PR rtl-optimization/79032
* lra-constraints.c (simplify_operand_subreg): In the MEM case, test
the alignment of the adjusted memory reference against that of MODE,
instead of the alignment of the original memory reference.
From-SVN: r244311
Reduce the size of variant, it doesn't need an index of
type size_t internally.
* include/std/variant (parse_numbers.h): New include.
(__select_index): New.
(_Variant_storage<false, _Types...>::_M_reset_impl): Use
_index_type for comparison with variant_npos.
(_Variant_storage<false, _Types...>::__index_type): New.
(_Variant_storage<false, _Types...>::_M_index): Change the
type from size_t to __index_type.
(_Variant_storage<true, _Types...>::__index_type): New.
(_Variant_storage<true, _Types...>::_M_index): Change the
type from size_t to __index_type.
(_Variant_base::_M_valid): Use _Storage::__index_type
for comparison with variant_npos.
(variant::index): Use _Base::_Storage::__index_type
for comparison with variant_npos.
* testsuite/20_util/variant/index_type.cc: New.
From-SVN: r244309
2017-01-11 Richard Biener <rguenther@suse.de>
* tree-vrp.c (evrp_dom_walker::before_dom_children): Also
set range/nonnull info for PHI results. Do not set it on
stmts marked for removal.
* gcc.dg/tree-ssa/pr61743-1.c: Adjust.
From-SVN: r244305
* expr.c (store_field): In the bitfield case, fetch the return value
from the registers before applying a single big-endian adjustment.
Always do a final load for a BLKmode value not larger than a word.
From-SVN: r244299
PR c++/77949 identifies an ICE when the C++ frontend attempts to emit a
fix-it hint inserting a missing semicolon at column 4097 of a source file.
This column value exceeds LINE_MAP_MAX_COLUMN_NUMBER and hence isn't
representable using a location_t.
Attempting to do so leads to these problems, which this patch fixes:
(a) when encountering a column number > LINE_MAP_MAX_COLUMN_NUMBER we
create a new linemap with m_column_and_range_bits == 0, but
linemap_position_for_column doesn't check for this, and hence can emit
a bogus location_t value that's calculated relative to the previous
linemap start, but which will be decoded relative to the new linemap,
leading to very large incorrect line values.
(b) when encountering a column number that can't be represented, and
for which the linemap was pre-existing, the code would hit this assertion:
if (linemap_assert_fails (column < (1u << map->m_column_and_range_bits)))
around a bail-out condition. The patch replaces this assertion with a
simple conditional, to stop the ICE when this occurs, and fixes the
bit count (effective column bits, vs column+range bits)
(c) the C++ frontend wasn't checking for failure of
linemap_position_for_loc_and_offset when considering emitting the fix-it
hint. The patch adds a conditional, so that no fix-it hint is emitted
if the location is bogus.
gcc/cp/ChangeLog:
PR c++/77949
* parser.c (cp_parser_class_specifier_1): Only suggest inserting
a missing semicolon if we have a valid insertion location for
the fix-it hint.
gcc/ChangeLog:
PR c++/77949
* input.c (selftest::test_accessing_ordinary_linemaps): Verify
that we correctly handle column numbers greater than
LINE_MAP_MAX_COLUMN_NUMBER.
gcc/testsuite/ChangeLog:
PR c++/77949
* g++.dg/diagnostic/pr77949.C: New test case.
libcpp/ChangeLog:
PR c++/77949
* line-map.c (linemap_position_for_column): When calling
linemap_start_line, detect if a new linemap was created with
0 column bits, and bail out early if this is the case.
(linemap_position_for_loc_and_offset): Replace overzealous
linemap_assert_fails with a simple conditional; use correct
bit count.
From-SVN: r244292
I looked at a diff of proc.go between Go 1.7 and gccgo, and copied
over all the easy stuff.
Reviewed-on: https://go-review.googlesource.com/35090
From-SVN: r244291
2017-01-10 Joe Seymour <joe.s@somniumtech.com>
* config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
from TI's devices.csv file as of September 2016.
* config/msp430/msp430.c (msp430_mcu_data): Likewise.
From-SVN: r244289
2016-01-10 Richard Biener <rguenther@suse.de>
PR tree-optimization/79034
* tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
Propagate out degenerate PHIs in the joiner.
* g++.dg/torture/pr79034.C: New testcase.
From-SVN: r244274
Call _M_extract_* functions family through temporary int objects, so
it doesn't convert from lvalue to rvalue through a temporary in AVR
because of the incompatible types used in AVR-Libc.
This fixes compilation errors with AVR-Libc while compiling libstdc++
for AVR target.
2017-01-10 Felipe Magno de Almeida <felipe@expertisesolutions.com.br>
* include/bits/locale_facets_nonio.tcc
(time_get::_M_extract_via_format): Avoid compilation errors with
non-standard struct tm.
From-SVN: r244271
2017-01-10 François Dumont <fdumont@gcc.gnu.org>
Jonathan Wakely <jwakely@redhat.com>
* python/libstdcxx/v6/printers.py (_versioned_namespace): Define.
(is_specialization, strip_versioned_namespace): New helpers functions
to work with symbols in the versioned namespace.
(Printer.add_version): Add second name using versioned namespace.
(add_one_template_type_printer, add_one_type_printer): Add second
type printers using versioned namespace.
(register_type_printers): Add template type printer for basic_string.
(build_libstdcxx_dictionary): Remove dead code.
* python/libstdcxx/v6/xmethods.py: Make all matchers look for
versioned namespace.
* testsuite/libstdc++-prettyprinters/48362.cc: Adjust expected
results.
* testsuite/libstdc++-prettyprinters/whatis.cc: Likewise.
Co-Authored-By: Jonathan Wakely <jwakely@redhat.com>
From-SVN: r244262
Drop the size arguments for the hash/equal functions stored in type
descriptors. Types know what size they are. To make this work,
generate hash/equal functions for types that can use an identity
comparison but are not a standard size and alignment.
Drop the multiplications by 33 in the generated hash code and the
reflect package hash code. They are not necessary since we started
passing a seed value around, as the seed includes the hash of the
earlier values.
Copy the algorithms for standard types from the Go 1.7 runtime,
replacing the C functions.
Reviewed-on: https://go-review.googlesource.com/34983
From-SVN: r244256
* expr.c (store_field): In the bitfield case, if the value comes from
a function call and is of an aggregate type returned in registers, do
not modify the field mode; extract the value in all cases if the mode
is BLKmode and the size is not larger than a word.
From-SVN: r244249
PR tree-optimization/79007
PR tree-optimization/67955
* tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
conservative for pt.null when flag_non_call_exceptions is on.
From-SVN: r244247
PR tree-optimization/78899
* tree-if-conv.c (version_loop_for_if_conversion): Instead of
returning bool return struct loop *, NULL for failure and the new
loop on success.
(versionable_outer_loop_p): Don't version outer loop if it has
dont_vectorized bit set.
(tree_if_conversion): When versioning outer loop, ensure
tree_if_conversion is performed also on the inner loop of the
non-vectorizable outer loop copy.
* tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
LOOP_VECTORIZED in inner loop of the scalar outer loop and
prevent vectorization of it.
(vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
the outer loop vectorization of the non-scalar version is attempted
before vectorization of the inner loop in scalar version. If
outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
vectorization of its inner loop.
* tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
has 2 inner loops, rename also on edges from bb whose single pred
is outer_loop->header. Fix typo in function comment.
* gcc.target/i386/pr78899.c: New test.
* gcc.dg/pr71077.c: New test.
From-SVN: r244238
This started by moving procresize from C to Go so that we can pass the
right type to the memory allocator when allocating a p, which forced
the gomaxprocs variable to move from C to Go, and everything else
followed from that.
Reviewed-on: https://go-review.googlesource.com/34916
From-SVN: r244236