2009-04-29 Vincent Celier <celier@adacore.com>
* prj-part.adb: Minor comment update
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sem_aggr.adb (Resolve_Record_Aggregate): handle properly
box-initialized records with discriminated subcomponents that are
constrained by discriminants of enclosing components. New subsidiary
procedures Add_Discriminant_Values, Propagate_Discriminants.
2009-04-29 Arnaud Charlet <charlet@adacore.com>
* g-socket.adb: Code clean up.
From-SVN: r146976
2009-04-29 Richard Guenther <rguenther@suse.de>
PR middle-end/39937
* tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Do not
loose type conversions.
(forward_propagate_addr_expr): Fix tuplification bug. Remove
stmts only if there are no uses of its definition.
* gcc.c-torture/compile/pr39937.c: New testcase.
From-SVN: r146973
* config/bfin/bfin.h (splitting_loops): Declare.
* config/bfin/bfin-protos.h (WA_05000257, WA_05000283, WA_05000315):
Reorder bit definitions to be ascending.
(WA_LOAD_LCREGS, ENABLE_WA_LOAD_LCREGS): New macros.
* config/bfin/bfin.c (splitting_loops): New variable.
(bfin_cpus): Add WA_LOAD_LCREGS as needed.
(struct loop_info): Remove members INIT and LOOP_INIT.
(bfin_optimize_loop): Don't set them. Reorder the code that generates
the LSETUP sequence. Allow LC to be loaded from any register, but also
add a case to push/pop a PREG scratch if ENABLE_WA_LOAD_LCREGS.
(bfin_reorg_loops): When done, split all BB_ENDs with splitting_loops
set to 1.
* config/bfin/bfin.md (loop_end splitter): Use splitting_loops instead
of reload_completed.
From Jie Zhang:
* config/bfin/bfin.md (movsi_insn): Refine constraints.
From-SVN: r146971
2009-04-29 Gary Dismukes <dismukes@adacore.com>
* sem_aggr.adb (Valid_Limited_Ancestor): Add test for the name of a
function entity, to cover the case of a parameterless function call
that has not been resolved.
2009-04-29 Robert Dewar <dewar@adacore.com>
* err_vars.ads, prj-part.adb, scans.ads, exp_tss.adb: Minor
reformatting and comment updates.
From-SVN: r146970
2009-04-29 Arnaud Charlet <charlet@adacore.com>
* gnat_ugn.texi: Update some documentation about interfacing with C++
Mention -fkeep-inline-functions.
* gnat_ugn.texi: Minor edits
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sem_aggr.adb (Resolve_Record_Aggregate): When building an aggregate
for a defaulted component of an enclosing aggregate, inherit the type
from the component declaration of the enclosing type.
2009-04-29 Albert Lee <lee@adacore.com>
* g-socthi-vms.ads, g-socthi-vxworks.ads, s-oscons-tmplt.c,
g-socthi-mingw.ads, g-socthi.ads, g-socket.adb, g-sothco.ads
(System.OS_Constants): New type Msg_Iovlen_T which follows whether the
msg_iovlen field in struct msghdr is 32 or 64 bits wide.
Relocate the Msghdr record type from GNAT.Sockets.Thin to
GNAT.Sockets.Common, and use System.OS_Constants.Msg_Iovlen_T as the
type for the Msg_Iovlen field.
From-SVN: r146969
2009-04-29 Vincent Celier <celier@adacore.com>
* sinput-l.adb (Load_File): When preprocessing, set temporarily the
Source_File_Index_Table entries for the source, to avoid crash when
reporting an error.
* gnatcmd.adb (Test_If_Relative_Path): Use
Makeutl.Test_If_Relative_Path.
* makeutl.adb:(Test_If_Relative_Path): Process switches --RTS= only if
Including_RTS is True.
* makeutl.ads (Test_If_Relative_Path): New Boolean parameter
Including_RTS defaulted to False.
* sinput.ads, scans.ads, err_vars.ads: Initialize some variables with
a default value.
2009-04-29 Javier Miranda <miranda@adacore.com>
* gnat_ugn.texi: Adding documentation for non-default C++ constructors.
From-SVN: r146967
2009-04-29 Javier Miranda <miranda@adacore.com>
* sem_ch3.adb (Analyze_Object_Declaration): Disable error message
associated with dyamically tagged expressions if the expression
initializing a tagged type corresponds with a non default CPP
constructor.
(OK_For_Limited_Init): CPP constructor calls are OK for initialization
of limited type objects.
* sem_ch5.adb (Analyze_Assignment): Improve the error message reported
when a CPP constructor is called in an assignment. Disable also the
error message associated with dyamically tagged expressions if the
exporession initializing a tagged type corresponds with a non default
CPP constructor.
* sem_prag.adb (Analyze_Pragma): Remove code disabling the use of
non-default C++ constructors.
* sem_util.ads, sem_util.adb (Is_CPP_Constructor_Call): New subprogram.
* exp_tss.ads, exp_tss.adb (Base_Init_Proc): Add support for
non-default constructors.
(Init_Proc): Add support for non-default constructors.
* exp_disp.adb (Set_Default_Constructor): Removed.
(Set_CPP_Constructors): Code based in removed Set_Default_Constructor
but extending its functionality to handle non-default constructors.
* exp_aggr.adb (Build_Record_Aggr_Code): Add support for non-default
constructors. Minor code cleanup removing unrequired label and goto
statement.
* exp_ch3.adb (Build_Initialization_Call): Add support for non-default
constructors.
(Build_Init_Statements): Add support for non-default constructors.
(Expand_N_Object_Declaration): Add support for non-default constructors.
(Freeze_Record_Type): Replace call to Set_Default_Constructor by call
to Set_CPP_Constructors.
* exp_ch5.adb (Expand_N_Assignment_Statement): Add support for
non-default constructors.
Required to handle its use in build-in-place statements.
* gnat_rm.texi (CPP_Constructor): Document new extended use of this
pragma for non-default C++ constructors and the new compiler support
that allows the use of these constructors in record components, limited
aggregates, and extended return statements.
From-SVN: r146966
2009-04-29 Vincent Celier <celier@adacore.com>
* prj-part.adb (Parse_Single_Project): Do not attempt to find a
project extending an abstract project.
2009-04-29 Eric Botcazou <ebotcazou@adacore.com>
* targparm.ads: Fix oversight.
From-SVN: r146965
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* lib-xref.adb (Output_Overridden_Op): Follow several levels of
derivation when necessary, to find the user-subprogram that is actally
being overridden.
From-SVN: r146964
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* exp_ch9.ads, exp_ch9.adb (Build_Wrapper_Spec): Use source line of
primitive operation, rather than source line of synchronized type, when
building the wrapper for a primitive operation that overrides an
operation inherited from a progenitor, to improve the error message on
duplicate declarations.
* sem_ch3.adb (Process_Full_View): Use new signature of
Build_Wrapper_Spec.
From-SVN: r146961
2009-04-29 Robert Dewar <dewar@adacore.com>
* prj-nmsc.ads: Minor reformatting
2009-04-29 Eric Botcazou <ebotcazou@adacore.com>
* exp_ch4.adb (Expand_N_Conditional_Expression): Set the SLOC of the
expression on the existing parent If statement.
From-SVN: r146960
2009-04-29 Robert Dewar <dewar@adacore.com>
* prj-nmsc.ads: Minor reformatting
2009-04-29 Eric Botcazou <ebotcazou@adacore.com>
* exp_ch4.adb (Expand_N_Conditional_Expression): Set the SLOC of the
expression on the existing parent If statement.
From-SVN: r146959
2009-04-29 Eric Botcazou <ebotcazou@adacore.com>
* exp_ch4.adb (Expand_N_Conditional_Expression): Set the SLOC of an
existing parent If statement on the newly created one.
From-SVN: r146957
* config/bfin/bfin.c (bfin_register_move_cost): Test for subsets of
DREGS rather than comparing directly. Remove code that tries to
account for latencies.
From-SVN: r146956
* gnatcmd.adb, prj-proc.adb, prj-proc.ads, make.adb, prj-part.adb,
prj-part.ads, prj.adb, prj.ads, clean.adb, prj-dect.adb, prj-dect.ads,
prj-nmsc.adb, prj-nmsc.ads, prj-pars.adb, prj-pars.ads, prj-makr.adb
(Set_In_Configuration, In_Configuration): Removed.
Replaced by an extra parameter Is_Config_File in several parameter to
avoid global variables to store the state of the parser.
From-SVN: r146955
2009-04-29 Albert Lee <lee@adacore.com>
* g-socthi-vxworks.ads: Change the spec of Msghdr to match the one in
the default version of GNAT.Sockets.Thin.
* g-socthi-vms.ads: Change the spec of Msghdr to match the one in the
default version of GNAT.Sockets.Thin.
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sem_ch6.adb (Analyze_Subprogram_Specification): If the subprogram is
an overriding operation of an inherited interface operation, and the
controlling type is a synchronized type, we replace the type with its
corresponding record, to match the proper signature of an overriding
operation. The same processing must be performed for an access
parameter whose designated type is derived from a synchronized
interface.
From-SVN: r146954
* pex-win32.c (pex_win32_pipe): Add _O_NOINHERIT.
(pex_win32_exec_child): Ensure each process has only one handle open
on pipe endpoints. Close standard input after creating child for
symmetry with standard output/standard error.
From-SVN: r146953
* config/bfin/bfin.c (bfin_optimize_loop): Unify handling of
problematic last insns. Test for TYPE_CALL rather than CALL_P.
Remove special case testing for last insn of inner loops. Don't fail if
the loop ends with a jump, emit an extra nop instead.
From-SVN: r146952
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sinfo.ads, sinfo.adb: New attribute Next_Implicit_With, to chain
with_clauses generated for the same unit through rtsfind, and that
appear in the context of different units.
* rtsfind.adb: New attribute First_Implicit_With, component of the
Unit_Record that stores information about a unit loaded through rtsfind.
From-SVN: r146951
2009-04-29 Gary Dismukes <dismukes@adacore.com>
* exp_ch3.adb (Stream_Operation_OK): Return True for limited interfaces
(other conditions permitting), so that abstract stream subprograms will
be declared for them.
2009-04-29 Bob Duff <duff@adacore.com>
* g-expect.adb (Expect_Internal): Fix check for overfull buffer.
* g-expect.ads: Minor comment fixes.
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* freeze.adb, lib-xref.adb (Check_Dispatching_Operation): if the
dispatching operation is a body without previous spec, update the list
of primitive operations to ensure that cross-reference information is
up-to-date.
2009-04-29 Albert Lee <lee@adacore.com>
* g-socthi-vms.adb, g-socthi-vms.ads, g-socthi-vxworks.adb,
g-socthi-vxworks.ads, g-socthi-mingw.adb, g-socthi-mingw.ads,
g-socthi.adb, g-socthi.ads, g-socket.adb, g-socket.ads
(GNAT.Sockets.Thin.C_Readv,
GNAT.Sockets.Thin.C_Writev): Remove unused subprograms.
(GNAT.Sockets.Thin.C_Recvmsg,
GNAT.Sockets.Thin.C_Sendmsg): New bindings to call recvmsg(2) and
sendmsg(2).
(GNAT.Sockets.Receive_Vector, GNAT.Sockets.Send_Vector): Use
C_Recvmsg/C_Sendmsg rather than Readv/C_Writev.
From-SVN: r146949
2009-04-29 Richard Guenther <rguenther@suse.de>
* tree-cfg.c (verify_types_in_gimple_reference): Add require_lvalue
parameter. Allow invariants as base if !require_lvalue.
(verify_gimple_assign_single): Adjust.
From-SVN: r146947
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sem_disp.adb (Check_Dispatching_Operation): if the dispatching
operation is a body without previous spec, update the list of
primitive operations to ensure that cross-reference information is
up-to-date.
* sem_ch12.adb (Build_Instance_Compilation_Unit_Nodes): When creating a
new compilation unit node for the instance declaration, keep the
context items of the original unit on it, so that the context of the
instance body only holds the context inherited from the generic body.
2009-04-29 Bob Duff <duff@adacore.com>
* sem_res.adb: Minor comment fix.
From-SVN: r146945
2009-04-29 Thomas Quinot <quinot@adacore.com>
* sem_elim.adb: Minor reformatting
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* exp_aggr.adb (Convert_To_Positional): if the current unit is a
predefined unit, allow arbitrary number of components in static
aggregate, to ensure that the same level of constant folding applies
for Ada 95 and Ada 05 versions of the file.
From-SVN: r146944
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sem_elim.adb (Check_Eliminated): Handle new improved eliminate
information: no need for full scope check.
(Eliminate_Error): Do not emit error in a generic context.
2009-04-29 Ed Falis <falis@adacore.com>
* adaint.c (__gnat_rmdir): return error code if VTHREADS is defined.
VxWorks 653 POS does not support rmdir.
2009-04-29 Matteo Bordin <bordin@adacore.com>
* s-stausa.adb, s-stausa.ads: Get_Usage_Range: changing the way
results are printed.
From-SVN: r146943
2009-04-29 Arnaud Charlet <charlet@adacore.com>
* s-taskin.adb (Initialize): Remove pragma Warnings Off and remove
unused assignment.
2009-04-29 Thomas Quinot <quinot@adacore.com>
* make.adb: Minor reformatting.
Minor code reorganization throughout.
2009-04-29 Matteo Bordin <bordin@adacore.com>
* s-stausa.ads: Changed visibility of type Task_Result: moved to
public part to give application visibility over it.
This is for future improvement and to build a public API on top of it.
Changed record components name of type Task_Result to reflect the new
way of reporting.
* s-stausa.adb: Actual_Size_Str changed to reflect the new way of
reporting Stack usage.
* gnat_ugn.texi: Update doc of stack usage report.
* g-tastus.ads, s-stusta.ads, s-stusta.adb: New files.
* Makefile.rtl: Add new run-time files.
From-SVN: r146942
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* exp_ch7.adb (Build_Final_List): If the designated type is a Taft
Amendment type, add the with_clause for Finalization.List_Controller
only if the current context is a package body.
From-SVN: r146939
2009-04-29 Thomas Quinot <quinot@adacore.com>
* sem_res.adb (Static_Concatenation): An N_Op_Concat with static
operands is static only if it is a predefined concatenation operator.
* sem_util.adb: Minor reformatting
* sem_ch12.adb (Save_References): When propagating semantic information
from generic copy back to generic template, for the case of an
identifier that has been rewritten to an explicit dereference whose
prefix is either an object name or a parameterless funcion call
denoting a global object or function, properly capture the denoted
global entity: perform the corresponding rewriting in the template,
and point the rewritten identifier to the correct global entity (not
to the associated identifier in the generic copy).
From-SVN: r146936
2009-04-29 Robert Dewar <dewar@adacore.com>
* rtsfind.adb, prj-env.adb: Minor reformatting
Minor code reorganization
2009-04-29 Emmanuel Briot <briot@adacore.com>
* make.adb: Fix comment
* prj.adb (Ultimate_Extending_Project_Of): Fix handling when no project
is given as argument, as might happen in gnatmake.
From-SVN: r146935
2009-04-29 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Check_Abstract_Overriding): Improve error message when
an abstract operation of a progenitor is not properly overridden by an
operation of a derived synchronized type.
From-SVN: r146934
2009-04-29 Bob Duff <duff@adacore.com>
* exp_ch7.adb (Build_Final_List): For an access type that designates a
Taft Amendment type, if the access type needs finalization, make sure
the implicit with clause for List_Controller occurs on the package spec.
* rtsfind.adb (Text_IO_Kludge): Fine tune the creation of implicit
with's created for the pseudo-children of Text_IO and friends. In
particular, avoid cycles, such as Ada.Wide_Text_IO.Integer_IO and
Ada.Text_IO.Integer_IO both with-ing each other.
* sem.adb (Walk_Library_Items): Suppress assertion failure in certain
oddball cases when pragma Extend_System is used.
* sem_ch12.adb (Get_Associated_Node): Prevent direct 'with' cycles in
the case where a package spec instantiates a generic whose body with's
this package, so Walk_Library_Items won't complain about cyclic with's.
2009-04-29 Emmanuel Briot <briot@adacore.com>
* gnatcmd.adb, prj-proc.adb, make.adb, mlib-prj.adb, prj.adb, prj.ads,
prj-pp.adb, prj-pp.ads, makeutl.adb, clean.adb, prj-nmsc.adb,
mlib-tgt.adb, mlib-tgt.ads, prj-util.adb, prj-env.adb, prj-env.ads
(Project_Id): now a real pointer to Project_Data, instead of an index
into the Projects_Table. This simplifies the API significantly, avoiding
extra lookups in this table and the need to pass the Project_Tree_Ref
parameter in several cases
From-SVN: r146931
gcc/
* config/bfin/bfin.md (sp_or_sm, spm_string, spm_name): New macro.
(ss<spm_name>hi3, ss<spm_name>hi3_parts, ss<spm_name>hi3_low_parts,
ss<spm_name_hi3_high_parts): New patterns, replacing ssaddhi3, ssubhi3,
ssaddhi3_parts and sssubhi3_parts.
(flag_mulhi3_parts): Produce a HImode output rather than trying to set
a VEC_SELECT.
* config/bfin/bfin.c (bfin_expand_builtin, case BFIN_BUILTIN_CPLX_SQU):
Adjust accordingly.
gcc/testsuite/
* gcc.target/bfin/20090411-1.c: New test.
From-SVN: r146929
2009-04-28 Richard Guenther <rguenther@suse.de>
* tree-vect-loop.c (get_initial_def_for_induction): Use
correct types for pointer increment.
From-SVN: r146927
