Marek Polacek [Fri, 28 Nov 2014 09:06:48 +0000 (09:06 +0000)]
re PR c/63862 (C frontend converts shift-count to int while standard wants integer promotions)
PR c/63862
c-family/
* c-ubsan.c (ubsan_instrument_shift): Change the type of a MINUS_EXPR
to op1_utype.
* c-gimplify.c (c_gimplify_expr): Convert right operand of a shift
expression to unsigned_type_node.
c/
* c-typeck.c (build_binary_op) <RSHIFT_EXPR, LSHIFT_EXPR>: Don't
convert the right operand to integer type.
cp/
* typeck.c (cp_build_binary_op) <RSHIFT_EXPR, LSHIFT_EXPR>: Don't
convert the right operand to integer type.
testsuite/
* gcc.c-torture/execute/shiftopt-1.c: Don't XFAIL anymore.
* c-c++-common/ubsan/shift-7.c: New test.
Richard Biener [Fri, 28 Nov 2014 08:57:43 +0000 (08:57 +0000)]
re PR tree-optimization/64084 (match-and-simplify prefers complex matches)
2014-11-28 Richard Biener <rguenther@suse.de>
PR middle-end/64084
* genmatch.c (dt_node::gen_kids_1): New function, split out
from dt_node::gen_kids.
(decision_tree::cmp_node): DT_TRUE are generally not equal.
(decision_tree::find_node): Treat DT_TRUE as barrier for
node CSE on the same level.
(dt_node::append_node): Do not keep DT_TRUE last.
(dt_node::gen_kids): Emit code after each DT_TRUE node seen.
* gcc.dg/tree-ssa/ssa-ccp-34.c: New testcase.
* gcc.dg/tree-ssa/forwprop-31.c: Likewise.
Uros Bizjak [Thu, 27 Nov 2014 21:46:32 +0000 (22:46 +0100)]
i386.md (preferred_for_size): New attribute
* config/i386/i386.md (preferred_for_size): New attribute
(*pushxf): Split Yx*r constraints to r,*r. Use preferred_for_size
attribute to conditionally disable alternative 1.
(*pushdf): Split Yd*r constraints to r,*r. Use preferred_for_size
and prefered_for_speed attributes to conditionally disable
alternative 1.
(*movxf_internal): Split Yx*r constraints to r,*r. Use
preferred_for_size attribute to conditionally disable
alternatives 3 and 4.
(*movdf_internal): Split Yd*r constraints to r,*r. Use
preferred_for_size and prefered_for_speed attributes to conditionally
disable alternatives 3 and 4.
* config/i386/constraints.md (Yd, Yx): Remove register constraints.
Tony Wang [Thu, 27 Nov 2014 13:38:51 +0000 (13:38 +0000)]
lib1funcs.S (FUNC_START): Add conditional section redefine for macro L_arm_muldivsf3 and L_arm_muldivdf3.
2014-11-27 Tony Wang <tony.wang@arm.com>
libgcc/
* config/arm/lib1funcs.S (FUNC_START): Add conditional section
redefine for macro L_arm_muldivsf3 and L_arm_muldivdf3.
(SYM_END, ARM_SYM_START): Add macros used to expose function Symbols.
Jakub Jelinek [Thu, 27 Nov 2014 11:04:16 +0000 (12:04 +0100)]
re PR middle-end/64067 (ICE in expand_expr_real_1, at expr.c:10540, involving compound literal shenanigans)
PR middle-end/64067
* expr.c (expand_expr_addr_expr_1) <case COMPOUND_LITERAL_EXPR>:
Handle it by returning address of COMPOUND_LITERAL_EXPR_DECL
not only if modifier is EXPAND_INITIALIZER, but whenever
COMPOUND_LITERAL_EXPR_DECL is non-NULL and TREE_STATIC.
Jakub Jelinek [Thu, 27 Nov 2014 11:03:12 +0000 (12:03 +0100)]
re PR tree-optimization/64024 (gcc.dg/vect/vect-simd-clone-6.c ICEs)
PR tree-optimization/64024
* tree-vectorizer.h (struct _stmt_vec_info): Remove simd_clone_fndecl
field. Add simd_clone_info field.
(STMT_VINFO_SIMD_CLONE_FNDECL): Remove.
(STMT_VINFO_SIMD_CLONE_INFO): Define.
* tree-vect-stmts.c (vectorizable_simd_clone_call): Adjust for
STMT_VINFO_SIMD_CLONE_FNDECL becoming first element of
STMT_VINFO_SIMD_CLONE_INFO vector. For linear arguments, remember
base and linear_step from analysis phase and use it during transform
phase, biased by the difference between LOOP_VINFO_NITERS{_UNCHANGED,}
multiplied by linear_step.
(free_stmt_vec_info): Release STMT_VINFO_SIMD_CLONE_INFO.
* gcc.dg/vect/vect-simd-clone-13.c: New test.
* gcc.dg/vect/vect-simd-clone-14.c: New test.
re PR target/59593 ([arm big-endian] using "ldrh" access a immediate which stored in a memory by word)
2014-11-27 Thomas Preud'homme <thomas.preudhomme@arm.com>
gcc/
PR target/59593
* config/arm/arm.c (dump_minipool): dispatch to consttable pattern
based on mode size.
* config/arm/arm.md (consttable_1): Move from config/arm/thumb1.md and
make it TARGET_EITHER.
(consttable_2): Move from config/arm/thumb1.md, make it TARGET_EITHER
and move HFmode handling from consttable_4 to it.
(consttable_4): Move HFmode handling to consttable_2 pattern.
* config/arm/thumb1.md (consttable_1): Move to config/arm/arm.md.
(consttable_2): Ditto.
gcc/testsuite/
PR target/59593
* gcc.target/arm/constant-pool.c: New test.
Richard Biener [Thu, 27 Nov 2014 09:33:32 +0000 (09:33 +0000)]
re PR tree-optimization/61634 (ICE in in vect_get_vec_def_for_operand, at tree-vect-stmts.c:1423)
2014-11-27 Richard Biener <rguenther@suse.de>
PR tree-optimization/61634
* tree-vect-slp.c: Include gimple-walk.h.
(vect_detect_hybrid_slp_stmts): Rewrite to propagate hybrid
down the SLP tree for one scalar statement.
(vect_detect_hybrid_slp_1): New walker function.
(vect_detect_hybrid_slp_2): Likewise.
(vect_detect_hybrid_slp): Properly handle pattern statements
in a pre-scan over all loop stmts.
Merge the ./config directory between GCC and Binutils
This pulls in the missing Binutils pieces into the ./config
directory. Contains these missing Binutils changes:
2014-08-14 Alan Modra <amodra@gmail.com>
* plugins.m4: Test for dlfcn.h or windows.h here to set default
for --enable-plugins. Report error if someone tries to enable
plugins on a host we don't support.
2014-08-19 Alan Modra <amodra@gmail.com>
* plugins.m4 (AC_PLUGINS): If plugins are enabled, add -ldl to
LIBS via AC_SEARCH_LIBS.
* configure.ac (target_makefile_frag): Set for nios2-*-elf*.
* configure: Regenerated.
config/
* mt-nios2-elf: New file.
Based on that file version, all GCC and Binutils commits to this file were
applied. This pulls in the following missing Binutils changes to configure.ac:
2014-07-21 Joel Sherrill <joel.sherrill@oarcorp.com>
Disable gdb for or1k*-*-* until supported
* configure.ac (or1k*-*-*): Disable gdb.
* configure: Regenerated.
2014-07-27 Joel Sherrill <joel.sherrill@oarcorp.com>
GDB not supported for or1k*-*-rtems*
* configure.ac (or1k*-*-rtems*): gdb not supported. The ordering
of the stanzas results in this not being caught by or1k*-*-* later.
* configure. Regenerated.
Jakub Jelinek [Wed, 26 Nov 2014 19:36:36 +0000 (20:36 +0100)]
e.53.4.c: Add -DITESTITERS=20 to dg-options unless expensive testing is on.
* testsuite/libgomp.c/examples-4/e.53.4.c: Add -DITESTITERS=20
to dg-options unless expensive testing is on.
(TESTITERS): Define to N if not defined.
(main): Use TESTITERS instead of N.
* testsuite/libgomp.c/examples-4/e.55.1.c: Define CHUNKSZ from
dg-additional-options depending on whether expensive testing is on.
* testsuite/libgomp.fortran/examples-4/e.55.1.f90 (e_55_1_mod):
Decrease N to 100000 and CHUNKSZ to 10000.
Diagnose string constant conversion to char* in c++11 and above as forbidden, not deprecated.
Diagnose string constant conversion to char* in c++11 and above
as forbidden, not deprecated.
* typeck.c (string_conv_p): Do a pedwarn in c++11 and above,
change the diagnostic for the Wwrite-strings case for c++11 and above.
Richard Biener [Wed, 26 Nov 2014 14:39:43 +0000 (14:39 +0000)]
fold-const.h (const_unop): Declare.
2014-11-26 Richard Biener <rguenther@suse.de>
* fold-const.h (const_unop): Declare.
(const_binop): Likewise.
* fold-const.c (const_binop): Export overload that expects
a type parameter and dispatches to fold_relational_const as well.
Check both operand kinds for guarding the transforms.
(const_unop): New function, with constant folding from fold_unary_loc.
(fold_unary_loc): Dispatch to const_unop for tcc_constant operand.
Remove constant folding done there from the simplifications.
(fold_binary_loc): Check for constants using CONSTANT_CLASS_P.
(fold_negate_expr): Remove dead code from the REAL_CST case.
Avoid building garbage in the COMPLEX_CST case.
* gimple-match-head.c (gimple_resimplify1): Dispatch to
const_unop.
(gimple_resimplify2): Dispatch to const_binop.
(gimple_simplify): Likewise.
Ilya Enkovich [Wed, 26 Nov 2014 14:37:07 +0000 (14:37 +0000)]
re PR bootstrap/63995 (Bootstrap error with -mmpx -fcheck-pointer-bounds)
gcc/
PR bootstrap/63995
* tree-chkp-opt.c (chkp_reduce_bounds_lifetime): Ignore
debug statement when searching for a new position for
bounds load/creation statement.
Mark Wielaard [Wed, 26 Nov 2014 10:10:27 +0000 (10:10 +0000)]
DWARF add DW_AT_noreturn on noreturn function subprogram.
This implements the DWARFv5 noreturn proposal:
http://dwarfstd.org/ShowIssue.php?issue=140331.1
TREE_THIS_VOLATILE on a FUNCTION_DECL node means the function does not
return normally. This catches the traditional noreturn GNU attribute,
the C11 _Noreturn keyword and the C++11 [[noreturn]] attribute.
This relies on the DW_AT_noreturn constant defined in the DWARFv5 DRAFT:
http://www.dwarfstd.org/doc/dwarf5.20141029.pdf
gcc/ChangeLog
* dwarf2out.c (gen_subprogram_die): Add DW_AT_noreturn when the
function decl has TREE_THIS_VOLATILE.
gcc/testsuite/ChangeLog
* g++.dg/debug/dwarf2/noreturn-function.C: New test.
* gcc.dg/debug/dwarf2/noreturn-function-attribute.c: Likewise.
* gcc.dg/debug/dwarf2/noreturn-function-keyword.c: Likewise.
include/ChangeLog
* dwarf2.def (DW_AT_noreturn): New DWARF5 attribute.
Richard Biener [Wed, 26 Nov 2014 09:43:39 +0000 (09:43 +0000)]
re PR tree-optimization/62238 (ICE with LTO on valid code on x86_64-linux-gnu in verify_ssa (in 64-bit mode))
2014-11-26 Richard Biener <rguenther@suse.de>
PR tree-optimization/62238
* tree-predcom.c (ref_at_iteration): Unshare the expression
before gimplifying it.
(prepare_initializers_chain): Discard unused seq.
Tobias Burnus [Tue, 25 Nov 2014 22:33:32 +0000 (23:33 +0100)]
gfortran.h (gfc_option_t): Remove flags moved as Var to .opt.
2014-11-25 Tobias Burnus <burnus@net-b.de>
gcc/fortran/
* gfortran.h (gfc_option_t): Remove flags moved as Var to .opt.
(gfc_error_now_1): Renamed from gfc_error_now.
(gfc_error_now): Renamed from gfc_error_now_2.
(gfc_warning_now_1): Renamed from gfc_warning_now.
(gfc_warning_now): Renamed from gfc_warning_now_2.
* error.c (gfc_error_now_1): Renamed from gfc_error_now.
(gfc_error_now): Renamed from gfc_error_now_2.
(gfc_warning_now_1): Renamed from gfc_warning_now.
(gfc_warning_now): Renamed from gfc_warning_now_2.
(gfc_get_errors): Include common diagnostic in count.
* lang.opt (Wc-binding-type, Wconversion, Wconversion-extra,
Wintrinsics-std): Create a Var for those warnings.
* check.c (gfc_check_cmplx): Pass warning flag to
diagnostic function.
* decl.c (get_proc_name, gfc_verify_c_interop_param, build_sym
gfc_set_constant_character_len, verify_bind_c_sym): Ditto; use
_1 for old diagnostic, remove _2 for new diagnostic.
* expr.c (gfc_check_assign, gfc_check_vardef_context): Ditto.
* frontend-passes.c (doloop_code, do_function): Ditto.
* intrinsic.c (gfc_is_intrinsic, gfc_convert_type_warn): Ditto.
* match.c (gfc_match_common): Ditto.
* module.c (use_iso_fortran_env_module, gfc_use_module): Ditto.
* parse.c (decode_statement, decode_gcc_attribute, next_free,
next_fixed, gfc_check_do_variable): Ditto.
* resolve.c (resolve_common_vars, resolve_ordinary_assign):
Ditto.
* scanner.c (add_path_to_list, skip_free_comments,
gfc_next_char_literal, gfc_gobble_whitespace, load_line,
preprocessor_line, load_file): Ditto.
* symbol.c (gfc_set_default_type, verify_bind_c_derived_type):
Ditto.
* options.c (gfc_post_options): Ditto.
(gfc_init_options, set_Wall, gfc_handle_option): Ditto; remove
flags which now have a Var.
* invoke.texi (Wconversion-extra): Make clear that the flag
does not imply -Wconversion.
gcc/testsuite/
* gfortran.dg/empty_label.f: Change test to continue testing
for -fmax-errors=1.
* gfortran.dg/empty_label.f90: Ditto.
* gfortran.dg/warnings_are_errors_1.f90: Update dg-*.
Matthew Fortune [Tue, 25 Nov 2014 14:57:33 +0000 (14:57 +0000)]
MIPS: Testsuite fixes for soft-float configurations
gcc/testuite/
* gcc.target/mips/mips.exp: Add support for -msoft-float
and -mhard-float options. Ensure that explicit -mfp*
options imply both -mhard-float and -mdouble-float.
* gcc.target/mips/call-clobbered-1.c: Add -mhard-float to the
compile options.
Jakub Jelinek [Tue, 25 Nov 2014 11:16:27 +0000 (12:16 +0100)]
re PR preprocessor/60436 (C preprocessor segfaults on assembly file)
PR preprocessor/60436
* line-map.c (linemap_line_start): If highest is above 0x60000000
and we are still tracking columns or highest is above 0x70000000,
force add_map.
Tim Shen [Tue, 25 Nov 2014 05:43:04 +0000 (05:43 +0000)]
re PR libstdc++/63920 (Any regular expression should not match an empty sequence if match_not_null is specified)
PR libstdc++/63920
* include/bits/regex_executor.h: Make _M_begin non const.
* include/bits/regex_executor.tcc (_Executor<>::_M_search): Increase
_M_begin in search algorithm, so that _M_begin is treated as
"current start position" for each search iteration.
* testsuite/28_regex/algorithms/regex_search/ecma/flags.cc: New
testcase.
Michael Meissner [Mon, 24 Nov 2014 19:27:29 +0000 (19:27 +0000)]
re PR target/63965 (ICE: in extract_constrain_insn, at recog.c:2230 on ppc64)
2014-11-24 Michael Meissner <meissner@linux.vnet.ibm.com>
PR target/63965
* config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Do not set
Altivec & -16 mask if the type is not valid for Altivec registers.
(rs6000_secondary_reload_memory): Add support for ((reg + const) +
reg) that occurs during push_reload processing.
* config/rs6000/altivec.md (altivec_mov<mode>): Add instruction
alternative for moving constant vectors which are easy altivec
constants to GPRs. Set the length attribute each of the
alternatives.
Michael Meissner [Mon, 24 Nov 2014 18:58:40 +0000 (18:58 +0000)]
re PR target/63965 (ICE: in extract_constrain_insn, at recog.c:2230 on ppc64)
2014-11-24 Michael Meissner <meissner@linux.vnet.ibm.com>
PR target/63965
* config/rs6000/rs6000-cpus.def: Undo November 21st changes, a
work in progress patch was committed instead of the fixes for
63965.
* config/rs6000/rs6000.c: Likewise.