David Malcolm [Tue, 30 Aug 2016 13:54:48 +0000 (13:54 +0000)]
rich_location: add convenience overloads for adding fix-it hints
Adding a fix-it hint to a diagnostic usually follows one of these
patterns:
(a) an insertion fix-its, with the insertion at the primary caret location
(b) a removals/replacements, affecting the range of the primary location
(other cases are possible, e.g. multiple fix-its, and affecting other
locations, but these are the common ones)
Given these common cases, this patch adds overloads of the rich_location
methods for adding fix-it hints, so that the location information can
be omitted if it matches that of the primary location within the
rich_location.
Similarly when adding "remove" and "replace" fix-it hints to a diagnostic,
it's tedious to have to extract the source_range from a location_t
(aka source_location). To make this more convenient, this patch
adds overload of the rich_location::add_fixit_remove/replace methods,
accepting a source_location directly.
The patch updates the various in-tree users of fix-it hints to use
the new simpler API where appropriate. I didn't touch the case where
there are multiple fix-its in one rich_location, as it seems better to
be more explicit about locations for this case (adding a pair of parens
in warn_logical_not_parentheses).
The above makes the gcc_rich_location::add_fixit_misspelled_id overload
taking a const char * rather redundant, so I eliminated it.
gcc/c/ChangeLog:
* c-decl.c (implicit_decl_warning): Use add_fixit_replace
rather than add_fixit_misspelled_id.
(undeclared_variable): Likewise.
* c-parser.c (c_parser_declaration_or_fndef): Likewise. Remove
now-redundant "here" params from add_fixit_insert method calls.
(c_parser_parameter_declaration): Likewise.
* c-typeck.c (build_component_ref): Remove now-redundant range
param from add_fixit_replace method calls.
gcc/cp/ChangeLog:
* name-lookup.c (suggest_alternatives_for): Use add_fixit_replace
rather than add_fixit_misspelled_id.
* parser.c (cp_parser_diagnose_invalid_type_name): Likewise.
gcc/ChangeLog:
* diagnostic-show-locus.c (test_one_liner_fixit_insert): Remove
redundant location param.
(test_one_liner_fixit_remove): Likewise.
(test_one_liner_fixit_replace): Likewise.
(test_one_liner_fixit_replace_equal_secondary_range): Likewise.
* gcc-rich-location.c
(gcc_rich_location::add_fixit_misspelled_id): Eliminate call to
get_range_from_loc. Drop overload taking a const char *.
* gcc-rich-location.h
(gcc_rich_location::add_fixit_misspelled_id): Drop overload taking
a const char *.
libcpp/ChangeLog:
* include/line-map.h (rich_location::add_fixit_insert): Add
comments. Add overload omitting the source_location param.
(rich_location::add_fixit_remove): Add comments. Add overloads
omitting the range, and accepting a source_location.
(rich_location::add_fixit_replace): Likewise.
* line-map.c (rich_location::add_fixit_insert): Add comments. Add
overload omitting the source_location param.
(rich_location::add_fixit_remove): Add comments. Add overloads
omitting the range, and accepting a source_location.
(rich_location::add_fixit_replace): Likewise.
Jakub Jelinek [Tue, 30 Aug 2016 06:54:02 +0000 (08:54 +0200)]
re PR tree-optimization/72866 (Compile time hog w/ -O3 (-Ofast))
PR tree-optimization/72866
* tree-vect-patterns.c (search_type_for_mask): Turn into
a small wrapper, move all code to ...
(search_type_for_mask_1): ... this new function. Add caching
and adjust recursive calls.
Jakub Jelinek [Tue, 30 Aug 2016 06:46:38 +0000 (08:46 +0200)]
re PR c++/77363 (Missing debug information in DWARF)
PR debug/77363
* dwarf2out.c (modified_type_die): Use lookup_type_die (type)
instead of lookup_type_die (type_main_variant (type)) even for array
types.
Ian Lance Taylor [Tue, 30 Aug 2016 03:27:43 +0000 (03:27 +0000)]
compiler: add -fgo-c-header=FILE to create a C header
The new -fgo-c-header=FILE option will write a C header file defining
all the struct types and numeric const values in package scope. This
will be used when building the Go runtime package (libgo/go/runtime) to
generate a C header file that may be included by the C code in the C
runtime package (libgo/runtime).
This will ensure that the Go code and C code are working with the same
data structures as we convert the runtime from C to Go to upgrade to the
current GC runtime, notably the concurrent garbage collector.
Nothing uses MQ anymore, but it still shows up in all the dump files.
This patch removes it from CALL_REALLY_USED_REGISTERS so that that does
not happen anymore (it is still a fixed register, there should be no
functional change).
* config/rs6000/rs6000.h (CALL_REALLY_USED_REGISTERS): Do not
include MQ.
David Malcolm [Mon, 29 Aug 2016 20:42:57 +0000 (20:42 +0000)]
Allow the use of ad-hoc locations for fix-it hints
Currently the fix-it validator rejects ad-hoc locations.
Fix this by calling get_pure_location on the input locations to
add_fixit_insert/replace. Doing so requires moving get_pure_location
from gcc to libcpp.
gcc/ChangeLog:
* diagnostic-show-locus.c
(selftest::test_one_liner_fixit_validation_adhoc_locations): New
function.
(selftest::test_diagnostic_show_locus_one_liner): Call it.
* input.c (get_pure_location): Move to libcpp/line-map.c.
* input.h (get_pure_location): Convert decl to an inline function
calling implementation in libcpp.
libcpp/ChangeLog:
* include/line-map.h (get_pure_location): New decl.
* line-map.c (get_pure_location): Move here, from gcc/input.c, adding
a line_maps * param.
(rich_location::add_fixit_insert): Call get_pure_location on "where".
(rich_location::add_fixit_replace): Call get_pure_location on the
end-points.
Michael Meissner [Mon, 29 Aug 2016 19:38:07 +0000 (19:38 +0000)]
vec-init-4.c: New runtime tests for various vector short/char initializations.
2016-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
* gcc.target/powerpc/vec-init-4.c: New runtime tests for various
vector short/char initializations.
* gcc.target/powerpc/vec-init-5.c: Likewise.
* gcc.target/powerpc/vec-init-6.c: New compile time test for
vector initialization optimizations.
* gcc.target/powerpc/vec-init-7.c: Likewise.
* gcc.target/powerpc/vec-init-8.c: New runtime tests for various
vector float/double initializations.
* gcc.target/powerpc/vec-init-9.c: Likewise.
David Malcolm [Mon, 29 Aug 2016 17:10:23 +0000 (17:10 +0000)]
make_location: ensure end-points are pure locations
gcc/ChangeLog:
* input.c (make_location): Call get_start and get_finish
on the endpoints to avoid storing packed ranges or ad-hoc
ranges in them.
(selftest::test_make_location_nonpure_range_endpoints): New function.
(selftest::input_c_tests): Call it.
* input.h (get_start): New inline function.
Fritz Reese [Mon, 29 Aug 2016 12:24:25 +0000 (12:24 +0000)]
Fix, reorganize, and clarify comparisons of anonymous types/components.
2016-08-29 Fritz Reese <fritzoreese@gmail.com>
Fix, reorganize, and clarify comparisons of anonymous types/components.
PR fortran/77327
* interface.c (is_anonymous_component, is_anonymous_dt): New functions.
* interface.c (compare_components, gfc_compare_derived_types): Use new
functions.
Tom de Vries [Sun, 28 Aug 2016 05:05:15 +0000 (05:05 +0000)]
Tag {ms,sysv}_va_list_type_node with '{ms,sysv}_abi va_list' attribute
2016-08-28 Tom de Vries <tom@codesourcery.com>
PR lto/70955
* config/i386/i386.c (ix86_build_builtin_va_list_64): Tag type with
'sysv_abi va_list' attribute.
(ix86_build_builtin_va_list): Tag type with 'ms_abi va_list' attribute.
(ix86_canonical_va_list_type): Handle 'sysv_abi/ms_abi va_list'
attributes.
* gcc.dg/pr70955.c: New test.
* gcc.dg/lto/pr70955_0.c: Same.
* gcc.dg/lto/pr70955_1.c: Same.
Patrick Palka [Sat, 27 Aug 2016 22:00:17 +0000 (22:00 +0000)]
Fix folding of VECTOR_CST comparisons
gcc/ChangeLog:
PR tree-optimization/71077
PR tree-optimization/68542
* fold-const.c (fold_relational_const): Fix folding of
VECTOR_CST comparisons that have a scalar boolean result type.
(selftest::test_vector_folding): New static function.
(selftest::fold_const_c_tests): Call it.
gcc/testsuite/ChangeLog:
PR tree-optimization/71077
* gcc.target/i386/pr71077.c: New test.
Tim Shen [Sat, 27 Aug 2016 02:03:23 +0000 (02:03 +0000)]
re PR libstdc++/77356 (regex error for a ECMAScript syntax string)
PR libstdc++/77356
* include/bits/regex_compiler.tcc(_M_insert_bracket_matcher,
_M_expression_term): Modify to support dash literal.
* include/bits/regex_scanner.h: Add dash as a token type to make
a different from the mandated dash literal by escaping.
* include/bits/regex_scanner.tcc(_M_scan_in_bracket): Emit dash
token in bracket expression parsing.
* testsuite/28_regex/regression.cc: Add new testcases.
David Malcolm [Fri, 26 Aug 2016 21:25:41 +0000 (21:25 +0000)]
Add validation and consolidation of fix-it hints
The first aspect of this patch is to add some checking of fix-it hints.
The idea is to put this checking within the rich_location machinery,
rather than requiring every diagnostic to implement it for itself.
The fixits within a rich_location are "atomic": all must be valid for
any to be applicable.
We reject any fixits involving locations above
LINE_MAP_MAX_LOCATION_WITH_COLS.
There's no guarantee that it's sane to modify a macro, so we reject
any fix-its that touch them.
For example, note the attempt to provide a fix-it for the definition
of the macro FIELD:
spellcheck-fields-2.c: In function ‘test_macro’:
spellcheck-fields-2.c:26:15: error: ‘union u’ has no member named ‘colour’; did you mean ‘color’?
#define FIELD colour
^
color
spellcheck-fields-2.c:27:15: note: in expansion of macro ‘FIELD’
return ptr->FIELD;
^~~~~
After this patch, the fixit is not displayed:
spellcheck-fields-2.c: In function ‘test_macro’:
spellcheck-fields-2.c:26:15: error: ‘union u’ has no member named ‘colour’; did you mean ‘color’?
#define FIELD colour
^
spellcheck-fields-2.c:27:15: note: in expansion of macro ‘FIELD’
return ptr->FIELD;
^~~~~
We might want some way for a diagnostic to opt-in to fix-its that
affect macros, but for now it's simplest to reject them.
The other aspect of this patch is fix-it consolidation: in some cases
neighboring fix-its can be merged. For example, in a diagnostic to
modernize old-style struct initializers from:
struct s example = {
- foo: 1,
+ .foo = 1,
};
one approach would be to replace the "foo" with ".foo" and the ":"
with " =". This would give two "replace" fix-its:
foo: 1,
--- FIXIT 1
.foo
- FIXIT 2
=
This patch allows them to be consolidated into a single "replace" fix-it:
foo: 1,
----
.foo =
gcc/ChangeLog:
* diagnostic-show-locus.c
(selftest::test_fixit_consolidation): New function.
(selftest::diagnostic_show_locus_c_tests): Call it.
* gcc-rich-location.h (gcc_rich_location): Eliminate unused
constructor based on source_range.
gcc/testsuite/ChangeLog:
* gcc.dg/spellcheck-fields-2.c (test): Move
dg-begin/end-multiline-output within function body.
(test_macro): New function.
libcpp/ChangeLog:
* include/line-map.h (rich_location): Eliminate unimplemented
constructor based on source_range.
(rich_location::get_last_fixit_hint): New method.
(rich_location::reject_impossible_fixit): New method.
(rich_location): Add fields m_line_table and
m_seen_impossible_fixit.
(fixit_hint::maybe_append_replace): New pure virtual function.
(fixit_insert::maybe_append_replace): New function.
(fixit_replace::maybe_append_replace): New function.
* line-map.c (rich_location::rich_location): Initialize
m_line_table and m_seen_impossible_fixit.
(rich_location::add_fixit_insert): Call
reject_impossible_fixit and bail out if true.
(column_before_p): New function.
(rich_location::add_fixit_replace): Call reject_impossible_fixit
and bail out if true. Attempt to consolidate with neighboring
fixits.
(rich_location::get_last_fixit_hint): New method.
(rich_location::reject_impossible_fixit): New method.
(fixit_insert::maybe_append_replace): New method.
(fixit_replace::maybe_append_replace): New method.
David Malcolm [Fri, 26 Aug 2016 17:59:08 +0000 (17:59 +0000)]
Tweak to colors of fix-it hints
Previous, fix-it hints were printed using the color of the severity
of the diagnostic (magenta for warnings, red for errors, cyan for
notes).
This patch updates fix-it hints so that replacement text is printed in
green, to better distinguish the suggested improvement from
the current code. For example:
spellcheck-fields.cc:52:13: error: 'struct s' has no member named 'colour'; did you mean 'color'?
return ptr->colour; <<< RED
^~~~~~ <<< RED
color <<< GREEN
It makes sense for the underlinings that indicate deletions to
be printed in red, so the patch changes that also. For example:
diagnostic-test-show-locus-color.c:179:9: warning: example of a removal hint
int a;; <<< MAGENTA
^ <<< MAGENTA
- <<< RED
gcc/ChangeLog:
* diagnostic-color.c (color_dict): Add "fixit-insert" and
"fixit-delete".
(parse_gcc_colors): Update description of default GCC_COLORS.
* diagnostic-show-locus.c (colorizer::set_fixit_hint): Delete.
(colorizer::set_fixit_insert): New method.
(colorizer::set_fixit_delete): New method.
(colorizer::get_color_by_name): New method.
(colorizer::STATE_FIXIT_INSERT): New constant.
(colorizer::STATE_FIXIT_DELETE): New constant.
(class colorizer): Drop "_cs" suffix from fields. Delete "_ce"
fields in favor of new field "m_stop_color". Add fields
"m_fixit_insert" and "m_fixit_delete".
(colorizer::colorizer): Update for above changes. Replace
colorize_start calls with calls to get_color_by_name.
(colorizer::begin_state): Handle STATE_FIXIT_INSERT and
STATE_FIXIT_DELETE. Update for field renamings.
(colorizer::finish_state): Simplify by using m_stop_color,
rather than multiple identical "*_ce" fields.
(colorizer::get_color_by_name): New method.
(layout::print_any_fixits): Print insertions and replacements
using the "fixit-insert" color, and deletions using the
"fixit-delete" color.
* doc/invoke.texi (-fdiagnostics-color): Update description of
default GCC_COLORS, and of the supported capabilities.
Max Filippov [Fri, 26 Aug 2016 17:23:37 +0000 (17:23 +0000)]
xtensa: report stack usage
This enables options -fstack-usage and -Wstack-usage.
2016-08-26 Max Filippov <jcmvbkbc@gmail.com>
gcc/
* config/xtensa/xtensa.c (xtensa_expand_prologue): Update
current_function_static_stack_size variable with the static
stack frame size of the current function when
flag_stack_usage_info is enabled.
Richard Biener [Fri, 26 Aug 2016 12:33:43 +0000 (12:33 +0000)]
re PR tree-optimization/69047 (memcpy is not as optimized as union is)
2016-08-26 Richard Biener <rguenther@suse.de>
PR tree-optimization/69047
* tree-ssa.c (maybe_rewrite_mem_ref_base): Handle general bitfield
extracts similar to what FRE does.
(non_rewritable_mem_ref_base): Likewise.
Jonathan Wakely [Fri, 26 Aug 2016 11:35:58 +0000 (12:35 +0100)]
Restore dg-interpreter-batch-mode for libstdc++ tests
2016-08-26 Jonathan Wakely <jwakely@redhat.com>
Pedro Alves <palves@redhat.com>
* testsuite/lib/gdb-test.exp (gdb-dg-runtest): Define wrapper to save
and restore dg-interpreter-batch-mode.
* testsuite/libstdc++-prettyprinters/prettyprinters.exp: Use
gdb-dg-runtest instead of dg-runtest.
* testsuite/libstdc++-xmethods/xmethods.exp: Likewise.
Co-Authored-By: Pedro Alves <palves@redhat.com>
From-SVN: r239776
Joseph Myers [Fri, 26 Aug 2016 11:13:42 +0000 (12:13 +0100)]
Always support float128 on x86.
In <https://gcc.gnu.org/ml/gcc-bugs/2016-08/msg03233.html>, Nick
reported i386-elf and ia64-elf failing to build because of
float128_type_node being NULL, but being used by the back end for
__float128.
The global float128_type_node is only available conditionally, if
target hooks indicate TFmode is not only available as a scalar mode
and of the right format, but also supported in libgcc. The back-end
support, however, expects the type always to be available for
__float128 even if the libgcc support is missing.
Although a target-specific node could be restored in the case where
libgcc support is missing, it seems better to address the missing
libgcc support. Thus, this patch enables TFmode soft-fp in libgcc
globally for all x86 targets - the only special cases needed being for
targets that use soft-fp for SFmode and DFmode, one of which already
had the support for TFmode as well (so I based the i[34567]86-*-rtems*
configuration on that present for i[34567]86-*-elfiamcu). The i386
implementation of TARGET_LIBGCC_FLOATING_MODE_SUPPORTED_P is then
removed as no longer needed.
I can provide such a patch for ia64 if useful, but am not in a
position to test it (and while I'm reasonably confident that enabling
this support would be right for ia64-elf and ia64-freebsd, I've no
real idea if enabling libgcc support for TFmode, with or without also
enabling it for XFmode, would be safe for ia64-vms).
Bootstrapped with no regressions on x86_64-pc-linux-gnu.
Patch for performing interprocedural bitwise constant propagation.
2016-08-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
Martin Jambhor <mjambor@suse.cz>
* common.opt: New option -fipa-bit-cp.
* doc/invoke.texi: Document -fipa-bit-cp.
* opts.c (default_options_table): Add entry for -fipa-bit-cp.
(enable_fdo_optimizations): Check for flag_ipa_bit_cp.
* tree-ssa-ccp.h: New header file.
* tree-ssa-ccp.c: Include tree-ssa-ccp.h
(bit_value_binop_1): Change to bit_value_binop_1 and export it.
Replace all occurences of tree parameter by two new params: signop, int.
(bit_value_unop_1): Change to bit_value_unop and export it.
Replace all occurences of tree parameter by two new params: signop,
int.
(bit_value_binop): Change call from bit_value_binop_1 to
bit_value_binop.
(bit_value_assume_aligned): Likewise.
(bit_value_unop): Change call from bit_value_unop_1 to bit_value_unop.
(do_ssa_ccp): Pass nonzero_p || flag_ipa_cp_bit instead of nonzero_p
to ccp_finalize.
(ccp_finalize): Skip processing if val->mask == 0.
* ipa-cp.c: Include tree-ssa-ccp.h
(ipcp_bits_lattice): New class.
(ipcp_param_lattice (bits_lattice): New member.
(print_all_lattices): Call ipcp_bits_lattice::print.
(set_all_contains_variable): Call ipcp_bits_lattice::set_to_bottom.
(initialize_node_lattices): Likewise.
(propagate_bits_accross_jump_function): New function.
(propagate_constants_accross_call): Call
propagate_bits_accross_jump_function.
(ipcp_propagate_stage): Store parameter types when in_lto_p is true.
(ipcp_store_bits_results): New function.
(ipcp_driver): Call ipcp_store_bits_results.
* ipa-prop.h (ipa_bits): New struct.
(ipa_jump_func): Add new member bits of type ipa_bits.
(ipa_param_descriptor): Change decl to decl_or_type.
(ipa_get_param): Change decl to decl_or_type and assert on
PARM_DECL.
(ipa_get_type): New function.
(ipcp_transformation_summary): New member bits.
* ipa-prop.c (ipa_get_param_decl_index_1): s/decl/decl_or_type.
(ipa_populate_param_decls): Likewise.
(ipa_dump_param): Likewise.
(ipa_print_node_jump_functions_for_edge): Pretty-print ipa_bits jump
function.
(ipa_set_jf_unknown): Set ipa_bits::known to false.
(ipa_compute_jump_functions_for_edge): Compute jump function for bits
propagation.
(ipa_node_params_t::duplicate): Copy src->bits into dst->bits.
(ipa_write_jump_function): Add streaming for ipa_bits.
(ipa_read_jump_function): Add support for reading streamed ipa_bits.
(write_ipcp_transformation_info): Add streaming for ipa_bits
summary for ltrans.
(read_ipcp_transfomration_info): Add support for reading streamed ipa_bits.
(ipcp_update_bits): New function.
(ipcp_transform_function): Call ipcp_update_bits.
Richard Biener [Thu, 25 Aug 2016 09:25:31 +0000 (09:25 +0000)]
dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes): Only add locations in late dwarf.
2016-08-25 Richard Biener <rguenther@suse.de>
* dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
Only add locations in late dwarf.
(gen_scheduled_generic_parms_dies): Do not set early dwarf here.
(dwarf2out_early_finish): But do it here.
Marek Polacek [Thu, 25 Aug 2016 08:22:56 +0000 (08:22 +0000)]
re PR c/77323 (Bad "defaults to 'int'" warning for unsupported types)
PR c/77323
* c-decl.c (declspecs_add_type): Set typespec_word even when __intN
or _FloatN or _FloatNx is not supported.
(finish_declspecs): Set type to integer_type_node when _FloatN or
_FloatNx is not supported.
Michael Collison [Wed, 24 Aug 2016 19:31:20 +0000 (19:31 +0000)]
arm-modes.def: Add new condition code mode CC_V to represent the overflow bit.
2016-08-24 Michael Collison <michael.collison@linaro.org>
Michael Collison <michael.collison@arm.com>
* config/arm/arm-modes.def: Add new condition code mode CC_V
to represent the overflow bit.
* config/arm/arm.c (maybe_get_arm_condition_code):
Add support for CC_Vmode.
(arm_gen_unlikely_cbranch): New function to generate common
rtl conditional branches for overflow patterns.
* config/arm/arm-protos.h: Add prototype for
arm_gen_unlikely_cbranch.
* config/arm/arm.md (addv<mode>4, add<mode>3_compareV,
addsi3_compareV_upper): New patterns to support signed
builtin overflow add operations.
(uaddv<mode>4, add<mode>3_compareC, addsi3_compareV_upper):
New patterns to support unsigned builtin add overflow operations.
(subv<mode>4, sub<mode>3_compare1): New patterns to support signed
builtin overflow subtract operations,
(usubv<mode>4): New patterns to support unsigned builtin subtract
overflow operations.
(negvsi3, negvdi3, negdi2_compare, negsi2_carryin_compare): New patterns
to support builtin overflow negate operations.
* gcc.target/arm/builtin_saddl.c: New testcase.
* gcc.target/arm/builtin_saddll.c: New testcase.
* gcc.target/arm/builtin_uaddl.c: New testcase.
* gcc.target/arm/builtin_uaddll.c: New testcase.
* gcc.target/arm/builtin_ssubl.c: New testcase.
* gcc.target/arm/builtin_ssubll.c: New testcase.
* gcc.target/arm/builtin_usubl.c: New testcase.
* gcc.target/arm/builtin_usubll.c: New testcase.
Co-Authored-By: Michael Collison <michael.collison@arm.com>
From-SVN: r239739
fragments.texi (MULTILIB_REUSE): Mention that only options in MULTILIB_OPTIONS should be used.
2016-08-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
* doc/fragments.texi (MULTILIB_REUSE): Mention that only options in
MULTILIB_OPTIONS should be used. Small wording fixes.
* genmultilib: Memorize set of all option combinations in
combination_space. Detect if RHS of MULTILIB_REUSE uses an option not
found in MULTILIB_OPTIONS by checking if option set is listed in
combination_space. Output new and existing error message to stderr.
t-aprofile (MULTILIB_MATCHES): Add mapping for -mcpu=cortex-a7, -mfpu=neon-fp16, -mfpu=fpv5-d16 and -mfpu=fp-armv8.
2016-08-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
* config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping for
-mcpu=cortex-a7, -mfpu=neon-fp16, -mfpu=fpv5-d16 and -mfpu=fp-armv8.
Fix typo in -mfpu=vfpv3-d16-fp16 mapping.
(MULTILIB_REUSE): Remove reuse rules for option set including
-mfpu=fp-armv8 and -mfpu=vfpv4
Michael Meissner [Tue, 23 Aug 2016 20:41:32 +0000 (20:41 +0000)]
rs6000.c (rs6000_expand_vector_init): Set initialization of all 0's to the 0 constant, instead of directly generating XOR.
[gcc]
2016-08-23 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/rs6000.c (rs6000_expand_vector_init): Set
initialization of all 0's to the 0 constant, instead of directly
generating XOR. Add support for V4SImode vector initialization on
64-bit systems with direct move, and rework the ISA 3.0 V4SImode
initialization. Change variables used in V4SFmode vector
intialization. For V4SFmode vector splat on ISA 3.0, make sure
any memory addresses are in index form. Add support for using
VSPLTH/VSPLTB to initialize vector short and vector char vectors
with all of the same element.
(regno_or_subregno): New helper function to return a register
number for either REG or SUBREG.
(rs6000_adjust_vec_address): Do not generate ADDI <reg>,R0,<num>.
Use regno_or_subregno where possible.
(rs6000_split_v4si_init_di_reg): New helper function to build up a
DImode value from two SImode values in order to generate V4SImode
vector initialization on 64-bit systems with direct move.
(rs6000_split_v4si_init): Split up the insns for a V4SImode vector
initialization.
(rtx_is_swappable_p): V4SImode vector initialization insn is not
swappable.
* config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Add
declaration.
* config/rs6000/vsx.md (VSX_SPLAT_I): New mode iterators and
attributes to initialize V8HImode and V16QImode vectors with the
same element.
(VSX_SPLAT_COUNT): Likewise.
(VSX_SPLAT_SUFFIX): Likewise.
(UNSPEC_VSX_VEC_INIT): New unspec.
(vsx_concat_v2sf): Eliminate using 'preferred' register classes.
Allow SFmode values to come from Altivec registers.
(vsx_init_v4si): New insn/split for V4SImode vector initialization
on 64-bit systems with direct move.
(vsx_splat_<mode>, VSX_W iterator): Rework V4SImode and V4SFmode
vector initializations, to allow V4SImode vector initializations
on 64-bit systems with direct move.
(vsx_splat_v4si): Likewise.
(vsx_splat_v4si_di): Likewise.
(vsx_splat_v4sf): Likewise.
(vsx_splat_v4sf_internal): Likewise.
(vsx_xxspltw_<mode>, VSX_W iterator): Eliminate using 'preferred'
register classes.
(vsx_xxspltw_<mode>_direct, VSX_W iterator): Likewise.
(vsx_vsplt<VSX_SPLAT_SUFFIX>_di): New insns to support
initializing V8HImode and V16QImode vectors with the same
element.
* config/rs6000/rs6000.h (TARGET_DIRECT_MOVE_64BIT): Disallow
optimization if -maltivec=be.
[gcc/testsuite]
2016-08-23 Michael Meissner <meissner@linux.vnet.ibm.com>
* gcc.target/powerpc/vec-init-1.c: Add tests where the vector is
being created from pointers to memory locations.
* gcc.target/powerpc/vec-init-2.c: Likewise.
Ian Lance Taylor [Tue, 23 Aug 2016 17:45:45 +0000 (17:45 +0000)]
compiler: revamp scheme for ordering calls to import init fcns.
Switch to a new method for determining the order in which import init
functions are invoked: build an init fcn dependence DAG and walk the DAG
to rewrite/adjust priorities to account for discrepancies introduced by
"go test".
This patch includes a change to the export data format generated
by gccgo. Older versions of gccgo will not be able to read object files
produced by a newer gccgo, but the new gcc will still be able to read
old object files.
Richard Biener [Tue, 23 Aug 2016 13:58:19 +0000 (13:58 +0000)]
re PR tree-optimization/77286 (ICE in fold_convert_loc, at fold-const.c:2248 building 435.gromacs)
2016-08-23 Richard Biener <rguenther@suse.de>
PR tree-optimization/77286
* tree-vect-loop.c (vect_analyze_loop_form_1): Do not modify
the CFG here.
(vect_transform_loop): Split exit edges of loop and scalar
loop if required and at the appropriate time.
Jonathan Wakely [Tue, 23 Aug 2016 13:15:12 +0000 (14:15 +0100)]
libstdc++/77334 move assign RB trees of non-copyable types
PR libstdc++/77334
* include/bits/stl_tree.h (_Rb_tree::_M_move_assign): New functions.
(_Rb_tree::operator=(_Rb_tree&&)): Dispatch to _M_move_assign.
* testsuite/23_containers/map/77334.cc: New test.
Jonathan Wakely [Tue, 23 Aug 2016 10:13:26 +0000 (11:13 +0100)]
Implement resolution of LWG DR 685 precisely
PR libstdc++/71771
* include/bits/stl_iterator.h
(operator-(reverse_iterator<Iter>, reverse_iterator<Iter>): Only
define for C++98 mode.
(operator-(move_iterator<Iter>, move_iterator<Iter>): Don't define.
* testsuite/24_iterators/headers/iterator/synopsis.cc: Use
-std=gnu++98.
* testsuite/24_iterators/headers/iterator/synopsis_c++11.cc: New test.
* testsuite/24_iterators/headers/iterator/synopsis_c++14.cc: New test.
* testsuite/24_iterators/headers/iterator/synopsis_c++17.cc: New test.
* testsuite/24_iterators/move_iterator/greedy_ops.cc: Don't test
difference operator.
* testsuite/24_iterators/reverse_iterator/greedy_ops.cc: Only test
difference operator for C++98.
* testsuite/24_iterators/reverse_iterator/71771.cc: New test.
Dominik Vogt [Tue, 23 Aug 2016 09:20:15 +0000 (09:20 +0000)]
Drop excess size used for run time allocated stack variables.
* get_dynamic_stack_size is passed a SIZE of a data block (which is
allocated elsewhere), the SIZE_ALIGN of the SIZE (i.e. the alignment
of the underlying memory units (e.g. 32 bytes split into 4 times 8
bytes = 64 bit alignment) and the REQUIRED_ALIGN of the data portion
of the allocated memory.
* Assuming the function is called with SIZE = 2, SIZE_ALIGN = 8 and
REQUIRED_ALIGN = 64 it first adds 7 bytes to SIZE -> 9. This is
what is needed to have two bytes 8-byte-aligned at some memory
location without any known alignment.
* Finally round_push is called to round up SIZE to a multiple of the
stack slot size.
The key to understanding this is that the function assumes that
STACK_DYNMAIC_OFFSET is completely unknown at the time its called
and therefore it does not make assumptions about the alignment of
STACKPOINTER + STACK_DYNMAIC_OFFSET. The latest patch simply
hard-codes that SP + SDO is supposed to be aligned to at least
stack slot size (and does that in a very complicated way). Since
there is no guarantee that this is the case on all targets, the
patch is broken. It may miscalculate a SIZE that is too small in
some cases.
However, on many targets there is some guarantee about the
alignment of SP + SDO even if the actual value of SDO is unknown.
On s390x it's always 8-byte-aligned (stack slot size). So the
right fix should be to add knowledge about the target's guaranteed
alignment of SP + SDO to the function. I'm right now testing a
much simpler patch that uses
REGNO_POINTER_ALIGN(VIRTUAL_STACK_DYNAMIC_REGNUM) as the
alignment.
gcc/ChangeLog:
2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
* explow.c (get_dynamic_stack_size): Take known alignment of stack
pointer + STACK_DYNAMIC_OFFSET into account when calculating the
size needed. Correct a typo in a comment.
projects / thirdparty / gcc.git / log
