate_Checks_Suppressed): New function.
* exp_util.ads, exp_util.adb (Make_Predicate_Check): Check setting of
Predicate_Check.
* snames.ads-tmpl (Name_Predicate_Check): New check name.
* types.ads (Predicate_Check): New definition.
* gnat_rm.texi: Add documentation for Predicate_Check.

From-SVN: r198226

11 years agoexp_ch8.adb (Expand_N_Subprogram_Renaming_Declaration): If this is a renaming of...
Ed Schonberg [Wed, 24 Apr 2013 13:10:01 +0000 (13:10 +0000)]
exp_ch8.adb (Expand_N_Subprogram_Renaming_Declaration): If this is a renaming of predefined equality for an untagged record...

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* exp_ch8.adb (Expand_N_Subprogram_Renaming_Declaration): If this
is a renaming of predefined equality for an untagged record,
add generated body to the freeze actions for the subprogram, to
prevent freezing issues when the record has incomplete components.
* exp_ch4.adb (Expand_Composite_Equality): If the type is a type
without completion, return a predefined comparison instead of
just False.  This may happen when building the expression for
record equality, when some component has a type whose completion
has not been seen yet. The operation will be analyzed an expanded
after the type has been frozen, at which point all component
types will have been completed, or an error reported.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb (Analyze_Aspect_Specifications): Do not delay
analysis of a Convention aspect.

From-SVN: r198225

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 13:08:31 +0000 (15:08 +0200)]
[multiple changes]

2013-04-24  Eric Botcazou  <ebotcazou@adacore.com>

* fe.h (Machine_Overflows_On_Target): New macro and declaration.
(Signed_Zeros_On_Target): Likewise.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch6.adb: Add with and use clause for Sem_Prag.
(Freeze_Subprogram): Analyze all delayed aspects for a null
procedure so that they are available when analyzing the
internally-generated _Postconditions routine.
* exp_ch13.adb: Remove with and use clause for Sem_Prag.
(Expand_N_Freeze_Entity): Move the code that analyzes delayed
aspects of null procedures to exp_ch6.Freeze_Subprogram.
* sem_prag.adb (Analyze_Abstract_State): Update the check on
volatile requirements.

2013-04-24  Bob Duff  <duff@adacore.com>

* ali-util.ads (Source_Record): New component Stamp_File
to record from whence the Stamp came.
* ali-util.adb (Set_Source_Table): Set Stamp_File component.
* bcheck.adb (Check_Consistency): Print additional information in
Verbose_Mode.
* gnatbind.adb (Gnatbind): Print additional information in
Verbose_Mode.

From-SVN: r198224

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 13:04:40 +0000 (15:04 +0200)]
[multiple changes]

2013-04-24  Robert Dewar  <dewar@adacore.com>

* exp_ch13.adb, sem_prag.adb: Update comments.
* sem_ch3.adb, exp_ch9.adb, g-socket.adb, sem_ch13.adb: Minor
reformatting.

2013-04-24  Doug Rupp  <rupp@adacore.com>

* vms_data.ads (/{NO}INHIBIT-EXEC): Document new default behavior.

2013-04-24  Yannick Moy  <moy@adacore.com>

* sinfo.ads: Minor correction of typo.

From-SVN: r198223

11 years ago[multiple changes]
Arnaud Charlet [Wed, 24 Apr 2013 13:01:09 +0000 (15:01 +0200)]
[multiple changes]

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb: Create packed array only when expander is
active.

2013-04-24  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Analyze_Depends_In_Decl_Part): Install the formals only
when the context warrants it.
(Analyze_Global_In_Decl_List): Install the formals only when
the context warrants it.
(Requires_Profile_Installation): New routine.

2013-04-24  Ed Schonberg  <schonberg@adacore.com>

* exp_ch6.adb (Expand_N_Simple_Return_Statement): When the return
type is a discriminated private type that does not require use
of the secondary stack, a constrained subtype of the underlying
type is created to convey the proper object size to the backend.
If the return type is originally a private type, the return
expression is wrapped in an unchecked_conversion. If the return
expression is used subsequently in a call to the postcondition
function, this conversion must be undone to prevent a spurious
error on the analysis of that call.

From-SVN: r198222

11 years agofix comment pasto
Alan Modra [Wed, 24 Apr 2013 11:09:55 +0000 (20:39 +0930)]
fix comment pasto

From-SVN: r198221

11 years agodwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT dependency behavio...
Julian Brown [Wed, 24 Apr 2013 10:07:24 +0000 (10:07 +0000)]
dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT dependency behavior in enumeration type DIE generation.

2013-04-24  Julian Brown  <julian@codesourcery.com>
    Chung-Lin Tang  <cltang@codesourcery.com>

* dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
dependency behavior in enumeration type DIE generation. Add TODO
note to comments about future DW_FORM_sdata/udata re-work of related
code.

Co-Authored-By: Chung-Lin Tang <cltang@codesourcery.com>
From-SVN: r198219

11 years agore PR testsuite/57050 (FAIL: gcc.c-torture/execute/pr56982.c compilation, -O0)
Richard Biener [Wed, 24 Apr 2013 08:21:04 +0000 (08:21 +0000)]
re PR testsuite/57050 (FAIL: gcc.c-torture/execute/pr56982.c compilation,  -O0)

2013-04-24  Richard Biener  <rguenther@suse.de>

PR testsuite/57050
* gcc.c-torture/execute/pr56982.c: Avoid sigjmp_buf use.

From-SVN: r198218

11 years agoDaily bump.
GCC Administrator [Wed, 24 Apr 2013 00:16:37 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r198216

11 years agoThis patch extracts approved portions of the hash_table patches to the...
Lawrence Crowl [Tue, 23 Apr 2013 22:00:12 +0000 (22:00 +0000)]
This patch extracts approved portions of the hash_table patches to the...

This patch extracts approved portions of the hash_table patches to
the cxx-conversion branch for files not under gcc/config.

Update various hash tables from htab_t to hash_table.
Modify types and calls to match.

* tree-ssa-coalesce.c'coalesce_list_d.list from htab_t to hash_table.

Fold coalesce_pair_map_hash and coalesce_pair_map_eq into new
struct coalesce_pair_hasher.

Removed struct coalesce_pair_iterator, as did not meet the hash_table
iterator interface and it provided no significant code reduction.
This leads to a change in the implementation of FOR_EACH_PARTITION_PAIR.

* statistics.c'statistics_hashes

Fold hash_statistics_eq into new struct stats_counter_hasher.

* hash-table.h'hash_table

Add documentation.
Add nested class iterator and methods to hash_table.
Add FOR_EACH_HASH_TABLE_ELEMENT implemented with those iterators.
Change uses of FOR_EACH_HTAB_ELEMENT to FOR_EACH_HASH_TABLE_ELEMENT.

* tree-ssa-sccvn.c'vn_tables_s.nary

Fold vn_nary_op_hash, vn_nary_op_eq into new struct vn_nary_op_hasher.
Add typedef vn_nary_op_table_type.
Add typedef vn_nary_op_iterator_type.

* tree-ssa-sccvn.c'vn_tables_s.phis

Fold vn_phi_hash, free_phi into new struct vn_phi_hasher.
Add typedef vn_phi_table_type.
Add typedef vn_phi_iterator_type.

* tree-ssa-sccvn.c'vn_tables_s.references

Fold vn_reference_hash, vn_reference_op_eq, free_reference
  into new struct vn_reference_hasher.
Add typedef vn_reference_table_type.
Add typedef vn_reference_iterator_type.

* tree-ssa-sccvn.c'constant_value_ids

Fold vn_constant_hash, vn_constant_eq into new struct vn_constant_hasher.

* tree-into-ssa.c'var_infos

Fold var_info_hash, var_info_eq into new struct var_info_hasher.

* tree-vectorizer.h'_loop_vec_info::peeling_htab

* tree-vectorizer.h

New struct peel_info_hasher.

* tree-vect-loop.c

Update dependent calls and types to match.

* tree-vect-data-refs.c

Fold vect_peeling_hash and vect_peeling_hash_eq into struct peel_info_hasher.

* tree-ssa-reassoc.c'undistribute_ops_list::ctable

Fold oecount_hash and oecount_eq into new struct oecount_hasher.

* tree-ssa-loop-im.c'memory_accesses.refs

Fold memref_hash and memref_eq into new struct mem_ref_hasher.

Tested on x86_64.

Index: gcc/ChangeLog

2013-04-23  Lawrence Crowl  <crowl@google.com>

* Makefile.in: Update as needed below.

* hash-table.h (class hash_table):
Correct many methods with parameter types compare_type to the correct
value_type.  (Correct code was unlikely to notice the change.)
(hash_table::elements_with_deleted) New.
(class hashtable::iterator): New.
(hashtable::begin()): New.
(hashtable::end()): New.
(FOR_EACH_HASH_TABLE_ELEMENT): New.

* statistics.c (statistics_hashes):
Change type to hash_table.  Update dependent calls and types.

* tree-into-ssa.c (var_infos):
Change type to hash_table.  Update dependent calls and types.

* tree-ssa-coalesce.c (struct coalesce_list_d.list):
Change type to hash_table.  Update dependent calls and types.

* tree-ssa-loop-im.c (struct mem_ref.refs):
Change type to hash_table.  Update dependent calls and types.

* tree-ssa-reassoc.c (undistribute_ops_list::ctable):
Change type to hash_table.  Update dependent calls and types.

* tree-ssa-sccvn.c (vn_tables_s::nary):
Change type to hash_table.  Update dependent calls and types.
(vn_tables_s::phis): Likewise.
(vn_tables_s::references): Likewise.

* tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
(vn_reference_eq): Update parameter and return types.

* tree-ssa-structalias.c (pointer_equiv_class_table):
Change type to hash_table.  Update dependent calls and types.
(location_equiv_class_table): Likewise.

* tree-vect-data-refs.c: Consequential changes for making
peeling a hash_table.

* tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
(destroy_loop_vec_info): Dependent hash_table update.

* tree-vectorizer.h (peeling_htab):
Change type to hash_table.  Update dependent calls and types.

From-SVN: r198213

11 years agore PR target/55445 (Always defined __SEH__ when build from trunk)
Kai Tietz [Tue, 23 Apr 2013 19:17:29 +0000 (21:17 +0200)]
re PR target/55445 (Always defined __SEH__ when build from trunk)

PR target/55445
* raise-gcc.c (__SEH__): Additional check that
SjLj isn't active.

From-SVN: r198205

11 years agolra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p to check the register...
Shiva Chen [Tue, 23 Apr 2013 17:38:59 +0000 (17:38 +0000)]
lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p to check the register content is equal or not.

2013-04-23  Shiva Chen  <shiva0217@gmail.com>

        * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
        to check the register content is equal or not.
        * lra-constraints.c (match_reload): Use lra_assign_reg_val
        to assign register content record.
        * lra-eliminations.c (update_reg_eliminate): Use
lra_update_reg_val_offset to update register content offset.
        * lra-int.h (struct lra_reg): Add offset member.
        (lra_reg_val_equal_p): New static inline function.
        (lra_update_reg_val_offset): New static inline function.
        (lra_assign_reg_val): New static inline function.
        * lra.c (lra_create_new_reg): Use lra_assign_reg_val
        to assign register content record.
        (initialize_lra_reg_info_element): Initial offset to zero.

From-SVN: r198204

11 years agomips.md (*movhi_internal, [...]): New operands.
Catherine Moore [Tue, 23 Apr 2013 17:17:08 +0000 (13:17 -0400)]
mips.md (*movhi_internal, [...]): New operands.

2013-04-23  Catherine Moore  <clm@codesourcery.com>

* config/mips/mips.md (*movhi_internal, *movqi_internal): New
operands.  Record compression.

From-SVN: r198202

11 years agoSupport slim switch for cfg graph dump
Xinliang David Li [Tue, 23 Apr 2013 16:44:17 +0000 (16:44 +0000)]
Support slim switch for cfg graph dump

From-SVN: r198201

11 years agoMakefile.in (targ): Fix target name check.
Eric Botcazou [Tue, 23 Apr 2013 16:16:03 +0000 (16:16 +0000)]
Makefile.in (targ): Fix target name check.

2013-04-23  Eric Botcazou  <ebotcazou@adacore.com>
    Pascal Obry    <obry@adacore.com>

* gcc-interface/Makefile.in (targ): Fix target name check.
(../../gnatmake$(exeext)): Add '+' for LTO.
(../../gnatlink$(exeext)): Likewise.

Co-Authored-By: Pascal Obry <obry@adacore.com>
From-SVN: r198200

11 years agoUpdate dependencies.
Arnaud Charlet [Tue, 23 Apr 2013 16:10:13 +0000 (18:10 +0200)]
Update dependencies.

From-SVN: r198199

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 16:07:33 +0000 (18:07 +0200)]
[multiple changes]

2013-04-23  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch9.adb (Build_PPC_Wrapper): Correct the traversal of
pre- and post-conditions.
(Expand_N_Task_Type_Declaration):
Use the correct attribute to check for pre- and post-conditions.
* exp_ch13.adb (Expand_N_Freeze_Entity): Correct the traversal of
pre- and post-conditions.  Analyze delayed classification items.
* freeze.adb (Freeze_Entity): Use the correct attribute to
check for pre- and post- conditions.
* sem_ch3.adb (Analyze_Declarations): Correct the traversal
of pre- and post-conditions as well as contract- and
test-cases. Analyze delayed pragmas Depends and Global.
* sem_ch6.adb (Check_Subprogram_Contract): Use the correct
attribute to check for pre- and post-conditions, as well as
contract-cases and test-cases. (List_Inherited_Pre_Post_Aspects):
Correct the traversal of pre- and post- conditions.
(Process_Contract_Cases): Update the comment on usage. Correct
the traversal of contract-cases.
(Process_Post_Conditions): Update the comment on usage. Correct the
traversal of pre- and post-conditions.
(Process_PPCs): Correct the traversal of pre- and post-conditions.
(Spec_Postconditions): Use the correct
attribute to check for pre- and post- conditions, as well as
contract-cases and test-cases.
* sem_ch13.adb (Analyze_Aspect_Specifications): Reimplement the
actions related to aspects Depends and Global. Code refactoring
for pre- and post-conditions.
(Insert_Delayed_Pragma): New routine.
* sem_prag.adb (Add_Item): New routine.
(Analyze_Depends_In_Decl_Part): New routine.
(Analyze_Global_In_Decl_Part): New routine.
(Analyze_Pragma): Reimplement the actions related to aspects Depends and
Global. Verify that a body acts as a spec for pragma Contract_Cases.
(Chain_PPC): Use Add_Contract_Item to chain a pragma.
(Chain_CTC): Correct the traversal of contract-
and test-cases. Use Add_Contract_Item to chain a pragma.
(Chain_Contract_Cases): Correct the traversal of contract-
and test-cases. Use Add_Contract_Item to chain a pragma.
(Check_Precondition_Postcondition): Update the comment on usage.
(Check_Test_Case): Update the comment on usage.
* sem_prag.ads (Analyze_Depends_In_Decl_Part): New routine.
(Analyze_Global_In_Decl_Part): New routine.
* sem_util.ads, sem_util.adb (Add_Contract_Item): New routine.
* sinfo.adb (Classifications): New routine.
(Contract_Test_Cases): New routine.
(Pre_Post_Conditions): New routine.
(Set_Classifications): New routine.
(Set_Contract_Test_Cases): New routine.
(Set_Pre_Post_Conditions): New routine.
(Set_Spec_CTC_List): Removed.
(Set_Spec_PPC_List): Removed.
(Spec_CTC_List): Removed.
(Spec_PPC_List): Removed.
* sinfo.ads: Update the structure of N_Contruct along with all
related comments.
(Classifications): New routine and pragma Inline.
(Contract_Test_Cases): New routine and pragma Inline.
(Pre_Post_Conditions): New routine and pragma Inline.
(Set_Classifications): New routine and pragma Inline.
(Set_Contract_Test_Cases): New routine and pragma Inline.
(Set_Pre_Post_Conditions): New routine and pragma Inline.
(Set_Spec_CTC_List): Removed.
(Set_Spec_PPC_List): Removed.
(Spec_CTC_List): Removed.
(Spec_PPC_List): Removed.

2013-04-23  Doug Rupp  <rupp@adacore.com>

* init.c (GNAT$STOP) [VMS]: Bump sigargs[0] count by 2
to account for LIB$STOP not having the chance to add the PC and
PSL fields.

From-SVN: r198198

11 years agosem_ch13.adb: Minor code reorganization (remove some redundant assignments).
Robert Dewar [Tue, 23 Apr 2013 14:59:59 +0000 (14:59 +0000)]
sem_ch13.adb: Minor code reorganization (remove some redundant assignments).

2013-04-23  Robert Dewar  <dewar@adacore.com>

* sem_ch13.adb: Minor code reorganization (remove some redundant
assignments).
* sem_ch3.adb, sem_prag.adb: Minor reformatting.

From-SVN: r198197

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 14:57:33 +0000 (16:57 +0200)]
[multiple changes]

2013-04-23  Yannick Moy  <moy@adacore.com>

* einfo.ads: Minor typo fix.
* sem_ch13.adb (Build_Predicate_Functions): Reject cases where
Static_Predicate is applied to a non-scalar or non-static type.
* sem_prag.adb: Minor typo fix.

2013-04-23  Doug Rupp  <rupp@adacore.com>

* init.c (GNAT$STOP) [VMS]: New function.

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb: Add exp_pakd to context.
(Constrain_Component_Type): If the component of the parent is
packed, and the record subtype being built is already frozen,
as is the case for an itype, the component type itself will not
be frozen, and the packed array type for it must be constructed
explicitly.

2013-04-23  Thomas Quinot  <quinot@adacore.com>

* g-socket.adb, g-socket.ads (Set_Close_On_Exec): New subprogram.

From-SVN: r198196

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 14:53:51 +0000 (16:53 +0200)]
[multiple changes]

2013-04-23  Yannick Moy  <moy@adacore.com>

* err_vars.ads (Error_Msg_Qual_Level): Set variable to zero
at declaration.
* opt.ads (Multiple_Unit_Index): Set variable to zero at declaration.
* sem_util.adb (NCT_Table_Entries): Set variable to zero at declaration.
* set_targ.ads (Num_FPT_Modes): Set variable to zero at declaration.
* stylesw.adb (Save_Style_Check_Options): Protect testing the
value of Style_Check_Comments_Spacing by a previous test that
Style_Check_Comments is True.

2013-04-23  Thomas Quinot  <quinot@adacore.com>

* sem_prag.adb, sem_prag.ads (Effective_Name): Rename to
Original_Name, and move declaration to package body as this
subprogram is not used from outside. Also clarify documentation.

From-SVN: r198195

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 14:52:28 +0000 (16:52 +0200)]
[multiple changes]

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* exp_ch6.adb (Expand_N_Subprogram_Body): When compiling with
initialize_scalars, disable predicate checks on the generated
assignment to an out scalar parameter.

2013-04-23  Gary Dismukes  <dismukes@adacore.com>

* sem_ch4.adb (Analyze_Allocator): Remove error
check for "constrained in partial view" constraints entirely.

2013-04-23  Robert Dewar  <dewar@adacore.com>

* einfo.ads, sem_prag.ads: Minor reformatting.
* errout.ads: Comment update.

From-SVN: r198194

11 years agore PR middle-end/57036 (ice in update_ssa_across_abnormal_edges)
Richard Biener [Tue, 23 Apr 2013 14:36:02 +0000 (14:36 +0000)]
re PR middle-end/57036 (ice in update_ssa_across_abnormal_edges)

2013-04-23  Richard Biener  <rguenther@suse.de>

PR middle-end/57036
* tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
parameter, only add abnormal goto edges from the copied body
if the call could perform abnormal gotos.
(copy_cfg_body): Adjust.

* gcc.dg/torture/pr57036-1.c: New testcase.
* gcc.dg/torture/pr57036-2.c: Likewise.

From-SVN: r198192

11 years agoaarch64.md (*mov<mode>_aarch64): Add simd attribute.
Sofiane Naci [Tue, 23 Apr 2013 13:42:25 +0000 (13:42 +0000)]
aarch64.md (*mov<mode>_aarch64): Add simd attribute.

gcc/
* config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.

testsuite/
* gcc.target/aarch64/scalar-mov.c: New test.

From-SVN: r198191

11 years agosjlj.S: New file.
Andreas Krebbel [Tue, 23 Apr 2013 11:33:50 +0000 (11:33 +0000)]
sjlj.S: New file.

2013-04-23  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>

* config/s390/sjlj.S: New file.
* config/s390/target.h: New file.
* configure.tgt: Set options for S/390.

From-SVN: r198190

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:59:41 +0000 (11:59 +0200)]
[multiple changes]

2013-04-23  Yannick Moy  <moy@adacore.com>

* exp_ch5.adb: Minor typo.

2013-04-23  Thomas Quinot  <quinot@adacore.com>

* gnat_ugn.texi: Fix typo.

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* einfo.ads: Minor documentation clarification.

2013-04-23  Bob Duff  <duff@adacore.com>

* types.ads: Fix incorrect comment.

From-SVN: r198189

11 years agosem_aux.adb [...] (Effectively_has_Constrained_Partial_View): Rename subprogram as...
Ed Schonberg [Tue, 23 Apr 2013 09:58:23 +0000 (09:58 +0000)]
sem_aux.adb [...] (Effectively_has_Constrained_Partial_View): Rename subprogram as Object_Type_Has_Constrained_Partial_View...

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* sem_aux.adb sem_aux.ads (Effectively_has_Constrained_Partial_View):
Rename subprogram as Object_Type_Has_Constrained_Partial_View, better
description of purpose.
* checks.adb (Apply_Discriminant_Check): Use above renaming.
* sem_ch4.adb (Analyze_Allocator): Check Has_Constrained_Partial_View
of the base type, rather than using the Object_Type predicate.
* sem_attr.adb (Analyze_Attribute, case 'Access): Use above renaming.
* sem_util.adb (Is_Dependent_Component_Of_Mutable_Object): ditto.
* exp_attr.adb (Expand_N_Attribute_Reference, case 'Constrained): Ditto.
* exp_ch4.adb (Expand_N_Allocator): Ditto.

From-SVN: r198188

11 years agoexp_prag.adb (Expand_Pragma_Check): Check for Assert rather than Assertion.
Robert Dewar [Tue, 23 Apr 2013 09:56:06 +0000 (09:56 +0000)]
exp_prag.adb (Expand_Pragma_Check): Check for Assert rather than Assertion.

2013-04-23  Robert Dewar  <dewar@adacore.com>

* exp_prag.adb (Expand_Pragma_Check): Check for Assert rather
than Assertion.
* sem_prag.adb (Is_Valid_Assertion_Kind): Moved to spec
(Effective_Name): New function (Analyze_Pragma, case Check):
Disallow [Statement_]Assertions (Check_Kind): Implement
Statement_Assertions (Check_Applicable_Policy): Use Effective_Name
(Is_Valid_Assertion_Kind): Allow Statement_Assertions.
* sem_prag.ads (Is_Valid_Assertion_Kind): Moved here from body
(Effective_Name): New function.
* sem_res.adb: Minor reformatting.
* snames.ads-tmpl (Name_Statement_Assertions): New entry.
* gnat_rm.texi: Add documentation of new assertion kind
Statement_Assertions.

From-SVN: r198187

11 years agosinfo.ads, [...]: Minor reformatting and code clean up.
Robert Dewar [Tue, 23 Apr 2013 09:54:14 +0000 (09:54 +0000)]
sinfo.ads, [...]: Minor reformatting and code clean up.

2013-04-23  Robert Dewar  <dewar@adacore.com>

* sinfo.ads, einfo.adb, sem_res.adb, exp_ch6.adb, aspects.adb: Minor
reformatting and code clean up.

From-SVN: r198186

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:53:23 +0000 (11:53 +0200)]
[multiple changes]

2013-04-23  Vincent Celier  <celier@adacore.com>

* prj-part.ads, prj-conf.ads: Minor comment updates.

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* einfo.adb (Predicate_Function): For a private type, retrieve
predicate function from full view.
* aspects.adb (Find_Aspect): Ditto.
* exp_ch6.adb (Expand_Actuals): If the formal is class-wide and
the actual is a definite type, apply predicate check after call.
* sem_res.adb: Do not apply a predicate check before the call to
a generated Init_Proc.

From-SVN: r198185

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:50:31 +0000 (11:50 +0200)]
[multiple changes]

2013-04-23  Robert Dewar  <dewar@adacore.com>

* sem_ch13.adb (Analyze_Aspect_Specifications): Significant
rewrite to make sure Is_Ignore is properly captured when aspect
is declared.
* sem_ch6.adb: Minor reformatting.
* sem_prag.adb (Analyze_Pragma): Do not test policy at time of
pragma for the case of a pragma coming from an aspect (already
tested when we analyzed the aspect).

2013-04-23  Vincent Celier  <celier@adacore.com>

* prj-conf.adb (Parse_Project_And_Apply_Config): New
Boolean parameter Implicit_Project, defaulted to False. Call
Prj.Part.Parse with Implicit_Project.
* prj-conf.ads (Parse_Project_And_Apply_Config): New Boolean
parameter Implicit_Project, defaulted to False.
* prj-part.adb (Parse_Single_Project): New Boolean parameter
Implicit_Project, defaulted to False. When Implicit_Project is
True, change the Directory of the project node to the Current_Dir.
* prj-part.ads (Parse): New Boolean parameter, defaulted to False

2013-04-23  Robert Dewar  <dewar@adacore.com>

* exp_util.adb: Minor reformatting.

From-SVN: r198184

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:48:55 +0000 (11:48 +0200)]
[multiple changes]

2013-04-23  Robert Dewar  <dewar@adacore.com>

* xoscons.adb: Minor reformatting.

2013-04-23  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Check_Mode): Ensure that a
self-referential output appears in both input and output lists of
the subprogram as categorized by aspect Global.
(Check_Usage): Rename formal parameters to better illustrate their
function. Update all uses of the said formals.

2013-04-23  Thomas Quinot  <quinot@adacore.com>

* exp_util.adb, exp_util.ads (Fully_Qualified_Name_String): New
parameter Append_NUL to make NUL-termination optional.
* exp_dist.adb: Consistently use the above throughout instead of
Get_Library_Unit_Name_String.

From-SVN: r198183

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:47:13 +0000 (11:47 +0200)]
[multiple changes]

2013-04-23  Robert Dewar  <dewar@adacore.com>

* sem_util.adb, sem_res.adb, prj-tree.adb, prj-tree.ads: Minor
reformatting.

2013-04-23  Pascal Obry  <obry@adacore.com>

* xoscons.adb: Remove unused use clause, minor code clean-up.

From-SVN: r198182

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:45:55 +0000 (11:45 +0200)]
[multiple changes]

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* sem_util.ads, sem_util.adb: Code cleanup for Is_Expression_Function
(can apply to any scope entity).
* sem_res.adb (Resolve_Call):  If the call is within another
expression function it does not constitute a freeze point.

2013-04-23  Yannick Moy  <moy@adacore.com>

* exp_ch6.adb (Expand_Actuals): Test that Subp
is overloadable before testing if it's an inherited operation.

From-SVN: r198181

11 years agoa-envvar.adb, [...]: Minor reformatting.
Robert Dewar [Tue, 23 Apr 2013 09:44:10 +0000 (09:44 +0000)]
a-envvar.adb, [...]: Minor reformatting.

2013-04-23  Robert Dewar  <dewar@adacore.com>

* a-envvar.adb, a-envvar.ads, exp_util.adb, sem_ch12.adb: Minor
reformatting.

From-SVN: r198180

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:42:45 +0000 (11:42 +0200)]
[multiple changes]

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb (Analyze_Object_Declarations): Undo previous patch.
* exp_util.adb (Expand_Subtype_From_Expr): If the expression
is a source entity and the declaration is for an aliased
unconstrained array, create a new subtype so that the flag
Is_Constr_Subt_For_UN_Aliased does not pollute other entities.

2013-04-23  Hristian Kirtchev  <kirtchev@adacore.com>

* aspects.adb: Move tables Base_Aspect and Inherited_Aspect
from the spec to the body.
(Find_Aspect): Update the call to Get_Aspect_Id.
(Get_Aspect_Id): New version that takes an aspect specification.
* aspects.ads: Reorganize all aspect related tables.
(Get_Aspect_Id): New version that takes an aspect specification.
* par_sco.adb (Traverse_Aspects): Update the call to Get_Aspect_Id.
* sem_ch12.adb (Analyze_Generic_Subprogram_Declaration): Update
the call to Get_Aspect_Id.
* sem_ch13.adb (Analyze_Aspect_At_Freeze_Point): Update the
call to Get_Aspect_Id. (Analyze_Aspect_Specifications): Update
the call to Get_Aspect_Id. Update the call to Impl_Defined_Aspect.

From-SVN: r198179

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:12:14 +0000 (11:12 +0200)]
[multiple changes]

2013-04-23  Robert Dewar  <dewar@adacore.com>

* sem_prag.adb (Fix_Error): Rewrite to do more accurate job
of getting proper name in the case where pragma comes from
aspect.
* sem_ch3.adb, sinfo.ads, par-ch6.adb, exp_ch6.adb: Minor reformatting.

2013-04-23  Yannick Moy  <moy@adacore.com>

* sem_ch6.adb (Process_PPCs): Do not filter postconditions based on
applicable policy.

2013-04-23  Thomas Quinot  <quinot@adacore.com>

* par_sco.adb (Traverse_Aux_Decls): Minor code reorganization.

2013-04-23  Doug Rupp  <rupp@adacore.com>

* init.c: Move facility macros outside IN_RTS.

2013-04-23  Thomas Quinot  <quinot@adacore.com>

* freeze.adb (Freeze_Entity): For the case of a bit-packed
array time that is known at compile time to have more that
Integer'Last+1 elements, issue an error, since such arrays are
not supported.

From-SVN: r198178

11 years ago[multiple changes]
Arnaud Charlet [Tue, 23 Apr 2013 09:06:42 +0000 (11:06 +0200)]
[multiple changes]

2013-04-23  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Analyze_Dependency_Clause): Update all calls to
Analyze_Input_Output.
(Analyze_Input_List): Update all calls to Analyze_Input_Output.
(Analyze_Input_Output): Add formal parameter Self_Ref along with
comment on its usage. Update all calls to Analyze_Input_Output.
(Analyze_Pragma): Add new local variable Self_Ref to capture
the presence of a self-referential dependency clause. Update
all calls to Analyze_Input_Output.
(Check_Mode): Add formal parameter Self_Ref along with comment on its
usage. Verify the legality of a self-referential output.

2013-04-23  Ed Schonberg  <schonberg@adacore.com>

* exp_ch6.adb: Add predicate checks on by-copy parameter.

2013-04-23  Vincent Celier  <celier@adacore.com>

* a-envvar.adb, a-envvar.ads (Value): New.

From-SVN: r198177

11 years agocoretypes.h (gimple_stmt_iterator): Add struct to make compatible with C.
Andreas Schwab [Tue, 23 Apr 2013 09:05:31 +0000 (09:05 +0000)]
coretypes.h (gimple_stmt_iterator): Add struct to make compatible with C.

* coretypes.h (gimple_stmt_iterator): Add struct to make
compatible with C.

From-SVN: r198176

11 years agore PR middle-end/57026 (ice: SSA corruption)
Richard Biener [Tue, 23 Apr 2013 08:08:25 +0000 (08:08 +0000)]
re PR middle-end/57026 (ice: SSA corruption)

2013-04-23  Richard Biener  <rguenther@suse.de>

PR tree-optimization/57026
* tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
from SSA names occuring in abnormal PHI nodes.

* gcc.dg/torture/pr57026.c: New testcase.

From-SVN: r198175

11 years agoFix lto report names
Andi Kleen [Tue, 23 Apr 2013 03:10:29 +0000 (03:10 +0000)]
Fix lto report names

Some of the hash tables in lto-report are misnamed in the report.
Fix this up.

gcc/:

2013-04-22  Andi Kleen  <ak@linux.intel.com>

* lto/lto.c (print_lto_report_1): Fix LTO report names.

From-SVN: r198174

11 years agoPrint lto report at the right place
Andi Kleen [Tue, 23 Apr 2013 03:10:16 +0000 (03:10 +0000)]
Print lto report at the right place

The LTO report is currently printed when the type merging hash tables
are already destroyed, which makes them always show up as empty.
Print it earlier. Right now it's printed twice.

gcc/:

2013-04-22  Andi Kleen  <ak@linux.intel.com>

* lto/lto.c (print_lto_report_1): Declare early.
(read_cgraph_and_symbols): Call print_lto_report_1 early.

From-SVN: r198173

11 years agoAdd -flto-report-wpa
Andi Kleen [Tue, 23 Apr 2013 03:10:02 +0000 (03:10 +0000)]
Add -flto-report-wpa

-flto-report is useful, but it prints for every LTRANS pass and
is very noisy and the main problem is often in WPA only.

Add a new -flto-report-wpa option that is only printed for WPA.

gcc/:

2013-04-22  Andi Kleen  <ak@linux.intel.com>

* common.opt (-flto-report-wpa): Add.
* doc/invoke.texi (-flto-report-wpa): Add.
* lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
(lto_main): dito.

From-SVN: r198172

11 years agoImprove pointer hash function to include all bits
Andi Kleen [Tue, 23 Apr 2013 03:09:47 +0000 (03:09 +0000)]
Improve pointer hash function to include all bits

The hashtab pointer hash function is not very good. It throws most of the
bits in the pointer away.

This changes pointer_hash to use the mix code from jhash function that mixes
all the bits on the pointer and makes them dependent on each other, before doing
the modulo.

libiberty/:

2013-04-22  Andi Kleen <ak@linux.intel.com>

* hashtab.c (hash_pointer): Move to end of file and reimplement.

From-SVN: r198171