gcc.git
15 years agoi376.md (UNSPEC_MS_TO_SYSV_CALL): New constant.
Jan Hubicka [Sun, 21 Dec 2008 13:28:26 +0000 (14:28 +0100)]
i376.md (UNSPEC_MS_TO_SYSV_CALL): New constant.

* i376.md (UNSPEC_MS_TO_SYSV_CALL): New constant.
(call_1_rex64_ms_sysv, call_value_0_rex64_ms_sysv,
call_value_1_rex64_ms_sysv): New patterns.
* i386.c (function_arg_ms_64): Pass magical value of -2 as callarg.
(ix86_expand_call): Emit extra clobbers for ms->sysv ABI calls.

Co-Authored-By: Kai Tietz <kai.tietz@onevision.com>
From-SVN: r142859

15 years agoalpha.c (alpha_pad_noreturn): New static function.
Uros Bizjak [Sun, 21 Dec 2008 11:50:59 +0000 (12:50 +0100)]
alpha.c (alpha_pad_noreturn): New static function.

* config/alpha/alpha.c (alpha_pad_noreturn): New static function.
(alpha_reorg): Call alpha_pad_noreturn.

From-SVN: r142858

15 years ago* config/avr/avr.md ("andsi3"): Fix wrong cc attribute.
Denis Chertykov [Sun, 21 Dec 2008 06:16:02 +0000 (06:16 +0000)]
* config/avr/avr.md ("andsi3"): Fix wrong cc attribute.

From-SVN: r142856

15 years agoDaily bump.
GCC Administrator [Sun, 21 Dec 2008 00:16:52 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r142853

15 years agore PR target/37610 (FAIL: g++.dg/eh/pr29166.C execution test)
Eric Botcazou [Sat, 20 Dec 2008 22:32:30 +0000 (22:32 +0000)]
re PR target/37610 (FAIL: g++.dg/eh/pr29166.C execution test)

PR target/37610
* configure.ac (gcc_cv_as_cfi_advance_working): Fall back to 'no'
if readelf is nowhere to be found.
* configure: Regenerate.

From-SVN: r142850

15 years agore PR c++/36921 (warning "comparison does not have mathematical meaning" is not corre...
Jakub Jelinek [Sat, 20 Dec 2008 18:46:12 +0000 (19:46 +0100)]
re PR c++/36921 (warning "comparison does not have mathematical meaning" is not correct for overloaded operators that do not return boolean)

PR c++/36921
* c-common.c (warn_about_parentheses): Remove ARG_UNUSED from
arg_left.  Don't warn about X<=Y<=Z if comparison's type isn't
integral.

* g++.dg/warn/pr36921.C: New.

Co-Authored-By: Manuel López-Ibáñez <manu@gcc.gnu.org>
From-SVN: r142849

15 years agotarget-supports.exp: Add *-*-rtems* to list of targets without profiling support.
Joel Sherrill [Sat, 20 Dec 2008 15:28:57 +0000 (15:28 +0000)]
target-supports.exp: Add *-*-rtems* to list of targets without profiling support.

2008-12-19  Joel Sherrill <joel.sherrill@oarcorp.com>

* lib/target-supports.exp: Add *-*-rtems* to list
of targets without profiling support.

From-SVN: r142848

15 years agoDaily bump.
GCC Administrator [Sat, 20 Dec 2008 00:17:04 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r142846

15 years agore PR c++/38577 (ICE: tree check: expected call_expr, have compound_expr in build_new...
Jakub Jelinek [Fri, 19 Dec 2008 19:33:28 +0000 (20:33 +0100)]
re PR c++/38577 (ICE: tree check: expected call_expr, have compound_expr in build_new_method_call, at cp/call.c:6000)

PR c++/38577
* call.c (build_new_method_call): Handle call being COMPOUND_EXPR
or NOP_EXPR.

* g++.dg/template/call6.C: New test.

From-SVN: r142842

15 years agorevert: re PR libfortran/24685 (real(16) formatted input is broken for huge values...
Janis Johnson [Fri, 19 Dec 2008 18:12:40 +0000 (18:12 +0000)]
revert: re PR libfortran/24685 (real(16) formatted input is broken for huge values (gfortran.dg/default_format_2.f90))

Revert:
2008-12-12  Janis Johnson  <janis187@us.ibm.com>
PR libgfortran/24685
* gfortran.dg/default_format_denormal_2.f90: Change XFAIL to check
for size of long double.

From-SVN: r142840

15 years agobuiltins.c, [...]: Fix comments.
Andrew Haley [Fri, 19 Dec 2008 17:53:51 +0000 (17:53 +0000)]
builtins.c, [...]: Fix comments.

2008-12-19  Andrew Haley  <aph@redhat.com>

* builtins.c, tree-ssa-loop-ivopts.c, fixed-value.c:
Fix comments.

From-SVN: r142839

15 years agore PR target/38548 (bootstrap broken on arm-linux-gnu (not gnueabi))
Richard Earnshaw [Fri, 19 Dec 2008 17:31:12 +0000 (17:31 +0000)]
re PR target/38548 (bootstrap broken on arm-linux-gnu (not gnueabi))

PR target/38548
* arm/t-linux (LIB1ASMFUNCS): Add _arm_addsubdf3 and
_arm_addsubsf3.
* arm/lib1funcs.asm (clzsi2): Use RET macro for return
instruction.

From-SVN: r142838

15 years agore PR bootstrap/38578 (fatal warning during bootstrap on arm.c for output_move_double...
Richard Earnshaw [Fri, 19 Dec 2008 17:22:58 +0000 (17:22 +0000)]
re PR bootstrap/38578 (fatal warning during bootstrap on arm.c for output_move_double and arm_expand_prologue)

PR bootstrap/38578
* arm.c (load_multiple_sequence): Initialize ORDER array.
(store_multiple_sequence): Likewise.
(output_move_double): Make reg0 unsigned.
(arm_output_epilogue): Make amount unsigned.
(arm_expand_prologue): Move declaration of dwarf before block
statements.

From-SVN: r142837

15 years ago* df-scan.c ( df_hard_reg_init): Move declaration of i.
Steve Ellcey [Fri, 19 Dec 2008 16:34:26 +0000 (16:34 +0000)]
* df-scan.c ( df_hard_reg_init): Move declaration of i.

From-SVN: r142836

15 years agore PR libffi/26048 (libffi doesn't build on Solaris 10/x86 with native assembler)
Rainer Orth [Fri, 19 Dec 2008 14:59:42 +0000 (14:59 +0000)]
re PR libffi/26048 (libffi doesn't build on Solaris 10/x86 with native assembler)

PR libffi/26048
* configure.ac (HAVE_AS_X86_PCREL): New test.
* configure: Regenerate.
* fficonfig.h.in: Regenerate.
* src/x86/sysv.S [!FFI_NO_RAW_API]: Precalculate
RAW_CLOSURE_CIF_OFFSET, RAW_CLOSURE_FUN_OFFSET,
RAW_CLOSURE_USER_DATA_OFFSET for the Solaris 10/x86 assembler.
(.eh_frame): Only use SYMBOL-. iff HAVE_AS_X86_PCREL.
* src/x86/unix64.S (.Lstore_table): Move to .text section.
(.Lload_table): Likewise.
(.eh_frame): Only use SYMBOL-. iff HAVE_AS_X86_PCREL.

From-SVN: r142835

15 years agore PR libgcj/38396 (ecj1 linked against both -lgcj and -lgcj_bc)
Jakub Jelinek [Fri, 19 Dec 2008 14:57:29 +0000 (15:57 +0100)]
re PR libgcj/38396 (ecj1 linked against both -lgcj and -lgcj_bc)

PR libgcj/38396
* configure.ac (use_libgcj_bc): Set to no if not enable_shared.
(LIBGCJ_SPEC): Use -lgcj instead of -lgcj_bc even for -static
or -static-libgcj.
* Makefile.am (ecjx_SOURCES): Add ecjx.cc.
(ecjx_LDADD): Don't add libgcj.la when
NATIVE && USE_LIBBGCJ_BC.
* ecjx.cc: New file.
* Makefile.in: Regenerated.
* configure: Regenerated.

From-SVN: r142834

15 years agore PR bootstrap/37739 (bootstrap broken with core gcc > gcc-4.2.x)
Jakub Jelinek [Fri, 19 Dec 2008 14:55:42 +0000 (15:55 +0100)]
re PR bootstrap/37739 (bootstrap broken with core gcc > gcc-4.2.x)

PR bootstrap/37739
* config.host: For powerpc*-*-linux* host with 32-bit GCC,
use rs6000/x-linux-relax snippet if ld is new enough,
otherwise use rs6000/x-linux-O1.
* config/rs6000/x-linux-relax: New file.
* config/x-cflags-O1: New file.

From-SVN: r142833

15 years agoacinclude.m4: Guard a variable test against an empty string.
Ben Elliston [Fri, 19 Dec 2008 09:18:41 +0000 (09:18 +0000)]
acinclude.m4: Guard a variable test against an empty string.

* acinclude.m4: Guard a variable test against an empty string.
* configure: Regenerate.

From-SVN: r142829

15 years agoDaily bump.
GCC Administrator [Fri, 19 Dec 2008 00:16:45 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r142827

15 years agors6000.c (rs6000_generate_compare): Condition choice of e500 comparison instructions...
Joseph Myers [Thu, 18 Dec 2008 22:13:05 +0000 (22:13 +0000)]
rs6000.c (rs6000_generate_compare): Condition choice of e500 comparison instructions on flag_finite_math_only &&...

* config/rs6000/rs6000.c (rs6000_generate_compare): Condition
choice of e500 comparison instructions on flag_finite_math_only &&
!flag_trapping_math, not flag_unsafe_math_optimizations.
* config/rs6000/rs6000.md (abstf2): Condition choice of e500
instructions on flag_finite_math_only && !flag_trapping_math, not
flag_unsafe_math_optimizations.
(bltgt, sltgt): Disable for TARGET_HARD_FLOAT && !TARGET_FPRS.
* config/rs6000/spe.md (cmpsfeq_gpr, tstsfeq_gpr, cmpsfgt_gpr,
tstsfgt_gpr, cmpsflt_gpr, tstsflt_gpr, cmpdfeq_gpr, tstdfeq_gpr,
cmpdfgt_gpr, tstdfgt_gpr, cmpdflt_gpr, tstdflt_gpr, cmptfeq_gpr,
tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr, tsttflt_gpr):
Condition choice of comparison instructions on
flag_finite_math_only && !flag_trapping_math, not
flag_unsafe_math_optimizations.

From-SVN: r142822

15 years agore PR middle-end/38565 (ICE with vector code)
Andrew Pinski [Thu, 18 Dec 2008 22:00:46 +0000 (22:00 +0000)]
re PR middle-end/38565 (ICE with vector code)

2008-12-18  Andrew Pinski  <andrew_pinski@playstation.sony.com>

        PR middle-end/38565
        * testsuite/g++.dg/torture/pr38565.C: New test.

From-SVN: r142821

15 years ago21674.cc: Add xfail on all targets except darwin.
Jack Howarth [Thu, 18 Dec 2008 21:08:28 +0000 (21:08 +0000)]
21674.cc: Add xfail on all targets except darwin.

2008-12-18  Jack Howarth  <howarth@bromo.med.uc.edu>

* testsuite/21_strings/basic_string/element_access/wchar_t/21674.cc:
Add xfail on all targets except darwin.
* libstdc++-v3/testsuite/21_strings/basic_string/element_access/char/
1674.cc: Same.
* libstdc++-v3/testsuite/22_locale/ctype/is/char/2.cc: Correct syntax
for xfailing on all targets except linux.

From-SVN: r142819

15 years agore PR c++/38427 (crash for reference init code)
Jakub Jelinek [Thu, 18 Dec 2008 20:51:07 +0000 (21:51 +0100)]
re PR c++/38427 (crash for reference init code)

PR c++/38427
* init.c (perform_member_init): For value-initialized
references call permerror instead of warning and don't emit any
INIT_EXPR.

* g++.dg/init/ctor9.C: New test.

From-SVN: r142818

15 years agoBackport from upstream Libtool:
Ralf Wildenhues [Thu, 18 Dec 2008 20:04:55 +0000 (20:04 +0000)]
Backport from upstream Libtool:

* libltdl.m4 (_LT_SYS_DYNAMIC_LINKER)
(_LT_LINKER_SHLIBS): Add cache variables to tests that
require the linker to work.  For shlibpath_overrides_runpath,
this also changes the semantics to let the result from the C
compiler take precedence.

boehm-gc/

* configure: Regenerate.

fixincludes/

* configure: Regenerate.

gcc/

* configure: Regenerate.

libffi/

* configure: Regenerate.

libgfortran/

* configure: Regenerate.

libgomp/

* configure: Regenerate.

libjava/classpath/

* configure: Regenerate.

libjava/

* configure: Regenerate.

libmudflap/

* configure: Regenerate.

libobjc/

* configure: Regenerate.

libssp/

* configure: Regenerate.

libstdc++-v3/

* configure: Regenerate.

zlib/

* configure: Regenerate.

From-SVN: r142817

15 years agore PR c++/38485 (GCC can't parse a parenthesized comma in a template-id within a...
Jason Merrill [Thu, 18 Dec 2008 15:54:59 +0000 (10:54 -0500)]
re PR c++/38485 (GCC can't parse a parenthesized comma in a template-id within a default argument)

        PR c++/38485
        * parser.c (cp_parser_token_starts_cast_expression): An EOF
        can't start a cast-expression.

From-SVN: r142815

15 years agore PR c++/38561 (Revision 142799 may cause C++ demangle failures)
Jason Merrill [Thu, 18 Dec 2008 14:51:56 +0000 (09:51 -0500)]
re PR c++/38561 (Revision 142799 may cause C++ demangle failures)

        PR c++/38561
        * cp-demangle.c (d_expression, d_print_comp): Revert
        cast changes.

From-SVN: r142814

15 years agoavoid target dependency
Jason Merrill [Thu, 18 Dec 2008 14:31:01 +0000 (09:31 -0500)]
avoid target dependency

From-SVN: r142813

15 years agodf-scan.c (persistent_obstack, [...]): Remove.
Jan Hubicka [Thu, 18 Dec 2008 13:52:45 +0000 (14:52 +0100)]
df-scan.c (persistent_obstack, [...]): Remove.

* df-scan.c (persistent_obstack, df_invalidated_by_call): Remove.
(df_scan_start_dump, df_get_call_refs, df_hard_reg_init): Rename
df_invalidated_by_call to invalidated_by_call_regset.
* df.h (df_invalidated_by_call): Remove.
* regclass.c (regs_invalidated_by_call_regset, persistent_obstack): New
variables.
(init_reg_sets_1): Initialize regs_invalidated_by_call_regset.
(globalize_reg): Likewise.
* df-problems.c (df_rd_local_compute, df_lr_confluence_n,
df_byte_lr_alloc): Rename df_invalidated_by_call to
invalidated_by_call_regset.
* basic-block.h (regs_invalidated_by_call_regset): Declare.

Co-Authored-By: Kai Tietz <kai.tietz@onevision.com>
From-SVN: r142812

15 years agoira-cost.c (copy_cost): Lazilly initialize move_cost if needed.
Jan Hubicka [Thu, 18 Dec 2008 13:49:47 +0000 (14:49 +0100)]
ira-cost.c (copy_cost): Lazilly initialize move_cost if needed.

* ira-cost.c (copy_cost): Lazilly initialize move_cost if needed.

Co-Authored-By: Kai Tietz <kai.tietz@onevision.com>
From-SVN: r142811

15 years agoi386.h (CONDITIONAL_REGISTER_USAGE): Initialize for current function ABI.
Jan Hubicka [Thu, 18 Dec 2008 13:48:36 +0000 (14:48 +0100)]
i386.h (CONDITIONAL_REGISTER_USAGE): Initialize for current function ABI.

* i386.h (CONDITIONAL_REGISTER_USAGE): Initialize for current function
ABI.
* i386.c (ix86_call_abi_override): Do not trigger target re-init and
do not try to modify call used regs.
(ix86_maybe_switch_abi): New function.
(TARGET_EXPAND_TO_RTL_HOOK): New macro.

Co-Authored-By: Kai Tietz <kai.tietz@onevision.com>
From-SVN: r142810

15 years agore PR rtl-optimization/37922 (code generation error)
Kenneth Zadeck [Thu, 18 Dec 2008 13:38:39 +0000 (13:38 +0000)]
re PR rtl-optimization/37922 (code generation error)

2008-12-18  Kenneth Zadeck <zadeck@naturalbridge.com>

PR rtl-optimization/37922
* dse.c (bb_info): Added regs_live field.
(look_for_hardregs): New function.
(replace_read): Added regs_live parameter and code to check that
shift sequence does not clobber live hardregs.
(check_mem_read_rtx): Added parameter to replace_read.
(dse_step1): Added regs_live bitmap and initialize it.
(rest_of_handle_dse): Added DF_NOTES problem and earlier call to
df_analyze.
* df-problems.c Renamed to
df_simulate_initialize_backwards.
(df_simulate_one_insn): Renamed to
df_simulate_one_insn_backwards.
(df_simulate_artificial_refs_at_top): Renamed to
df_simulate_finalize_backwards.
(df_simulate_initialized_forwards,
df_simulate_one_insn_forwards,
df_simulate_finalize_backwards): New functions.
* df.h (df_simulate_artificial_refs_at_end): Renamed to
df_simulate_initialize_backwards.
(df_simulate_one_insn): Renamed to
df_simulate_one_insn_backwards.
(df_simulate_artificial_refs_at_top): Renamed to
df_simulate_finalize_backwards.
(df_simulate_initialized_forwards,
df_simulate_one_insn_forwards,
df_simulate_finalize_backwards): New functions.
* ra-conflict.c (global_conflicts): Renamed
df_simulate_artificial_refs_at_end to
df_simulate_initialize_backwards.
* sel-sched.c (propagate_lv_set): Renamed df_simulate_one_insn to
df_simulate_one_insn_backwards.
* ifcvt.c (dead_or_predicable): Renamed
df_simulate_artificial_refs_at_end to
df_simulate_initialize_backwards. Renamed df_simulate_one_insn to
df_simulate_one_insn_backwards.
* recog.c (peephole2_optimize): Ditto.
* rtl-factoring (collect_pattern_seqs, clear_regs_live_in_seq): Ditto.

2008-12-18  Kenneth Zadeck <zadeck@naturalbridge.com>

PR rtl-optimization/37922
* g++.dg/torture/pr37922.C: New test.

From-SVN: r142809

15 years agore PR fortran/31822 (Missing run-time bound checks for character pointer => target)
Daniel Kraft [Thu, 18 Dec 2008 10:05:54 +0000 (11:05 +0100)]
re PR fortran/31822 (Missing run-time bound checks for character pointer => target)

2008-12-18  Daniel Kraft  <d@domob.eu>

PR fortran/31822
* gfortran.h (gfc_check_same_strlen): Made public.
* trans.h (gfc_trans_same_strlen_check): Made public.
* check.c (gfc_check_same_strlen): Made public and adapted error
message output to be useful not only for intrinsics.
(gfc_check_merge): Adapt to gfc_check_same_strlen change.
* expr.c (gfc_check_pointer_assign): Use gfc_check_same_strlen for
string length compile-time check.
* trans-expr.c (gfc_trans_pointer_assignment): Add runtime-check for
equal string lengths using gfc_trans_same_strlen_check.
* trans-intrinsic.c (gfc_trans_same_strlen_check): Renamed and made
public from conv_same_strlen_check.
(gfc_conv_intrinsic_merge): Adapted accordingly.

2008-12-18  Daniel Kraft  <d@domob.eu>

PR fortran/31822
* gfortran.dg/char_pointer_assign_2.f90: Updated expected error message
to be more detailed.
* gfortran.dg/char_pointer_assign_4.f90: New test.
* gfortran.dg/char_pointer_assign_5.f90: New test.

From-SVN: r142808

15 years agore PR middle-end/38533 (tree-ssa-reassoc.c increases register pressure several times)
Jakub Jelinek [Thu, 18 Dec 2008 07:54:43 +0000 (08:54 +0100)]
re PR middle-end/38533 (tree-ssa-reassoc.c increases register pressure several times)

PR middle-end/38533
* tree-ssa-reassoc.c (remove_visited_stmt_chain): New function.
(rewrite_expr_tree): Add moved argument, move stmts together if
needed.  Call remove_visited_stmt_chain.
(linearize_expr_tree): Don't move stmts here.
(reassociate_bb): Call remove_visited_stmt_chain if num ops is 1.
Adjust rewrite_expr_tree caller.

* gcc.dg/tree-ssa/pr38533.c: New test.
* gcc.c-torture/execute/pr38533.c: New test.

From-SVN: r142807

15 years agore PR middle-end/38505 (Revision 142061 caused ICE on __builtin_memcpy)
Jakub Jelinek [Thu, 18 Dec 2008 07:52:07 +0000 (08:52 +0100)]
re PR middle-end/38505 (Revision 142061 caused ICE on  __builtin_memcpy)

PR middle-end/38505
* tree-ssa.c (useless_type_conversion_p_1): Return
false if inner_type is incomplete and outer_type is complete.

* gcc.c-torture/compile/pr38505.c: New test.

From-SVN: r142806

15 years ago* config.sub, config.guess: Update from upstream sources.
Ben Elliston [Thu, 18 Dec 2008 03:27:27 +0000 (03:27 +0000)]
* config.sub, config.guess: Update from upstream sources.

From-SVN: r142805

15 years agoDaily bump.
GCC Administrator [Thu, 18 Dec 2008 00:16:47 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r142803

15 years agosemantics.c (describable_type): New function.
Jason Merrill [Wed, 17 Dec 2008 19:08:14 +0000 (14:08 -0500)]
semantics.c (describable_type): New function.

gcc/cp:
        * semantics.c (describable_type): New function.
        (finish_decltype_type): Use it for dependent exprs.
        * cp-tree.h: Declare it.
        * mangle.c (write_type) [DECLTYPE_TYPE]: Set skip_evaluation.
        (write_expression): If skip_evaluation, use type stubs.
        * tree.c (cp_tree_equal): Handle PARM_DECLs from different
        declarations of a function.
        * init.c (build_new): Do auto deduction if type is describable.
        * decl.c (cp_finish_decl): Likewise.
        * parser.c (cp_parser_omp_for_loop): Likewise.
gcc/testsuite:
        * g++.dg/cpp0x/auto6.C: Test more stuff.
        * g++.dg/cpp0x/auto12.C: New test.
libiberty:
        * cp-demangle.c (d_expression): Handle rvalue stubs too.
        [DEMANGLE_COMPONENT_CAST]: Update mangling.
        (d_print_comp): Avoid extra ", " with empty template argument packs.
        Remove handling for obsolete T() mangling.

From-SVN: r142799

15 years agoinstall.texi (Prerequisites): Document PPL and CLooG-PPL dependences and the configur...
Sebastian Pop [Wed, 17 Dec 2008 18:56:33 +0000 (18:56 +0000)]
install.texi (Prerequisites): Document PPL and CLooG-PPL dependences and the configure options.

2008-12-17  Sebastian Pop  <sebastian.pop@amd.com>

        * doc/install.texi (Prerequisites): Document PPL and CLooG-PPL
        dependences and the configure options.
        (Configuration): Document --with-cloog, --with-ppl, --with-cloog-lib,
         --with-ppl-lib, --with-cloog-incude, --with-ppl-include.

From-SVN: r142798

15 years agoc99_classification_macros_c.cc: Don't xfail test on darwin8 or later.
Jack Howarth [Wed, 17 Dec 2008 17:51:49 +0000 (17:51 +0000)]
c99_classification_macros_c.cc: Don't xfail test on darwin8 or later.

2008-12-17  Jack Howarth  <howarth@bromo.med.uc.edu>

        * testsuite/26_numerics/headers/cmath/c99_classification_macros_c.cc:
Don't xfail test on darwin8 or later.

From-SVN: r142797

15 years agore PR middle-end/38556 (Revision 142792 breaks bootstrap)
H.J. Lu [Wed, 17 Dec 2008 17:05:05 +0000 (09:05 -0800)]
re PR middle-end/38556 (Revision 142792 breaks bootstrap)

2008-12-17  H.J. Lu  <hongjiu.lu@intel.com>

PR middle-end/38556
* calls.c (expand_call): Check outgoing_reg_parm_stack_space
only if REG_PARM_STACK_SPACE is defined.

From-SVN: r142794

15 years agocalls.c (expand_call): Do not sibcall if outgoing_reg_parm_stack_space does not match.
Jan Hubicka [Wed, 17 Dec 2008 12:03:04 +0000 (13:03 +0100)]
calls.c (expand_call): Do not sibcall if outgoing_reg_parm_stack_space does not match.

* calls.c (expand_call): Do not sibcall if
outgoing_reg_parm_stack_space does not match.

Co-Authored-By: Kai Tietz <kai.tietz@onevision.com>
From-SVN: r142792

15 years agore PR fortran/38137 (MERGE: -fbounds-check runtime check for same string length)
Daniel Kraft [Wed, 17 Dec 2008 10:16:28 +0000 (11:16 +0100)]
re PR fortran/38137 (MERGE: -fbounds-check runtime check for same string length)

2008-12-17  Daniel Kraft  <d@domob.eu>

PR fortran/38137
* trans-intrinsic.c (conv_same_strlen_check): New method.
(gfc_conv_intrinsic_merge): Call it here to actually do the check.

2008-12-17  Daniel Kraft  <d@domob.eu>

PR fortran/38137
* gfortran.dg/merge_char_3.f90: New test.

From-SVN: r142791

15 years agoDaily bump.
GCC Administrator [Wed, 17 Dec 2008 00:17:41 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r142789

15 years agoavr.c (avr_mcu_t): Add attiny87...
Anatoly Sokolov [Tue, 16 Dec 2008 21:05:59 +0000 (00:05 +0300)]
avr.c (avr_mcu_t): Add attiny87...

* config/avr/avr.c (avr_mcu_t): Add attiny87, attiny327, at90pwm81,
atmega16m1, at90scr100, atmega128rfa1, m3000f, m3000s and m3001b
devices.
* config/avr/avr.h (LINK_SPEC, CRT_BINUTILS_SPECS): (Ditto.).
* config/avr/t-avr (MULTILIB_MATCHES): (Ditto.).

From-SVN: r142783

15 years agoc99_classification_macros_c.cc: Don't xfail excessive errors on darwin8 or later.
Jack Howarth [Tue, 16 Dec 2008 18:54:30 +0000 (18:54 +0000)]
c99_classification_macros_c.cc: Don't xfail excessive errors on darwin8 or later.

2008-12-16  Jack Howarth  <howarth@bromo.med.uc.edu>

        * testsuite/26_numerics/headers/cmath/c99_classification_macros_c.cc:
Don't xfail excessive errors on darwin8 or later.

From-SVN: r142782

15 years agore PR preprocessor/38842 (Problem with SystemC compilation using GCC 4.3.2)
Hariharan Sandanagobalane [Tue, 16 Dec 2008 16:30:26 +0000 (16:30 +0000)]
re PR preprocessor/38842 (Problem with SystemC compilation using GCC 4.3.2)

        PR target/38842
        * config/picochip/picochip.opt (mae): Squeezed the comments onto one
        line.

From-SVN: r142781

15 years agoconfigure.ac (gcc_cv_ld_hidden): Sun ld on Solaris 9 and up supports hidden.
Rainer Orth [Tue, 16 Dec 2008 13:09:57 +0000 (13:09 +0000)]
configure.ac (gcc_cv_ld_hidden): Sun ld on Solaris 9 and up supports hidden.

* configure.ac (gcc_cv_ld_hidden): Sun ld on Solaris 9 and up
supports hidden.
* configure: Regenerate.

From-SVN: r142779

15 years agore PR target/37436 (arm-cross-g++. internal compiler error: in extract_insn, at recog...
Richard Earnshaw [Tue, 16 Dec 2008 12:03:41 +0000 (12:03 +0000)]
re PR target/37436 (arm-cross-g++. internal compiler error: in extract_insn, at recog.c:1990)

PR target/37436
* arm.c (arm_legitimate_index): Only accept addresses that are in
canonical form.
* predicates.md (arm_reg_or_extendqisi_mem_op): New predicate.
* arm.md (extendqihi2): Use arm_reg_or_extendqisi_mem_op predicate
for operand1.
(extendqisi2): Likewise.
(arm_extendqisi, arm_extendqisi_v6): Use arm_extendqisi_mem_op
predicate for operand1.

From-SVN: r142778

15 years agocstdatomic: Remove assert include.
Benjamin Kosnik [Tue, 16 Dec 2008 01:28:46 +0000 (01:28 +0000)]
cstdatomic: Remove assert include.

2008-12-15  Benjamin Kosnik  <bkoz@redhat.com>

* include/c_global/cstdatomic: Remove assert include.
* include/debug/debug.h: Consolidate __replacement_assert...
* include/parallel/base.h: Consolidate __replacement_assert...
* include/bits/c++config: ...here.
* include/bits/atomic_0.h: Use it.
* include/bits/atomic_2.h: Same.
* testsuite/29_atomics/atomic/cons/assign_neg.cc: Adjust line numbers.
* testsuite/29_atomics/atomic/cons/copy_neg.cc: Same.

From-SVN: r142777

15 years agoDaily bump.
GCC Administrator [Tue, 16 Dec 2008 00:17:29 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r142775

15 years agomips.c (mips_output_conditional_branch): Assert that openrands[1] is a CODE_LABEL.
Adam Nemet [Mon, 15 Dec 2008 21:59:51 +0000 (21:59 +0000)]
mips.c (mips_output_conditional_branch): Assert that openrands[1] is a CODE_LABEL.

* config/mips/mips.c (mips_output_conditional_branch): Assert that
openrands[1] is a CODE_LABEL.

From-SVN: r142772

15 years agomips.c (mips_expand_builtin_direct): Set TARGET to the result of mips_prepare_builtin...
Richard Sandiford [Mon, 15 Dec 2008 21:23:37 +0000 (21:23 +0000)]
mips.c (mips_expand_builtin_direct): Set TARGET to the result of mips_prepare_builtin_target.

gcc/
* config/mips/mips.c (mips_expand_builtin_direct): Set TARGET to
the result of mips_prepare_builtin_target.

From-SVN: r142771

15 years agomips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_abicalls if TARGET_ABICALLS is true.
Richard Sandiford [Mon, 15 Dec 2008 21:11:09 +0000 (21:11 +0000)]
mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_abicalls if TARGET_ABICALLS is true.

gcc/
* config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_abicalls
if TARGET_ABICALLS is true.

From-SVN: r142770

15 years agomips.md (move_doubleword_fpr<mode>): Use TARGET_FLOAT64 && !TARGET_64BIT to detect...
Richard Sandiford [Mon, 15 Dec 2008 21:10:00 +0000 (21:10 +0000)]
mips.md (move_doubleword_fpr<mode>): Use TARGET_FLOAT64 && !TARGET_64BIT to detect the mxhc1 case.

gcc/
* config/mips/mips.md (move_doubleword_fpr<mode>): Use
TARGET_FLOAT64 && !TARGET_64BIT to detect the mxhc1 case.

From-SVN: r142769

15 years agoinclhack.def (hpux_spu_info): Match PA as well as IA64.
Steve Ellcey [Mon, 15 Dec 2008 18:11:19 +0000 (18:11 +0000)]
inclhack.def (hpux_spu_info): Match PA as well as IA64.

* inclhack.def (hpux_spu_info): Match PA as well as IA64.
* fixincl.x: Regenerate.

From-SVN: r142767

15 years agore PR fortran/38487 (Bogus Warning: INTENT(INOUT) actual argument might interfere...
Mikael Morin [Mon, 15 Dec 2008 18:08:42 +0000 (19:08 +0100)]
re PR fortran/38487 (Bogus Warning: INTENT(INOUT) actual argument might interfere with actual argument)

2008-12-15  Mikael Morin  <mikael.morin@tele2.fr>

PR fortran/38487
* gfortran.dg/elemental_dependency_2.f90: New test.

2008-12-15  Mikael Morin  <mikael.morin@tele2.fr>

PR fortran/38487
* dependency.c (gfc_is_data_pointer): New function.
(gfc_check_argument_var_dependency): Disable the warning
in the pointer case.
(gfc_check_dependency): Use gfc_is_data_pointer.

From-SVN: r142766

15 years agopicochip.c (picochip_override_options): Disable CFI asm and change the signature...
Hariharan Sandanagobalane [Mon, 15 Dec 2008 16:20:44 +0000 (16:20 +0000)]
picochip.c (picochip_override_options): Disable CFI asm and change the signature of brev and byteSwap functions to use...

        * config/picochip/picochip.c (picochip_override_options): Disable CFI
        asm and change the signature of brev and byteSwap functions to use
        unsigned values.
        * config/picochip/picochip.md (commsTestPort): This is a complex
        instruction and should not be vliwed. Dont set insn type.

From-SVN: r142765

15 years agore PR fortran/38113 (on warning/error: skip whitespaces, move position marker to...
Mikael Morin [Mon, 15 Dec 2008 14:46:22 +0000 (15:46 +0100)]
re PR fortran/38113 (on warning/error: skip whitespaces, move position marker to actual variable name)

2008-12-15  Mikael Morin  <mikael.morin@tele2.fr>

PR fortran/38113
* error.c (show_locus): Start counting columns at 0.
* primary.c (match_actual_arg): Eat spaces
before copying the current locus.
(match_variable): Copy the locus before matching.

From-SVN: r142763

15 years agos390.c (s390_swap_cmp): New function.
Wolfgang Gellerich [Mon, 15 Dec 2008 14:24:03 +0000 (14:24 +0000)]
s390.c (s390_swap_cmp): New function.

2008-12-15  Wolfgang Gellerich  <gellerich@de.ibm.com>

* config/s390/s390.c (s390_swap_cmp): New function.
(s390_non_addr_reg_read_p): New function.
(s390_z10_optimize_cmp): New function.
(s390_reorg): Added call to s390_optimize_cmp.
* config/s390/s390.md (nop1): New insn.

From-SVN: r142762

15 years agoDaily bump.
GCC Administrator [Mon, 15 Dec 2008 00:16:51 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r142760

15 years agoi386-5.C: Include x86intrin.h.
Uros Bizjak [Sun, 14 Dec 2008 19:51:03 +0000 (20:51 +0100)]
i386-5.C: Include x86intrin.h.

* g++.dg/other/i386-5.C: Include x86intrin.h.
* g++.dg/other/i386-6.C: Ditto.

From-SVN: r142752

15 years agore PR fortran/35937 (Wrong type for charlength of function)
Paul Thomas [Sun, 14 Dec 2008 16:00:25 +0000 (16:00 +0000)]
re PR fortran/35937 (Wrong type for charlength of function)

2008-12-14  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/35937
* trans-expr.c (gfc_finish_interface_mapping): Fold convert the
character length to gfc_charlen_type_node.

2008-12-14  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/35937
* gfortran.dg/char_length_14.f90: New test.

From-SVN: r142750

15 years agore PR fortran/38504 (double minus sign when printing integer?)
Jerry DeLisle [Sun, 14 Dec 2008 06:58:38 +0000 (06:58 +0000)]
re PR fortran/38504 (double minus sign when printing integer?)

2008-12-13  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR libfortran/38504
* gfortran.dg/fmt_int_sign.f90: New test.

From-SVN: r142748

15 years agore PR fortran/38504 (double minus sign when printing integer?)
Jerry DeLisle [Sun, 14 Dec 2008 06:50:53 +0000 (06:50 +0000)]
re PR fortran/38504 (double minus sign when printing integer?)

2008-12-13  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR libfortran/38504
io/write.c (write_decimal): Skip extra sign '-' at beginning of string
returned by gfc_itoa.

From-SVN: r142747

15 years agoDaily bump.
GCC Administrator [Sun, 14 Dec 2008 00:16:49 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r142745

15 years agoatomic.cc: Revert last change.
Benjamin Kosnik [Sat, 13 Dec 2008 20:01:57 +0000 (20:01 +0000)]
atomic.cc: Revert last change.

2008-12-13  Benjamin Kosnik  <bkoz@redhat.com>

* src/atomic.cc: Revert last change.
* include/bits/atomic_2.h: Move cassert include to...
* include/c_global/cstdatomic: ...here.
* testsuite/29_atomics/atomic/cons/assign_neg.cc: Adjust line numbers.
* testsuite/29_atomics/atomic/cons/copy_neg.cc: Same.

From-SVN: r142742

15 years agopr38409.c: Fixed commit problem.
Sebastian Pop [Sat, 13 Dec 2008 19:53:29 +0000 (19:53 +0000)]
pr38409.c: Fixed commit problem.

* gcc.dg/graphite/pr38409.c: Fixed commit problem.

From-SVN: r142741

15 years agosse-12.c: Include x86intrin.h.
Uros Bizjak [Sat, 13 Dec 2008 14:10:28 +0000 (15:10 +0100)]
sse-12.c: Include x86intrin.h.

* gcc.target/i386/sse-12.c: Include x86intrin.h.
(dg-options): Use -mavx instead of -msse4. Add -maes -mpclmul.
* gcc.target/i386/sse-13.c: Include x86intrin.h. Define immintrin.h
intrinsics that require immediate arguments.
(dg-options): Use -mavx instead of -msse4.
* gcc.target/i386/sse-14.c: Include x86intrin.h. Define immintrin.h
intrinsics that require immediate arguments.
(test_3): New macro.
(dg-options): Use -mavx instead of -msse4.
* g++.dg/other/i386-2.C: Include x86intrin.h.
(dg-options): Use -mavx instead of -msse4.
* g++.dg/other/i386-3.C: Include x86intrin.h.
(dg-options): Use -mavx instead of -msse4.

From-SVN: r142739

15 years agoMakefile.am: Remove -x c++ for atomic build rules.
Benjamin Kosnik [Sat, 13 Dec 2008 07:57:51 +0000 (07:57 +0000)]
Makefile.am: Remove -x c++ for atomic build rules.

2008-12-12  Benjamin Kosnik  <bkoz@redhat.com>

* src/Makefile.am: Remove -x c++ for atomic build rules.
* src/Makefile.in: Regenerate.
* src/atomic.cc: Move mutex include before cstdatomic.

From-SVN: r142738

15 years agore PR libstdc++/37144 (A bug in include/ext/pb_ds/detail/pat_trie_/constructors_destr...
H.J. Lu [Sat, 13 Dec 2008 01:40:15 +0000 (01:40 +0000)]
re PR libstdc++/37144 (A bug in include/ext/pb_ds/detail/pat_trie_/constructors_destructor_fn_imps.hpp)

2008-12-12  H.J. Lu  <hongjiu.lu@intel.com>

PR libstdc++/37144
* testsuite/util/regression/trait/assoc/type_trait.hpp
(regression_test_type_traits): Add const to pair_type_rebind.

From-SVN: r142736

15 years agoatomic.cc: Correct guards to match mutex.cc.
Benjamin Kosnik [Sat, 13 Dec 2008 00:40:06 +0000 (00:40 +0000)]
atomic.cc: Correct guards to match mutex.cc.

2008-12-12  Benjamin Kosnik  <bkoz@redhat.com>

* src/atomic.cc: Correct guards to match mutex.cc.

From-SVN: r142735

15 years agoDaily bump.
GCC Administrator [Sat, 13 Dec 2008 00:17:03 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r142732

15 years agore PR bootstrap/38383 (fails to build cross gcc for target hppa64-hp-hpux11.00)
Rainer Emrich [Sat, 13 Dec 2008 00:14:15 +0000 (00:14 +0000)]
re PR bootstrap/38383 (fails to build cross gcc for target hppa64-hp-hpux11.00)

PR bootstrap/38383
* pa64-hpux.h (LINK_GCC_C_SEQUENCE_SPEC): Don't hardcode search path
for the milli.a library.

From-SVN: r142729

15 years agore PR target/24779 (Python miscompilation - TOC reload)
Andrew Pinski [Fri, 12 Dec 2008 22:54:09 +0000 (14:54 -0800)]
re PR target/24779 (Python miscompilation - TOC reload)

2008-12-12  Andrew Pinski  <andrew_pinskia@playstation.sony.com>
            Peter Bergner <bergner@vnet.ibm.com>

        PR target/24779
        * config/rs6000/rs6000.md (call_indirect_aix32): Move the load of the
        TOC into the call pattern.
        (call_indirect_aix64): Likewise.
        (call_value_indirect_aix32): Likewise.
        (call_value_indirect_aix64): Likewise.
        (call_indirect_nonlocal_aix32_internal): New insn and split patterns
        to split off the load of the TOC.
        (call_indirect_nonlocal_aix32): Enable only after reload.
        (call_indirect_nonlocal_aix64_internal): New insn and split patterns
        to split off the load of the TOC.
        (call_indirect_nonlocal_aix64): Enable only after reload.
        (call_value_indirect_nonlocal_aix32_internal): New insn and split
        patterns to split off the load of the TOC.
        (call_value_indirect_nonlocal_aix32): Enable only after reload.
        (call_value_indirect_nonlocal_aix64_internal): New insn and split
        patterns to split off the load of the TOC.
        (call_value_indirect_nonlocal_aix64): Enable only after reload.

Co-Authored-By: Peter Bergner <bergner@vnet.ibm.com>
From-SVN: r142727

15 years agore PR testsuite/38163 (gcc.dg/tree-ssa/loop-3.c failure at -m64 on i686-apple-darwin9)
Uros Bizjak [Fri, 12 Dec 2008 22:39:43 +0000 (23:39 +0100)]
re PR testsuite/38163 (gcc.dg/tree-ssa/loop-3.c failure at -m64 on i686-apple-darwin9)

PR testsuite/38163
* gcc.dg/tree-ssa/loop-3.c: Compile only on nonpic x86 targets.
Remove dg-skip-if directive.

From-SVN: r142726

15 years agore PR libfortran/24685 (real(16) formatted input is broken for huge values (gfortran...
Janis Johnson [Fri, 12 Dec 2008 22:17:31 +0000 (22:17 +0000)]
re PR libfortran/24685 (real(16) formatted input is broken for huge values (gfortran.dg/default_format_2.f90))

PR libgfortran/24685
* gfortran.dg/default_format_denormal_2.f90: Change XFAIL to check
for size of long double.

From-SVN: r142724

15 years agore PR objc++/31032 (expected tree that contains 'decl with RTL' structure, have ...
Janis Johnson [Fri, 12 Dec 2008 22:14:39 +0000 (22:14 +0000)]
re PR objc++/31032 (expected tree that contains 'decl with RTL' structure, have 'field_decl'  in assemble_external_real, at varasm.c:2225)

PR objc++/31032
* obj-c++.dg/bitfield-1.mm: Remove XFAIL for ICE.

From-SVN: r142723

15 years agorevert: cfgrtl.c (rtl_verify_flow_info_1): Don't apply BLOCK_FOR_INSN on a BARRIER...
Andreas Schwab [Fri, 12 Dec 2008 21:16:09 +0000 (21:16 +0000)]
revert: cfgrtl.c (rtl_verify_flow_info_1): Don't apply BLOCK_FOR_INSN on a BARRIER insn.

Revert:
2008-12-12  Andreas Schwab  <schwab@suse.de>
* cfgrtl.c (rtl_verify_flow_info_1): Don't apply BLOCK_FOR_INSN on
a BARRIER insn.

From-SVN: r142721

15 years ago* gcc.dg/pr35442.c: Ignore a possible warning.
Janis Johnson [Fri, 12 Dec 2008 20:39:06 +0000 (20:39 +0000)]
* gcc.dg/pr35442.c: Ignore a possible warning.

From-SVN: r142720

15 years agore PR tree-optimization/32044 (final value replacement too aggressive for e.g. target...
Zdenek Dvorak [Fri, 12 Dec 2008 20:32:47 +0000 (21:32 +0100)]
re PR tree-optimization/32044 (final value replacement too aggressive for e.g. targets with no native div/mod insns)

PR tree-optimization/32044
* tree-scalar-evolution.h (expression_expensive_p): Declare.
* tree-scalar-evolution.c (expression_expensive_p): New function.
(scev_const_prop): Avoid introducing expensive expressions.
* tree-ssa-loop-ivopts.c (may_eliminate_iv): Ditto.

* gcc.dg/pr34027-1.c: Change outcome.
* gcc.dg/tree-ssa/pr32044.c: New test.

From-SVN: r142719

15 years agore PR target/11594 (testcase gcc.dg/20020103-1.c fails with "scan-assembler-not LC")
Janis Johnson [Fri, 12 Dec 2008 19:34:48 +0000 (19:34 +0000)]
re PR target/11594 (testcase gcc.dg/20020103-1.c fails with "scan-assembler-not LC")

PR target/11594
* gcc.dg/20020103-1.c: Remove XFAIL for powerpc.

From-SVN: r142718

15 years agore PR middle-end/38409 ([graphite] ICE : in canonicalize_loop_ivs, at tree-parloops...
Sebastian Pop [Fri, 12 Dec 2008 17:28:06 +0000 (17:28 +0000)]
re PR middle-end/38409 ([graphite] ICE : in canonicalize_loop_ivs, at tree-parloops.c:1384)

2008-12-12  Sebastian Pop  <sebastian.pop@amd.com>

PR middle-end/38409
* gcc.dg/graphite/pr38409.c: New.
* graphite.c (nb_reductions_in_loop): Use simple_iv.

From-SVN: r142716

15 years agoconfigure.ac (ppllibs): Add by default the lib flags.
Sebastian Pop [Fri, 12 Dec 2008 17:24:23 +0000 (17:24 +0000)]
configure.ac (ppllibs): Add by default the lib flags.

2008-12-12  Sebastian Pop <sebastian.pop@amd.com>

* configure.ac (ppllibs): Add by default the lib flags.
* configure: Regenerate.

From-SVN: r142715

15 years agocstdatomic: Update to N2798.
Benjamin Kosnik [Fri, 12 Dec 2008 17:10:16 +0000 (17:10 +0000)]
cstdatomic: Update to N2798.

2008-12-11  Benjamin Kosnik  <bkoz@redhat.com>
    Richard Henderson  <rth@redhat.com>

* include/c_global/cstdatomic: Update to N2798.
(atomic): Remove explicit constructors as per DR 845.
* include/bits/atomic_0.h: New. Switchable implementation.
* include/bits/atomic_2.h: New. Lock-free implementation.
* include/c_compatibility/stdatomic.h: Use foward headers.
* include/bits/atomicfwd_cxx.h: New.
* include/bits/atomicfwd_c.h: New.
* src/atomic.cc: Adjust.
* acinclude.m4 (GLIBCXX_CHECK_STANDARD_LAYOUT): Remove,
unconditionally use default/deleted syntax.
(GLIBCXX_ENABLE_ATOMIC_BUILTINS): Check for 2, 8.
* include/Makefile.am (bits_headers): Add atomicfwd_c.h,
atomicfwd_cxx.h, atomic_0.h, atomic_2.h.
* include/Makefile.in: Regenerate.
* configure: Regenerate.
* config.h.in: Regenerate.
* config/abi/pre/gnu.ver: Adjust exports.

* testsuite/27_io/ios_base/types/fmtflags/bitmask_operators.cc: Adjust.
* testsuite/27_io/ios_base/types/openmode/bitmask_operators.cc: Same.
* testsuite/27_io/ios_base/types/iostate/bitmask_operators.cc: Same.
* testsuite/29_atomics/atomic_address/cons/assign_neg.cc: Same.
* testsuite/29_atomics/atomic_address/cons/explicit_value.cc: Move to..
* testsuite/29_atomics/atomic_address/cons/single_value.cc: ...this.
* testsuite/29_atomics/atomic_address/cons/copy_neg.cc
* testsuite/29_atomics/atomic_integral/cons/single_value.cc: New.
* testsuite/29_atomics/atomic_integral/cons/assign_neg.cc: New.
* testsuite/29_atomics/atomic_integral/cons/copy_neg.cc: New.
* testsuite/29_atomics/atomic_integral/cons/default.cc: New.
* testsuite/29_atomics/atomic_integral/cons/direct_list.cc: New.
* testsuite/29_atomics/atomic_integral/cons/copy_list.cc: New.
* testsuite/29_atomics/atomic_integral/requirements/
standard_layout.cc: New.
* testsuite/29_atomics/atomic_integral/operators/
integral_assignment.cc: New.
* testsuite/29_atomics/atomic_integral/operators/increment_neg.cc: New.
* testsuite/29_atomics/atomic_integral/operators/bitwise_neg.cc: New.
* testsuite/29_atomics/atomic_integral/operators/decrement_neg.cc: New.
* testsuite/29_atomics/atomic_integral/operators/increment.cc: New.
* testsuite/29_atomics/atomic_integral/operators/decrement.cc: New.
* testsuite/29_atomics/atomic_integral/operators/bitwise.cc: New.
* testsuite/29_atomics/atomic_integral/operators/
integral_conversion.cc: New.
* testsuite/29_atomics/atomic_flag/cons/assign_neg.cc: Adjust.
* testsuite/29_atomics/atomic_flag/cons/copy_neg.cc: Same.
* testsuite/29_atomics/atomic_flag/requirements/
standard_layout.cc: Same.
* testsuite/29_atomics/atomic_flag/
atomic_global_fence_compatibility.cc: Kill.
* testsuite/29_atomics/headers/cstdatomic/types_std_c++0x.cc: Adjust.
* testsuite/29_atomics/headers/cstdatomic/functions_std_c++0x.cc: Same.
* testsuite/29_atomics/headers/cstdatomic/macros.cc: Same.
* testsuite/29_atomics/headers/stdatomic.h/macros.c: Same.
* testsuite/29_atomics/headers/stdatomic.h/types.c: Same.
* testsuite/29_atomics/atomic/cons/assign_neg.cc: Same.
* testsuite/29_atomics/atomic/cons/explicit_value.cc: Move to...
* testsuite/29_atomics/atomic/cons/single_value.cc: ...this.
* testsuite/29_atomics/atomic/cons/copy_neg.cc
* testsuite/29_atomics/atomic/cons/direct_list.cc: New.
* testsuite/29_atomics/atomic/cons/copy_list.cc: New.
* testsuite/29_atomics/atomic/requirements/standard_layout.cc: New.
* testsuite/29_atomics/atomic/requirements/base_classes.cc: New.
* testsuite/29_atomics/atomic/operators/integral_assignment.cc: New.
* testsuite/29_atomics/atomic/operators/integral_conversion.cc: New.
* testsuite/util/testsuite_hooks.h (bitmask_operators): Move...
* testsuite/util/testsuite_common_types.h: ...here.
(atomic_integrals_no_bool): New.
(atomic_integrals): New.
(has_increment_operators, has_decrement_operators)
(direct_list_initializable, single_value_constructible)
(standard_layout, has_bitwise_operators, integral_convertable)
(integral_assignable): Add.

Co-Authored-By: Richard Henderson <rth@redhat.com>
From-SVN: r142714

15 years agox86intrin.h: New header file to support all x86 intrinsics
Dwarakanath Rajagopal [Fri, 12 Dec 2008 16:53:27 +0000 (16:53 +0000)]
x86intrin.h: New header file to support all x86 intrinsics

2008-12-12  Dwarakanath Rajagopal  <dwarak.rajagopal@amd.com>

* config/i386/x86intrin.h: New header file to support all x86
intrinsics
* config.gcc (extra_headers): For x86 and x86-64, add x86intrin.h

From-SVN: r142713

15 years agore PR target/38402 (Undocumented Yz constraint)
H.J. Lu [Fri, 12 Dec 2008 14:32:00 +0000 (14:32 +0000)]
re PR target/38402 (Undocumented Yz constraint)

2008-12-12  H.J. Lu  <hongjiu.lu@intel.com>

PR target/38402
* gcc/doc/md.texi: Remove Y and document Yz, Y2, Yi and Ym
constraints for x86.

From-SVN: r142710

15 years agore PR fortran/36355 (matmul argument-check: wrong error messages)
Daniel Franke [Fri, 12 Dec 2008 13:22:55 +0000 (08:22 -0500)]
re PR fortran/36355 (matmul argument-check: wrong error messages)

gcc/fortran:
2008-12-12  Daniel Franke  <franke.daniel@gmail.com>

        PR fortran/36355
        * check.c (gfc_check_matmul): Fixed error message for invalid
        types to correctly identify the offending argument, added check
        for mismatching types.

gcc/testsuite:
2008-12-12 Daniel Franke  <franke.daniel@gmail.com>

        PR fortran/36355
        * gfortran.dg/matmul_argument_types.f90: New.

From-SVN: r142709

15 years ago(rtl_verify_flow_info_1): Don't apply BLOCK_FOR_INSN on
Andreas Schwab [Fri, 12 Dec 2008 12:28:39 +0000 (12:28 +0000)]
(rtl_verify_flow_info_1): Don't apply BLOCK_FOR_INSN on
a BARRIER insn.

From-SVN: r142708

15 years agore PR libstdc++/37582 (std::pow strange overload resolution)
Jakub Jelinek [Fri, 12 Dec 2008 11:22:33 +0000 (12:22 +0100)]
re PR libstdc++/37582 (std::pow strange overload resolution)

PR c++/37582
* testsuite/26_numerics/headers/cmath/37582.cc: New.

From-SVN: r142707

15 years agosync.md (memory_barrier): Remove mem:BLK from operands.
Uros Bizjak [Fri, 12 Dec 2008 10:25:55 +0000 (11:25 +0100)]
sync.md (memory_barrier): Remove mem:BLK from operands.

* config/alpha/sync.md (memory_barrier): Remove mem:BLK from operands.
Use Pmode for scratch reg.
(*mb_internal): Use (match_dup 0) for unspec operand.

From-SVN: r142705

15 years agotree-vrp.c (extract_range_from_binary_expr): Don't shift by floor_log2 of zero.
Alexandre Oliva [Fri, 12 Dec 2008 04:56:29 +0000 (04:56 +0000)]
tree-vrp.c (extract_range_from_binary_expr): Don't shift by floor_log2 of zero.

* tree-vrp.c (extract_range_from_binary_expr): Don't shift by
floor_log2 of zero.  Negate widened zero.

From-SVN: r142702

15 years agofp-bit.c (nan): Rename from this ..
Ben Elliston [Fri, 12 Dec 2008 02:23:35 +0000 (02:23 +0000)]
fp-bit.c (nan): Rename from this ..

* config/fp-bit.c (nan): Rename from this ..
(makenan): .. to this.

From-SVN: r142701

15 years agoDaily bump.
GCC Administrator [Fri, 12 Dec 2008 00:16:57 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r142699

15 years agore PR testsuite/29071 (gcc.dg/20020919-1.c compilation test fails on powerpc-apple...
Janis Johnson [Thu, 11 Dec 2008 23:37:03 +0000 (23:37 +0000)]
re PR testsuite/29071 (gcc.dg/20020919-1.c compilation test fails on powerpc-apple-darwin8 at -m64)

PR testsuite/29071
* gcc.dg/20020919-1.c: Fix a preprocessor check.

From-SVN: r142696

15 years ago* config/mips/mips.md (*branch_bit<bbv><mode>,
Adam Nemet [Thu, 11 Dec 2008 22:50:12 +0000 (22:50 +0000)]
* config/mips/mips.md (*branch_bit<bbv><mode>,

*branch_bit<bbv><mode>_inverted): Renumber operands so that the
label becomes operands[1].

From-SVN: r142694

15 years agore PR middle-end/38413 ([graphite] ICE: in build2_stat, at tree.c:3293)
Sebastian Pop [Thu, 11 Dec 2008 20:19:54 +0000 (20:19 +0000)]
re PR middle-end/38413 ([graphite] ICE: in build2_stat, at tree.c:3293)

2008-12-11  Sebastian Pop  <sebastian.pop@amd.com>

PR middle-end/38413
* gcc.dg/graphite/pr38413.c: New.

From-SVN: r142691

15 years agore PR middle-end/38459 ([graphite] SEGFAULT in cloog_clast_create)
Sebastian Pop [Thu, 11 Dec 2008 19:57:18 +0000 (19:57 +0000)]
re PR middle-end/38459 ([graphite] SEGFAULT in cloog_clast_create)

2008-12-11  Sebastian Pop  <sebastian.pop@amd.com>

PR middle-end/38459
* gfortran.dg/graphite/pr37857.f90: New.

From-SVN: r142690

15 years agore PR libgcj/38006 (Incorrect proplist on inherit.png)
H.J. Lu [Thu, 11 Dec 2008 19:48:16 +0000 (11:48 -0800)]
re PR libgcj/38006 (Incorrect proplist on inherit.png)

2008-12-11  H.J. Lu  <hongjiu.lu@intel.com>

PR libgcj/38006
* classpath/tools/resource/gnu/classpath/tools/gjdoc/htmldoclet/inherit.png:
Set svn:mime-type to application/octet-stream.

From-SVN: r142689

15 years ago[multiple changes]
Sebastian Pop [Thu, 11 Dec 2008 19:35:41 +0000 (19:35 +0000)]
[multiple changes]

2008-12-11  Harsha Jagasia  <harsha.jagasia@amd.com>

PR tree-optimization/38446
* graphite.c (register_bb_in_sese): New.
(bb_in_sese_p): Check if bb belongs to sese region by explicitly
looking at the bbs in the region.
* graphite.h (sese): Add region_basic_blocks pointer set to
structure and initialize at the time of defining new scop.

2008-12-11  Tobias Grosser  <grosser@fim.uni-passau.de>

* graphite.c (new_graphite_bb): Initialize GBB_STATIC_SCHEDULE.
(find_params_in_bb): Do not free data refs.
(free_graphite_bb): Add FIXME on disabled free_data_refs.

2008-12-11  Sebastian Pop  <sebastian.pop@amd.com>

* graphite.c (struct ivtype_map_elt): New.
(debug_ivtype_elt, debug_ivtype_map_1, debug_ivtype_map,
new_ivtype_map_elt, ivtype_map_elt_info, eq_ivtype_map_elts,
gcc_type_for_cloog_iv): New.
(loop_iv_stack_patch_for_consts): Use the type of the induction
variable from the original loop, except for the automatically
generated loops, i.e., in the case of a strip-mined loop, in
which case there is no original loop: in that case just use
integer_type_node.
(new_graphite_bb): Initialize GBB_CLOOG_IV_TYPES.
(free_graphite_bb): Free GBB_CLOOG_IV_TYPES.
(clast_name_to_gcc): Accept params to be NULL.
(clast_to_gcc_expression): Take an extra parameter for the type.
Convert to that type all the expressions built by this function.
(gcc_type_for_clast_expr, gcc_type_for_clast_eq): New.
(graphite_translate_clast_equation): Compute the type of the
clast_equation before translating its LHS and RHS.
(clast_get_body_of_loop, gcc_type_for_iv_of_clast_loop): New.
(graphite_create_new_loop): Compute the type of the induction
variable before translating the lower and upper bounds and before
creating the induction variable.
(rename_variables_from_edge, rename_phis_end_scop): New.
(copy_bb_and_scalar_dependences): Call rename_phis_end_scop.
(sese_add_exit_phis_edge): Do not use integer_zero_node.
(find_cloog_iv_in_expr, compute_cloog_iv_types_1,
compute_cloog_iv_types): New.
(gloog): Call compute_cloog_iv_types before starting the
translation of the clast.

* graphite.h (struct graphite_bb): New field cloog_iv_types.
(GBB_CLOOG_IV_TYPES): New.
(debug_ivtype_map): Declared.
(oldiv_for_loop): New.

2008-12-10  Tobias Grosser  <grosser@fim.uni-passau.de>

PR middle-end/38459
* graphite.c (new_scop): Initialize SCOP_ADD_PARAMS.
(param_index): Assert if parameter is not know after parameter
detection.
(find_params_in_bb): Detect params directly in GBB_CONDITIONS.
(find_scop_parameters): Mark, that we have finished parameter
detection.
(graphite_transform_loops): Move condition detection before parameter
detection.
* graphite.h (struct scop): Add SCOP_ADD_PARAMS.

2008-12-11  Harsha Jagasia  <harsha.jagasia@amd.com>

PR tree-optimization/38446
* gcc.dg/graphite/pr38446.c: New.

2008-12-11  Sebastian Pop  <sebastian.pop@amd.com>

* gcc.dg/graphite/scop-16.c: Test only scop specific info.
* gcc.dg/graphite/scop-17.c: Same.
* gcc.dg/graphite/block-5.c: New.
* gcc.dg/graphite/block-6.c: New.
* gcc.dg/graphite/pr37485.c: Clean dump file after.
* gcc.dg/graphite/pr37684.c: Same.
* gcc.dg/graphite/block-2.c: Same.

2008-12-10  Tobias Grosser  <grosser@fim.uni-passau.de>

PR middle-end/38459
* gfortran.dg/graphite/pr38459.f90: New.

From-SVN: r142688

15 years agore PR testsuite/35677 (Intermitent failure "FAIL: libgomp.fortran/crayptr2.f90")
John David Anglin [Thu, 11 Dec 2008 18:33:48 +0000 (18:33 +0000)]
re PR testsuite/35677 (Intermitent failure "FAIL: libgomp.fortran/crayptr2.f90")

PR testsuite/35677
* emutls.c (__emutls_get_address): Make sure offset is really zero
before initializing the object's offset.

From-SVN: r142687