rguenth [Mon, 14 Dec 2015 15:26:24 +0000 (15:26 +0000)]
2015-12-14 Richard Biener <rguenther@suse.de>
PR tree-optimization/68852
* tree-vectorizer.h (struct _slp_tree): Add def_type member.
(SLP_TREE_DEF_TYPE): New accessor.
* tree-vect-stmts.c (vect_is_simple_use): Remove BB vectorization
hack.
* tree-vect-slp.c (vect_create_new_slp_node): Initialize
SLP_TREE_DEF_TYPE.
(vect_build_slp_tree): When a node is to be built up from scalars
do not push a NULL as child but instead set its def_type to
vect_external_def.
(vect_analyze_slp_cost_1): Check for child def-type instead
of NULL.
(vect_detect_hybrid_slp_stmts): Likewise.
(vect_bb_slp_scalar_cost): Likewise.
(vect_get_slp_defs): Likewise.
(vect_slp_analyze_node_operations): Likewise. Before
processing node push the children def-types to the underlying
stmts vinfo and restore it afterwards.
(vect_schedule_slp_instance): Likewise.
(vect_slp_analyze_bb_1): Do not mark stmts not in SLP instances
as not vectorizable.
ville [Mon, 14 Dec 2015 11:08:30 +0000 (11:08 +0000)]
Document the implementation of Logical Operator Type Traits.
* doc/html/index.html: Regenerate.
* doc/html/manual/status.html: Likewise.
* doc/xml/manual/status_cxx2017.xml: Add P0013R1 to C++ 201z
and to Library Fundamentals 2 TS.
thopre01 [Mon, 14 Dec 2015 02:05:16 +0000 (02:05 +0000)]
2015-12-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
PR testsuite/68629
* lib/target-supports.exp (check_effective_target_cilkplus): Also
check that compiling with -fcilkplus does not give an error.
* c-c++-common/attr-simd-3.c: Require cilkplus effective target.
aoliva [Sun, 13 Dec 2015 20:37:26 +0000 (20:37 +0000)]
[PR67355] drop dummy zero from reverse VTA ops, fix infinite recursion
VTA's cselib expression hashing compares expressions with the same
hash before adding them to the hash table. When there is a collision
involving a self-referencing expression, we could get infinite
recursion, in spite of the cycle breakers already in place. The
problem is currently latent in the trunk, because by chance we don't
get a collision.
Such value cycles are often introduced by reverse_op; most often,
they're indirect, and then value canonicalization takes care of the
cycle, but if the reverse operation simplifies to the original value,
we used to issue a (plus V (const_int 0)), because at some point
adding a plain value V to a location list as a reverse_op equivalence
caused other problems.
This dummy zero, in turn, caused the value canonicalizer to not fully
realize the equivalence, leading to more complex graphs and,
occasionally, to infinite recursion when comparing such
value-plus-zero expressions recursively.
Simply using V solves the infinite recursion from the PR testcase,
since the extra equivalence and the preexisting value canonicalization
together prevent recursion while the unrecognized equivalence
wouldn't, but it exposed another infinite recursion in
memrefs_conflict_p: get_addr had a cycle breaker in place, to skip RTL
referencing values introduced after the one we're examining, but it
wouldn't break the cycle if the value itself appeared in the
expression being examined.
After removing the dummy zero above, this kind of cycle in the
equivalence graph is no longer introduced by VTA itself, but dummy
zeros are also present in generated code, such as in the 32-bit x86's
pro_epilogue_adjust_stack_si_add epilogue insn generated as part of
the builtin longjmp in _Unwind_RaiseException building libgcc's
unwind-dw2.o. So, break the recursion cycle for them too.
for gcc/ChangeLog
PR debug/67355
* var-tracking.c (reverse_op): Don't add dummy zero to reverse
ops that simplify back to the original value.
* alias.c (refs_newer_value_p): Cut off recursion for
expressions containing the original value.
jbeulich [Fri, 11 Dec 2015 15:53:24 +0000 (15:53 +0000)]
avoid alignment of static variables affecting stack's
Function (or more narrow) scope static variables (as well as others not
placed on the stack) should also not have any effect on the stack
alignment. I noticed the issue first with Linux'es dynamic_pr_debug()
construct using an 8-byte aligned sub-file-scope local variable.
According to my checking bad behavior started with 4.6.x (4.5.3 was
still okay), but generated code got quite a bit worse as of 4.9.0.
gcc/
2015-12-11 Jan Beulich <jbeulich@suse.com>
* cfgexpand.c (expand_one_var): Exit early for static and
external variables when adjusting stack alignment related.
gcc/testsuite/
2015-12-11 Jan Beulich <jbeulich@suse.com>
ienkovich [Fri, 11 Dec 2015 14:38:53 +0000 (14:38 +0000)]
libmpx/
2015-12-11 Tsvetkova Alexandra <aleksandra.tsvetkova@intel.com>
* mpxrt/Makefile.am (libmpx_la_LDFLAGS): Add -version-info
option.
* libmpxwrap/Makefile.am (libmpx_la_LDFLAGS): Likewise and
fix include path.
* libmpx/Makefile.in: Regenerate.
* mpxrt/Makefile.in: Regenerate.
* libmpxwrap/Makefile.in: Regenerate.
* mpxrt/libtool-version: New version.
* libmpxwrap/libtool-version: Likewise.
* mpxrt/libmpx.map: Add new version and a new symbol.
* mpxrt/mpxrt.h: New file.
* mpxrt/mpxrt.c (NUM_L1_BITS): Moved to mpxrt.h.
(REG_IP_IDX): Moved to mpxrt.h.
(REX_PREFIX): Moved to mpxrt.h.
(XSAVE_OFFSET_IN_FPMEM): Moved to mpxrt.h.
(MPX_L1_SIZE): Moved to mpxrt.h.
* libmpxwrap/mpx_wrappers.c (mpx_pointer): New type.
(mpx_bt_entry): New type.
(alloc_bt): New function.
(get_bt): New function.
(copy_if_possible): New function.
(copy_if_possible_from_end): New function.
(move_bounds): New function.
(__mpx_wrapper_memmove): Use move_bounds to copy bounds.
gcc/testsuite/
2015-12-11 Tsvetkova Alexandra <aleksandra.tsvetkova@intel.com>
* gcc.target/i386/mpx/memmove-1.c: New test.
* gcc.target/i386/mpx/memmove-2.c: New test.
krebbel [Fri, 11 Dec 2015 11:16:13 +0000 (11:16 +0000)]
S/390: Fix warnings in "*setmem_long..." patterns.
gcc/ChangeLog
* config/s390/s390.c (s390_expand_setmem): Use new expanders.
* config/s390/s390.md ("*setmem_long")
("*setmem_long_and", "*setmem_long_31z"): Fix warnings.
("*setmem_long_and_31z"): New define_insn.
("setmem_long_<P:mode>"): New expanders.
* (<modesize>): New mode attribute
marxin [Fri, 11 Dec 2015 10:59:29 +0000 (10:59 +0000)]
Fix PR target/67484
Uros Bizjak <ubizjak@gmail.com>
PR target/67484
* config/i386/i386.c (ix86_valid_target_attribute_tree):
Use ggc_strdup to copy option_strings to opts->x_ix86_arch_string and
opts->x_ix86_tune_string.
segher [Fri, 11 Dec 2015 00:27:14 +0000 (00:27 +0000)]
Fix shrink-wrapping bug (PR67778, PR68634)
After shrink-wrapping has found the "tightest fit" for where to place
the prologue, it tries move it earlier (so that frame saves are run
earlier) -- but without copying any more basic blocks.
Unfortunately a candidate block we select can be inside a loop, and we
will still allow it (because the loop always exits via our previously
chosen block). We can do that just fine if we make a duplicate of the
block, but we do not want to here.
So we need to detect this situation. We can place the prologue at a
previous block PRE only if PRE dominates every block reachable from
it, because then we will never need to duplicate that block (it will
always be executed with prologue).
PR rtl-optimization/67778
PR rtl-optimization/68634
* shrink-wrap.c (try_shrink_wrapping): Add a comment about why we want
to put the prologue earlier. When determining if an earlier block is
suitable, make sure it dominates every block reachable from it.
hubicka [Thu, 10 Dec 2015 23:57:15 +0000 (23:57 +0000)]
PR ipa/61886
* lto-symtab.c (lto_symtab_merge_p): Avoid merging across different
values of error and warning attributes.
* gcc.dg/lto/pr61886_0.c: New testcase
hubicka [Thu, 10 Dec 2015 19:20:14 +0000 (19:20 +0000)]
* ipa-cp.c (ipcp_cloning_candidate_p): Use node->optimize_for_size_p.
(good_cloning_opportunity_p): Likewise.
(gather_context_independent_values): Do not return true when
polymorphic call context is known or when we have known aggregate
value of unused parameter.
(estimate_local_effects): Try to create clone for all context
when either some params are substituted or devirtualization is possible
or some params can be removed; use local flag instead of
node->will_be_removed_from_program_if_no_direct_calls_p.
(identify_dead_nodes): Likewise.
dmalcolm [Thu, 10 Dec 2015 18:21:00 +0000 (18:21 +0000)]
Add -Wmisleading-indentation to -Wall
gcc/c-family/ChangeLog:
* c.opt (Wmisleading-indentation): Add to -Wall for C and C++.
gcc/ChangeLog:
* doc/invoke.texi (-Wall): Add -Wmisleading-indentation to the
list.
(-Wmisleading-indentation): Update documentation to reflect
being enabled by -Wall in C/C++.
vmakarov [Thu, 10 Dec 2015 18:00:47 +0000 (18:00 +0000)]
2015-12-10 Vladimir Makarov <vmakarov@redhat.com>
PR rtl-optimization/68691
* lra-spills.c (lra_final_code_change): Check pseudo occurrence
number in non-debug insns and remove debug insns if necessary.
dmalcolm [Thu, 10 Dec 2015 17:12:38 +0000 (17:12 +0000)]
testsuite/lib/multline.exp: show test name and line numbers
gcc/testsuite/ChangeLog:
* lib/multiline.exp (_multiline_expected_outputs): Update comment.
(dg-end-multiline-output): Capture line numbers within
_multiline_expected_outputs.
(handle-multiline-outputs): Access global $testname_with_flags
and add it as a prefix to pass/fail results. Extract line numbers
from $_multiline_expected_outputs and print them within pass/fail
results, replacing the printing of $index. Consolidate the
string prefix shared between pass/fail into a new local ($title).
law [Thu, 10 Dec 2015 16:34:43 +0000 (16:34 +0000)]
2015-12-10 Jeff Law <law@redhat.com>
PR tree-optimization/68619
* tree-ssa-dom.c (dom_opt_dom_walker::before_dom_children): Propgate
return value from optimize_stmt.
(dom_opt_dom_walker): Add new argument to dom_walker constructor.
(pass_dominator:execute): If a block has an unreachable edge,
remove all jump threads through any successor of the affected block.
(record_equivalences_from_phis): Ignore alternative if the edge
does not have EDGE_EXECUTABLE set.
(single_incoming_edge_ignoring_loop_edges): Similarly.
(optimize_stmt): If a gimple_code has a compile-time constant
condition, return the edge taken for that constant value. Also
change the condition to true/false as necessary.
* domwalk.h (dom_walker::dom_walker): Add new argument
skip_unreachable_blocks. Don't provide empty constructor body.
(dom_walker::before_dom_children): Change return type.
(dom_walker::bb_reachable): Declare new private method.
(dom_walker::propagate_unreachable_to_edges): Likewise.
(dom_walker::m_unreachable_dom): Declare new private data member.
(dom_walker::m_skip_unreachable_blocks): Likewise.
* domwalk.c: Include dumpfile.h.
(dom_walker::dom_walker): New constructor. Initialize private data
members. If needed, set EDGE_EXECUTABLE for all edges in the CFG,
extracted from tree-ssa-sccvn.c.
(dom_walker::bb_reachable): New method extracted from tree-ssa-sccvn.c
(dom_walker::propagate_unreachable_to_edges): Likewise.
(dom_walker::walk): Only call before_dom_children on reachable
blocks. If before_dom_children returns an edge, then clear
EDGE_EXECUTABLE for all other outgoing edges from the same block.
For unreachable blocks, call propagate_unreachable_to_edges.
Similarly, only call after_dom_children on reachable blocks. For
unreachable blocks, conditionally clear m_unreachable_dom.
* tree-ssa-sccvn.c (sccvn_dom_walker::unreachable_dom): Remove
private data member.
(sccvn_dom_walker::after_dom_children): Use methods from dom_walker
class.
(run_scc_vn): Likewise.
(sccvn_dom_walker::before_dom_children): Likewise. Return the taken
outgoing edge if a COND, SWITCH, or GOTO are optimized.
* compare-elim.c (find_comparison_dom_walker::before_dom_children):
Change return type to an edge. Always return NULL.
* fwprop.c (single_def_use_dom_walker::before_dom_children): Likewise.
* gimple-ssa-strength-reduction.c
(find_candidates_dom_walker::before_dom_children): Likewise.
* ipa-prop.c (analysis_dom_walker::before_dom_children): Likewise.
(ipcp_modif_dom_walker::before_dom_children): Likewise.
* tree-into-ssa.c (rewrite_dom_walker::before_dom_children): Likewise.
(rewrite_update_dom_walker::before_dom_children): Likewise.
(mark_def_dom_children::before_dom_children): Likewise.
* tree-ssa-dse.c (dse_dom_walker::before_dom_children): Likewise.
* tree-ssa-loop-im.c
(invariantness_dom_walker::before_dom_children): Likewise.
(move_computations_dom_walker::before_dom_walker): Likewise.
* tree-ssa-phiopt.c
(nontrapping_dom_walker::before_dom_children): Likewise.
* tree-ssa-pre.c
(eliminate_dom_walker::before_dom_children): Likewise.
* tree-ssa-propagate.c
(substitute_and_fold_dom_walker::before_dom_children): Likewise.
* tree-ssa-strlen.c
(strlen_dom_walker::before_dom_children): Likewise.
* tree-ssa-uncprop.c
(uncprop_dom_walker::before_dom_children): Likewise.
PR tree-optimization/68619
* gcc.dg/tree-ssa/pr68619-1.c: New test.
* gcc.dg/tree-ssa/pr68619-2.c: New test.
* gcc.dg/tree-ssa/pr68619-3.c: New test.
* gcc.dg/tree-ssa/pr68619-4.c: New test.
* gcc.dg/tree-ssa/pr68619-5.c: New test.
jakub [Thu, 10 Dec 2015 16:11:49 +0000 (16:11 +0000)]
PR rtl-optimization/68376
PR rtl-optimization/68670
* ifcvt.c (noce_try_abs): For one_cmpl allow < 0, >= 0
or > -1 conditions regardless of negate, and disallow
all other conditions.
* gcc.c-torture/execute/pr68376-2.c (f5, f6, f7, f8): New
tests.
(main): Call them.
* gcc.dg/pr68670-1.c: New test.
* gcc.dg/pr68670-2.c: New test.
dmalcolm [Thu, 10 Dec 2015 15:20:53 +0000 (15:20 +0000)]
Fix -Wmisleading-indentation warning in graphite-optimize-isl.c
../../../src/gcc/graphite-optimize-isl.c: In function ‘isl_union_set* scop_get_domains(scop_p)’:
../../../src/gcc/graphite-optimize-isl.c:362:5: error: statement is indented as if it were guarded by... [-Werror=misleading-indentation]
return res;
^~~~~~
In file included from ../../../src/gcc/hash-table.h:236:0,
from ../../../src/gcc/coretypes.h:348,
from ../../../src/gcc/graphite-optimize-isl.c:28:
../../../src/gcc/vec.h:1343:3: note: ...this ‘for’ clause, but it is not
for (I = 0; (V).iterate ((I), &(P)); ++(I))
^
../../../src/gcc/graphite-optimize-isl.c:359:3: note: in expansion of macro ‘FOR_EACH_VEC_ELT’
FOR_EACH_VEC_ELT (scop->pbbs, i, pbb)
^~~~~~~~~~~~~~~~
dmalcolm [Thu, 10 Dec 2015 15:18:26 +0000 (15:18 +0000)]
Fix -Wmisleading-indentation warning in ifcvt.c
../../../src/gcc/ifcvt.c: In function ‘bool noce_try_inverse_constants(noce_if_info*)’:
../../../src/gcc/ifcvt.c:1233:2: error: statement is indented as if it were guarded by... [-Werror=misleading-indentation]
seq = end_ifcvt_sequence (if_info);
^~~
../../../src/gcc/ifcvt.c:1230:7: note: ...this ‘if’ clause, but it is not
if (target != if_info->x)
^~
due to the lack of an outdent after the conditional at lines 1230-1:
dmalcolm [Thu, 10 Dec 2015 15:13:49 +0000 (15:13 +0000)]
Fix -Wmisleading-indentation warning in gcc/regrename.c
../../../src/gcc/regrename.c: In function ‘void scan_rtx_address(rtx_insn*, rtx_def**, reg_class, scan_actions, machine_mode, addr_space_t)’:
../../../src/gcc/regrename.c:1314:7: error: statement is indented as if it were guarded by... [-Werror=misleading-indentation]
break;
^
../../../src/gcc/regrename.c:1311:5: note: ...this ‘if’ clause, but it is not
if (!AUTO_INC_DEC)
^
dmalcolm [Thu, 10 Dec 2015 15:11:37 +0000 (15:11 +0000)]
Fix misleading indentation in gcc/fortran/io.c
Fix this warning:
../../../src/gcc/fortran/io.c: In function ‘match gfc_match_open()’:
../../../src/gcc/fortran/io.c:2003:4: error: statement is indented as if it were guarded by... [-Werror=misleading-indentation]
if (!compare_to_allowed_values ("DELIM", delim, NULL, NULL,
^
../../../src/gcc/fortran/io.c:2000:2: note: ...this ‘if’ clause, but it is not
if (!is_char_type ("DELIM", open->delim))
^
dmalcolm [Thu, 10 Dec 2015 15:06:05 +0000 (15:06 +0000)]
Fix -Wmisleading-indentation warning in function.c
Fix this warning:
../../../src/gcc/function.c: In function ‘void locate_and_pad_parm(machine_mode, tree, int, int, int, tree, args_size*, locate_and_pad_arg_data*)’:
../../../src/gcc/function.c:4123:2: error: statement is indented as if it were guarded by... [-Werror=misleading-indentation]
{
^
../../../src/gcc/function.c:4119:7: note: ...this ‘if’ clause, but it is not
if (initial_offset_ptr->var)
^
redi [Thu, 10 Dec 2015 14:32:16 +0000 (14:32 +0000)]
Fix dejagnu directives in shared_ptr test
PR libstdc++/68825
* include/experimental/bits/shared_ptr.h (__shared_ptr, __weak_ptr,
experimental::shared_ptr, experimental::weak_ptr): Constrain
assignment operators.
* testsuite/experimental/memory/shared_ptr/cons/copy_ctor_neg.cc:
Change to a compile-only test and change dg-excess-errors to dg-error.
ebotcazou [Thu, 10 Dec 2015 14:20:50 +0000 (14:20 +0000)]
PR ada/48013
* gcc-interface/trans.c (empty_stmt_list_p): New predicate.
(gigi): Invoke it to compute the No_Elaboration_Code property.
(insert_code_for): Do not insert the code if it's empty.
rguenth [Thu, 10 Dec 2015 09:10:40 +0000 (09:10 +0000)]
2015-12-10 Richard Biener <rguenther@suse.de>
* tree-if-conv.c: Include params.h.
(ifcvt_memrefs_wont_trap): Use PARAM_ALLOW_STORE_DATA_RACES
instead of flag_tree_loop_if_convert_stores to guard cases
we'd introduce store-data-races.
danglin [Thu, 10 Dec 2015 01:08:13 +0000 (01:08 +0000)]
PR 68115/libfortran
* configure.ac: Set libbacktrace_cv_sys_sync to no on hppa*-*-hpux*.
* configure: Regenerate.
* elf.c (backtrace_initialize): Cast __sync_bool_compare_and_swap call
to void.
projects / thirdparty / gcc.git / log
