gcc.git
16 years agomips.md (UNSPEC_SYNC_NEW_OP_12, [...]): New define_constants.
David Daney [Tue, 20 May 2008 23:13:13 +0000 (23:13 +0000)]
mips.md (UNSPEC_SYNC_NEW_OP_12, [...]): New define_constants.

2008-05-20  David Daney  <ddaney@avtrex.com>

* config/mips/mips.md (UNSPEC_SYNC_NEW_OP_12,
UNSPEC_SYNC_OLD_OP_12,
UNSPEC_SYNC_EXCHANGE_12): New define_constants.
(UNSPEC_SYNC_EXCHANGE, UNSPEC_MEMORY_BARRIER,
UNSPEC_SET_GOT_VERSION,
UNSPEC_UPDATE_GOT_VERSION): Renumber.
(optab, insn): Add 'plus' and 'minus' to define_code_attr.
(atomic_hiqi_op): New define_code_iterator.
(sync_compare_and_swap<mode>): Call
mips_expand_atomic_qihi instead of
mips_expand_compare_and_swap_12.
(compare_and_swap_12): Use MIPS_COMPARE_AND_SWAP_12 instead of
MIPS_COMPARE_AND_SWAP_12_0.  Pass argument to
MIPS_COMPARE_AND_SWAP_12.
(sync_<optab><mode>, sync_old_<optab><mode>,
sync_new_<optab><mode>, sync_nand<mode>, sync_old_nand<mode>,
sync_new_nand<mode>): New define_expands for HI and QI mode
operands.
(sync_<optab>_12, sync_old_<optab>_12, sync_new_<optab>_12,
sync_nand_12, sync_old_nand_12, sync_new_nand_12): New insns.
(sync_lock_test_and_set<mode>): New define_expand for HI and QI
modes.
(test_and_set_12): New insn.
(sync_old_add<mode>, sync_new_add<mode>, sync_old_<optab><mode>,
sync_new_<optab><mode>, sync_old_nand<mode>,
sync_new_nand<mode>, sync_lock_test_and_set<mode>):  Add early
clobber to operand 0 for SI and DI mode insns.
* config/mips/mips-protos.h (mips_gen_fn_6, mips_gen_fn_5,
mips_gen_fn_4): New typedefs.
(mips_gen_fn_ptrs): Define new union type.
(mips_expand_compare_and_swap_12): Remove declaration.
(mips_expand_atomic_qihi): Declare function.
* config/mips/mips.c (mips_expand_compare_and_swap_12): Rename to...
(mips_expand_atomic_qihi): ... this.  Use new generator function
parameter.
* config/mips/mips.h (MIPS_COMPARE_AND_SWAP_12): Add OPS parameter.
(MIPS_COMPARE_AND_SWAP_12_0): Delete macro.
(MIPS_COMPARE_AND_SWAP_12_ZERO_OP,
MIPS_COMPARE_AND_SWAP_12_NONZERO_OP,
MIPS_SYNC_OP_12, MIPS_SYNC_OP_12_NOT_NOP,
MIPS_SYNC_OP_12_NOT_NOT, MIPS_SYNC_OLD_OP_12,
MIPS_SYNC_OLD_OP_12_NOT_NOP, MIPS_SYNC_OLD_OP_12_NOT_NOP_REG,
MIPS_SYNC_OLD_OP_12_NOT_NOT, MIPS_SYNC_OLD_OP_12_NOT_NOT_REG,
MIPS_SYNC_NEW_OP_12, MIPS_SYNC_NEW_OP_12_NOT_NOP,
MIPS_SYNC_NEW_OP_12_NOT_NOT, MIPS_SYNC_EXCHANGE_12,
MIPS_SYNC_EXCHANGE_12_ZERO_OP,
MIPS_SYNC_EXCHANGE_12_NONZERO_OP): New macros.

From-SVN: r135684

16 years agoi386.c (ix86_expand_vector_init_one_nonzero): Add the missing break.
H.J. Lu [Tue, 20 May 2008 21:39:45 +0000 (21:39 +0000)]
i386.c (ix86_expand_vector_init_one_nonzero): Add the missing break.

2008-05-20  H.J. Lu  <hongjiu.lu@intel.com>

* config/i386/i386.c (ix86_expand_vector_init_one_nonzero): Add
the missing break.

From-SVN: r135682

16 years agoavr.h (machine_function): Add 'is_OS_main' field.
Anatoly Sokolov [Tue, 20 May 2008 21:23:08 +0000 (01:23 +0400)]
avr.h (machine_function): Add 'is_OS_main' field.

* config/avr/avr.h (machine_function): Add 'is_OS_main' field.
* config/avr/avr.c (avr_OS_main_function_p): Add new function.
(avr_attribute_table): Add 'OS_main' function attribute.
(avr_regs_to_save, expand_prologue, expand_epilogue): Handle
functions with 'OS_main' attribute.

From-SVN: r135681

16 years agomodular.adb: Remove test...
Samuel Tardieu [Tue, 20 May 2008 20:59:44 +0000 (20:59 +0000)]
modular.adb: Remove test...

    gcc/testsuite/
* gnat.dg/modular.adb: Remove test, gnat.dg/modular1.adb already
checks that the bug is fixed and is more concise.

From-SVN: r135679

16 years agore PR ada/35791 ([Ada] V-table messed up with interface composition)
Samuel Tardieu [Tue, 20 May 2008 20:45:49 +0000 (20:45 +0000)]
re PR ada/35791 ([Ada] V-table messed up with interface composition)

    gcc/testsuite/
PR ada/35791
* gnat.dg/check_displace_generation.adb: New.

From-SVN: r135677

16 years agore PR middle-end/35204 (crash by too deep recursion in DFS tree-ssa-sccvn.c:1898)
Richard Guenther [Tue, 20 May 2008 20:40:23 +0000 (20:40 +0000)]
re PR middle-end/35204 (crash by too deep recursion in DFS tree-ssa-sccvn.c:1898)

2008-05-20  Richard Guenther  <rguenther@suse.de>

PR tree-optimization/35204
* tree-ssa-sccvn.c (extract_and_process_scc_for_name): New
helper, split out from ...
(DFS): ... here.  Make the DFS walk non-recursive.

From-SVN: r135676

16 years agore PR ada/30740 (Improper semantics in gnat's compilation of certain expressions...
Samuel Tardieu [Tue, 20 May 2008 20:24:33 +0000 (20:24 +0000)]
re PR ada/30740 (Improper semantics in gnat's compilation of certain expressions involving modular arithmetic)

    gcc/testsuite/
PR ada/30740
* gnat.dg/modular.adb: New test.

From-SVN: r135675

16 years agore PR tree-optimization/36181 (Simple for loop generates ICE with -ftree-parallelize...
Sebastian Pop [Tue, 20 May 2008 19:17:12 +0000 (19:17 +0000)]
re PR tree-optimization/36181 (Simple for loop generates ICE with -ftree-parallelize-loops=2)

2008-05-20  Sebastian Pop  <sebastian.pop@amd.com>
    Jan Sjodin  <jan.sjodin@amd.com>

PR tree-optimization/36181
* tree-parloops.c (loop_has_vector_phi_nodes): New.
(parallelize_loops): Don't parallelize when the loop has vector
phi nodes.

* gcc.dg/tree-ssa/pr36181.c: New.

From-SVN: r135673

16 years agotree-loop-linear.c (gather_interchange_stats): Look in the access matrix...
Jan Sjodin [Tue, 20 May 2008 19:11:56 +0000 (19:11 +0000)]
tree-loop-linear.c (gather_interchange_stats): Look in the access matrix...

2008-05-20  Jan Sjodin  <jan.sjodin@amd.com>
    Sebastian Pop  <sebastian.pop@amd.com>

* tree-loop-linear.c (gather_interchange_stats): Look in the access matrix,
and never look at the tree representation of the memory accesses.
(linear_transform_loops): Computes parameters and access matrices.
* tree-data-ref.c (compute_data_dependences_for_loop): Returns false when fails.
(access_matrix_get_index_for_parameter): New.
* tree-data-ref.h (struct access_matrix): New.
(AM_LOOP_NEST_NUM, AM_NB_INDUCTION_VARS, AM_PARAMETERS, AM_MATRIX,
AM_NB_PARAMETERS, AM_CONST_COLUMN_INDEX, AM_NB_COLUMNS,
AM_GET_SUBSCRIPT_ACCESS_VECTOR, AM_GET_ACCESS_MATRIX_ELEMENT,
am_vector_index_for_loop): New.
(struct data_reference): Add field access_matrix.
(DR_ACCESS_MATRIX): New.
(compute_data_dependences_for_loop): Update declaration.
(lambda_collect_parameters, lambda_compute_access_matrices): Declared.
* lambda.h (lambda_vector_vec_p): Declared.
* lambda-code.c: Depend on pointer-set.h.
(lambda_collect_parameters_from_af, lambda_collect_parameters,
av_for_af_base, av_for_af, build_access_matrix,
lambda_compute_access_matrices): New.
* Makefile.in (lambda-code.o): Depend on pointer-set.h.

Co-Authored-By: Sebastian Pop <sebastian.pop@amd.com>
From-SVN: r135672

16 years agore PR testsuite/36057 (struct-layout-1_generate.c uses non-ISO C '%n$' operand...
Uros Bizjak [Tue, 20 May 2008 19:01:46 +0000 (21:01 +0200)]
re PR testsuite/36057 (struct-layout-1_generate.c  uses non-ISO C  '%n$'  operand number formats)

PR testsuite/36057
* g++.dg/compat/struct-layout-1_generate.c (DG_OPTIONS): Remove.
(const char *dg_options): New array.
(switchfiles): Loop through dg_options array to generate dg-options
directives.  Remove numbered arguments usage from fprintf
format strings.

From-SVN: r135671

16 years agoPR c++/33979 (partial)
Paolo Carlini [Tue, 20 May 2008 18:11:34 +0000 (18:11 +0000)]
PR c++/33979 (partial)

2008-05-20  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/33979 (partial)
* include/tr1_impl/functional_hash.h (hash<char16_t>,
hash<char32_t>): Add specializations.
* include/tr1_impl/type_traits (is_integral<char16_t>,
is_integral<char32_t>): Likewise.
* include/std/limits (numeric_limits<char16_t>,
numeric_limits<char32_t>): Likewise.
* src/limits_c++0x.cc: Add.
* src/Makefile.am: Update.
* testsuite/20_util/hash/requirements/explicit_instantiation.cc:
Update.
* testsuite/20_util/is_integral/value.cc: New.
* testsuite/20_util/is_integral/requirements/typedefs.cc: Likewise.
* testsuite/20_util/is_integral/requirements/
explicit_instantiation.cc: Likewise.
* testsuite/18_support/numeric_limits/char16_32_t.cc: Likewise.
* config/abi/pre/gnu.ver: Export new numeric_limits symbols at
GLIBCXX_3.4.11.
* configure: Regenerate.
* src/Makefile.in: Likewise.
* config.h.in: Likewise.

From-SVN: r135668

16 years agoinstall.texi2html: Generate gcc-vers.texi in $DESTDIR not $SOURCEDIR/include.
Joseph Myers [Tue, 20 May 2008 16:21:31 +0000 (17:21 +0100)]
install.texi2html: Generate gcc-vers.texi in $DESTDIR not $SOURCEDIR/include.

* doc/install.texi2html: Generate gcc-vers.texi in $DESTDIR not
$SOURCEDIR/include.

From-SVN: r135664

16 years agore PR middle-end/36206 (ice for legal code with -O3)
Jan Sjodin [Tue, 20 May 2008 16:05:09 +0000 (16:05 +0000)]
re PR middle-end/36206 (ice for legal code with -O3)

2008-05-20  Jan Sjodin  <jan.sjodin@amd.com>
    Sebastian Pop  <sebastian.pop@amd.com>

PR tree-optimization/36206
* tree-scalar-evolution.c: Remove enum INSERT_SUPERLOOP_CHRECS,
FOLD_CONVERSIONS.
(instantiate_scev_1): Rename flags to fold_conversions.
Do not check for INSERT_SUPERLOOP_CHRECS, keep SSA_NAMEs defined
outeside instantiation_loop.
* tree-chrec.h (evolution_function_is_affine_in_loop): New.
(evolution_function_is_affine_or_constant_p): Removed.
* tree-data-ref.c (dr_analyze_indices): Replace resolve_mixers with
instantiate_scev.
(analyze_siv_subscript): Pass in the loop nest number.
Call evolution_function_is_affine_in_loop instead of
evolution_function_is_affine_p.
(analyze_overlapping_iterations): Pass in the loop nest number.

* tree-chrec.h (chrec_fold_op): New.
* tree-data-ref.c (initialize_matrix_A): Traverse NOP_EXPR, PLUS_EXPR, and
other trees.

* testsuite/gfortran.dg/pr36206.f: New.

Co-Authored-By: Sebastian Pop <sebastian.pop@amd.com>
From-SVN: r135663

16 years agoc-incpath.c (INO_T_EQ): Do not define on non-inode systems.
Nathan Sidwell [Tue, 20 May 2008 15:22:31 +0000 (15:22 +0000)]
c-incpath.c (INO_T_EQ): Do not define on non-inode systems.

* c-incpath.c (INO_T_EQ): Do not define on non-inode systems.
(DIRS_EQ): New.
(remove_duplicates): Do not set inode on non-inode systems.  Use
DIRS_EQ.

From-SVN: r135661

16 years agogen_disp.ad[sb]: New test.
Arnaud Charlet [Tue, 20 May 2008 14:02:20 +0000 (14:02 +0000)]
gen_disp.ad[sb]: New test.

* gnat.dg/gen_disp.ad[sb]: New test.
* gnat.dg/specs/empty_variants.ads: Adjust.

From-SVN: r135658

16 years agoconfig.gcc (tm_file): Update comments about relative pathnames.
Sandra Loosemore [Tue, 20 May 2008 13:32:47 +0000 (09:32 -0400)]
config.gcc (tm_file): Update comments about relative pathnames.

2008-05-20  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* config.gcc (tm_file): Update comments about relative pathnames.

From-SVN: r135656

16 years agoResync.
Arnaud Charlet [Tue, 20 May 2008 13:03:26 +0000 (15:03 +0200)]
Resync.

From-SVN: r135654

16 years agoMinor reformatting.
Arnaud Charlet [Tue, 20 May 2008 13:00:35 +0000 (13:00 +0000)]
Minor reformatting.

Update comments.

From-SVN: r135653

16 years agos-taspri-vxworks.ads (Task_Address, [...]): New type/constant.
Arnaud Charlet [Tue, 20 May 2008 12:59:41 +0000 (12:59 +0000)]
s-taspri-vxworks.ads (Task_Address, [...]): New type/constant.

* s-taspri-vxworks.ads (Task_Address, Task_Address_Size): New
type/constant.

* g-socthi-vxworks.ads: Update to latest socket changes.

* a-caldel-vms.adb: Resync with a-caldel spec.

From-SVN: r135652

16 years agog-byorma.adb, [...]: Update comments.
Robert Dewar [Tue, 20 May 2008 12:52:53 +0000 (14:52 +0200)]
g-byorma.adb, [...]: Update comments.

2008-05-20  Robert Dewar  <dewar@adacore.com>

* g-byorma.adb, gnatlink.adb, prepcomp.adb, sinfo.ads,
sem_ch12.adb: Update comments. Minor reformatting.

* exp_ch2.adb: Typo

* s-unstyp.ads: Fixed some typos in comments.

From-SVN: r135651

16 years ago2008-05-20 Ed Schonberg <schonberg@adacore.com>
Ed Schonberg [Tue, 20 May 2008 12:52:41 +0000 (14:52 +0200)]
2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* exp_ch7.adb
(Expand_Ctrl_Function_Call): Do not attach result to finalization list
if expression is aggregate component.

From-SVN: r135650

16 years agos-winext.ads: Replace representation clause by pragma Pack.
Arnaud Charlet [Tue, 20 May 2008 12:52:23 +0000 (14:52 +0200)]
s-winext.ads: Replace representation clause by pragma Pack.

* s-winext.ads: Replace representation clause by pragma Pack. Gives
equivalent representation, but has the advantage of allowing
compilation of this file under 64 bits platforms.

* s-os_lib.adb (Normalize_Pathname): Mark Cur_Dir constant.

* s-osinte-irix.ads: (Alternate_Stack_Size): Add dummy declaration.

* adaint.c:
Don't define dummy implementation of convert_addresses on ppc-linux.

From-SVN: r135649

16 years ago2008-05-20 Bob Duff <duff@adacore.com>
Bob Duff [Tue, 20 May 2008 12:52:09 +0000 (14:52 +0200)]
2008-05-20  Bob Duff  <duff@adacore.com>

* a-rttiev.adb
(Set_Handler): Remove code from both of these that implements
RM-D.15(15/2), because it causes a race condition and potential
deadlock.
(Process_Queued_Events): Add comment explaining "exception when others
=> null". Add clarifying ".all", even though implicit .all is legal
here.

From-SVN: r135648

16 years agovms_data.ads: Add entry for -gnatr Put GNAT SYNC section in proper alpha order Add...
Robert Dewar [Tue, 20 May 2008 12:51:51 +0000 (14:51 +0200)]
vms_data.ads: Add entry for -gnatr Put GNAT SYNC section in proper alpha order Add VMS...

2008-05-20  Robert Dewar  <dewar@adacore.com>
    Vincent Celier  <celier@adacore.com>

* vms_data.ads: Add entry for -gnatr
Put GNAT SYNC section in proper alpha order
Add VMS qualifier /DISPLAY_PROGRESS equivalent to gnatmake switch -d

* gnat_ugn.texi: Add documentation for new gnatname switch --and
Update the style checks section
Add documentation of -gnatr
Add to the "Adding the Results of Compiler Checks to gnatcheck Output"
subsection the explanation how compiler checks should be disabled for
gnatcheck.
Update the list of Ada 95 reserved words used by in the project language
Add documentation for project qualifiers.
Document that abstract projects may be extended by different projects in
the same project tree.
Add documentation for gnatmake switch -d

* ug_words: Add -gnatyy VMS equivalence string.
Add entry for -gnatr

From-SVN: r135647

16 years agouintp.adb, urealp.adb: Replace calls to Increment_Last + Set with Append.
Thomas Quinot [Tue, 20 May 2008 12:51:40 +0000 (14:51 +0200)]
uintp.adb, urealp.adb: Replace calls to Increment_Last + Set with Append.

2008-05-20  Thomas Quinot  <quinot@adacore.com>

* uintp.adb, urealp.adb: Replace calls to Increment_Last + Set with
Append.

From-SVN: r135646

16 years ago2008-05-20 Ed Schonberg <schonberg@adacore.com>
Ed Schonberg [Tue, 20 May 2008 12:51:29 +0000 (14:51 +0200)]
2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* sprint.adb
(Print_Itype): Do not modify the sloc of the component type of a
(packed) array itype, because it is an unrelated type whose source
location is independent of the point of creation of the itype itself.

From-SVN: r135645

16 years agosem_res.adb (Resolve_Op_Not): Warn on double negation
Robert Dewar [Tue, 20 May 2008 12:51:17 +0000 (14:51 +0200)]
sem_res.adb (Resolve_Op_Not): Warn on double negation

2008-05-20  Robert Dewar  <dewar@adacore.com>

* sem_res.adb (Resolve_Op_Not): Warn on double negation

From-SVN: r135644

16 years ago2008-05-20 Ed Schonberg <schonberg@adacore.com>
Ed Schonberg [Tue, 20 May 2008 12:51:06 +0000 (14:51 +0200)]
2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* sem_eval.adb
(Eval_Slice): Warn when a slice whose discrete range is a subtype name
denotes the whole array of its prefix.

From-SVN: r135643

16 years agosem_dist.ads, [...] (Is_RACW_Stub_Type_Operation): New subprogram.
Thomas Quinot [Tue, 20 May 2008 12:50:52 +0000 (14:50 +0200)]
sem_dist.ads, [...] (Is_RACW_Stub_Type_Operation): New subprogram.

2008-05-20  Thomas Quinot  <quinot@adacore.com>

* sem_dist.ads, sem_dist.adb (Is_RACW_Stub_Type_Operation): New
subprogram.

* sem_type.adb
(Add_One_Interp): Ignore any interpretation that is a primitive
operation of an RACW stub type (these primitives are only executed
through dispatching, never through static calls).
(Collect_Interps): When only one interpretation has been found, set N's
Entity and Etype to that interpretation, otherwise Entity and Etype may
still refer to an interpretation that was ignored by Add_One_Interp,
in which case would end up with being marked as not overloaded but with
an Entity attribute not pointing to its (unique) correct interpretation.

From-SVN: r135642

16 years ago2008-05-20 Ed Schonberg <schonberg@adacore.com>
Ed Schonberg [Tue, 20 May 2008 12:50:43 +0000 (14:50 +0200)]
2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* sem_ch8.adb
(Note_Redundant_Use): Diagnose a redundant use within a subprogram body
when there is a use clause for the same entity in the context.
(Analyze_Subprogram_Renaming): A renaming_as_body is legal if it is
created for a stream attribute of an abstract type or interface type.

From-SVN: r135641

16 years ago2008-05-20 Ed Schonberg <schonberg@adacore.com>
Ed Schonberg [Tue, 20 May 2008 12:50:26 +0000 (14:50 +0200)]
2008-05-20  Ed Schonberg  <schonberg@adacore.com>
    Thomas Quinot  <quinot@adacore.com>

* sem_ch4.adb
(Try_Indexed_Call): Handle properly a construct of the form F(S) where
F is a parameterless function that returns an array, and S is a subtype
mark.
(Analyze_Call): Insert dereference when the prefix is a parameterless
function that returns an access to subprogram and the call has
parameters.
Reject a non-overloaded call whose name resolves to denote
a primitive operation of the stub type generated to support a remote
access-to-class-wide type.

From-SVN: r135640

16 years agoMakefile.in (GNATTOOLS2_FILES): Replaced by common-tools target in gcc/ada/Makefile.in
Arnaud Charlet [Tue, 20 May 2008 12:50:26 +0000 (12:50 +0000)]
Makefile.in (GNATTOOLS2_FILES): Replaced by common-tools target in gcc/ada/Makefile.in

* Makefile.in (GNATTOOLS2_FILES): Replaced by common-tools target
in gcc/ada/Makefile.in

From-SVN: r135639

16 years ago2008-05-20 Javier Miranda <miranda@adacore.com>
Javier Miranda [Tue, 20 May 2008 12:50:03 +0000 (14:50 +0200)]
2008-05-20  Javier Miranda  <miranda@adacore.com>
    Ed Schonberg  <schonberg@adacore.com>
    Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch3.adb
(Analyze_Object_Declaration): Fix over-conservative condition
restricting use of predefined assignment with tagged types that have
convention CPP.
(Analyze_Object_Declaration): Relax the check regarding deferred
constants declared in scopes other than packages since they can be
completed with pragma Import.
Add missing escaping of all-caps word 'CPP' in error messages.
(Build_Discriminated_Subtype): Do not inherit representation clauses
from parent type if subtype already carries them, because they are
inherited earlier during derivation and already include those that may
come from a partial view.

* sem_ch9.adb, sem_ch5.adb, sem_ch6.adb (Analyze_Subprogram_Body):
Check the declarations of a subprogram body for proper deferred
constant completion.

* sem_ch7.ads, sem_ch7.adb
(Inspect_Deferred_Constant_Completion): Moved to sem_util.

From-SVN: r135638

16 years ago2008-05-20 Thomas Quinot <quinot@adacore.com>
Thomas Quinot [Tue, 20 May 2008 12:49:41 +0000 (14:49 +0200)]
2008-05-20  Thomas Quinot  <quinot@adacore.com>

* sem_cat.adb
(Set_Categorization_From_Scope): Do not set In_Remote_Types unless in
the visible part of the spec of a remote types unit.
(Validate_Remote_Access_Object_Type_Declaration):
New local subprogram Is_Valid_Remote_Object_Type, replaces
Is_Recursively_Limited_Private.
(Validate_RACW_Primitives): Enforce E.2.2(14) rules: the types of all
non-controlling formals (and the return type, even though this is not
explicit in the standard) must support external streaming.
(Validate_RCI_Subprogram_Declaration): Enforce E.2.3(14) rules: same
as above for of RAS types and RCI subprograms. (The return type is not
checked yet).
Update comments related to RACWs designating limited interfaces per
ARG ruling on AI05-060.

* sem_util.ads, sem_util.adb
(Is_Remote_Access_To_Class_Wide_Type): Only rely on Is_Remote_Types and
Is_Remote_Call_Interface to identify RACW types in a stable and
consistent way. We used to rely in this predicate on the privateness of
the designated type and its ancestors, but depending on the currently
visible private parts, this caused false negatives. We now uniformly
rely on checks made at the point where the RACW type is declared.
(Inspect_Deferred_Constant_Completion): Moved from Sem_Ch7.

From-SVN: r135637

16 years agore PR ada/17985 (GNAT accepts extension aggregate where expexted type is not extension)
Ed Schonberg [Tue, 20 May 2008 12:49:21 +0000 (14:49 +0200)]
re PR ada/17985 (GNAT accepts extension aggregate where expexted type is not extension)

2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* sem_aggr.adb: Update comments.
Improve previous change for PR ada/17985

From-SVN: r135636

16 years agotestint.adb: New test.
Arnaud Charlet [Tue, 20 May 2008 12:49:20 +0000 (12:49 +0000)]
testint.adb: New test.

* gnat.dg/testint.adb: New test.
* gnat.dg/modular1.adb: New test.
* gnat.dg/test_iface_aggr.adb: New test.
* gnat.dg/specs/tag2.ads: Adjust.

From-SVN: r135635

16 years ago2008-05-20 Hristian Kirtchev <kirtchev@adacore.com>
Hristian Kirtchev [Tue, 20 May 2008 12:48:30 +0000 (14:48 +0200)]
2008-05-20  Hristian Kirtchev  <kirtchev@adacore.com>

* par-ch9.adb
(P_Protected): Update the error message on missing "-gnat05" switch when
using interfaces in conjunction with protected types. Remove the
incorrect error message associated with the presence of "private" after
a "with".

From-SVN: r135634

16 years ago2008-05-20 Hristian Kirtchev <kirtchev@adacore.com>
Hristian Kirtchev [Tue, 20 May 2008 12:48:03 +0000 (14:48 +0200)]
2008-05-20  Hristian Kirtchev  <kirtchev@adacore.com>

* par-ch3.adb
(P_Access_Definition): Change the error message when parsing "access
all" in Ada 95 mode. The message no longer forces the user to recompile
in 05 mode only to discover that anonymous access types are not allowed
to have "all".

From-SVN: r135633

16 years agoopt.ads (Treat_Restrictions_As_Warnings): New switch
Robert Dewar [Tue, 20 May 2008 12:47:49 +0000 (14:47 +0200)]
opt.ads (Treat_Restrictions_As_Warnings): New switch

2008-05-20  Robert Dewar  <dewar@adacore.com>

* opt.ads (Treat_Restrictions_As_Warnings): New switch

* sem_prag.adb, par-prag.adb, restrict.ads: Implement flag
Treat_Restrictions_As_Warnings.

* switch-c.adb: Recognize new switch -gnatr

* usage.adb: Add line for -gnatr

From-SVN: r135632

16 years ago* Makefile.in
Arnaud Charlet [Tue, 20 May 2008 12:47:34 +0000 (14:47 +0200)]
* Makefile.in
(common-tools): New rule, to avoid parallel build failure on gnat tools.
Reenable parallel builds on this Makefile.

* Make-lang.in: Update dependencies.

From-SVN: r135631

16 years agoinit.c: Enable stack probing on ppc-linux.
Tristan Gingold [Tue, 20 May 2008 12:47:13 +0000 (14:47 +0200)]
init.c: Enable stack probing on ppc-linux.

2008-05-20  Tristan Gingold  <gingold@adacore.com>

* init.c: Enable stack probing on ppc-linux.

* tracebak.c: Add symbolic traceback for ppc-linux.

* system-linux-ppc.ads: Enable stack probing on ppc-linux.

From-SVN: r135630

16 years ago2008-05-20 Vincent Celier <celier@adacore.com>
Vincent Celier [Tue, 20 May 2008 12:47:03 +0000 (14:47 +0200)]
2008-05-20  Vincent Celier  <celier@adacore.com>

* gnatname.adb
(Scan_Args): Rewrite to take into account new switch --and to separate
arguments into sections.
(Gnatname): Call Prj.Makr.Initialize, then Prj.Makr.Process for each
section, then Finalize.

From-SVN: r135629

16 years ago2008-05-20 Ed Schonberg <schonberg@adacore.com>
Ed Schonberg [Tue, 20 May 2008 12:46:54 +0000 (14:46 +0200)]
2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* freeze.adb
(Freeze_Enumeration_Type): For a subtype that inherits a foreign
convention from its base type, do not set the type to that of integer,
because it may inherit a size clause.
Warn on a size clause with a size different
from that of Integer, if the type has convention C.

From-SVN: r135628

16 years agoexp_smem.ads, [...]: Construction of access and assign routines has been replaced...
Kevin Pouget [Tue, 20 May 2008 12:46:42 +0000 (14:46 +0200)]
exp_smem.ads, [...]: Construction of access and assign routines has been replaced by an...

2008-05-20  Kevin Pouget  <pouget@adacore.com>

* exp_smem.ads, exp_smem.adb: Construction of access and assign
routines has been replaced by an instantiation of
System.Shared_Storage.Shared_Var_Procs generic package, while expanding
shared variable declaration.
Calls to access and assign routines have been replaced by calls to
Read/Write routines of System.Shared_Storage.Shared_Var_Procs
instantiated package.

* rtsfind.ads: RE_Shared_Var_Procs entry has been added in RE_Unit_Table
It identifies the new generic package added in s-shasto.

* s-shasto.adb, s-shasto.ads: A new generic package has been added, it
is instantiated for each shared passive variable. It provides
supporting procedures called upon each read or write access by the
expanded code.

* sem_attr.adb:
For this runtime unit (always compiled in GNAT mode), we allow
stream attributes references for limited types for the case where
shared passive objects are implemented using stream attributes,
which is the default in GNAT's persistent storage implementation.

From-SVN: r135627

16 years ago2008-05-20 Thomas Quinot <quinot@adacore.com>
Thomas Quinot [Tue, 20 May 2008 12:46:31 +0000 (14:46 +0200)]
2008-05-20  Thomas Quinot  <quinot@adacore.com>

* exp_dist.adb
(GARLIC_Support.Add_RACW_Read_Attribute): When a zero value is received,
and the RACW is null-excluding, raise CONSTRAINT_ERROR instead of
assigning NULL into the result, to avoid a spurious warning.
(Add_RACW_Features, case Same_Scope): Add assertion that designated type
is not frozen.
(Add_Stub_Type): Set entity flag Is_RACW_Stub_Type on generated stub
type.
(Build_From_Any_Function, Build_To_Any_Function,
Build_TypeCode_Function): For a type that has user-specified stream
attributes, use an opaque sequence of octets as the representation.

From-SVN: r135626

16 years agoexp_disp.adb (Make_DT, [...]): Avoid generating dispatch tables of locally defined...
Javier Miranda [Tue, 20 May 2008 12:46:19 +0000 (14:46 +0200)]
exp_disp.adb (Make_DT, [...]): Avoid generating dispatch tables of locally defined tagged types statically.

2008-05-20  Javier Miranda  <miranda@adacore.com>
    Ed Schonberg  <schonberg@adacore.com>

* exp_disp.adb (Make_DT, Make_Secondary_DT, Make_Tags): Avoid
generating dispatch tables of locally defined tagged types statically.
Remove implicit if-statement that is no longer required.
(Expand_Dispatching_Call): If this is a call to an instance of the
generic dispatching constructor, the type of the first argument may be
a subtype of Tag, so always use the base type to recognize this case.

From-SVN: r135625

16 years agoexp_ch6.adb (Expand_Call): Add guard to ensure that both the parent and the derived...
Hristian Kirtchev [Tue, 20 May 2008 12:46:06 +0000 (14:46 +0200)]
exp_ch6.adb (Expand_Call): Add guard to ensure that both the parent and the derived type are of the same kind.

2008-05-20  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch6.adb (Expand_Call): Add guard to ensure that both the parent
and the derived type are of the same kind.
(Expand_Call): Generate type conversions for actuals of
record or array types when the parent and the derived types differ in
size and/or packed status.

From-SVN: r135624

16 years agoprj.adb (Hash (Project_Id)): New function
Vincent Celier [Tue, 20 May 2008 12:45:54 +0000 (14:45 +0200)]
prj.adb (Hash (Project_Id)): New function

2008-05-20  Vincent Celier  <celier@adacore.com>

* prj.adb (Hash (Project_Id)): New function
(Project_Empty): Add new component Interfaces_Defined

* prj.ads (Source_Data): New component Object_Linked
(Language_Config): New components Object_Generated and Objects_Linked
(Hash (Project_Id)): New function
(Source_Data): New Boolean components In_Interfaces and
Declared_In_Interfaces.
(Project_Data): New Boolean component Interfaces_Defined

* prj-attr.adb:
New project level attribute Object_Generated and Objects_Linked
Add new project level attribute Interfaces

* prj-dect.adb: Use functions Present and No throughout
(Parse_Variable_Declaration): If a string type is specified as a simple
name and is not found in the current project, look for it also in the
ancestors of the project.

* prj-makr.adb:
Replace procedure Make with procedures Initialize, Process and Finalize
to implement H414-023: process different directories with different
patterns.
Use functions Present and No throughout

* prj-makr.ads:
Replace procedure Make with procedures Initialize, Process and Finalize

* prj-nmsc.adb
(Add_Source): Set component Object_Exists and Object_Linked accordnig to
the language configuration.
(Process_Project_Level_Array_Attributes): Process new attributes
Object_Generated and Object_Linked.
(Report_No_Sources): New Boolean parameter Continuation, defaulted to
False, to indicate that the erreor/warning is a continuation.
(Check): Call Report_No_Sources with Contnuation = True after the first
call.
(Error_Msg): Process successively contnuation character and warning
character.
(Find_Explicit_Sources): Check that all declared sources have been found
(Check_File): Indicate in hash table Source_Names when a declared source
is found.
(Check_File): Set Other_Part when found
(Find_Explicit_Sources): In multi language mode, check if all exceptions
to the naming scheme have been found. For Ada, report an error if an
exception has not been found. Otherwise, disregard the exception.
(Check_Interfaces): New procedure
(Add_Source): When Other_Part is defined, set mutual pointers in spec
and body.
(Check): In multi-language mode, call Check_Interfaces
(Process_Sources_In_Multi_Language_Mode): Set In_Interfaces to False
for an excluded source.
(Remove_Source): A source replacing a source in the interfaces is also
in the interfaces.

* prj-pars.adb: Use function Present

* prj-part.adb: Use functions Present and No throughout
(Parse_Single_Project): Set the parent project for child projects
(Create_Virtual_Extending_Project): Register project with no qualifier
(Parse_Single_Project): Allow an abstract project to be extend several
times. Do not allow an abstract project to extend a non abstract
project.

* prj-pp.adb: Use functions Present and No throughout
(Print): Take into account the full associative array attribute
declarations.

* prj-proc.adb: Use functions Present and No throughout
(Expression): Call itself with the same From_Project_Node for the
default value of an external reference.

* prj-strt.adb: Use functions Present and No throughout
(Parse_Variable_Reference): If a variable is specified as a simple name
and is not found in the current project, look for it also in the
ancestors of the project.

* prj-tree.ads, prj-tree.adb (Present): New function
(No): New function
Use functions Present and No throughout
(Parent_Project_Of): New function
(Set_Parent_Project_Of): New procedure

* snames.ads, snames.adb:
Add new standard names Object_Generated and Objects_Linked

From-SVN: r135623

16 years agoexp_ch5.adb (Expand_Assign_Record): Within an initialization procedure for a derived...
Ed Schonberg [Tue, 20 May 2008 12:45:38 +0000 (14:45 +0200)]
exp_ch5.adb (Expand_Assign_Record): Within an initialization procedure for a derived type retrieve the discriminant...

2008-05-20  Ed Schonberg  <schonberg@adacore.com>

* exp_ch5.adb (Expand_Assign_Record): Within an initialization
procedure for a derived type retrieve the discriminant values from the
parent using the corresponding discriminant.
(Expand_N_Assignment_Statement): Skip generation of implicit
if-statement associated with controlled types if we are
compiling with restriction No_Finalization.

From-SVN: r135622

16 years ago2008-05-20 Gary Dismukes <dismukes@adacore.com>
Gary Dismukes [Tue, 20 May 2008 12:45:27 +0000 (14:45 +0200)]
2008-05-20  Gary Dismukes  <dismukes@adacore.com>
    Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch3.adb
(Expand_N_Object_Declaration): Correct the condition which triggers the
generation of a call to Displace when initializing a class-wide object.
(Build_Dcheck_Functions): Build discriminant-checking for null variants
when Frontend_Layout_On_Target is true to ensure that they're available
for calling when a record variant size function is built in Layout.

From-SVN: r135621

16 years ago2008-05-20 Javier Miranda <miranda@adacore.com>
Javier Miranda [Tue, 20 May 2008 12:45:15 +0000 (14:45 +0200)]
2008-05-20  Javier Miranda  <miranda@adacore.com>

* exp_aggr.adb
(Build_Record_Aggr_Code): Fix wrong tests checking progenitors. Previous
tests did not covered the case in which the type of the aggregate has
no progenitors but some its parents has progenitors.

From-SVN: r135620

16 years agore PR ada/30740 (Improper semantics in gnat's compilation of certain expressions...
Robert Dewar [Tue, 20 May 2008 12:44:55 +0000 (14:44 +0200)]
re PR ada/30740 (Improper semantics in gnat's compilation of certain expressions involving modular arithmetic)

2008-05-20  Robert Dewar  <dewar@adacore.com>

PR ada/30740
* einfo.ads, einfo.adb (Non_Binary_Modulus): Applies to all types and
subtypes, always False for non-modular types.
Shared_Var_Assign_Proc (node22) and Shared_Var_Read_Proc (node 15)
entry nodes have been replaced by Shared_Var_Procs_Instance (node22)
for Shared_Storage package.
(Is_RACW_Stub_Type): New entity flag.

* exp_ch4.adb
(Expand_N_Op_Expon): Avoid incorrect optimization of a*(2**b) in the
case where we have a modular type with a non-binary modules.
Comments reformattings.

* sem_intr.adb: Simplify code not that Non_Binary_Modulus applies to
all types.

From-SVN: r135619

16 years agodecl.c: Do not emit a variable for a object that has an address representation clause...
Tristan Gingold [Tue, 20 May 2008 12:44:43 +0000 (14:44 +0200)]
decl.c: Do not emit a variable for a object that has an address representation clause...

2008-05-20  Tristan Gingold  <gingold@adacore.com>

* decl.c: Do not emit a variable for a object that has an address
representation clause whose value is known at compile time.
When a variable has an address clause whose value is known at compile
time, refer to this variable by using directly the address instead of
dereferencing a pointer.

From-SVN: r135618

16 years agocstand.adb (Print_Standard): Issue the correct Size clause for type Wide_Wide_Character.
Vincent Celier [Tue, 20 May 2008 12:44:33 +0000 (14:44 +0200)]
cstand.adb (Print_Standard): Issue the correct Size clause for type Wide_Wide_Character.

2008-05-20  Vincent Celier  <celier@adacore.com>

* cstand.adb (Print_Standard): Issue the correct Size clause for type
Wide_Wide_Character.

From-SVN: r135617

16 years agochecks.adb (Apply_Arithmetic_Overflow_Check): Avoid intermediate overflow if result...
Robert Dewar [Tue, 20 May 2008 12:44:23 +0000 (14:44 +0200)]
checks.adb (Apply_Arithmetic_Overflow_Check): Avoid intermediate overflow if result converted to wider integer type.

2008-05-20  Robert Dewar  <dewar@adacore.com>
    Gary Dismukes  <dismukes@adacore.com>

* checks.adb (Apply_Arithmetic_Overflow_Check): Avoid intermediate
overflow if result converted to wider integer type.
(Apply_Type_Conversion_Checks): Don't emit checks on conversions to
discriminated types when discriminant checks are suppressed.

From-SVN: r135616

16 years agoali.adb: Correct casing of ASCII.NUL
Robert Dewar [Tue, 20 May 2008 12:44:12 +0000 (14:44 +0200)]
ali.adb: Correct casing of ASCII.NUL

2008-05-20  Robert Dewar  <dewar@adacore.com>

* ali.adb: Correct casing of ASCII.NUL

* styleg-c.adb (Check_Identifier): Handle case of names in ASCII
properly.

From-SVN: r135615

16 years agore PR ada/24533 (FAIL: a85013b: *** glibc detected *** free(): invalid pointer...
Arnaud Charlet [Tue, 20 May 2008 12:43:59 +0000 (14:43 +0200)]
re PR ada/24533 (FAIL:   a85013b: *** glibc detected *** free(): invalid pointer: 0x00062a00 ***)

2008-05-20  Arnaud Charlet  <charlet@adacore.com>

* s-linux-hppa.ads (atomic_lock_t): Put back proper alignment now that
the underlying issue with malloc/free has been fixed. Remove associated
comments.
Minor reformatting.
Related to PR ada/24533

From-SVN: r135614

16 years agoi-cobol.ads: Interfaces.COBOL should be preelaborate.
Santiago Uruena [Tue, 20 May 2008 12:43:44 +0000 (14:43 +0200)]
i-cobol.ads: Interfaces.COBOL should be preelaborate.

2008-05-20  Santiago Uruena  <uruena@adacore.com>

* i-cobol.ads: Interfaces.COBOL should be preelaborate.

From-SVN: r135613

16 years agoMakefile.rtl (GNAT.Sockets.Thin_Common): New unit.
Thomas Quinot [Tue, 20 May 2008 12:43:29 +0000 (14:43 +0200)]
Makefile.rtl (GNAT.Sockets.Thin_Common): New unit.

2008-05-20  Thomas Quinot  <quinot@adacore.com>

* Makefile.rtl (GNAT.Sockets.Thin_Common): New unit.

* g-sttsne-vxworks.adb: Add missing dependency on Sockets.Constants.
Add missing "with" of Ada.Unchecked_Conversion

* g-soccon-linux-ppc.ads, g-soccon-linux-64.ads, g-soccon-lynxos.ads,
g-soccon-linux-x86.ads, g-soccon-hpux-ia64.ads,
g-soccon-solaris-64.ads, g-soccon-tru64.ads, g-soccon-aix.ads,
g-soccon-irix.ads, g-soccon-hpux.ads, g-soccon-solaris.ads,
g-soccon-vms.ads, g-soccon-mingw.ads, g-soccon-vxworks.ads,
g-socthi-vxworks.adb, g-soccon-freebsd.ads, g-soccon.ads:
Move common code out of GNAT.Sockets.Thin implementations and into
Thin_Common.
New constant SIZEOF_fd_set
New flag Has_Sockaddr_Len
New constants SIZEOF_sockaddr_in, SIZEOF_sockaddr_in6

* g-stsifd-sockets.adb
(Create): Remove call to Set_Length; use Set_Family to set the family
and (on appropriate platforms) length fields in struct sockaddr.

* g-socthi.adb, g-socthi.ads, g-socthi-vms.ads, g-socthi-vms.adb,
g-socthi-mingw.adb, g-socthi-mingw.ads, g-socthi-vxworks.adb,
g-soccon-darwin.ads, g-soccon-darwin.ads: New constant SIZEOF_fd_set
Move common code out of GNAT.Sockets.Thin implementations and into
Thin_Common.

* g-socket.ads, g-socket.adb:
Move common code out of GNAT.Sockets.Thin implementations and into
Thin_Common.
(Connect_Socket, Accept_Socket): Provide new versions of these two
routines that operate with a user specified timeout.
(Bind_Socket, Connect_Socket, Send_Socket): Remove calls to Set_Length,
this is now handled automatically by Set_Family on platforms that
require it.

* gen-soccon.c:
Move common code out of GNAT.Sockets.Thin implementations and into
Thin_Common.
(SIZEOF_sockaddr_in6): On platforms where IPv6 is not supported, define
this constant to 0 (not -1) because we use it to initialize an
unsigned_char value.
Align values for numeric constants only.
Handle the case of systems that do not support AF_INET6.
New constant SIZEOF_fd_set
New flag Has_Sockaddr_Len
New constants SIZEOF_sockaddr_in, SIZEOF_sockaddr_in6

* gsocket.h: New flag Has_Sockaddr_Len
New constants SIZEOF_sockaddr_in, SIZEOF_sockaddr_in6

From-SVN: r135612

16 years agotree-ssa-reassoc.c (fini_reassoc): Use the statistics infrastructure.
Richard Guenther [Tue, 20 May 2008 12:34:06 +0000 (12:34 +0000)]
tree-ssa-reassoc.c (fini_reassoc): Use the statistics infrastructure.

2008-05-20  Richard Guenther  <rguenther@suse.de>

* tree-ssa-reassoc.c (fini_reassoc): Use the statistics
infrastructure.
* tree-ssa-sccvn.c (process_scc): Likewise.
* tree-ssa-sink.c (execute_sink_code): Likewise.
* tree-ssa-threadupdate.c (thread_through_all_blocks): Likewise.
* tree-vrp.c (process_assert_insertions): Likewise.
* tree-ssa-dce.c (eliminate_unnecessary_stmts): Likewise.
(perform_tree_ssa_dce): Likewise.
* tree-ssa-dom.c (tree_ssa_dominator_optimize): Likewise.
(dump_dominator_optimization_stats): Likewise.
* tree-vectorizer.c (vectorize_loops): Likewise.

* gcc.dg/tree-ssa/ssa-sink-1.c: Adjust.
* gcc.dg/tree-ssa/ssa-sink-2.c: Likewise.
* gcc.dg/tree-ssa/ssa-sink-3.c: Likewise.
* gcc.dg/tree-ssa/ssa-sink-4.c: Likewise.

From-SVN: r135611

16 years ago20061214-1.c: New test.
Sandra Loosemore [Tue, 20 May 2008 12:14:02 +0000 (08:14 -0400)]
20061214-1.c: New test.

2008-05-20  Sandra Loosemore  <sandra@codesourcery.com>

gcc/testsuite/
* gcc.c-torture/compile/20061214-1.c: New test.

From-SVN: r135610

16 years agoFinally reverted parts of my escaped patch. (quilt sucks).
Kai Tietz [Tue, 20 May 2008 10:24:43 +0000 (12:24 +0200)]
Finally reverted parts of my escaped patch. (quilt sucks).

From-SVN: r135607

16 years ago2008-05-20 Kai Tietz <kai.tietz@onevision.com>
Kai Tietz [Tue, 20 May 2008 10:08:42 +0000 (12:08 +0200)]
2008-05-20  Kai Tietz  <kai.tietz@onevision.com>

Reverted escaped patch.

From-SVN: r135606

16 years agotree-vn.c (vn_lookup_with_vuses): Do not use the alias oracle.
Richard Guenther [Tue, 20 May 2008 10:03:27 +0000 (10:03 +0000)]
tree-vn.c (vn_lookup_with_vuses): Do not use the alias oracle.

2008-05-20  Richard Guenther  <rguenther@suse.de>

* tree-vn.c (vn_lookup_with_vuses): Do not use the alias oracle.

* gcc.c-torture/execute/20080519-1.c: New testcase.

From-SVN: r135605

16 years agoi386-protos.h (ix86_return_in_memory): Removed.
Kai Tietz [Tue, 20 May 2008 10:00:20 +0000 (10:00 +0000)]
i386-protos.h (ix86_return_in_memory): Removed.

2008-05-20  Kai Tietz  <kai.tietz@onevision.com>

* config/i386/i386-protos.h (ix86_return_in_memory): Removed.
(ix86_i386elf_return_in_memory): Likewise.
(ix86_i386interix_return_in_memory): Likewise.
* config/i386/i386-interix.h (TARGET_RETURN_IN_MEMORY): Removed.
(SUBTARGET_RETURN_IN_MEMORY): New.
* config/i386/i386elf.h: Likewise.
* config/i386/ptx4-i.h: Likewise.
* config/i386/sol2-10.h: Likewise.
* config/i386/sysv4.h: Likewise.
* config/i386/vx-common.h: Likewise.
* config/i386/i386.h (TARGET_RETURN_IN_MEMORY): Removed.
* config/i386/i386.c (ix86_return_in_memory): Made static and
make use of optional SUBTARGET_RETURN_IN_MEMORY macro.
(ix86_i386elf_return_in_memory): Removed.
(ix86_i386interix_return_in_memory): Removed.
(TARGET_RETURN_IN_MEMORY): Declared within i386.c only.
* target-def.h (TARGET_RETURN_IN_MEMORY): Remove protection #ifdef.

From-SVN: r135604

16 years ago* cselib.c (cselib_record_sets): Use correct mode for IF_THEN_ELSE.
Alexandre Oliva [Tue, 20 May 2008 06:44:59 +0000 (06:44 +0000)]
* cselib.c (cselib_record_sets): Use correct mode for IF_THEN_ELSE.

From-SVN: r135601

16 years agoRevert r135493 & r135463
Xinliang David Li [Tue, 20 May 2008 04:28:57 +0000 (04:28 +0000)]
Revert r135493 & r135463

From-SVN: r135599

16 years agoDaily bump.
GCC Administrator [Tue, 20 May 2008 00:17:39 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r135596

16 years agotree.c (substitute_placeholder_in_expr) <tcc_vl_exp>: Minor tweak.
Eric Botcazou [Mon, 19 May 2008 22:24:38 +0000 (22:24 +0000)]
tree.c (substitute_placeholder_in_expr) <tcc_vl_exp>: Minor tweak.

From-SVN: r135591

16 years agoFix a typo.
H.J. Lu [Mon, 19 May 2008 22:07:24 +0000 (15:07 -0700)]
Fix a typo.

From-SVN: r135589

16 years agotree.c (substitute_in_expr): Fix thinko.
Eric Botcazou [Mon, 19 May 2008 21:42:48 +0000 (21:42 +0000)]
tree.c (substitute_in_expr): Fix thinko.

* tree.c (substitute_in_expr) <tcc_vl_exp>: Fix thinko.

From-SVN: r135582

16 years agoi386.c (ix86_expand_vector_init_concat): Change sizes of operand array from 8/4 to...
H.J. Lu [Mon, 19 May 2008 19:42:11 +0000 (19:42 +0000)]
i386.c (ix86_expand_vector_init_concat): Change sizes of operand array from 8/4 to 4/2.

2008-05-19  H.J. Lu  <hongjiu.lu@intel.com>

* gcc/config/i386/i386.c (ix86_expand_vector_init_concat): Change
sizes of operand array from 8/4 to 4/2.
(ix86_expand_vector_init_general): Change size of operand array
from 32 to 16.  Remove op0, op1 and half_mode.

From-SVN: r135579

16 years agoi386.c (ix86_expand_vector_init_concat): New.
H.J. Lu [Mon, 19 May 2008 18:54:44 +0000 (18:54 +0000)]
i386.c (ix86_expand_vector_init_concat): New.

2008-05-19  H.J. Lu  <hongjiu.lu@intel.com>

* config/i386/i386.c (ix86_expand_vector_init_concat): New.
(ix86_expand_vector_init_interleave): Likewise.
(ix86_expand_vector_init_general): Use them.  Assert word_mode
== SImode when n_words == 4.

From-SVN: r135578

16 years agore PR fortran/36265 (character issue with gfortran)
Francois-Xavier Coudert [Mon, 19 May 2008 15:57:28 +0000 (15:57 +0000)]
re PR fortran/36265 (character issue with gfortran)

PR fortran/36265

* trans-expr.c (gfc_conv_string_tmp): Pick the correct type for
the temporary variable.

* gfortran.dg/char_length_11.f90: New test.

From-SVN: r135576

16 years agowidechar_intrinsics_5.f90: Add dg-do directive and use -fbackslash option.
Francois-Xavier Coudert [Mon, 19 May 2008 15:19:05 +0000 (15:19 +0000)]
widechar_intrinsics_5.f90: Add dg-do directive and use -fbackslash option.

* gfortran.dg/widechar_intrinsics_5.f90: Add dg-do directive and
use -fbackslash option.

From-SVN: r135575

16 years agochar_cast_2.f90: Adjust count in scanning the tree dump file.
Francois-Xavier Coudert [Mon, 19 May 2008 14:28:19 +0000 (14:28 +0000)]
char_cast_2.f90: Adjust count in scanning the tree dump file.

* gfortran.dg/char_cast_2.f90: Adjust count in scanning the tree
dump file.

From-SVN: r135574

16 years ago* gfortran.dg/bind_c_module.f90: Adjust expected error messages.
Francois-Xavier Coudert [Mon, 19 May 2008 14:01:40 +0000 (14:01 +0000)]
* gfortran.dg/bind_c_module.f90: Adjust expected error messages.

From-SVN: r135530

16 years agoforgot to save changelog back when updating the doc on saturday.
Kenneth Zadeck [Mon, 19 May 2008 14:01:28 +0000 (14:01 +0000)]
forgot to save changelog back when updating the doc on saturday.

From-SVN: r135529

16 years agosimplify.c (gfc_simplify_dble, [...]): Initialize result variable to avoid warnings.
Francois-Xavier Coudert [Mon, 19 May 2008 12:22:39 +0000 (12:22 +0000)]
simplify.c (gfc_simplify_dble, [...]): Initialize result variable to avoid warnings.

* simplify.c (gfc_simplify_dble, gfc_simplify_real): Initialize
result variable to avoid warnings.

From-SVN: r135528

16 years agochar_cast_1.f90: Adjust count in scanning the tree dump file.
Francois-Xavier Coudert [Mon, 19 May 2008 10:58:39 +0000 (10:58 +0000)]
char_cast_1.f90: Adjust count in scanning the tree dump file.

* gfortran.dg/char_cast_1.f90: Adjust count in scanning the tree
dump file.

From-SVN: r135527

16 years agoi386.c (ix86_secondary_reload): New static function.
Uros Bizjak [Mon, 19 May 2008 08:14:04 +0000 (10:14 +0200)]
i386.c (ix86_secondary_reload): New static function.

        * config/i386/i386.c (ix86_secondary_reload): New static function.
        (TARGET_SECONDARY_RELOAD): New define.
        * config/i386/i386.h (SECONDARY_OUTPUT_RELOAD_CLASS): Remove.
        * config/i386/i386.md (reload_outqi): Remove.

From-SVN: r135526

16 years agoAdd file omitted from last checkin.
Kaveh Ghazi [Mon, 19 May 2008 06:57:43 +0000 (06:57 +0000)]
Add file omitted from last checkin.

From-SVN: r135525

16 years agoDaily bump.
GCC Administrator [Mon, 19 May 2008 00:17:38 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r135521

16 years agore PR middle-end/35509 (builtin isinf() mismatch to compile-time substitution)
Kaveh R. Ghazi [Sun, 18 May 2008 23:19:38 +0000 (23:19 +0000)]
re PR middle-end/35509 (builtin isinf() mismatch to compile-time substitution)

PR middle-end/35509

* builtins.c (mathfn_built_in_1): Renamed from mathfn_built_in.
Add `implicit' parameter.  Handle BUILT_IN_SIGNBIT.
(mathfn_built_in): Rewrite in terms of mathfn_built_in_1.
(fold_builtin_classify): Handle BUILT_IN_ISINF_SIGN.
(fold_builtin_1): Likewise.
* builtins.def (BUILT_IN_ISINF_SIGN): New.
c-common.c (check_builtin_function_arguments): Handle
BUILT_IN_ISINF_SIGN.
* doc/extend.texi: Document __builtin_isinf_sign.
* fold-const.c (operand_equal_p): Handle COND_EXPR.

testsuite:
* gcc.dg/builtins-error.c: Test __builtin_isinf_sign.
* gcc.dg/tg-tests.h: Likewise.  Mark variables volatile.
* gcc.dg/torture/builtin-isinf_sign-1.c: New test.

From-SVN: r135517

16 years agointrinsic.c (char_conversions, ncharconv): New static variables.
Francois-Xavier Coudert [Sun, 18 May 2008 22:45:05 +0000 (22:45 +0000)]
intrinsic.c (char_conversions, ncharconv): New static variables.

* intrinsic.c (char_conversions, ncharconv): New static variables.
(find_char_conv): New function.
(add_functions): Add simplification functions for ADJUSTL and
ADJUSTR. Don't check the kind of their argument. Add checking for
LGE, LLE, LGT and LLT.
(add_subroutines): Fix argument type for SLEEP. Fix argument name
for SYSTEM.
(add_char_conversions): New function.
(gfc_intrinsic_init_1): Call add_char_conversions.
(gfc_intrinsic_done_1): Free char_conversions.
(check_arglist): Use kind == 0 as a signal that we don't want
the kind value to be checked.
(do_simplify): Also simplify character functions.
(gfc_convert_chartype): New function
* trans-array.c (gfc_trans_array_ctor_element): Don't force the
use of default character type.
(gfc_trans_array_constructor_value): Likewise.
(get_array_ctor_var_strlen): Use integer kind to build an integer
instead of a character kind!
(gfc_build_constant_array_constructor): Don't force the use of
default character type.
(gfc_conv_loop_setup): Likewise.
* trans-expr.c (gfc_conv_string_tmp): Don't force the use of
default character type. Allocate enough memory for wide strings.
(gfc_conv_concat_op): Make sure operand kind are the same.
(string_to_single_character): Remove gfc_ prefix. Reindent.
Don't force the use of default character type.
(gfc_conv_scalar_char_value): Likewise.
(gfc_build_compare_string): Call string_to_single_character.
(fill_with_spaces): New function
(gfc_trans_string_copy): Add kind arguments. Use them to deal
with wide character kinds.
(gfc_conv_statement_function): Whitespace fix. Call
gfc_trans_string_copy with new kind arguments.
(gfc_conv_substring_expr): Call gfc_build_wide_string_const
instead of using gfc_widechar_to_char.
(gfc_conv_string_parameter): Don't force the use of default
character type.
(gfc_trans_scalar_assign): Pass kind args to gfc_trans_string_copy.
* intrinsic.h (gfc_check_lge_lgt_lle_llt, gfc_convert_char_constant,
gfc_resolve_adjustl, gfc_resolve_adjustr): New prototypes.
* decl.c (gfc_set_constant_character_len): Don't assert the
existence of a single character kind.
* trans-array.h (gfc_trans_string_copy): New prototype.
* gfortran.h (gfc_check_character_range, gfc_convert_chartype):
New prototypes.
* error.c (print_wide_char_into_buffer): New function lifting
code from gfc_print_wide_char. Fix order to output '\x??' instead
of 'x\??'.
(gfc_print_wide_char): Call print_wide_char_into_buffer.
(show_locus): Call print_wide_char_into_buffer with buffer local
to this function.
* trans-const.c (gfc_build_wide_string_const): New function.
(gfc_conv_string_init): Deal with wide characters strings
constructors.
(gfc_conv_constant_to_tree): Call gfc_build_wide_string_const.
* trans-stmt.c (gfc_trans_label_assign): Likewise.
(gfc_trans_character_select): Deal with wide strings.
* expr.c (gfc_check_assign): Allow conversion between character
kinds on assignment.
* trans-const.h (gfc_build_wide_string_const): New prototype.
* trans-types.c (gfc_get_character_type_len_for_eltype,
gfc_get_character_type_len): Create too variants of the old
gfc_get_character_type_len, one getting kind argument and the
other one directly taking a type tree.
* trans.h (gfor_fndecl_select_string_char4,
gfor_fndecl_convert_char1_to_char4,
gfor_fndecl_convert_char4_to_char1): New prototypes.
* trans-types.h (gfc_get_character_type_len_for_eltype): New
prototype.
* resolve.c (resolve_operator): Exit early when kind mismatches
are detected, because that makes us issue an error message later.
(validate_case_label_expr): Fix wording of error message.
* iresolve.c (gfc_resolve_adjustl, gfc_resolve_adjustr): New
functions.
(gfc_resolve_pack): Call _char4 variants of library function
when dealing with wide characters.
(gfc_resolve_reshape): Likewise.
(gfc_resolve_spread): Likewise.
(gfc_resolve_transpose): Likewise.
(gfc_resolve_unpack): Likewise.
* target-memory.c (size_character): Take character kind bit size
correctly into account (not that it changes anything for now, but
it's more generic).
(gfc_encode_character): Added gfc_ prefix. Encoding each
character of a string by calling native_encode_expr for the
corresponding unsigned integer.
(gfc_target_encode_expr): Add gfc_ prefix to encode_character.
* trans-decl.c (gfc_build_intrinsic_function_decls): Build
gfor_fndecl_select_string_char4, gfor_fndecl_convert_char1_to_char4
and gfor_fndecl_convert_char4_to_char1.
* target-memory.h (gfc_encode_character): New prototype.
* arith.c (gfc_check_character_range): New function.
(eval_intrinsic): Allow non-default character kinds.
* check.c (gfc_check_access_func): Only allow default
character kind arguments.
(gfc_check_chdir): Likewise.
(gfc_check_chdir_sub): Likewise.
(gfc_check_chmod): Likewise.
(gfc_check_chmod_sub): Likewise.
(gfc_check_lge_lgt_lle_llt): New function.
(gfc_check_link): Likewise.
(gfc_check_link_sub): Likewise.
(gfc_check_symlnk): Likewise.
(gfc_check_symlnk_sub): Likewise.
(gfc_check_rename): Likewise.
(gfc_check_rename_sub): Likewise.
(gfc_check_fgetputc_sub): Likewise.
(gfc_check_fgetput_sub): Likewise.
(gfc_check_stat): Likewise.
(gfc_check_stat_sub): Likewise.
(gfc_check_date_and_time): Likewise.
(gfc_check_ctime_sub): Likewise.
(gfc_check_fdate_sub): Likewise.
(gfc_check_gerror): Likewise.
(gfc_check_getcwd_sub): Likewise.
(gfc_check_getarg): Likewise.
(gfc_check_getlog): Likewise.
(gfc_check_hostnm): Likewise.
(gfc_check_hostnm_sub): Likewise.
(gfc_check_ttynam_sub): Likewise.
(gfc_check_perror): Likewise.
(gfc_check_unlink): Likewise.
(gfc_check_unlink_sub): Likewise.
(gfc_check_system_sub): Likewise.
* primary.c (got_delim): Perform correct character range checking
for all kinds.
* trans-intrinsic.c (gfc_conv_intrinsic_conversion): Generate
calls to library functions convert_char4_to_char1 and
convert_char1_to_char4 for character conversions.
(gfc_conv_intrinsic_char): Allow all character kings.
(gfc_conv_intrinsic_strcmp): Fix whitespace.
(gfc_conv_intrinsic_repeat): Take care of all character kinds.
* intrinsic.texi: For all GNU intrinsics accepting character
arguments, mention that they're restricted to the default kind.
* simplify.c (simplify_achar_char): New function.
(gfc_simplify_achar, gfc_simplify_char): Call simplify_achar_char.
gfc_simplify_ichar): Don't error out for wide characters.
(gfc_convert_char_constant): New function.

* gfortran.dg/achar_3.f90: Adjust error messages.
* gfortran.dg/achar_5.f90: New test.
* gfortran.dg/achar_6.F90: New test.
* gfortran.dg/widechar_1.f90: New test.
* gfortran.dg/widechar_2.f90: New test.
* gfortran.dg/widechar_3.f90: New test.
* gfortran.dg/widechar_4.f90: New test.
* gfortran.dg/widechar_intrinsics_1.f90: New test.
* gfortran.dg/widechar_intrinsics_2.f90: New test.
* gfortran.dg/widechar_intrinsics_3.f90: New test.
* gfortran.dg/widechar_intrinsics_4.f90: New test.
* gfortran.dg/widechar_intrinsics_5.f90: New test.
* gfortran.dg/widechar_select_1.f90: New test.
* gfortran.dg/widechar_select_2.f90: New test.

From-SVN: r135515

16 years agotree-ssa-dom.c (tree_ssa_dominator_optimize): If some blocks need EH cleanup at the...
Eric Botcazou [Sun, 18 May 2008 22:36:30 +0000 (22:36 +0000)]
tree-ssa-dom.c (tree_ssa_dominator_optimize): If some blocks need EH cleanup at the end of the pass...

* tree-ssa-dom.c (tree_ssa_dominator_optimize): If some blocks need
EH cleanup at the end of the pass, search for those that have been
turned into forwarder blocks and do the cleanup on their successor.

From-SVN: r135514

16 years agoUpdate ChangeLog which was forgotten in the check in:
Tobias Burnus [Sun, 18 May 2008 21:44:42 +0000 (23:44 +0200)]
Update ChangeLog which was forgotten in the check in:

> r135512 | tkoenig | 2008-05-18 22:39:04 +0200 (Sun, 18 May 2008) | 158 lines

From-SVN: r135513

16 years agoin_pack.m4 (internal_pack_'rtype_code`): Destination pointer is restrict.
Thomas Koenig [Sun, 18 May 2008 20:39:04 +0000 (20:39 +0000)]
in_pack.m4 (internal_pack_'rtype_code`): Destination pointer is restrict.

2008-05-18  Thomas Koenig  <tkoenig@gcc.gnu.org>

* m4/in_pack.m4 (internal_pack_'rtype_code`):  Destination
pointer is restrict.
* m4/transpose.m4 (transpose_'rtype_code`):  Likewise.
* m4/pack.m4 (pack_'rtype_code`):  Likewise.
* m4/spread.m4 (spread_'rtype_code`):  Likewise.
(spread_scalar_'rtype_code`): Likewise.
* m4/iforeach.m4 (name`'rtype_qual`_'atype_code):  Likewise.
* m4/eoshift1.m4 (eoshift1):  Likewise.
* m4/eoshift3.m4 (eoshift3):  Likewise.
* m4/in_unpack.m4 (internal_unpack_'rtype_ccode`):  Likewise.
* m4/unpack.m4 (unpack0_'rtype_code`):  Likewise.
(unpack1_'rtype_code`):  Likewise.
* intrinsics/pack_generic.c (pack_generic.c):  Likewise.
* intrinsics/unpack_generic.c (unpack_internal):  Likewise.
* intrinsics/eoshift0.c (eoshift0):  Likewise.
* intrinsics/eoshift2.c (eoshift2):  Likewise.
* intrinsics/reshape_generic.c (reshape_internal):  Likewise.
* intrinsics/reshape_packed.c (reshape_packed):  Likewise.
* generated/eoshift1_16.c: Regenerated.
* generated/eoshift1_4.c: Regenerated.
* generated/eoshift1_8.c: Regenerated.
* generated/eoshift3_16.c: Regenerated.
* generated/eoshift3_4.c: Regenerated.
* generated/eoshift3_8.c: Regenerated.
* generated/in_pack_c10.c: Regenerated.
* generated/in_pack_c16.c: Regenerated.
* generated/in_pack_c4.c: Regenerated.
* generated/in_pack_c8.c: Regenerated.
* generated/in_pack_i1.c: Regenerated.
* generated/in_pack_i16.c: Regenerated.
* generated/in_pack_i2.c: Regenerated.
* generated/in_pack_i4.c: Regenerated.
* generated/in_pack_i8.c: Regenerated.
* generated/in_pack_r10.c: Regenerated.
* generated/in_pack_r16.c: Regenerated.
* generated/in_pack_r4.c: Regenerated.
* generated/in_pack_r8.c: Regenerated.
* generated/in_unpack_c10.c: Regenerated.
* generated/in_unpack_c16.c: Regenerated.
* generated/in_unpack_c4.c: Regenerated.
* generated/in_unpack_c8.c: Regenerated.
* generated/in_unpack_i1.c: Regenerated.
* generated/in_unpack_i16.c: Regenerated.
* generated/in_unpack_i2.c: Regenerated.
* generated/in_unpack_i4.c: Regenerated.
* generated/in_unpack_i8.c: Regenerated.
* generated/in_unpack_r10.c: Regenerated.
* generated/in_unpack_r16.c: Regenerated.
* generated/in_unpack_r4.c: Regenerated.
* generated/in_unpack_r8.c: Regenerated.
* generated/maxloc0_16_i1.c: Regenerated.
* generated/maxloc0_16_i16.c: Regenerated.
* generated/maxloc0_16_i2.c: Regenerated.
* generated/maxloc0_16_i4.c: Regenerated.
* generated/maxloc0_16_i8.c: Regenerated.
* generated/maxloc0_16_r10.c: Regenerated.
* generated/maxloc0_16_r16.c: Regenerated.
* generated/maxloc0_16_r4.c: Regenerated.
* generated/maxloc0_16_r8.c: Regenerated.
* generated/maxloc0_4_i1.c: Regenerated.
* generated/maxloc0_4_i16.c: Regenerated.
* generated/maxloc0_4_i2.c: Regenerated.
* generated/maxloc0_4_i4.c: Regenerated.
* generated/maxloc0_4_i8.c: Regenerated.
* generated/maxloc0_4_r10.c: Regenerated.
* generated/maxloc0_4_r16.c: Regenerated.
* generated/maxloc0_4_r4.c: Regenerated.
* generated/maxloc0_4_r8.c: Regenerated.
* generated/maxloc0_8_i1.c: Regenerated.
* generated/maxloc0_8_i16.c: Regenerated.
* generated/maxloc0_8_i2.c: Regenerated.
* generated/maxloc0_8_i4.c: Regenerated.
* generated/maxloc0_8_i8.c: Regenerated.
* generated/maxloc0_8_r10.c: Regenerated.
* generated/maxloc0_8_r16.c: Regenerated.
* generated/maxloc0_8_r4.c: Regenerated.
* generated/maxloc0_8_r8.c: Regenerated.
* generated/minloc0_16_i1.c: Regenerated.
* generated/minloc0_16_i16.c: Regenerated.
* generated/minloc0_16_i2.c: Regenerated.
* generated/minloc0_16_i4.c: Regenerated.
* generated/minloc0_16_i8.c: Regenerated.
* generated/minloc0_16_r10.c: Regenerated.
* generated/minloc0_16_r16.c: Regenerated.
* generated/minloc0_16_r4.c: Regenerated.
* generated/minloc0_16_r8.c: Regenerated.
* generated/minloc0_4_i1.c: Regenerated.
* generated/minloc0_4_i16.c: Regenerated.
* generated/minloc0_4_i2.c: Regenerated.
* generated/minloc0_4_i4.c: Regenerated.
* generated/minloc0_4_i8.c: Regenerated.
* generated/minloc0_4_r10.c: Regenerated.
* generated/minloc0_4_r16.c: Regenerated.
* generated/minloc0_4_r4.c: Regenerated.
* generated/minloc0_4_r8.c: Regenerated.
* generated/minloc0_8_i1.c: Regenerated.
* generated/minloc0_8_i16.c: Regenerated.
* generated/minloc0_8_i2.c: Regenerated.
* generated/minloc0_8_i4.c: Regenerated.
* generated/minloc0_8_i8.c: Regenerated.
* generated/minloc0_8_r10.c: Regenerated.
* generated/minloc0_8_r16.c: Regenerated.
* generated/minloc0_8_r4.c: Regenerated.
* generated/minloc0_8_r8.c: Regenerated.
* generated/pack_c10.c: Regenerated.
* generated/pack_c16.c: Regenerated.
* generated/pack_c4.c: Regenerated.
* generated/pack_c8.c: Regenerated.
* generated/pack_i1.c: Regenerated.
* generated/pack_i16.c: Regenerated.
* generated/pack_i2.c: Regenerated.
* generated/pack_i4.c: Regenerated.
* generated/pack_i8.c: Regenerated.
* generated/pack_r10.c: Regenerated.
* generated/pack_r16.c: Regenerated.
* generated/pack_r4.c: Regenerated.
* generated/pack_r8.c: Regenerated.
* generated/spread_c10.c: Regenerated.
* generated/spread_c16.c: Regenerated.
* generated/spread_c4.c: Regenerated.
* generated/spread_c8.c: Regenerated.
* generated/spread_i1.c: Regenerated.
* generated/spread_i16.c: Regenerated.
* generated/spread_i2.c: Regenerated.
* generated/spread_i4.c: Regenerated.
* generated/spread_i8.c: Regenerated.
* generated/spread_r10.c: Regenerated.
* generated/spread_r16.c: Regenerated.
* generated/spread_r4.c: Regenerated.
* generated/spread_r8.c: Regenerated.
* generated/transpose_c10.c: Regenerated.
* generated/transpose_c16.c: Regenerated.
* generated/transpose_c4.c: Regenerated.
* generated/transpose_c8.c: Regenerated.
* generated/transpose_i16.c: Regenerated.
* generated/transpose_i4.c: Regenerated.
* generated/transpose_i8.c: Regenerated.
* generated/transpose_r10.c: Regenerated.
* generated/transpose_r16.c: Regenerated.
* generated/transpose_r4.c: Regenerated.
* generated/transpose_r8.c: Regenerated.
* generated/unpack_c10.c: Regenerated.
* generated/unpack_c16.c: Regenerated.
* generated/unpack_c4.c: Regenerated.
* generated/unpack_c8.c: Regenerated.
* generated/unpack_i1.c: Regenerated.
* generated/unpack_i16.c: Regenerated.
* generated/unpack_i2.c: Regenerated.
* generated/unpack_i4.c: Regenerated.
* generated/unpack_i8.c: Regenerated.
* generated/unpack_r10.c: Regenerated.
* generated/unpack_r16.c: Regenerated.
* generated/unpack_r4.c: Regenerated.
* generated/unpack_r8.c: Regenerated.

From-SVN: r135512

16 years agore PR target/36090 (ppc64 cacoshl miscompilation)
Jakub Jelinek [Sun, 18 May 2008 20:19:03 +0000 (22:19 +0200)]
re PR target/36090 (ppc64 cacoshl miscompilation)

PR target/36090
* gcc.c-torture/execute/20080502-1.c: New test.

From-SVN: r135508

16 years agostat_1.f90: Skip on mingw.
Francois-Xavier Coudert [Sun, 18 May 2008 17:55:39 +0000 (17:55 +0000)]
stat_1.f90: Skip on mingw.

* gfortran.dg/stat_1.f90: Skip on mingw.
* gfortran.dg/stat_2.f90: Skip on mingw.

From-SVN: r135506

16 years agorevert: i386.c (ix86_secondary_reload): New static function.
Uros Bizjak [Sun, 18 May 2008 17:00:00 +0000 (19:00 +0200)]
revert: i386.c (ix86_secondary_reload): New static function.

Revert:
2008-05-18  Uros Bizjak  <ubizjak@gmail.com>

        * config/i386/i386.c (ix86_secondary_reload): New static function.
        (TARGET_SECONDARY_RELOAD): New define.
        * config/i386/i386.h (SECONDARY_OUTPUT_RELOAD_CLASS): Remove.
        * config/i386/i386.md (reload_outqi): Remove.

From-SVN: r135505

16 years agotree-cfg.c (verify_gimple_expr): Allow conversions from pointers to sizetype and...
Richard Guenther [Sun, 18 May 2008 16:34:39 +0000 (16:34 +0000)]
tree-cfg.c (verify_gimple_expr): Allow conversions from pointers to sizetype and vice versa.

2008-05-18  Richard Guenther  <rguenther@suse.de>

* tree-cfg.c (verify_gimple_expr): Allow conversions from
pointers to sizetype and vice versa.

From-SVN: r135504

16 years agoi386.c (ix86_secondary_reload): New static function.
Uros Bizjak [Sun, 18 May 2008 13:42:01 +0000 (15:42 +0200)]
i386.c (ix86_secondary_reload): New static function.

* config/i386/i386.c (ix86_secondary_reload): New static function.
(TARGET_SECONDARY_RELOAD): New define.
* config/i386/i386.h (SECONDARY_OUTPUT_RELOAD_CLASS): Remove.
* config/i386/i386.md (reload_outqi): Remove.

From-SVN: r135499

16 years agoselect.c: Moved content to select_inc.c.
Francois-Xavier Coudert [Sun, 18 May 2008 12:00:20 +0000 (12:00 +0000)]
select.c: Moved content to select_inc.c.

        * runtime/select.c: Moved content to select_inc.c. Include it.
        Add macros for different character types.
        * runtime/select_inc.c: New file.
        * runtime/convert_char.c: New file.
        * intrinsics/pack_generic.c (pack_char4, pack_s_char4): New
        functions.
        * intrinsics/transpose_generic.c (transpose_char4): New function.
        * intrinsics/spread_generic.c (spread_char4, spread_char4_scalar):
        New functions.
        * intrinsics/unpack_generic.c (unpack1_char4, unpack0_char4):
        New functions.
        * intrinsics/reshape_generic.c (reshape_char): Use
        gfc_charlen_type as type for length variables.
        (reshape_char4): New function.
        * gfortran.map (GFORTRAN_1.1): Add _gfortran_select_string_char4,
        _gfortran_convert_char1_to_char4, _gfortran_convert_char4_to_char1,
        _gfortran_transpose_char4, _gfortran_spread_char4,
        _gfortran_spread_char4_scalar, _gfortran_reshape_char4,
        _gfortran_pack_char4, _gfortran_pack_s_char4,
        _gfortran_unpack0_char4 and _gfortran_unpack1_char4.
        * Makefile.am: Add runtime/convert_char.c.
        * Makefile.in: Regenerate.

From-SVN: r135496

16 years agore PR fortran/36251 (PUBLIC and PRIVATE abuse)
Tobias Burnus [Sun, 18 May 2008 11:10:11 +0000 (13:10 +0200)]
re PR fortran/36251 (PUBLIC and PRIVATE abuse)

2008-05-18  Steven G. Kargl  <kargls@comcast.net>

        PR fortran/36251
        * symbol.c (check_conflict): Issue errors for abuse of PUBLIC,
        * PRIVATE,
        and BIND(C).
        * resolve.c (gfc_verify_binding_labels): Fix NULL pointer
        * dereference.

2008-05-18  Steven G. Kargl  <kargls@comcast.net>

        PR fortran/36251
        gfortran.dg/public_private_module.f90: new test.
        gfortran.dg/bind_c_module.f90: new test.

From-SVN: r135495

16 years agocomment/style fixes
Xinliang David Li [Sun, 18 May 2008 04:37:59 +0000 (04:37 +0000)]
comment/style fixes

From-SVN: r135493

16 years agoDaily bump.
GCC Administrator [Sun, 18 May 2008 00:18:07 +0000 (00:18 +0000)]
Daily bump.

From-SVN: r135490

16 years agosh.c (sh_output_mi_thunk): Update the use of init_flow.
Kaz Kojima [Sat, 17 May 2008 23:27:27 +0000 (23:27 +0000)]
sh.c (sh_output_mi_thunk): Update the use of init_flow.

* config/sh/sh.c (sh_output_mi_thunk): Update the use of
init_flow.

From-SVN: r135483

16 years agointrinsic.texi: Correct description of GET_COMMAND_ARGUMENT and GET_ENVIRONMENT_VARIA...
Tobias Burnus [Sat, 17 May 2008 22:28:51 +0000 (00:28 +0200)]
intrinsic.texi: Correct description of GET_COMMAND_ARGUMENT and GET_ENVIRONMENT_VARIABLE...

2008-05-16  Tobias Burnus  <burnus@net-b.de>

        * intrinsic.texi: Correct description of GET_COMMAND_ARGUMENT
        and GET_ENVIRONMENT_VARIABLE; fix keyword= name for GETENV,
        GETLOG, GMTIME, HOSTNM, IRAND, ITIME, KILL.
        Move LOG_GAMMA after LOG10.

From-SVN: r135482

16 years agolist_read.c (list_formatted_read_scalar): Declare type as volatile to shut up compile...
Thomas Koenig [Sat, 17 May 2008 22:24:06 +0000 (22:24 +0000)]
list_read.c (list_formatted_read_scalar): Declare type as volatile to shut up compiler warning.

2008-05-17  Thomas Koenig  <tkoenig@gcc.gnu.org>

* io/list_read.c (list_formatted_read_scalar):  Declare
type as volatile to shut up compiler warning.

From-SVN: r135480