Jeff Law [Sat, 5 Mar 2016 05:10:58 +0000 (22:10 -0700)]
re PR tree-optimization/69196 (code size regression with jump threading at -O2)
PR tree-optimization/69196
* tree-ssa-threadbackward.c (fsm_find_control_statement_thread_paths):
If the both SSA_NAMEs are anonymous, then consider them unassociated
and include the PHI in the statement count.
Patrick Palka [Sat, 5 Mar 2016 01:59:04 +0000 (01:59 +0000)]
Fix PR c++/66786 (ICE with nested lambdas in variable template)
gcc/cp/ChangeLog:
PR c++/66786
* pt.c (template_class_depth): Given a lambda type, iterate
into its LAMBDA_TYPE_EXTRA_SCOPE field instead of its
TYPE_CONTEXT. Given a VAR_DECL, iterate into its
CP_DECL_CONTEXT.
gcc/testsuite/ChangeLog:
PR c++/66786
* g++.dg/cpp1y/var-templ48.C: New test.
* g++.dg/cpp1y/var-templ49.C: New test.
Jason Merrill [Fri, 4 Mar 2016 22:53:29 +0000 (17:53 -0500)]
re PR c++/69203 (ICE in potential_constant_expression_1, at cp/constexpr.c:4754)
PR c++/69203
* cp-tree.h (COND_EXPR_IS_VEC_DELETE): New.
* init.c (build_vec_delete_1): Set it.
* constexpr.c (potential_constant_expression_1) [COND_EXPR]: Check it.
Cary Coutant [Fri, 4 Mar 2016 22:15:55 +0000 (14:15 -0800)]
Sync include/plugin-api.h with binutils.
2016-03-03 Than McIntosh <thanm@google.com>
* plugin-api.h: Add new hooks to the plugin transfer vector to
to support querying section alignment and section size.
(ld_plugin_get_input_section_alignment): New hook.
(ld_plugin_get_input_section_size): New hook.
(ld_plugin_tag): Add LDPT_GET_INPUT_SECTION_ALIGNMENT
and LDPT_GET_INPUT_SECTION_SIZE.
(ld_plugin_tv): Add tv_get_input_section_alignment and
tv_get_input_section_size.
Jakub Jelinek [Fri, 4 Mar 2016 22:10:49 +0000 (23:10 +0100)]
re PR c++/70035 (Calling a non-virtual member in base-class constructor call with ubsan causes segfault when superclass has virtual member with same name)
PR c++/70035
* cp-tree.h (cp_ubsan_maybe_initialize_vtbl_ptrs): New prototype.
* decl.c (start_preparsed_function): Call
cp_ubsan_maybe_initialize_vtbl_ptrs if needed.
* cp-ubsan.c (cp_ubsan_dfs_initialize_vtbl_ptrs,
cp_ubsan_maybe_initialize_vtbl_ptrs): New functions.
gcc/c-family/ChangeLog:
PR c/68187
* c-indentation.c (get_visual_column): Move code to determine next
tab stop to...
(next_tab_stop): ...this new function.
(line_contains_hash_if): Delete function.
(detect_preprocessor_logic): Delete function.
(get_first_nws_vis_column): New function.
(detect_intervening_unindent): New function.
(should_warn_for_misleading_indentation): Replace call to
detect_preprocessor_logic with a call to
detect_intervening_unindent.
gcc/testsuite/ChangeLog:
PR c/68187
* c-c++-common/Wmisleading-indentation.c (fn_42_a): New test
function.
(fn_42_b): Likewise.
(fn_42_c): Likewise.
gcc/c-family/ChangeLog:
PR c/68187
* c-indentation.c (should_warn_for_misleading_indentation): When
suppressing warnings about cases where the guard and body are on
the same column, only use the first non-whitespace column in place
of the guard token column when dealing with "else" clauses.
When rejecting aligned BODY and NEXT, loosen the requirement
from equality with the first non-whitespace of guard to simply
that they not be indented relative to it.
gcc/testsuite/ChangeLog:
PR c/68187
* c-c++-common/Wmisleading-indentation.c (fn_40_a): New test
function.
(fn_40_b): Likewise.
(fn_41_a): Likewise.
(fn_41_b): Likewise.
Jakub Jelinek [Fri, 4 Mar 2016 07:27:15 +0000 (08:27 +0100)]
re PR debug/69947 (DW_OP_GNU_implicit_pointer broken on the trunk)
PR debug/69947
* dwarf2out.c (prune_unused_types_walk_loc_descr): Handle
all other ops that have dw_val_class_die_ref operands,
and DW_OP_GNU_entry_value.
Jakub Jelinek [Thu, 3 Mar 2016 14:32:15 +0000 (15:32 +0100)]
re PR target/70021 (Test miscompiled with -O3 option for -march=core-avx2.)
PR target/70021
* tree-vect-stmts.c (vect_mark_relevant): Remove USED_IN_PATTERN
argument, if STMT_VINFO_IN_PATTERN_P (stmt_info), always mark
the pattern no matter if it is used just by non-pattern, pattern
or mix thereof.
(process_use, vect_mark_stmts_to_be_vectorized): Adjust callers.
* tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern): If
oprnd1 def_stmt is in pattern, don't look through it.
* gcc.dg/vect/pr70021.c: New test.
* gcc.target/i386/pr70021.c: New test.
Marek Polacek [Thu, 3 Mar 2016 11:42:19 +0000 (11:42 +0000)]
re PR middle-end/70050 (ICE: tree check: expected integer_type or enumeral_type or boolean_type or real_type or fixed_point_type, have vector_type in generic_simplify_162, at generic-match.c:6175)
Martin Liska [Thu, 3 Mar 2016 10:08:09 +0000 (10:08 +0000)]
Skip properly debug stmt in optimize_mask_stores (PR
PR tree-optimization/70043
* tree-vect-loop.c (optimize_mask_stores): Move iterator to
previous statement if we see a debug statement.
* gfortran.dg/vect/pr70043.f90: New test.
The attached patch adds the a target specific attribute via the
new target macro TARGET_BEGIN_TRANSACTION_ATTRIBUTE to the
function begin_transaction(). S/390 uses this to set the
soft-float target attribute which is needed to fix a crash with
-m31.
As there seems to be no place in libitm to document internal macros like
USE_HTM_FASTPATH or the new macro, I've put the documentation in a
comment where the macro is used.
2016-03-03 Dominik Vogt <vogt@linux.vnet.ibm.com>
* config/s390/target.h (TARGET_BEGIN_TRANSACTION_ATTRIBUTE): Define
function attribute to disable floating point in begin_transaction() on
S/390.
* beginend.cc (begin_transaction): Use
TARGET_BEGIN_TRANSACTION_ATTRIBUTE.
Richard Biener [Thu, 3 Mar 2016 09:12:53 +0000 (09:12 +0000)]
re PR tree-optimization/55936 (Missed VRP optimization)
2016-03-03 Richard Biener <rguenther@suse.de>
PR tree-optimization/55936
* tree-vrp.c (compare_name_with_value): Add use_equiv_p
parameter and guard unsafe equivalence use.
(vrp_evaluate_conditional_warnv_with_ops): Always use
safe equivalences but not via the quadratic compare_names
helper.
re PR target/70014 ([ARM] Predicate does not match constraint (*subsi3_carryin_const))
2016-03-03 Michael Collison <michael.collison@linaro.org>
PR target/70014
* config/arm/arm.md (*subsi3_carryin_const): Change predicate
for operand 1 to s_register_operand. Change predicate for operand
2 to arm_not_immediate_operand.
re PR libffi/70024 (libffi ABI change w/o SONAME bump)
PR libffi/70024
* Makefile.am (libffi_version_script): Look in cwd for libffi.map.
(libffi_version_dep, libffi.map-sun): Likewise.
(libffi.map): New target.
* libffi.map.in: Rename from libffi.map. Add required defines,
includes, and conditionals.
Jakub Jelinek [Wed, 2 Mar 2016 19:16:14 +0000 (20:16 +0100)]
re PR libgomp/69555 (libgomp.c++/target-6.C fails because of undefined behaviour)
PR libgomp/69555
* gimplify.c (gimplify_decl_expr): For decls with REFERENCE_TYPE, also
gimplify_type_sizes the type they refer to.
(omp_notice_variable): Handle reference vars to VLAs.
* omp-low.c (lower_omp_target): Emit setup of OMP_CLAUSE_PRIVATE reference
to VLA decls in the second pass instead of first pass.
* testsuite/libgomp.c++/pr69555-1.C: New test.
* testsuite/libgomp.c++/pr69555-2.C: New test.
* elf.c (backtrace_initialize): Properly initialize elf_fileline_fn to
avoid possible crash.
(elf_add): Don't set *fileline_fn to elf_nodebug value in case of
missing debug info anymore.
Bin Cheng [Wed, 2 Mar 2016 14:10:56 +0000 (14:10 +0000)]
re PR rtl-optimization/69052 (Performance regression after r229402.)
PR tree-optimization/69052
* loop-invariant.c (canonicalize_address): New function.
(inv_can_prop_to_addr_use): Check validity of address expression
which is canonicalized by above function.
gcc/testsuite/ChangeLog
PR tree-optimization/69052
* gcc.target/i386/pr69052.c: New test.
Alan Modra [Wed, 2 Mar 2016 14:05:21 +0000 (00:35 +1030)]
decl alignment not respected
This patch cures a problem with ICF of read-only variables at the
intersection of -fsection-anchors, -ftree-loop-vectorize, and targets
with alignment restrictions.
What happens with the testcase is:
- "c" is referenced in a constructor, thus make_decl_rtl for "c",
- make_decl_rtl puts "c" in an anchor block (-fsection-anchors),
- anchor block contents can't move, so "c" alignment can't change by
ipa_increase_alignment (-ftree-loop-vectorize),
- however "a" alignment can be increased,
- ICF aliases "a" to "c".
So we have a decl for "a" saying it is aligned to 128 bits, using mem
for "c" which is only 16 bit aligned.
PR ipa/69990
gcc/
* ipa-icf.c (sem_variable::merge): Do not merge an alias with
larger alignment.
gcc/testsuite/
gcc.dg/pr69990.c: New.
Jakub Jelinek [Wed, 2 Mar 2016 13:10:36 +0000 (14:10 +0100)]
re PR target/70028 (Error: operand size mismatch for `kmovw' (wrong assembly generated) with -mavx512bw -masm=intel)
PR target/70028
* config/i386/i386.md (kmovw): Move m constraint to 2nd alternative.
(*movhi_internal): Put mask moves from and to memory separately
from moves from/to GPRs.
Jeff Law [Tue, 1 Mar 2016 23:12:10 +0000 (16:12 -0700)]
re PR tree-optimization/69196 (code size regression with jump threading at -O2)
PR tree-optimization/69196
* tree-ssa-threadbackward.c (fsm_find_control_statement_thread_paths):
Appropriately clamp the number of statements to copy when the
thread path does not traverse a loop backedge.
PR tree-optimization/69196
* gcc.dg/tree-ssa/pr69196.c: New test.
Eric Botcazou [Tue, 1 Mar 2016 22:31:52 +0000 (22:31 +0000)]
re PR rtl-optimization/70007 (wrong code with -mbmi2)
PR rtl-optimization/70007
* gcse.c (compute_ld_motion_mems): Tidy up and also invalidate memory
references present in REG_EQUAL notes attached to non-SET patterns.
Jeff Law [Tue, 1 Mar 2016 21:46:58 +0000 (14:46 -0700)]
re PR tree-optimization/69196 (code size regression with jump threading at -O2)
PR tree-optimization/69196
* tree-ssa-threadbackward.c (fsm_find_control_statement_thread_paths):
Do count some PHIs in the thread path against the insn count. Decrease
final statement count by one as the control statement in the last
block will get removed. Remove special cased code for handling PHIs in the last block.
PR tree-optimization/69196
* gcc.dg/tree-ssa/vrp46.c: Twiddle threading params to keep it from
duplicating code and spoiling the expected output.
projects / thirdparty / gcc.git / log
