Jakub Jelinek [Tue, 17 Jan 2017 09:38:48 +0000 (10:38 +0100)]
re PR other/79046 (g++ -print-file-name=plugin uses full version number in path)
PR other/79046
* configure: Regenerated.
config/
* acx.m4 (GCC_BASE_VER): New m4 function.
(ACX_TOOL_DIRS): Require GCC_BASE_VER, for
--with-gcc-major-version-only use just major number from BASE-VER.
gcc/
* configure.ac: Add GCC_BASE_VER.
* Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
version from BASE-VER file.
(CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
(gcc.o): Depend on $(BASEVER).
* common.opt (dumpfullversion): New option.
* gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
* doc/invoke.texi: Document -dumpfullversion.
* doc/install.texi: Document --with-gcc-major-version-only.
* configure: Regenerated.
libatomic/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* testsuite/Makefile.in: Regenerated.
* configure: Regenerated.
* Makefile.in: Regenerated.
libgomp/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* testsuite/Makefile.in: Regenerated.
* configure: Regenerated.
* Makefile.in: Regenerated.
libgcc/
* configure.ac: Add GCC_BASE_VER.
* Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
version from BASE-VER file.
* configure: Regenerated.
libssp/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
* Makefile.in: Regenerated.
liboffloadmic/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* aclocal.m4: Include ../config/acx.m4.
* configure: Regenerated.
* Makefile.in: Regenerated.
libquadmath/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
* Makefile.in: Regenerated.
libmpx/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
* Makefile.in: Regenerated.
libada/
* configure.ac: Add GCC_BASE_VER.
* Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
version from BASE-VER file.
* configure: Regenerated.
lto-plugin/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
* Makefile.in: Regenerated.
libitm/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* testsuite/Makefile.in: Regenerated.
* configure: Regenerated.
* Makefile.in: Regenerated.
fixincludes/
* configure.ac: Add GCC_BASE_VER.
* Makefile.in (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
libcilkrts/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* aclocal.m4: Include ../config/acx.m4.
* configure: Regenerated.
* Makefile.in: Regenerated.
libcc1/
* configure.ac: Add GCC_BASE_VER. For --with-gcc-major-version-only
use just major number from BASE-VER.
* configure: Regenerated.
* Makefile.in: Regenerated.
libobjc/
* configure.ac: Add GCC_BASE_VER.
* Makefile.in (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
libstdc++-v3/
* configure.ac: Add GCC_BASE_VER.
* fragment.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* po/Makefile.in: Regenerated.
* libsupc++/Makefile.in: Regenerated.
* testsuite/Makefile.in: Regenerated.
* src/Makefile.in: Regenerated.
* configure: Regenerated.
* Makefile.in: Regenerated.
* include/Makefile.in: Regenerated.
* doc/Makefile.in: Regenerated.
* python/Makefile.in: Regenerated.
* src/c++11/Makefile.in: Regenerated.
* src/c++98/Makefile.in: Regenerated.
* src/filesystem/Makefile.in: Regenerated.
libvtv/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* testsuite/Makefile.in: Regenerated.
* configure: Regenerated.
* Makefile.in: Regenerated.
libsanitizer/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* libbacktrace/Makefile.in: Regenerated.
* interception/Makefile.in: Regenerated.
* asan/Makefile.in: Regenerated.
* ubsan/Makefile.in: Regenerated.
* configure: Regenerated.
* sanitizer_common/Makefile.in: Regenerated.
* lsan/Makefile.in: Regenerated.
* Makefile.in: Regenerated.
* tsan/Makefile.in: Regenerated.
libgfortran/
* configure.ac: Add GCC_BASE_VER.
* Makefile.am (gcc_version): Use @get_gcc_base_ver@ instead of cat to
get version from BASE-VER file.
* configure: Regenerated.
* Makefile.in: Regenerated.
Richard Biener [Tue, 17 Jan 2017 08:38:59 +0000 (08:38 +0000)]
re PR tree-optimization/71433 (-Warray-bounds false positive with -O2)
2017-01-17 Richard Biener <rguenther@suse.de>
PR tree-optimization/71433
* tree-vrp.c (register_new_assert_for): Merge same asserts
on all incoming edges.
(process_assert_insertions_for): Handle insertions at the
beginning of BBs.
Jeff Law [Mon, 16 Jan 2017 23:51:57 +0000 (16:51 -0700)]
revert: Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection for several include directories that may be relative...
Revert:
2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
* Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
for several include directories that may be relative to sysroot.
* config/i386/x-mingw32 (gplus_includedir): Define.
(gplus_tool_includedir, gplus_backward_include_dir): Likewise.
(native_system_includedir): Likewise.
* config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
override if TARGET_SYSTEM_ROOT is defined.
(NATIVE_SYSTEM_HEADER_DIR): Likewise.
David Malcolm [Mon, 16 Jan 2017 18:08:45 +0000 (18:08 +0000)]
Fix testcases for PR c/78304
The testcases as written made assumptions about size_t and long
being invalid for use with "%u".
We only need some invalid type, so this patch converts them to
attempt a "const char *" with "%u", which should be invalid for
every target.
gcc/testsuite/ChangeLog:
PR c/78304
* gcc.dg/format/pr78304.c: Convert argument from integral type
to a pointer.
* gcc.dg/format/pr78304-2.c: Likewise.
Carl Love [Mon, 16 Jan 2017 17:18:05 +0000 (17:18 +0000)]
rs6000-c (altivec_overloaded_builtins): Add support for built-in functions vector signed char vec_nabs (vector signed...
gcc/ChangeLog:
2017-01-16 Carl Love <cel@us.ibm.com>
* config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
for built-in functions
vector signed char vec_nabs (vector signed char)
vector signed short vec_nabs (vector signed short)
vector signed int vec_nabs (vector signed int)
vector signed long long vec_nabs (vector signed long long)
vector float vec_nabs (vector float)
vector double vec_nabs (vector double)
* config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
and NABS overload.
* config/rs6000/altivec.md: New define_expand nabs<mode>2 types
* config/rs6000/altivec.h: New define for vec_nabs built-in function.
* doc/extend.texi: Update the documentation file for the new built-in
functions.
gcc/testsuite/ChangeLog:
2017-01-16 Carl Love <cel@us.ibm.com>
* gcc.target/powerpc/builtins-3.c: New vec_nabs testcase.
* gcc.target/powerpc/builtins-3-p8.c: New vec_nabs testcase.
Jonathan Wakely [Mon, 16 Jan 2017 11:41:41 +0000 (11:41 +0000)]
PR78702 fix accessibility of locale::facet::__shim
PR libstdc++/78702
* include/bits/locale_classes.h (locale::facet::__shim): Change from
private to protected.
* src/c++11/cxx11-shim_facets.cc (__shim_accessor): Define helper to
make locale::facet::__shim accessible.
Richard Biener [Mon, 16 Jan 2017 09:33:12 +0000 (09:33 +0000)]
re PR tree-optimization/77283 (Revision 238005 disables loop unrolling)
2017-01-13 Richard Biener <rguenther@suse.de>
PR tree-optimization/77283
* gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
and ssa-iterators.h.
(is_feasible_trace): Implement a cost model based on joiner
PHI node uses.
* gcc.dg/tree-ssa/split-path-7.c: Adjust.
* gcc.dg/tree-ssa/split-path-8.c: New testcase.
* gcc.dg/tree-ssa/split-path-9.c: Likewise.
Make rtl_split_edge work for jumps that fall through (PR72749)
If a jump always falls through but that cannot be optimised away, like
is the case with the PowerPC bdnz insn if its jump target is the same as
the fallthrough, sched gets confused if it schedules some instructions
from before that jump instruction to behind it: it splits the
fallthrough branch, but the jump target isn't updated, and things fall
apart as in the PR. This patch fixes it.
The second patch fragment makes -fsched-verbose=N work for N>=4; the
currently scheduled fragment can now contain a label. Everything else
seems to work fine with that.
PR target/72749
* cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
fallthrough.
* haifa-sched.c (dump_insn_stream): Don't crash if there is a label
in the currently scheduled RTL fragment.
Jeff Law [Sun, 15 Jan 2017 09:34:01 +0000 (02:34 -0700)]
re PR tree-optimization/79090 (DSE wrongly removes store at variable offset)
PR tree-optimization/79090
* tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
variable length stores.
(compute_trims): Delete dead assignment to *trim_tail.
(dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
zero length.
Bernd Schmidt [Sat, 14 Jan 2017 16:52:18 +0000 (09:52 -0700)]
re PR rtl-optimization/78626 (ICE in rtl_verify_bb_insns, at cfgrtl.c:2656 (error: flow control insn inside a basic block))
PR rtl-optimization/78626
PR rtl-optimization/78727
* cprop.c (one_cprop_pass): Collect unconditional traps in the middle
of a block, and split such blocks after everything else is finished.
PR rtl-optimization/78626
PR rtl-optimization/78727
* gcc.dg/torture/pr78626.c: New test.
* gcc.dg/torture/pr78727.c: New test.
Alan Modra [Sat, 14 Jan 2017 13:24:46 +0000 (23:54 +1030)]
Avoid PR72749 by not using unspecs
Rather than using unspecs in doloop insns to stop combine creating
these insns, use legitimate_combined_insn.
PR target/72749
* combine.c (recog_for_combine_1): Set INSN_CODE before calling
target legitimate_combined_insn.
* config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
(rs6000_legitimate_combined_insn): New function.
* config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
all uses.
(ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
(ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
(ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
Jeff Law [Sat, 14 Jan 2017 06:16:23 +0000 (23:16 -0700)]
re PR tree-optimization/33562 (aggregate DSE disabled)
PR tree-optimization/33562
PR tree-optimization/61912
PR tree-optimization/77485
* tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
a statement.
(delete_dead_assignment): Likewise.
(dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
statement to delete_dead_call and delete_dead_assignment.
Ian Lance Taylor [Sat, 14 Jan 2017 01:50:31 +0000 (01:50 +0000)]
compiler: implement type aliases
This is a start of implementing type aliases (`type T1 = T2`) in the
Go frontend. This is incomplete, in that the reflection information
is not updated for an embedded type alias. It is also not well
tested. Finally, the change to the language has not been approved.
This should be regarded as preliminary work for experimental use.
Ian Lance Taylor [Sat, 14 Jan 2017 00:05:42 +0000 (00:05 +0000)]
libgo: update to Go 1.8 release candidate 1
Compiler changes:
* Change map assignment to use mapassign and assign value directly.
* Change string iteration to use decoderune, faster for ASCII strings.
* Change makeslice to take int, and use makeslice64 for larger values.
* Add new noverflow field to hmap struct used for maps.
Unresolved problems, to be fixed later:
* Commented out test in go/types/sizes_test.go that doesn't compile.
* Commented out reflect.TestStructOf test for padding after zero-sized field.
David Malcolm [Fri, 13 Jan 2017 19:27:43 +0000 (19:27 +0000)]
Don't suppress bogus usage of macros from system headers in -Wformat (PR c/78304)
gcc/ChangeLog:
PR c/78304
* substring-locations.c (format_warning_va): Strengthen case 1 so
that both endpoints of the substring must be within the format
range for just the substring to be printed.
gcc/testsuite/ChangeLog:
PR c/78304
* gcc.dg/format/diagnostic-ranges.c (test_macro): Undef INT_FMT.
(test_macro_2): New test.
(test_macro_3): New test.
(test_macro_4): New test.
(test_non_contiguous_strings): Convert line number to line offset.
* gcc.dg/format/pr78304-2.c: New test case.
* gcc.dg/format/pr78304.c: New test case.
Uros Bizjak [Fri, 13 Jan 2017 18:32:44 +0000 (19:32 +0100)]
i386.opt (msgx): Use ix86_isa_flags2 variable.
* config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
* config/i386/i386.c (ix86_target_string): Add missing options
to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
flags_other and ix86_target_other to flags2_other. Display unknown
isa2 options.
(ix86_valid_target_attribute_inner_p): Add missing options and
reorder options by implied ISAs, as in ix86_target_string.
testsuite/ChangeLog:
* gcc.target/i386/funcspec-56.inc: Add missing options and
reorder options by implied ISAs, as in ix86_target_string.
Janne Blomqvist [Fri, 13 Jan 2017 17:05:48 +0000 (19:05 +0200)]
PR 78534 Change character length from int to size_t
In order to handle large character lengths on (L)LP64 targets, switch
the GFortran character length from an int to a size_t.
This is an ABI change, as procedures with character arguments take
hidden arguments with the character length.
I also changed the _size member in vtables from int to size_t, as
there were some cases where character lengths and sizes were
apparently mixed up and caused regressions otherwise. Although I
haven't tested, this might enable very large derived types as well.
Also, as there are some places in the frontend were negative character
lengths are used as special flag values, in the frontend the character
length is handled as a signed variable of the same size as a size_t,
although in the runtime library it really is size_t.
I haven't changed the character length variables for the co-array
intrinsics, as this is something that may need to be synchronized with
OpenCoarrays.
This is v4 of the patch. v3 was applied but had to reverted due to
breaking bootstrap. The fix is in resolve.c:resolve_charlen, where
it's necessary to check that an expression is constant before using
mpz_sgn.
Overview of v3 of the patch: All the issues pointed out by FX's review
of v2 have been fixed. In particular, there are now new functions
gfc_mpz_get_hwi and gfc_mpz_set_hwi, similar to the GMP functions
mpz_get_si and mpz_set_si, except that they get/set a HOST_WIDE_INT
instead of a long value. Similarly, gfc_get_int_expr now takes a
HOST_WIDE_INT instead of a long, gfc_extract_long is replaced by
gfc_extract_hwi. Also, the preliminary work to handle
gfc_charlen_type_node being unsigned has been removed.
Regtested on x86_64-pc-linux-gnu and i686-pc-linux-gnu.
frontend:
2017-01-13 Janne Blomqvist <jb@gcc.gnu.org>
PR fortran/78534
PR fortran/66310
* class.c (gfc_find_derived_vtab): Use gfc_size_kind instead of
hardcoded kind.
(find_intrinsic_vtab): Likewise.
* expr.c (gfc_get_character_expr): Length parameter of type
gfc_charlen_t.
(gfc_get_int_expr): Value argument of type HOST_WIDE_INT.
(gfc_extract_hwi): New function.
(simplify_const_ref): Make string_len of type gfc_charlen_t.
(gfc_simplify_expr): Use HOST_WIDE_INT for substring refs.
* gfortran.h (gfc_mpz_get_hwi): New prototype.
(gfc_mpz_set_hwi): Likewise.
(gfc_charlen_t): New typedef.
(gfc_expr): Use gfc_charlen_t for character lengths.
(gfc_size_kind): New extern variable.
(gfc_extract_hwi): New prototype.
(gfc_get_character_expr): Use gfc_charlen_t for character length.
(gfc_get_int_expr): Use HOST_WIDE_INT type for value argument.
* iresolve.c (gfc_resolve_repeat): Pass string length directly without
temporary, use gfc_charlen_int_kind.
* match.c (select_intrinsic_set_tmp): Use HOST_WIDE_INT for charlen.
* misc.c (gfc_mpz_get_hwi): New function.
(gfc_mpz_set_hwi): New function.
* module.c (atom_int): Change type from int to HOST_WIDE_INT.
(parse_integer): Don't complain about large integers.
(write_atom): Use HOST_WIDE_INT for integers.
(mio_integer): Handle integer type mismatch.
(mio_hwi): New function.
(mio_intrinsic_op): Use HOST_WIDE_INT.
(mio_array_ref): Likewise.
(mio_expr): Likewise.
* resolve.c (resolve_select_type): Use HOST_WIDE_INT for charlen,
use snprintf.
(resolve_substring_charlen): Use gfc_charlen_int_kind.
(resolve_charlen): Use mpz_sgn to determine sign.
* simplify.c (gfc_simplify_repeat): Use HOST_WIDE_INT/gfc_charlen_t
instead of long.
* target-memory.c (size_character): Length argument of type
gfc_charlen_t.
(gfc_encode_character): Likewise.
(gfc_interpret_character): Use gfc_charlen_t.
* target-memory.h (gfc_encode_character): Modify prototype.
* trans-array.c (get_array_ctor_var_strlen): Use
gfc_conv_mpz_to_tree_type.
* trans-const.c (gfc_conv_mpz_to_tree_type): New function.
* trans-const.h (gfc_conv_mpz_to_tree_type): New prototype.
* trans-expr.c (gfc_class_len_or_zero_get): Build const of type
gfc_charlen_type_node.
(gfc_conv_intrinsic_to_class): Use gfc_charlen_int_kind instead of
4, fold_convert to correct type.
(gfc_conv_class_to_class): Build const of type size_type_node for
size.
(gfc_copy_class_to_class): Likewise.
(gfc_conv_string_length): Use same type in expression.
(gfc_conv_substring): Likewise, use HOST_WIDE_INT for charlen.
(gfc_conv_string_tmp): Make sure len is of the right type.
(gfc_conv_concat_op): Use same type in expression.
(gfc_conv_procedure_call): Likewise.
(alloc_scalar_allocatable_for_subcomponent_assignment):
fold_convert to right type.
(gfc_trans_subcomponent_assign): Likewise.
(trans_class_vptr_len_assignment): Build const of correct type.
(gfc_trans_pointer_assignment): Likewise.
(alloc_scalar_allocatable_for_assignment): fold_convert to right
type in expr.
(trans_class_assignment): Build const of correct type.
* trans-intrinsic.c (gfc_conv_associated): Likewise.
(gfc_conv_intrinsic_repeat): Do calculation in sizetype.
* trans-io.c (gfc_build_io_library_fndecls): Use
gfc_charlen_type_node for character lengths.
* trans-stmt.c (gfc_trans_label_assign): Build const of
gfc_charlen_type_node.
(gfc_trans_character_select): Likewise.
(gfc_trans_allocate): Likewise, don't typecast strlen result.
(gfc_trans_deallocate): Don't typecast strlen result.
* trans-types.c (gfc_size_kind): New variable.
(gfc_init_types): Determine gfc_charlen_int_kind and gfc_size_kind
from size_type_node.
testsuite:
2017-01-13 Janne Blomqvist <jb@gcc.gnu.org>
PR fortran/78534
PR fortran/66310
* gfortran.dg/repeat_4.f90: Use integers of kind C_SIZE_T.
* gfortran.dg/repeat_7.f90: New test for PR 66310.
* gfortran.dg/scan_2.f90: Handle potential cast in assignment.
* gfortran.dg/string_1.f90: Limit to ilp32 targets.
* gfortran.dg/string_1_lp64.f90: New test.
* gfortran.dg/string_3.f90: Limit to ilp32 targets.
* gfortran.dg/string_3_lp64.f90: New test.
libgfortran:
2017-01-13 Janne Blomqvist <jb@gcc.gnu.org>
PR fortran/78534
* intrinsics/args.c (getarg_i4): Use gfc_charlen_type.
(get_command_argument_i4): Likewise.
(get_command_i4): Likewise.
* intrinsics/chmod.c (chmod_internal): Likewise.
* intrinsics/env.c (get_environment_variable_i4): Likewise.
* intrinsics/extends_type_of.c (struct vtype): Use size_t for size
member.
* intrinsics/gerror.c (gerror): Use gfc_charlen_type.
* intrinsics/getlog.c (getlog): Likewise.
* intrinsics/hostnm.c (hostnm_0): Likewise.
* intrinsics/string_intrinsics_inc.c (string_len_trim): Rework to
work if gfc_charlen_type is unsigned.
(string_scan): Likewise.
* io/transfer.c (transfer_character): Modify prototype.
(transfer_character_write): Likewise.
(transfer_character_wide): Likewise.
(transfer_character_wide_write): Likewise.
(transfer_array): Typecast to avoid signed-unsigned comparison.
* io/unit.c (is_trim_ok): Use gfc_charlen_type.
* io/write.c (namelist_write): Likewise.
* libgfortran.h (gfc_charlen_type): Change typedef to size_t.
Avoid excessively-big hash tables in empty-add cycles
A big source of cache misses when compiling a recent version of
gimple-match.ii was the call to cv_cache.empty () in clear_cv_cache.
The problem was that at one early point the hash table had grown
to 8191 entries (128k on LP64 hosts). It then stayed at that size
for the rest of the compilation, even though subsequent uses needed
only a small number of entries (usually fewer than ten). We would
still clear the whole 128k each time clear_cv_cache was called.
empty() already looks for cases where the hash table is very big
and cuts it down. At the moment it fires when the table is 1M
in size and reduces it to the next selected prime above 1K (so
almost 2K in practice). One fix would have been to lower the
threshold, but that didn't feel like the right approach. Reducing
the current limit of 1M by a factor of 8 would be pretty significant
on its own, but I think this cv_cache behaviour would have been a
problem even with 64k or 32k tables.
I think the existing check is really for cases in which even a
well-populated table would need to be shrunk rather than cleared.
Here the problem isn't that the table is excessively big in
absolute terms, more that one outlier has made the table much
too big for the general case.
traverse() already shrinks the table if it's "too empty",
which is taken to be if:
no. elements * 8 < capacity && capacity > 32
So an alternative would be to apply the same test (and the same choice
of shrunken size) to empty_slow too. The patch below does this.
It gives a 2.5% improvement in gimple-match.ii compile time at -O0 -g
and doesn't seem to adversely affect any other tests I've tried.
Of course, there's a theoretical risk of a table alternating between
one large element count and one small element count. If there was a
factor of eight difference between the two, we could shrink the table
on seeing each small element count, then grow it again when adding the
large number of elements. That seems pretty unlikely in practice
though.
Also, empty_slow() does involve a traversal if some form of manual
gc is needed on active elements, so trying to recover from an outlier
should have even more benefit there. (cv_cache uses automatic gc and so
the traversal gets optimised away.)
The calculation of the existing 1M threshold was assuming that each
entry was pointer-sized. This patch makes it use the actual size of the
entry instead.
Tested on aarch64-linux-gnu and x86_64-linux-gnu.
gcc/
* hash-table.h (hash_table::too_empty_p): New function.
(hash_table::expand): Use it.
(hash_table::traverse): Likewise.
(hash_table::empty_slot): Use sizeof (value_type) instead of
sizeof (PTR) to convert bytes to elements. Shrink the table
if the current size is excessive for the current number of
elements.
record_reg_classes is often the hottest function when generating
unoptimised output. It seems typical for over 60% of the instructions
it handles to be moves, and of course moves tend to be the instructions
with the longest constraint strings.
Maybe we should avoid using move constraints to set costs in unoptimised
output and instead use the "natural" class for the mode being moved.
That's too invasive for stage 3 though.
However, seeing so many moves means that we see many "failing"
alternatives, usually because of '*' or because of hard registers
in function call sequences. The frequency of alternatives that are
detected as failures after the first operand tends again to be more
than 60%. Previously we would continue to process the other operands
of the alternative regardless. This patch instead adds a short-cut.
As well as avoiding unnecessary work, it means that the alt_fail
variable can be jump-threaded away.
Tested on aach64-linux-gnu and x86_64-linux-gnu. It reduces compile
time by about 1% on some tests with "-g -O0".
gcc/
* ira-costs.c (record_reg_classes): Break from the inner loop
early once alt_fail is known to be true. Update outer loop
handling accordingly.
Jeff Law [Fri, 13 Jan 2017 15:50:11 +0000 (08:50 -0700)]
tree-ssa-dse.c (decrement_count): New function.
* tree-ssa-dse.c (decrement_count): New function.
(increment_start_addr, maybe_trim_memstar_call): Likewise.
(dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
when we know the partially dead statement is a mem* function.
Jeff Law [Fri, 13 Jan 2017 15:42:08 +0000 (08:42 -0700)]
re PR tree-optimization/33562 (aggregate DSE disabled)
PR tree-optimization/33562
PR tree-optimization/61912
PR tree-optimization/77485
* doc/invoke.texi: Document new dse-max-object-size param.
* params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
* tree-ssa-dse.c: Include params.h.
(dse_store_status): New enum.
(initialize_ao_ref_for_dse): New, partially extracted from
dse_optimize_stmt.
(valid_ao_ref_for_dse, normalize_ref): New.
(setup_live_bytes_from_ref, compute_trims): Likewise.
(clear_bytes_written_by, maybe_trim_complex_store): Likewise.
(maybe_trim_partially_dead_store): Likewise.
(maybe_trim_complex_store): Likewise.
(dse_classify_store): Renamed from dse_possibly_dead_store_p.
Track what bytes live from the original store. Return tri-state
for dead, partially dead or live.
(dse_dom_walker): Add constructor, destructor and new private members.
(delete_dead_call, delete_dead_assignment): New extracted from
dse_optimize_stmt.
(dse_optimize_stmt): Make a member of dse_dom_walker.
Use initialize_ao_ref_for_dse.
re PR libstdc++/78389 (list::merge and list::sort are not exception safe)
PR libstdc++/78389
* include/bits/list.tcc (merge(list&&)):
Adjust list sizes if the comparator throws.
(merge(list&&, _StrictWeakOrdering)): Likewise.
(sort()): Splice elements back from the scratch buffers
if the comparator throws.
(sort(_StrictWeakOrdering)): Likewise.
* testsuite/23_containers/list/operations/78389.cc: New.
Martin Liska [Fri, 13 Jan 2017 12:56:54 +0000 (13:56 +0100)]
Reload global options when strict aliasing is dropped (PR ipa/79043).
2017-01-13 Martin Liska <mliska@suse.cz>
PR ipa/79043
* function.c (set_cfun): Add new argument force.
* function.h (set_cfun): Likewise.
* ipa-inline-transform.c (inline_call): Use the function when
strict alising from is dropped for function we inline to.
2017-01-13 Martin Liska <mliska@suse.cz>
PR ipa/79043
* gcc.c-torture/execute/pr79043.c: New test.
Arnaud Charlet [Fri, 13 Jan 2017 11:16:56 +0000 (12:16 +0100)]
[multiple changes]
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* einfo.ads: minor grammar fixes in comment of Normalized_Position_Max.
* scil_ll.adb: Minor style fix in comment.
* sem_ch8.adb (Analyze_Expanded_Name): Perform dimension analysis
even if entity is already set, because the node may be renalyzed
after inlining transformations.
2017-01-13 Javier Miranda <miranda@adacore.com>
* sem_res.adb (Resolve_Call): Do not establish a transient scope
for a call to inlinable expression function (since the call will
be replaced by its returned object).
* exp_ch6.ads (Is_Inlinable_Expression_Function): New subprogram.
* exp_ch6.adb (Expression_Of_Expression_Function): New subprogram.
(Expand_Call): For inlinable expression function call replace the
call by its returned object.
(Is_Inlinable_Expression_Function): New subprogram.
Arnaud Charlet [Fri, 13 Jan 2017 11:13:00 +0000 (12:13 +0100)]
[multiple changes]
2017-01-13 Gary Dismukes <dismukes@adacore.com>
* checks.adb: Minor typo fix and reformatting.
2017-01-13 Javier Miranda <miranda@adacore.com>
* contracts.adb (Contract_Only_Subprograms): Remove formal.
(Copy_Original_Specification): Removed.
(Skip_Contract_Only_Subprogram): Move here checks previously
located in the caller of this routine (to leave the code more clean).
(Build_Contract_Only_Subprogram): Code cleanup.
* scil_ll.ads, scil_ll.adb (Get_Contract_Only_Body_Name): Removed.
(Get_Contract_Only_Missing_Body_Name): Removed.
Arnaud Charlet [Fri, 13 Jan 2017 11:08:57 +0000 (12:08 +0100)]
[multiple changes]
2017-01-13 Javier Miranda <miranda@adacore.com>
* sem_ch6.adb (Cloned_Expression): New subprogram.
(Freeze_Expr_Types): Complete previous patch since the expression
of an expression-function may have iterators and loops with
defining identifiers which, as part of the preanalysis of the
expression, may be left decorated with itypes that will not be
available in the tree passed to the backend.
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* checks.adb (Apply_Type_Conversion_Checks): Optimize a type
conversion to Integer of an expression that is an attribute
reference 'Pos on an enumeration type.
* sem_ch6.adb (Analyze_Subprogram_Body_Helper):
Ensure that the input body is a subprogram body before trying to
determine whether it denoted an expression function. Note that
subprogram body stubs cannot denote expression functions.
Arnaud Charlet [Fri, 13 Jan 2017 10:54:43 +0000 (11:54 +0100)]
[multiple changes]
2017-01-13 Javier Miranda <miranda@adacore.com>
* einfo.ads (Component_Bit_Offset): Fix documentation.
* sem_ch13.adb (Check_Record_Representation_Clause): Skip check
on record holes for components with unknown compile-time offsets.
* binde.adb (Elab_New): New elaboration order algorithm
that is expected to cause fewer ABE issues. This is a work in
progress. The new algorithm is currently disabled, and can be
enable by the -dp switch, or by modifying the Do_Old and Do_New
etc. flags and rebuilding. Experimental code is included to
compare the results of the old and new algorithms.
* binde.ads: Use GNAT.Dynamic_Tables instead of Table, so we
can have multiple of these tables, so the old and new algorithms
can coexist.
* bindgen.ads (Gen_Output_File): Pass Elab_Order as an 'in'
parameter of type array. This avoids the global variable, and
allows bounds checking (which is normally defeated by the tables
packages). It also ensures that the Elab_Order is read-only
to Bindgen.
* bindgen.adb: Pass Elab_Order as an 'in' parameter to all
subprograms that need it, as above.
* debug.adb: Document new -dp switch. Modify doc of old -do
switch.
* gnatbind.adb (Gnatbind): Make use of new interfaces to Binde
and Bindgen. Move writing of closure (-R and -Ra switches)
to Binde; that's more convenient.
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* sem_ch6.adb (Analyze_Expression_Function): If the expression
function is a completion, all entities referenced in the
expression are frozen. As a consequence, a reference to an
uncompleted private type from an enclosing scope is illegal.
Arnaud Charlet [Fri, 13 Jan 2017 10:51:45 +0000 (11:51 +0100)]
[multiple changes]
2017-01-13 Javier Miranda <miranda@adacore.com>
* sem_ch6.adb (Freeze_Expr_Types): New subprogram.
(Analyze_Subprogram_Body_Helper): At the occurrence of an
expression function declaration that is a completion, its
expression causes freezing (AI12-0103).
2017-01-13 Vadim Godunko <godunko@adacore.com>
* a-coinho-shared.adb: Fix memory leaks in Constant_Reference and
Reference functions of Ada.Containers.Indefinite_Holders.
* exp_ch3.adb (Default_Initialize_Object): Do not default
initialize an object when it is of a task type and restriction
No_Tasking is in effect because the initialization is obsolete.
* exp_ch9.adb (Build_Master_Entity): Do not generate a master when
restriction No_Tasking is in effect.
(Build_Master_Renaming): Do not rename a master when restriction
No_Tasking is in effect.
Arnaud Charlet [Fri, 13 Jan 2017 10:48:57 +0000 (11:48 +0100)]
[multiple changes]
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* sem_aggr.adb (Resolve_Array_Aggregate): The code that verifies
the legality of An others clause applies as well to a choice in
an Iterated_component_ association.
(Resolve_Iterated_Component_Association): An others choice
is legal.
* exp_aggr.adb (Aggr_Assignment_OK_For_Backend): An
Iterated_Component_Association is not static.
* exp_ch3.adb (Freeze_Type): Mark the Ghost mode as set in case
control is passed to the expresion handler before the new mode
is set.
* sem_ch12.adb (Analyze_Package_Instantiation,
Analyze_Subprogram_Instantiation): Mark the Ghost mode as set
in case control is passed to the expresion handler before the
new mode is set.
Arnaud Charlet [Fri, 13 Jan 2017 10:42:37 +0000 (11:42 +0100)]
[multiple changes]
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* sem_util.ads, sem_util.adb (Choice_List): Move function here
from sem_aggr.adb, for use elsewhere.
* sem_ch3.adb (Analyze_Object_Declaration): Use Choice_List.
* sem_aggr.adb (Resolve_Array_Aggregate): Remove
Iterated_Component_Present.
* exp_aggr.adb: Use Choice_List throughout, to handle
Iterated_Component_Associations.
(Gen_Loop): Generate proper loop for an
Iterated_Component_Association: loop variable has the identifier
of the original association. Generate a loop even for a single
component choice, in order to make loop parameter visible in
expression.
(Flatten): An Iterated_Component_Association is not static.
2017-01-13 Yannick Moy <moy@adacore.com>
* exp_ch4.adb (Expand_N_Op_Expon): Ensure that the value of
float exponentiation for statically known small negative values
is the reciprocal of the exponentiation for the opposite value
of the exponent.
* s-exnllf.adb (Exn_Float, Exn_Long_Float, Exn_Long_Long_Float):
Ensure that the value of float exponentiation for negative values
is the reciprocal of the exponentiation for the opposite value
of the exponent.
* inline.adb (Expand_Inlined_Call): Fix the count
for the number of generated gotos.
Arnaud Charlet [Fri, 13 Jan 2017 10:38:56 +0000 (11:38 +0100)]
[multiple changes]
2017-01-13 Yannick Moy <moy@adacore.com>
* inline.adb: Code cleanup.
* sem_util.adb (Is_OK_Volatile_Context): Add
expression in delay statement as OK for volatile context.
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* sem_aggr.adb (Resolve_Array_Aggregate): In normal compilation
mode a choice that is a subtype with a static predicate is
replaced by the values it covers. This transformation must not
be performed in ASIS mode, to preserve the source for analysis.
Arnaud Charlet [Fri, 13 Jan 2017 10:33:37 +0000 (11:33 +0100)]
[multiple changes]
2017-01-13 Gary Dismukes <dismukes@adacore.com>
* sem_ch13.adb: Minor reformatting and typo fix.
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* par-ch4.adb (P_Aggregate_Or_Parent_Expr): An
Iterated_Component_Association is a named association in an
array aggregate.
* sem_aggr.adb (Resolve_Iterated_Component_Association): New
procedure, subsidiary of Resolve_Array_Aggregate, to analyze
and resolve the discrete choices and the expression of the
new construct.
* sinfo.adb, sinfo.ads: In analogy with N_Component_Association,
Loop_Actions and Box_Present are attributes of
N_Iterated_Component_Association nodes. Box_Present is always
False in this case.
* sprint.adb (Sprint_Node): An Iterated_Component_Association
has a Discrete_Choices list, as specified in the RM. A
Component_Association for aggregate uses instead a Choices list.
We have to live with this small inconsistency because the new
construct also has a defining identifier, and there is no way
to merge the two node structures.
Arnaud Charlet [Fri, 13 Jan 2017 10:24:28 +0000 (11:24 +0100)]
[multiple changes]
2017-01-13 Yannick Moy <moy@adacore.com>
* inline.adb (Remove_Aspects_And_Pragmas): Add Unused to the
list of pragmas to remove. Remove pragmas from the list of
statements in the body to inline.
* namet.adb, namet.ads (Nam_In): New version with 12 parameters.
2017-01-13 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Resolve_Aspects): New procedure, subsidiary of
Analyze_Declarations, to analyze and resolve the expressions of
aspect specifications in the current declarative list, so that
the expressions have proper entity and type info. This is needed
for ASIS when there is no subsequent expansion to generate this
semantic information.
* sem_ch13.adb (Check_Aspect_At_End_Of_Declarations): Use Etype of
original expression, to suppress cascaded errors when expression
has been constant-folded.
(Resolve_Aspect_Expressions, Resolve_Name): Preserve entities in
ASIS mode, because there is no subsequent expansion to decorate
the tree.
Yannick Moy [Fri, 13 Jan 2017 10:22:23 +0000 (10:22 +0000)]
inline.adb, inline.ads (Call_Can_Be_Inlined_In_GNATprove_Mode): New function to detect when a call may be inlined or not in GNATprove mode.
2017-01-13 Yannick Moy <moy@adacore.com>
* inline.adb, inline.ads (Call_Can_Be_Inlined_In_GNATprove_Mode):
New function to detect when a call may be inlined or not in
GNATprove mode.
(Expand_Inlined_Call): Ensure that a temporary
is always created in the cases where a type conversion may be
needed on an input parameter in GNATprove mode, so that GNATprove
sees the check to perform.
* sem_res.adb (Resolve_Call): In GNATprove mode, skip inlining
when not applicable due to actual requiring type conversion
with possible check but no temporary value can be copied for
GNATprove to see the check.