Richard Biener [Wed, 17 Aug 2016 08:18:47 +0000 (08:18 +0000)]
tree-ssa.c: Include tree-cfg.h and tree-dfa.h.
2016-08-17 Richard Biener <rguenther@suse.de>
* tree-ssa.c: Include tree-cfg.h and tree-dfa.h.
(verify_vssa): New function verifying virtual SSA form.
(verify_ssa): Call it.
* tree-ssa-loop-manip.c (slpeel_update_phi_nodes_for_guard2):
Do not apply loop-closed SSA handling to virtuals.
* ssa-iterators.h (op_iter_init): Handle GIMPLE_TRANSACTION.
* tree-into-ssa.c (prepare_use_sites_for): Skip virtual SSA names
when rewriting their symbol.
(prepare_def_site_for): Likewise.
* tree-chkp-opt.c (chkp_reduce_bounds_lifetime): Clear virtual
operands of moved stmts.
Richard Biener [Wed, 17 Aug 2016 08:11:32 +0000 (08:11 +0000)]
re PR tree-optimization/23855 (loop header should also be pulled out of the inner loop too)
2016-08-17 Richard Biener <rguenther@suse.de>
PR tree-optimization/23855
* tree-ssa-loop-unswitch.c: Include tree-ssa-loop-manip.h.
(tree_unswitch_outer_loop): Iterate find_loop_guard as long as we
find guards to hoist. Do not update SSA form but rewrite virtuals
into loop closed SSA.
(find_loop_guard): Adjust to skip already hoisted guards. Do
not mark virtuals for renaming or update SSA form.
Martin Liska [Wed, 17 Aug 2016 06:55:35 +0000 (08:55 +0200)]
Do not use LONG_LONG_TYPE_SIZE in #if macro condition
* coverage.c (get_gcov_type): Replace GCOV_TYPE_SIZE with
a LONG_LONG_TYPE_SIZE comparison.
* gcov-io.h: Remove macro definitions.
* tree-profile.c (gimple_gen_edge_profiler): Replace usage
of GCOV_TYPE_ATOMIC_FETCH_ADD with a LONG_LONG_TYPE_SIZE
comparison.
Joseph Myers [Tue, 16 Aug 2016 21:30:56 +0000 (22:30 +0100)]
Update soft-fp from glibc (PR libgcc/77265).
This patch updates soft-fp from glibc, bringing in the fix for PR
libgcc/77265, XFmode extension to TFmode wrongly turning an infinity
into a NaN. A test for that bug is added.
Bootstrapped with no regressions on x86_64-pc-linux-gnu.
PR libgcc/77265
gcc/testsuite:
* gcc.dg/torture/float128-extend-inf.c: New test.
* config/i386/i386.c (enum ix86_builtins): Reorder enumerators, so
that builtins not mentioned in bdesc_* arrays come first, then
the ones mentioned in bdesc_* arrays in the order they appear in
the arrays in between IX86_BUILTIN__BDESC_*_FIRST and
IX86_BUILTIN__BDESC_*_LAST enumerator.
(bdesc_mpx): Fix up a comment typo.
(bdesc_multi_arg): Remove __builtin_ia32_vpcomne[bwdq] and
__builtin_ia32_vpcomneu[bwdq] builtins.
(BDESC_VERIFY, BDESC_VERIFYS): Define.
(ix86_init_mmx_sse_builtins, ix86_init_mpx_builtins): Verify
enum ix86_builtins ordering.
(ix86_expand_builtin): Use enum ix86_builtins ordering assumption
for direct bdesc_* array member access instead of searching all the
arrays until an fcode match is found.
Uros Bizjak [Tue, 16 Aug 2016 18:23:47 +0000 (20:23 +0200)]
i386.md (*ashl<mode>3_mask): Rewrite define_insn pattern as define_insn_and_split.
* config/i386/i386.md (*ashl<mode>3_mask): Rewrite define_insn
pattern as define_insn_and_split. Split insn before reload to
ashl<mode>3_1.
(*<shift_insn><mode>3_mask): Ditto. Split insn before reload to
<shift_insn><mode>3_1.
(*<rotate_insn><mode>3_mask): Ditto. Split insn before reload to
<rotate_insn><mode>3_1.
David Malcolm [Tue, 16 Aug 2016 18:19:34 +0000 (18:19 +0000)]
Fix caret locations in format_type_warning (PR c/72857)
gcc/c-family/ChangeLog:
PR c/72857
* c-common.c (substring_loc::get_range): Rename to...
(substring_loc::get_location): ...this, converting param from a
source_range * to a location_t *. Call
get_source_location_for_substring rather than
get_source_range_for_substring, and pass in m_caret_idx.
* c-common.h (substring_loc::substring_loc): Add param "caret_idx".
(substring_loc::get_range): Replace with...
(substring_loc::get_location): ...this.
(substring_loc::set_caret_index): New method.
(substring_loc): Add field m_caret_idx.
* c-format.c (format_warning_va): Update for above changes.
Rename local "substring_loc" to "fmt_substring_loc" to avoid
clashing with type name.
(format_warning_at_char): Add caret_idx param to substring_loc ctor.
(check_argument_type): Likewise.
(format_type_warning): Rename param "fmt_loc" to "whole_fmt_loc"
Use a copy when emitting warnings, setting the caret index from TYPE.
gcc/ChangeLog:
PR c/72857
* input.c (get_source_range_for_substring): Rename to...
(get_source_location_for_substring): ...this, adding param
"caret_idx", and converting output param from source_range * to
location_t *.
(get_source_range_for_char): New function.
(get_num_source_ranges_for_substring): Update comment to reflect
above renaming.
(assert_char_at_range): Update to use get_source_range_for_char
rather than get_source_range_for_substring.
(test_lexer_string_locations_concatenation_2): Likewise.
* substring-locations.h (get_source_range_for_substring): Rename
to...
(get_source_location_for_substring): ...this, and adding param
"caret_idx", and converting output param from source_range * to
location_t *.
gcc/testsuite/ChangeLog:
PR c/72857
* gcc.dg/format/asm_fprintf-1.c: Restore column numbers
for embedded NUL.
* gcc.dg/format/c90-printf-1.c: Restore column numbers.
* gcc.dg/format/diagnostic-ranges.c (test_hex): Update expected
caret placement.
(test_oct): Likewise.
(test_multiple): Likewise.
(test_field_width_specifier): Likewise.
(test_field_width_specifier_2): New function.
(test_field_precision_specifier): New function.
(test_embedded_nul): Update expected caret placement.
(test_non_contiguous_strings): Update line number.
* gcc.dg/plugin/diagnostic-test-string-literals-1.c
(__emit_string_literal_range): Add "caret_idx" param.
(test_simple_string_literal): Add value for new param, updating
expected output..
(test_concatenated_string_literal): Likewise.
(test_multiline_string_literal): Likewise.
(test_hex): Likewise.
(test_oct): Likewise.
(test_multiple): Likewise.
(test_ucn4): Likewise.
(test_ucn8): Likewise.
(test_u8): Likewise.
(test_u): Likewise; update expected message, from "range" to
"location".
(test_U): Likewise.
(test_L): Likewise.
(test_macro): Add value for new param.
* gcc.dg/plugin/diagnostic-test-string-literals-2.c
(__emit_string_literal_range): Add "caret_idx" param.
(test_stringified_token_1): Add value for new param. Update
expected message, from "range" to "location".
(test_stringized_token_2): Likewise, adding param to macro.
(test_stringified_token_3): Likewise.
* gcc.dg/plugin/diagnostic_plugin_test_string_literals.c
(emit_warning): Convert param from source_range to location_t.
(test_string_literals): Add caret_idx param, and use it when
constructing a substring_loc. Update error message, from
"range" to "location".
Jakub Jelinek [Tue, 16 Aug 2016 16:50:47 +0000 (18:50 +0200)]
re PR target/71910 (ICE on valid OpenMP code)
PR target/71910
* tree-cfg.c (execute_fixup_cfg): Add node variable, use it. Before inlining,
add cgraph edge for the added __builtin_unreachable call.
Martin Liska [Tue, 16 Aug 2016 14:14:18 +0000 (16:14 +0200)]
gcov: add new option (--hash-filenames) (PR
PR gcov-profile/36412
* doc/gcov.texi: Document --hash-filenames(-x).
* gcov.c (print_usage): Add the option.
(process_args): Process the option, sort options alphabetically.
(md5sum_to_hex): New function.
(make_gcov_file_name): Do the md5sum and append it to a
filename.
Bin Cheng [Tue, 16 Aug 2016 13:09:40 +0000 (13:09 +0000)]
re PR tree-optimization/69848 (poor vectorization of a loop from SPEC2006 464.h264ref)
PR tree-optimization/69848
* config/aarch64/aarch64-simd.md (vcond<mode><mode>): Invert NE
and swtich operands to avoid additional NOT instruction.
(vcond<v_cmp_mixed><mode>): Ditto.
(vcondu<mode><mode>, vcondu<mode><v_cmp_mixed>): Ditto.
gcc/testsuite
* gcc.target/aarch64/simd/vcond-ne-bit.c: New test.
Jonathan Wakely [Tue, 16 Aug 2016 11:33:16 +0000 (12:33 +0100)]
PR 72847 Prevent double-free in std::vector<bool>
PR libstdc++/72847
* include/bits/stl_bvector.h (_Bvector_base::_M_deallocate): Zero
pointers to start and end of storage.
* testsuite/23_containers/vector/bool/72847.cc: New test.
* include/bits/vector.tcc (vector<bool>::_M_reallocate): Only update
_M_finish after deallocating.
(vector<bool>::_M_fill_insert): Likewise.
(vector<bool>::_M_insert_range): Likewise.
(vector<bool>::_M_insert_aux): Likewise.
Thomas Koenig [Mon, 15 Aug 2016 19:34:48 +0000 (19:34 +0000)]
frontend-passes.c (create_var): Set ts.deferred for deferred-length character variables.
2016-08-15 Thomas Koenig <tkoenig@gcc.gnu.org>
* frontend-passes.c (create_var): Set ts.deferred for
deferred-length character variables.
* dump-parse-tree.c (show_typespec): Also dump
is_c_interop, is_iso_c and deferred flags.
Uros Bizjak [Mon, 15 Aug 2016 18:46:53 +0000 (20:46 +0200)]
re PR target/72867 (SSE/AVX/AVX512: incorrect optimization of VMINPS/VMAXPS at compile time)
PR target/72867
* config/i386/sse.md (<code><mode>3<mask_name><round_saeonly_name>):
Emit ieee_<ieee_maxmin><mode>3<mask_name><round_saeonly_name>
for !flag_finite_math_only or flag_signed_zeros.
(*<code><mode>3<mask_name><round_saeonly_name>): Rename from
*<code><mode>3_finite<mask_name><round_saeonly_name>. Do not
depend on flag_finite_math_only.
(ieee_<ieee_maxmin><mode>3<mask_name><round_saeonly_name>):
New insn pattern.
(*<code><mode>3<mask_name><round_saeonly_name>): Remove.
(*ieee_smin<mode>3): Ditto.
(*ieee_smax<mode>3): Ditto.
* config/i386/mmx.md (mmx_<code>v2sf3): Emit
mmx_ieee_<ieee_maxmin>v2sf3 for !flag_finite_math_only or
flag_signed_zeros.
(*mmx_<code>v2sf3): Rename from *mmx_<code>v2sf3_finite. Do not
depend on flag_finite_math_only.
(mmx_ieee_<ieee_maxmin>v2sf3): New insn pattern.
(*mmx_<code>v2sf3): Remove.
* config/i386/subst.md (round_saeonly_mask_arg3): New subst attribute.
* config/i386/i386.c (ix86_expand_sse_fp_mimnax): Check
flag_signed_zeros instead of !flag_unsafe_math_optimizations.
testsuite/ChangeLog:
PR target/72867
* gcc.target/i386/pr72867.c: New test.
Ian Lance Taylor [Mon, 15 Aug 2016 18:05:24 +0000 (18:05 +0000)]
cmd/go: ignore errors from go/build for standard packages
The go/build package does not know that gccgo's standard packages don't
have source, and will report an error saying that it can not find them.
Work around that in the cmd/go sources, since the go/build sources don't
currently have a list of standard packages.
This should get a real fix in the master sources, somehow.
lra: A multiple_sets is not a simple_move_p (PR73650)
In the PR we have a PARALLEL of a move and a compare (a "mr." instruction).
The compare is dead, so single_set on it returns just the move. Then,
simple_move_p returns true; but the instruction does need reloads in this
case. This patch solves this by making simple_move_p return false for
every multiple_sets instruction.
PR rtl-optimization/73650
* lra-constraints.c (simple_move_p): If the insn is multiple_sets
it is not a simple move.
Uros Bizjak [Mon, 15 Aug 2016 11:40:37 +0000 (13:40 +0200)]
ops.cc: Always include ostream and ext/stdio_filebuf.h.
* src/filesystem/ops.cc: Always include ostream and
ext/stdio_filebuf.h.
(do_copy_file): Check if _GLIBCXX_USE_FCHMODAT is defined.
[_GLIBCXX_USE_SENDFILE]: Fallback to read/write operations in case
sendfile fails with ENOSYS or EINVAL.
Richard Biener [Mon, 15 Aug 2016 09:56:00 +0000 (09:56 +0000)]
re PR tree-optimization/73434 (Wrong code with casting, branches and aliasing)
2016-08-15 Richard Biener <rguenther@suse.de>
PR tree-optimization/73434
* tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Preserve
TBAA info on the base when forwarding a non-invariant address.
Jakub Jelinek [Mon, 15 Aug 2016 09:50:33 +0000 (11:50 +0200)]
re PR debug/71906 (Fortran allocatable strings debug info type size regression)
PR debug/71906
* dwarf2out.c (string_types): New variable.
(gen_array_type_die): Change early_dwarf handling of
DW_AT_string_length, create DW_OP_call4 referencing the
length var temporarily. Handle parameters that are pointers
to string length.
(adjust_string_types): New function.
(gen_subprogram_die): Temporarily set string_types to local var,
call adjust_string_types if needed.
(non_dwarf_expression, copy_deref_exprloc, optimize_string_length):
New functions.
(resolve_addr): Adjust DW_AT_string_length if it is DW_OP_call4.
* trans-decl.c (gfc_get_symbol_decl): Call gfc_finish_var_decl
for decl's character length before gfc_finish_var_decl on the
decl itself.
gcc/fortran/
* openmp.c (resolve_omp_clauses): Adjust use_device clause
handling to only allow pointers and arrays.
gcc/testsuite/
* gfortran.dg/goacc/host_data-tree.f95: Adjust to use pointers
in use_device clause.
* gfortran.dg/goacc/uninit-use-device-clause.f95: Likewise.
* gfortran.dg/goacc/list.f95: Adjust to catch
"neither a POINTER nor an array" error messages.
libgomp/
* testsuite/libgomp.oacc-fortran/host_data-1.f90: New test.
Ian Lance Taylor [Sat, 13 Aug 2016 02:52:42 +0000 (02:52 +0000)]
libgo: don't unset in shell script
Reportedly ksh fails to unset a variable that was not previously set.
Change match.sh and gotest to not unset LANG, but instead set LANG=C.
Also don't combine exporting and setting variable in a single statement.
Ian Lance Taylor [Sat, 13 Aug 2016 00:19:56 +0000 (00:19 +0000)]
crypto/aes, hash/crc32: ignore s390x specific files for now
These files are used to select s390x assembler support in the gc
toolchain. We don't currently have that support, as it is written in
the cmd/asm syntax rather than gas syntax. Mark the files to be ignored
for now, falling back to the default implementations.
Jakub Jelinek [Fri, 12 Aug 2016 19:56:26 +0000 (21:56 +0200)]
re PR c/71512 (ICE: verify_gimple failed with UBSAN)
PR c/71512
* ubsan.c (instrument_si_overflow): Pass true instead of false
to gsi_replace.
(pass_ubsan::execute): Call gimple_purge_dead_eh_edges at the end
of bbs. Return TODO_cleanup_cfg if any returned true.
* g++.dg/ubsan/pr71512.C: New test.
* c-c++-common/ubsan/pr71512-1.c: New test.
* c-c++-common/ubsan/pr71512-2.c: New test.
Michael Meissner [Fri, 12 Aug 2016 19:40:37 +0000 (19:40 +0000)]
vsx.md (vsx_concat_<mode>): Add support for the ISA 3.0 MTVSRDD instruction.
[gcc]
2016-08-12 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/vsx.md (vsx_concat_<mode>): Add support for the
ISA 3.0 MTVSRDD instruction.
(vsx_splat_<mode>): Change cpu type of MTVSRDD instruction to
vecperm.
[gcc/testsuite]
2016-08-12 Michael Meissner <meissner@linux.vnet.ibm.com>
* gcc.target/powerpc/vec-init-1.c: New tests to test various
vector initialization options.
* gcc.target/powerpc/vec-init-2.c: Likewise.
* gcc.target/powerpc/vec-init-3.c: New test to make sure MTVSRDD
is generated on ISA 3.0.
Jakub Jelinek [Fri, 12 Aug 2016 16:04:33 +0000 (18:04 +0200)]
re PR c/67410 (c/c-typeck.c references out of bounds array)
PR c/67410
* c-typeck.c (set_nonincremental_init_from_string): Use / instead of
% to determine val element to change. Assert that
wchar_bytes * charwidth fits into val array.
* gcc.dg/pr67410.c: New test.
Co-Authored-By: Martin Liska <mliska@suse.cz>
From-SVN: r239419
Jakub Jelinek [Fri, 12 Aug 2016 15:52:03 +0000 (17:52 +0200)]
sse.md (avx512dq_vextract<shuffletype>64x2_1_maskm, [...]): Use Yk constraint instead of k.
* config/i386/sse.md (avx512dq_vextract<shuffletype>64x2_1_maskm,
vec_extract_hi_<mode>_mask, vec_extract_lo_<mode>_mask,
vec_extract_hi_<mode>_mask): Use Yk constraint instead of k.
Bin Cheng [Fri, 12 Aug 2016 14:58:20 +0000 (14:58 +0000)]
re PR tree-optimization/69848 (poor vectorization of a loop from SPEC2006 464.h264ref)
PR tree-optimization/69848
* tree-vectorizer.h (enum vect_def_type): New condition reduction
type CONST_COND_REDUCTION.
* tree-vect-loop.c (vectorizable_reduction): Support new condition
reudction type CONST_COND_REDUCTION.
gcc/testsuite
PR tree-optimization/69848
* gcc.dg/vect/vect-pr69848.c: New test.
Uros Bizjak [Fri, 12 Aug 2016 13:44:13 +0000 (15:44 +0200)]
re PR testsuite/71008 (FAIL: gcc.target/i386/avx512f-vprord-1.c scan-assembler-times vprord)
PR testsuite/71008
* gcc.target/i386/avx512f-vprord-1.c: Scan also for vprold.
* gcc.target/i386/avx512f-vprorq-1.c: Scan also for vprolq.
* gcc.target/i386/avx512vl-vprord-1.c: Scan also for vprold.
* gcc.target/i386/avx512vl-vprorq-1.c: Scan also for vprolq.
Richard Biener [Fri, 12 Aug 2016 12:58:15 +0000 (12:58 +0000)]
re PR tree-optimization/57326 (Piecewise folding of operations on PHI nodes)
2016-08-12 Richard Biener <rguenther@suse.de>
PR tree-optimization/57326
* tree-ssa-pre.c (fully_constant_expression): Handle simplification
returning an SSA name.
(phi_translate_1): When fully_constant_expression returns a NAME
make sure we have a leader for it.
Alexandre Oliva [Fri, 12 Aug 2016 07:11:50 +0000 (07:11 +0000)]
[PR63240] generate debug info for defaulted member functions
This implements <http://dwarfstd.org/ShowIssue.php?issue=141215.3>, a
proposal already accepted for inclusion in DWARF-5, but using
DW_AT_GNU_defaulted instead of DW_AT_defaulted as the attribute name,
because the attribute id for DW_AT_defaulted is not yet publicly
available.
Alexandre Oliva [Fri, 12 Aug 2016 07:11:36 +0000 (07:11 +0000)]
[PR55641] drop spurious const_type from reference_type variables
Although C++ reference types, denoted by DW_TAG_reference_type in
DWARFv2+ debug info, are unchangeable, we output names of reference type
with DW_TAG_const_type, because internally we mark such variables as
TREE_READONLY. That's an internal implementation detail that shouldn't
leak to debug information. This patch fixes this.
for gcc/ChangeLog
PR debug/55641
* dwarf2out.c (decl_quals): Don't map TREE_READONLY to
TYPE_QUAL_CONST in reference-typed decls.
Alexandre Oliva [Fri, 12 Aug 2016 07:11:23 +0000 (07:11 +0000)]
[PR49366] emit loc exprs for C++ non-virtual pmf template value parms
We used to emit, in debug information, the values bound to pointer to
member function template parameters only when they were NULL or
virtual member functions, because those can be represented with
DW_AT_const_value.
In order to represent the symbolic pointer to member function
constants for non-virtual member functions, we'd need to be able to
emit relocations for part of DW_AT_const_value, which we don't. The
more viable alternative is to use DW_AT_location to represent such
values, as slated for inclusion in DWARFv5, according to
<URL:http://www.dwarfstd.org/ShowIssue.php?issue=130412.1>.
With this patch, when we can't emit a DW_AT_const_value, we emit each
"member" of the pointer to member function "record" as a
DW_OP_stack_value DW_OP_piece, as long as the referenced member
function is output in the same translation unit, otherwise we'd get
relocations to external symbols, something to avoid in debug sections.
for gcc/ChangeLog
PR debug/49366
* dwarf2out.c (loc_list_from_tree_1): Expand some CONSTRUCTORs
in DW_OP_pieces, just enough to handle pointers to member
functions.
(gen_remaining_tmpl_value_param_die_attribute): Use a location
expression on DWARFv5 if a constant value doesn't work.
Ian Lance Taylor [Thu, 11 Aug 2016 21:36:13 +0000 (21:36 +0000)]
os: fix build tags for dir_regfile.go
We want to build dir_regfile.go if not GNU/linux, and not solaris/386,
and not solaris/sparc. The latter two conditions were incorrect. To
write ! solaris/386 we have to write !solaris !386. I forgot De
Morgan's Law.
H.J. Lu [Thu, 11 Aug 2016 17:36:52 +0000 (17:36 +0000)]
Support TImode CONST_WIDE_INT store in 64-bit STV
Support TImode CONST_WIDE_INT store generated from piecewise store.
Need to verify performance impact before enabling TImode CONST_INT
store for __int128.
H.J. Lu [Thu, 11 Aug 2016 15:51:01 +0000 (15:51 +0000)]
Use TImode for piecewise move in 64-bit mode
Use TImode for piecewise move in 64-bit mode. We should use TImode in
32-bit mode and use OImode or XImode if they are available. But since
by_pieces_ninsns determines the widest mode with MAX_FIXED_MODE_SIZE,
we can only use TImode in 64-bit mode.
gcc/
* config/i386/i386.h (MOVE_MAX_PIECES): Use TImode in 64-bit
mode if unaligned SSE load and store are optimal.
Jakub Jelinek [Thu, 11 Aug 2016 14:50:09 +0000 (16:50 +0200)]
re PR c/72816 (ICE on x86_64-linux-gnu (tree check: expected tree that contains ‘decl with RTL’ structure, have ‘field_decl’ in set_decl_rtl, at emit-rtl.c:1282))
PR c/72816
* c-decl.c (grokdeclarator): When adding TYPE_DOMAIN for flexible
array member through typedef, for orig_qual_indirect == 0 clear
orig_qual_type.
PR tree-optimization/71083
* gcc.c-torture/execute/pr71083.c: New test.
* gnat.dg/loop_optimization23.adb: New test.
* gnat.dg/loop_optimization23_pkg.ads: New test.
* gnat.dg/loop_optimization23_pkg.adb: New test.
Janne Blomqvist [Thu, 11 Aug 2016 08:58:55 +0000 (11:58 +0300)]
Replace KISS PRNG with xorshift1024* using per-thread state.
frontend:
2016-08-11 Janne Blomqvist <jb@gcc.gnu.org>
* check.c (gfc_check_random_seed): Use new seed size in check.
* intrinsic.texi (RANDOM_NUMBER): Updated documentation.
(RANDOM_SEED): Likewise.
testsuite:
2016-08-11 Janne Blomqvist <jb@gcc.gnu.org>
* gfortran.dg/random_7.f90: Take into account that the last seed
value is the special p value.
* gfortran.dg/random_seed_1.f90: Seed size is now constant.
libgfortran:
2016-08-11 Janne Blomqvist <jb@gcc.gnu.org>
* intrinsics/random.c: Replace KISS with xorshift1024* using
per-thread state.
* runtime/main.c (init): Don't call random_seed_i4.
* tree-ssa-threadbackward.c (pass_data_thread_jumps): Remove
unconditional TODO_cleanup_cfg.
(pass_thread_jumps::execute): Initialize loops, perform a CFG
cleanup only if we threaded a jump.