Skip to content

R
riscv-gcc-1
Switch branch/tag
  1. 03 Mar, 2015 20 commits
  3. 02 Mar, 2015 20 commits
    • i386.md (*udivmod<mode>4_pow2): Remove reload_completed. · 29e27c0b 
      	* config/i386/i386.md (*udivmod<mode>4_pow2): Remove
	reload_completed.

From-SVN: r221127
      Aldy Hernandez committed
    • invoke.texi (Options for Code Generation Conventions): Fix URL of DSO paper. · 0ce03965 
              * doc/invoke.texi (Options for Code Generation Conventions):
        Fix URL of DSO paper.

From-SVN: r221125
      Ulrich Drepper committed
    • re PR lto/65130 (ICE with LTO on valid code on x86_64-linux-gnu) · 1af8bfe5 
      
	PR ipa/65130
	* ipa-inline.c (check_callers): Looks for recursion.
	(inline_to_all_callers): Give up on uninlinable or recursive edges.
	* ipa-inline-analysis.c (inline_summary_t::duplicate): Do not update
	summary of inline clones.
	(do_estimate_growth_1): Fix recursion check.

	* gcc.dg/lto/pr65130_0.c: New testcase.
	* gcc.dg/lto/pr65130_1.c: New testcase.

From-SVN: r221124
      Jan Hubicka committed
    • re PR ipa/64988 (ICE: verify_cgraph_node failed: Alias and target's comdat… · b91b562c 
      re PR ipa/64988 (ICE: verify_cgraph_node failed: Alias and target's comdat groups differs with -fdeclone-ctor-dtor)


	PR ipa/64988
	* ipa-inline-transform.c (clone_inlined_nodes): Do not dissolve
	comdat groups.
	* g++.dg/torture/pr64988.C: New testcase.

From-SVN: r221123
      Jan Hubicka committed
    • check.c (gfc_check_atomic): Properly check for coarrayness and for being coindexed. · bc81b5ce 
      2015-03-02  Tobias Burnus  <burnus@net-b.de>

        * check.c (gfc_check_atomic): Properly check for coarrayness
        and for being coindexed.

2015-03-02  Tobias Burnus  <burnus@net-b.de>

        * gfortran.dg/coarray_atomic_6.f90: New.

From-SVN: r221122
      Tobias Burnus committed
    • re PR lto/65276 (LTO: ICE add_type_duplicate, at ipa-devirt.c:1462) · af5513e9 
      	PR lto/65276
	* ipa-devirt.c (add_type_duplicate): Remove odr_violated assert
	when checking TYPE_BINFO.

Co-Authored-By: Aldy Hernandez <aldyh@redhat.com>

From-SVN: r221121
      Jan Hubicka committed
    • re PR libgcc/64885 (libstdc++ all_attributes failure) · 7727f8f2 
      	PR libgcc/64885
	* gthr-single.h: Use __unused__ attribute instead of unused.
	* config/gthr-vxworks.h: Likewise.
	* config/i386/gthr-win32.h: Likewise.

From-SVN: r221120
      Jonathan Wakely committed
    • re PR libstdc++/65279 (std::scoped_allocator_adaptor is not assignable) · 9d8dbe73 
      	PR libstdc++/65279
	* include/std/scoped_allocator (__inner_type_impl,
	scoped_allocator_adaptor): Add defaulted copy assignment and move
	assignment operators.
	* testsuite/20_util/scoped_allocator/65279.cc: New.

From-SVN: r221119
      Jonathan Wakely committed
    • re PR libstdc++/64367 (g++-v5/stdexcept:52:28: error: invalid use of non-static data member '_M_p') · e59a2e94 
      	PR libstdc++/64367
	* include/std/stdexcept (__sso_string): Don't use non-static member
	in sizeof.

From-SVN: r221118
      Jonathan Wakely committed
    • re PR ipa/65270 (issues with merging memory accesses from different code paths) · d366a1a7 
      2015-03-02  Richard Biener  <rguenther@suse.de>

	PR ipa/65270
	* ipa-icf-gimple.c: Include builtins.h.
	(func_checker::compare_memory_operand): Compare base alignment.

From-SVN: r221117
      Richard Biener committed
    • [multiple changes] · cd1a470a 
      2015-03-02  Robert Dewar  <dewar@adacore.com>

	* scng.adb (Scan): Ignore illegal character in relaxed
	semantics mode.

2015-03-02  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch4.adb (Analyze_Set_Membership); Retain Overloaded flag
	on left operand, so it can be properly resolved with type of
	alternatives of right operand.
	* sem_res.adb (Resolve_Set_Membership): Handle properly an
	overloaded left-hand side when the alternatives on the right
	hand side are literals of some universal type.	Use first
	non-overloaded alternative to find expected type.

2015-03-02  Ed Schonberg  <schonberg@adacore.com>

	* exp_ch7.adb (Make_Set_Finalize_Address_Call): Use underlying
	type to retrieve designated type, because the purported access
	type may be a partial (private) view, when it is declared in
	the private part of a nested package, and finalization actions
	are generated when completing compilation of enclosing unit.

From-SVN: r221116
      Arnaud Charlet committed
    • back_end.adb (Call_Back_End): Remove previous patch... · 89f0276a 
      2015-03-02  Robert Dewar  <dewar@adacore.com>

	* back_end.adb (Call_Back_End): Remove previous patch,
	the back end now gets to see the result of -gnatd.1
	(Unnest_Subprogram_Mode) processing.
	* elists.ads, elists.adb (List_Length): New function.
	* exp_unst.ads, exp_unst.adb: Major changes, first complete version.
	* sem_util.adb (Check_Nested_Access): Handle formals in
	Unnest_Subprogram_Mode.
	(Adjust_Named_Associations): Minor reformatting.
	* sprint.adb (Sprint_Node_Actual): Fix failure to print aliased
	for parameters.

From-SVN: r221115
      Robert Dewar committed
    • [multiple changes] · 3830827c 
      2015-03-02  Robert Dewar  <dewar@adacore.com>

	* atree.ads, atree.adb (Uint24): New function
	(Set_Uint24): New procedure.
	* atree.h (Uint24): New macro for field access.
	* back_end.adb (Call_Back_End): For now, don't call back end
	if unnesting subprogs.
	* einfo.adb (Activation_Record_Component): New field
	(Subps_Index): New field.
	* einfo.ads (Activation_Record_Component): New field
	(Subps_Index): New field Minor reordering of comments into alpha order.
	* exp_unst.ads, exp_unst.adb: Continued development.

2015-03-02  Gary Dismukes  <dismukes@adacore.com>

	* exp_disp.ads: Minor reformatting.

2015-03-02  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch8.adb (Chain_Use_Clause): Do not chain use clause from
	ancestor to list of use clauses active in descendant unit if we
	are within the private part of an intervening parent, to prevent
	circularities in use clause list.

From-SVN: r221114
      Arnaud Charlet committed
    • exp_ch9.adb (Build_Corresponding_Record): Propagate type invariants to the… · 024d33d8 
      exp_ch9.adb (Build_Corresponding_Record): Propagate type invariants to the corresponding record type.

2015-03-02  Javier Miranda  <miranda@adacore.com>

	* exp_ch9.adb (Build_Corresponding_Record): Propagate type
	invariants to the corresponding record type.
	* exp_disp.ad[sb] (Set_DT_Position_Value): New subprogram
	which sets the value of the DTC_Entity associated with a given
	primitive of a tagged type and propagates the value to the
	wrapped subprogram.
	(Set_DTC_Entity_Value): Propagate the DTC
	value to the wrapped entity.
	* sem_ch13.adb (Build_Invariant_Procedure): Append the code
	associated with invariants of progenitors.
	* sem_ch3.adb (Build_Derived_Record_Type): Inherit type invariants
	of parents and progenitors.
	(Process_Full_View): Check hidden inheritance of class-wide type
	invariants.
	* sem_ch7.adb (Analyze_Package_Specification): Do not generate
	the invariant procedure for interface types; build the invariant
	procedure for tagged types inheriting invariants from their
	progenitors.
	* sem_prag.adb (Pragma_Invariant) Allow invariants in interface
	types but do not build their invariant procedure since their
	invariants will be propagated to the invariant procedure of
	types covering the interface.
	* exp_ch6.adb, exp_disp.adb, sem_ch3.adb, sem_ch7.adb,
	sem_ch8.adb, sem_disp.adb: Replace all calls to Set_DT_Position
	by calls to Set_DT_Position_Value.

From-SVN: r221113
      Javier Miranda committed
    • sem_attr.adb (Analyze_Attribute): Factor out heavily indented code in Denote_Same_Function. · 7c76aa3f 
      2015-03-02  Hristian Kirtchev  <kirtchev@adacore.com>

	* sem_attr.adb (Analyze_Attribute): Factor out heavily indented
	code in Denote_Same_Function.  Do not analyze attribute 'Result
	when it is inside procedure _Postconditions.  Remove a misplaced
	warning diagnostic. Code cleanup.
	(Denote_Same_Function): New routine.
	* sem_prag.adb (Analyze_Contract_Cases_In_Decl_Part): Code
	cleanup. Warn on pre/postconditions on an inlined subprogram.
	(Analyze_Pragma, Refined_Post case): Warn on pre/postconditions on
	an inlined subprogram.
	(Analyze_Pre_Post_Condition_In_Decl_Part): Code cleanup. Warn on
	pre/post condition on an inlined subprogram.
	(Analyze_Test_Case_In_Decl_Part): Code cleanup. Warn on
	pre/postconditions on an inlined subprogram.
	(Check_Postcondition_Use_In_Inlined_Subprogram): New routine.

From-SVN: r221112
      Hristian Kirtchev committed
    • [multiple changes] · aaeb3b3a 
      2015-03-02  Hristian Kirtchev  <kirtchev@adacore.com>

	* sem_prag.adb (Ensure_Aggregate_Form):
	Ensure that the name denoted by the Chars of a pragma argument
	association has the proper Sloc when converted into an aggregate.

2015-03-02  Bob Duff  <duff@adacore.com>

	* sem_ch6.adb (Check_Private_Overriding): Capture
	Incomplete_Or_Partial_View in a constant. This is cleaner and
	more efficient.

2015-03-02  Gary Dismukes  <dismukes@adacore.com>

	* einfo.ads, exp_unst.ads: Minor reformatting.

2015-03-02  Ed Schonberg  <schonberg@adacore.com>

	* a-strsea.adb (Find_Token): Ensure that the range of iteration
	does not perform any improper character access. This prevents
	erroneous access in the unusual case of an empty string target
	and a From parameter less than Source'First.

2015-03-02  Robert Dewar  <dewar@adacore.com>

	* elists.adb (List_Length): Fix incorrect result.

From-SVN: r221111
      Arnaud Charlet committed
    • sem_ch6.adb (Check_Private_Overriding): Refine the legality checks here. · acf624f2 
      2015-03-02  Bob Duff  <duff@adacore.com>

	* sem_ch6.adb (Check_Private_Overriding): Refine the legality
	checks here. It used to check that the function is merely
	overriding SOMEthing. Now it checks that the function is
	overriding a corresponding public operation. This is a correction
	to the implementation of the rule in RM-3.9.3(10).

From-SVN: r221110
      Bob Duff committed
    • debug.adb: Document new debug flag -gnatd.1. · 5a271a7f 
      2015-03-02  Robert Dewar  <dewar@adacore.com>

	* debug.adb: Document new debug flag -gnatd.1.
	* einfo.ads, einfo.adb (Has_Nested_Subprogram): New flag.
	(Has_Uplevel_Reference): New flag (Is_Static_Type): New flag.
	(Uplevel_Reference_Noted):New flag (Uplevel_References): New field.
	* elists.ads elists.adb (List_Length): New function.
	* exp_ch6.adb (Expand_N_Subprogram_Body): Call Unnest_Subprogram
	when appropriate (Process_Preconditions): Minor code
	reorganization and reformatting
	* exp_unst.ads, exp_unst.adb: New files.
	* gnat1drv.adb (Adjust_Global_Switches): Set
	Unnest_Subprogram_Mode if -gnatd.1
	* namet.ads, namet.adb (Name_Find_Str): New version of Name_Find with
	string argument.
	* opt.ads (Unnest_Subprogram_Mode): New flag.
	* par-ch3.adb (P_Identifier_Declarations): Fixes to -gnatd.2 handling.
	* sem_ch6.adb (Analyze_Subprogram_Body_Helper): Set
	Has_Nested_Subprogram flag.
	* sem_ch8.adb (Find_Direct_Name): New calling sequence for
	Check_Nested_Access.
	(Find_Selected_Component): Minor comment addition.
	* sem_util.adb (Check_Nested_Access): New version for use with Exp_Unst.
	(Note_Possible_Modification): New calling sequence for
	Check_Nested_Access.
	* sem_util.ads (Check_Nested_Access): New version for use with Exp_Unst.
	* gcc-interface/Make-lang.in (GNAT1_OBJS): Add exp_unst.o

From-SVN: r221109
      Robert Dewar committed
    • utils.c (gnat_pushdecl): For non-artificial pointer types... · aef308d0 
      2015-03-02  Pierre-Marie de Rodat  <derodat@adacore.com>                  
                                                                          
	* gcc-interface/utils.c (gnat_pushdecl): For non-artificial pointer     
	types, preserve the original type and create copies just like the C
	front-end does.  For artificial ones, do not define a name for
	the original type.
	(create_type_decl): When gnat_pushdecl made the input type the
	original type for the new declaration, do not define a stub   
	declaration for it.
	* gcc-interface/utils2.c (build_binary_op): Accept two different        
	pointer types when they point to the same type.

From-SVN: r221108
      Pierre-Marie de Rodat committed
    • exp_util.adb (Possible_Bit_Aligned_Component): Do not process an unanalyzed node. · 184a23e9 
      2015-03-02  Hristian Kirtchev  <kirtchev@adacore.com>

	* exp_util.adb (Possible_Bit_Aligned_Component): Do not process
	an unanalyzed node.
	* sem_util.adb (Kill_Current_Values): Do not invalidate and
	de-null a constant.

From-SVN: r221107
      Hristian Kirtchev committed