Here, we weren't recognizing that the template parameter of A is used by the
reference to d in the requires-clause of f. Fixed by passing down the
active template parameters in the context of normalization, and adding to
the mapping any such parameters shared by a member template used in the
constraint-expression.
* pt.c (struct find_template_parameter_info): Add ctx_parms.
(any_template_parm_r): Handle TEMPLATE_DECL.
(find_template_parameters): Take parms instead of their depth.
* constraint.cc (build_parameter_mapping): Pass them.
v8:
1. Rebase to master with Martin's static function (r280043) comments merge.
Boostrap/testsuite/SPEC2017 tested pass on Power8-LE.
2. TODO:
2.1. C++ devirt for multiple speculative call targets.
2.2. ipa-icf ipa_merge_profiles refine with COMDAT inline testcase.
This patch aims to fix PR69678 caused by PGO indirect call profiling
performance issues.
The bug that profiling data is never working was fixed by Martin's pull
back of topN patches, performance got GEOMEAN ~1% improvement(+24% for
511.povray_r specifically).
Still, currently the default profile only generates SINGLE indirect target
that called more than 75%. This patch leverages MULTIPLE indirect
targets use in LTO-WPA and LTO-LTRANS stage, as a result, function
specialization, profiling, partial devirtualization, inlining and
cloning could be done successfully based on it.
Performance can get improved from 0.70 sec to 0.38 sec on simple tests.
Details are:
1. PGO with topn is enabled by default now, but only one indirect
target edge will be generated in ipa-profile pass, so add variables to enable
multiple speculative edges through passes, speculative_id will record the
direct edge index bind to the indirect edge, indirect_call_targets length
records how many direct edges owned by the indirect edge, postpone gimple_ic
to ipa-profile like default as inline pass will decide whether it is benefit
to transform indirect call.
2. Use speculative_id to track and search the reference node matched
with the direct edge's callee for multiple targets. Actually, it is the
caller's responsibility to handle the direct edges mapped to same indirect
edge. speculative_call_info will return one of the direct edge specified,
this will leverage current IPA edge process framework mostly.
3. Enable LTO WPA/LTRANS stage multiple indirect call targets analysis for
profile full support in ipa passes and cgraph_edge functions. speculative_id
can be set by make_speculative id when multiple targets are binded to
one indirect edge, and cloned if new edge is cloned. speculative_id
is streamed out and stream int by lto like lto_stmt_uid.
4. Create and duplicate all speculative direct edge's call summary
in ipa-fnsummary.c with auto_vec.
5. Add 1 in module testcase and 2 cross module testcases.
6. Bootstrap and regression test passed on Power8-LE. No function
and performance regression for SPEC2017.
gcc/ChangeLog
2020-01-14 Xiong Hu Luo <luoxhu@linux.ibm.com>
PR ipa/69678
* cgraph.c (symbol_table::create_edge): Init speculative_id and
target_prob.
(cgraph_edge::make_speculative): Add param for setting speculative_id
and target_prob.
(cgraph_edge::speculative_call_info): Update comments and find reference
by speculative_id for multiple indirect targets.
(cgraph_edge::resolve_speculation): Decrease the speculations
for indirect edge, drop it's speculative if not direct target
left. Update comments.
(cgraph_edge::redirect_call_stmt_to_callee): Likewise.
(cgraph_node::dump): Print num_speculative_call_targets.
(cgraph_node::verify_node): Don't report error if speculative
edge not include statement.
(cgraph_edge::num_speculative_call_targets_p): New function.
* cgraph.h (int common_target_id): Remove.
(int common_target_probability): Remove.
(num_speculative_call_targets): New variable.
(make_speculative): Add param for setting speculative_id.
(cgraph_edge::num_speculative_call_targets_p): New declare.
(target_prob): New variable.
(speculative_id): New variable.
* ipa-fnsummary.c (analyze_function_body): Create and duplicate
call summaries for multiple speculative call targets.
* cgraphclones.c (cgraph_node::create_clone): Clone speculative_id.
* ipa-profile.c (struct speculative_call_target): New struct.
(class speculative_call_summary): New class.
(class speculative_call_summaries): New class.
(call_sums): New variable.
(ipa_profile_generate_summary): Generate indirect multiple targets summaries.
(ipa_profile_write_edge_summary): New function.
(ipa_profile_write_summary): Stream out indirect multiple targets summaries.
(ipa_profile_dump_all_summaries): New function.
(ipa_profile_read_edge_summary): New function.
(ipa_profile_read_summary_section): New function.
(ipa_profile_read_summary): Stream in indirect multiple targets summaries.
(ipa_profile): Generate num_speculative_call_targets from
profile summaries.
* ipa-ref.h (speculative_id): New variable.
* ipa-utils.c (ipa_merge_profiles): Update with target_prob.
* lto-cgraph.c (lto_output_edge): Remove indirect common_target_id and
common_target_probability. Stream out speculative_id and
num_speculative_call_targets.
(input_edge): Likewise.
* predict.c (dump_prediction): Remove edges count assert to be
precise.
* symtab.c (symtab_node::create_reference): Init speculative_id.
(symtab_node::clone_references): Clone speculative_id.
(symtab_node::clone_referring): Clone speculative_id.
(symtab_node::clone_reference): Clone speculative_id.
(symtab_node::clear_stmts_in_references): Clear speculative_id.
* tree-inline.c (copy_bb): Duplicate all the speculative edges
if indirect call contains multiple speculative targets.
* value-prof.h (check_ic_target): Remove.
* value-prof.c (gimple_value_profile_transformations):
Use void function gimple_ic_transform.
* value-prof.c (gimple_ic_transform): Handle topn case.
Fix comment typos. Change it to a void function.
gcc/testsuite/ChangeLog
2020-01-14 Xiong Hu Luo <luoxhu@linux.ibm.com>
PR ipa/69678
* gcc.dg/tree-prof/indir-call-prof-topn.c: New testcase.
* gcc.dg/tree-prof/crossmodule-indir-call-topn-1.c: New testcase.
* gcc.dg/tree-prof/crossmodule-indir-call-topn-1a.c: New testcase.
* gcc.dg/tree-prof/crossmodule-indir-call-topn-2.c: New testcase.
* lib/scandump.exp: Dump executable file name.
* lib/scanwpaipa.exp: New scan-pgo-wap-ipa-dump.
This adds octeontx2 naming. It currently uses the cortexa57
cost model and schedule model until I submit this. This is
more a place holder to get the naming of the cores in GCC 10.
I will submit the cost model in the next couple of days.
ChangeLog:
* config/aarch64/aarch64-cores.def (octeontx2): New define.
(octeontx2t98): New define.
(octeontx2t96): New define.
(octeontx2t93): New define.
(octeontx2f95): New define.
(octeontx2f95n): New define.
(octeontx2f95mm): New define.
* config/aarch64/aarch64-tune.md: Regenerate.
* doc/invoke.texi (-mcpu=): Document the new cpu types.
This patch replaces the update_web_docs_libstdcxx_svn script, that
updates online documentation from its sources in the GCC repository, run
once a day from cron, with update_web_docs_libstdcxx_git.
* update_web_docs_libstdcxx_git: New file.
* update_web_docs_libstdcxx_svn: Remove.
* crontab: Use update_web_docs_libstdcxx_git.
When we add a push rule, the default rule gets removed, so add that in
explicitly. This needs to come last since otherwise it would match
the custom redirecting rules we have for personal and vendor
sub-spaces.
I also noticed that the push rule for the vendor subspace still had
a force push default. We don't want that so remove it.
* gcc-git-customization.sh: Add back the default rule that
is lost by adding a custom push rule.
* git-fetch-vendor.sh: Likewise, also remove '+' from push specs.
The library has already worked around this issue, but I was curious about
why it wasn't working. The answer: because we were passing &var to fold,
which doesn't know about the constexpr values hash table. Fixed by passing
&"str" instead.
* constexpr.c (cxx_eval_builtin_function_call): Expose STRING_CST
to str/mem builtins.
This is a pretty rare situation since the C++11 change to make all
destructors default to noexcept, but it is still possible to define throwing
destructors, and if a destructor for a local variable throws during the
return, we've already constructed the return value, so now we need to
destroy it. I handled this somewhat like the new-expression cleanup; as in
that case, this cleanup can't properly nest with the cleanups for local
variables, so I introduce a cleanup region around the whole function and a
flag variable to indicate whether the return value actually needs to be
destroyed.
Setting the flag requires giving a COMPOUND_EXPR as the operand of a
RETURN_EXPR, so I adjust gimplify_return_expr to handle that.
This doesn't currently work with deduced return type because we don't know
the type when we're deciding whether to introduce the cleanup region.
gcc/
* gimplify.c (gimplify_return_expr): Handle COMPOUND_EXPR.
gcc/cp/
* cp-tree.h (current_retval_sentinel): New macro.
* decl.c (start_preparsed_function): Set up cleanup for retval.
* typeck.c (check_return_expr): Set current_retval_sentinel.
My earlier patch to reintroduce short_shift failed to adjust for the C++
front end use of const_op1 rather than op1.
* typeck.c (cp_build_binary_op): Use folded op1 for short_shift.
This patch replaces the update_web_docs_svn script, that updates
online documentation from its sources in the GCC repository, run once
a day from cron, with update_web_docs_git.
* update_web_docs_git: New file.
* update_web_docs_svn: Remove.
* crontab: Use update_web_docs_svn.
Bug 93241 reports a case where certain C expressions involving casts,
that would not be valid in an evaluated part of an integer constant
expression (because of e.g. involving integer overflow), are wrongly
rejected in an unevaluated part of an integer constant expression even
though all the operands and operations are ones that are valid in that
context. This is a rejects-valid regression in GCC 4.5 and later
relative to 4.4 (for some testcases; the one in the bug uses
_Static_assert which isn't supported in those older releases).
The rule in the C front end is that an expression with those
properties (valid in an unevaluated part of an integer constant
expression but not an evaluated part) must be represented either as an
INTEGER_CST with TREE_OVERFLOW set or as a C_MAYBE_CONST_EXPR with
C_MAYBE_CONST_EXPR_INT_OPERANDS set. This patch fixes build_c_cast to
check for that case and call note_integer_operands as needed.
Bootstrapped with no regressions for x86_64-pc-linux-gnu.
PR c/93241
gcc/c:
* c-typeck.c (build_c_cast): Check for expressions with integer
operands that can occur in an unevaluated part of an integer
constant expression and call note_integer_operands as needed.
gcc/testsuite:
* gcc.dg/c11-static-assert-10.c, gcc.dg/c99-const-expr-15.c: New
tests.
This simple script is intended to setup a new git configuration to
pull the branches and tags for a specific vendor. This should
simplify some of the steps needed for working with a vendor's
branches.
* git-fetch-vendor.sh: New file.
This patch is intended to help with folks setting up a git work
environment for use with GCC following the transition to git. It
currently does a couple of things.
1) Add an alias 'svn-rev' to git so that you can look up a legacy
commit by its svn revision number. This enables you to type
git svn-rev 1234
and git will show the commit log entry relating to SVN r1234.
2) Sets up tracking information for the user's personal area in
the git repo. It tries to figure out some sensible answers to the
data it needs, but allows the user to override the values. It then
creates the fetch and push entries that are needed for tracking the
extra refs. This implements one part of the recommendations that I've
proposed in svnwrite.html for dealing with private branches.
It should be possible to run the script more than once and for it to
DTRT. If you change your answers the configuration should be
correctly updated.
2020-01-13 Richard Earnshaw <rearnsha@arm.com>
* gcc-git-customization: New file.
PR libstdc++/93244
* include/bits/fs_path.h (path::generic_string<C,A>)
[_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Convert root-dir to forward-slash.
* testsuite/27_io/filesystem/path/generic/generic_string.cc: Check
root-dir is converted to forward slash in generic pathname.
* testsuite/27_io/filesystem/path/generic/utf.cc: New test.
* testsuite/27_io/filesystem/path/generic/wchar_t.cc: New test.
This implements the new requirements for C++20 that std::atomic should
initialize the atomic variable in its default constructor.
This patch does not add the deprecated attribute to atomic_init, but
that should be done at some point as it's deprecated in C++20.
The paper also deprecates the ATOMIC_FLAG_INIT macro, although we can't
apply the deprecated attribute to a macro.
PR libstdc++/58605
* include/bits/atomic_base.h (__cpp_lib_atomic_value_initialization):
Define.
(__atomic_flag_base, __atomic_base, __atomic_base<_PTp*>)
(__atomic_float): Add default member initializer for C++20.
* include/std/atomic (atomic): Likewise.
(atomic::atomic()): Remove noexcept-specifier on default constructor.
* include/std/version (__cpp_lib_atomic_value_initialization): Define.
* testsuite/29_atomics/atomic/cons/assign_neg.cc: Adjust dg-error line
number.
* testsuite/29_atomics/atomic/cons/copy_neg.cc: Likewise.
* testsuite/29_atomics/atomic/cons/value_init.cc: New test.
* testsuite/29_atomics/atomic_flag/cons/value_init.cc: New test.
* testsuite/29_atomics/atomic_flag/requirements/trivial.cc: Adjust
expected result for is_trivially_default_constructible.
* testsuite/29_atomics/atomic_float/requirements.cc: Likewise.
* testsuite/29_atomics/atomic_float/value_init.cc: New test.
* testsuite/29_atomics/atomic_integral/cons/assign_neg.cc: Likewise.
* testsuite/29_atomics/atomic_integral/cons/copy_neg.cc: Likewise.
* testsuite/29_atomics/atomic_integral/cons/value_init.cc
* testsuite/29_atomics/atomic_integral/requirements/trivial.cc: Adjust
expected results for is_trivially_default_constructible.
* testsuite/util/testsuite_common_types.h (has_trivial_dtor): Add
new test generator.
2020-01-13 Jakub Jelinek <jakub@redhat.com>
PR tree-optimization/90838
* tree-ssa-forwprop.c (simplify_count_trailing_zeroes): Use
SCALAR_INT_TYPE_MODE directly in CTZ_DEFINED_VALUE_AT_ZERO macro
argument rather than to initialize temporary for targets that
don't use the mode argument at all. Initialize ctzval to avoid
warning at -O0.
This patch updates the gcc_release script, used for snapshots and
releases, to use git instead of SVN.
The parts of the changes used in building snapshots have been tested.
The parts used for building releases have not, so it's likely further
fixes may turn out to be needed there when releases are first built
from git.
* gcc_release: Use git instead of SVN.
* crontab: Update gcc_release calls.
This patch replaces the update_version_svn script, that updates the
DATESTAMP files from cron, with update_version_git.
* update_version_git: New file.
* update_version:svn: Remove.
* crontab: Use update_version_git.
This fixes a typo and also explains why test_container is not a range
when used with output_iterator_wrapper or input_iterator_wrapper.
* testsuite/util/testsuite_iterators.h: Improve comment.
From-SVN: r280146
This patch adds support for associating a "diagnostic_path" with a
diagnostic: a sequence of events predicted by the compiler that leads to
the problem occurring, with their locations in the user's source,
text descriptions, and stack information (for handling interprocedural
paths).
For example, the following (hypothetical) error has a 3-event
intraprocedural path:
test.c: In function 'demo':
test.c:29:5: error: passing NULL as argument 1 to 'PyList_Append' which
requires a non-NULL parameter
29 | PyList_Append(list, item);
| ^~~~~~~~~~~~~~~~~~~~~~~~~
'demo': events 1-3
|
| 25 | list = PyList_New(0);
| | ^~~~~~~~~~~~~
| | |
| | (1) when 'PyList_New' fails, returning NULL
| 26 |
| 27 | for (i = 0; i < count; i++) {
| | ~~~
| | |
| | (2) when 'i < count'
| 28 | item = PyLong_FromLong(random());
| 29 | PyList_Append(list, item);
| | ~~~~~~~~~~~~~~~~~~~~~~~~~
| | |
| | (3) when calling 'PyList_Append', passing NULL from (1) as argument 1
|
The patch adds a new "%@" format code for printing event IDs, so that
in the above, the description of event (3) mentions event (1), showing
the user where the bogus NULL value comes from (the event IDs are
colorized to draw the user's attention to them).
There is a separation between data vs presentation: the above shows how
the diagnostic-printing code has consolidated the path into a single run
of events, since all the events are near each other and within the same
function; more complicated examples (such as interprocedural paths)
might be printed as multiple runs of events.
Examples of how interprocedural paths are printed can be seen in the
test suite (which uses a plugin to exercise the code without relying
on specific warnings using this functionality).
Other output formats include
- JSON,
- printing each event as a separate "note", and
- to not emit paths.
gcc/ChangeLog:
* Makefile.in (OBJS): Add tree-diagnostic-path.o.
* common.opt (fdiagnostics-path-format=): New option.
(diagnostic_path_format): New enum.
(fdiagnostics-show-path-depths): New option.
* coretypes.h (diagnostic_event_id_t): New forward decl.
* diagnostic-color.c (color_dict): Add "path".
* diagnostic-event-id.h: New file.
* diagnostic-format-json.cc (json_from_expanded_location): Make
non-static.
(json_end_diagnostic): Call context->make_json_for_path if it
exists and the diagnostic has a path.
(diagnostic_output_format_init): Clear context->print_path.
* diagnostic-path.h: New file.
* diagnostic-show-locus.c (colorizer::set_range): Special-case
when printing a run of events in a diagnostic_path so that they
all get the same color.
(layout::m_diagnostic_path_p): New field.
(layout::layout): Initialize it.
(layout::print_any_labels): Don't colorize the label text for an
event in a diagnostic_path.
(gcc_rich_location::add_location_if_nearby): Add
"restrict_to_current_line_spans" and "label" params. Pass the
former to layout.maybe_add_location_range; pass the latter
when calling add_range.
* diagnostic.c: Include "diagnostic-path.h".
(diagnostic_initialize): Initialize context->path_format and
context->show_path_depths.
(diagnostic_show_any_path): New function.
(diagnostic_path::interprocedural_p): New function.
(diagnostic_report_diagnostic): Call diagnostic_show_any_path.
(simple_diagnostic_path::num_events): New function.
(simple_diagnostic_path::get_event): New function.
(simple_diagnostic_path::add_event): New function.
(simple_diagnostic_event::simple_diagnostic_event): New ctor.
(simple_diagnostic_event::~simple_diagnostic_event): New dtor.
(debug): New overload taking a diagnostic_path *.
* diagnostic.def (DK_DIAGNOSTIC_PATH): New.
* diagnostic.h (enum diagnostic_path_format): New enum.
(json::value): New forward decl.
(diagnostic_context::path_format): New field.
(diagnostic_context::show_path_depths): New field.
(diagnostic_context::print_path): New callback field.
(diagnostic_context::make_json_for_path): New callback field.
(diagnostic_show_any_path): New decl.
(json_from_expanded_location): New decl.
* doc/invoke.texi (-fdiagnostics-path-format=): New option.
(-fdiagnostics-show-path-depths): New option.
(-fdiagnostics-color): Add "path" to description of default
GCC_COLORS; describe it.
(-fdiagnostics-format=json): Document how diagnostic paths are
represented in the JSON output format.
* gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
Add optional params "restrict_to_current_line_spans" and "label".
* opts.c (common_handle_option): Handle
OPT_fdiagnostics_path_format_ and
OPT_fdiagnostics_show_path_depths.
* pretty-print.c: Include "diagnostic-event-id.h".
(pp_format): Implement "%@" format code for printing
diagnostic_event_id_t *.
(selftest::test_pp_format): Add tests for "%@".
* selftest-run-tests.c (selftest::run_tests): Call
selftest::tree_diagnostic_path_cc_tests.
* selftest.h (selftest::tree_diagnostic_path_cc_tests): New decl.
* toplev.c (general_init): Initialize global_dc->path_format and
global_dc->show_path_depths.
* tree-diagnostic-path.cc: New file.
* tree-diagnostic.c (maybe_unwind_expanded_macro_loc): Make
non-static. Drop "diagnostic" param in favor of storing the
original value of "where" and re-using it.
(virt_loc_aware_diagnostic_finalizer): Update for dropped param of
maybe_unwind_expanded_macro_loc.
(tree_diagnostics_defaults): Initialize context->print_path and
context->make_json_for_path.
* tree-diagnostic.h (default_tree_diagnostic_path_printer): New
decl.
(default_tree_make_json_for_path): New decl.
(maybe_unwind_expanded_macro_loc): New decl.
gcc/c-family/ChangeLog:
* c-format.c (local_event_ptr_node): New.
(PP_FORMAT_CHAR_TABLE): Add entry for "%@".
(init_dynamic_diag_info): Initialize local_event_ptr_node.
* c-format.h (T_EVENT_PTR): New define.
gcc/testsuite/ChangeLog:
* gcc.dg/format/gcc_diag-10.c (diagnostic_event_id_t): New
typedef.
(test_diag): Add coverage of "%@".
* gcc.dg/plugin/diagnostic-path-format-default.c: New test.
* gcc.dg/plugin/diagnostic-path-format-inline-events-1.c: New test.
* gcc.dg/plugin/diagnostic-path-format-inline-events-2.c: New test.
* gcc.dg/plugin/diagnostic-path-format-inline-events-3.c: New test.
* gcc.dg/plugin/diagnostic-path-format-none.c: New test.
* gcc.dg/plugin/diagnostic-test-paths-1.c: New test.
* gcc.dg/plugin/diagnostic-test-paths-2.c: New test.
* gcc.dg/plugin/diagnostic-test-paths-3.c: New test.
* gcc.dg/plugin/diagnostic-test-paths-4.c: New test.
* gcc.dg/plugin/diagnostic_plugin_test_paths.c: New.
* gcc.dg/plugin/plugin.exp: Add the new plugin and test cases.
libcpp/ChangeLog:
* include/line-map.h (class diagnostic_path): New forward decl.
(rich_location::get_path): New accessor.
(rich_location::set_path): New function.
(rich_location::m_path): New field.
* line-map.c (rich_location::rich_location): Initialize m_path.
From-SVN: r280142
PR tree-optimization/93210
* fold-const.h (native_encode_initializer,
can_native_interpret_type_p): Declare.
* fold-const.c (native_encode_string): Fix up handling with off != -1,
simplify.
(native_encode_initializer): New function, moved from dwarf2out.c.
Adjust to native_encode_expr compatible arguments, including dry-run
and partial extraction modes. Don't handle STRING_CST.
(can_native_interpret_type_p): No longer static.
* gimple-fold.c (fold_ctor_reference): For native_encode_expr, verify
offset / BITS_PER_UNIT fits into int and don't call it if
can_native_interpret_type_p fails. If suboff is NULL and for
CONSTRUCTOR fold_{,non}array_ctor_reference returns NULL, retry with
native_encode_initializer.
(fold_const_aggregate_ref_1): Formatting fix.
* dwarf2out.c (native_encode_initializer): Moved to fold-const.c.
(tree_add_const_value_attribute): Adjust caller.
* gcc.dg/pr93210.c: New test.
* g++.dg/opt/pr93210.C: New test.
From-SVN: r280141
PR tree-optimization/90838
* tree-ssa-forwprop.c (simplify_count_trailing_zeroes): Use
SCALAR_INT_TYPE_MODE instead of TYPE_MODE as operand of
CTZ_DEFINED_VALUE_AT_ZERO.
From-SVN: r280140
2020-01-10 Vladimir Makarov <vmakarov@redhat.com>
PR inline-asm/93027
* gcc.target/i386/pr93027.c: Use the right PR number in the test.
From-SVN: r280138
PR libgomp/93219
* libgomp.h (gomp_print_string): Change return type from void to int.
* affinity-fmt.c (gomp_print_string): Likewise. Return true if
not all characters have been written.
From-SVN: r280137
This patch adds support for obscuring the line numbers printed in the
left-hand margin when printing the source code, converting them to "NN",
e.g from:
7111 | if (!(flags & 0x0001)) {
| ^
| |
| (1) following 'true' branch...
7112 |
to:
NN | if (!(flags & 0x0001)) {
| ^
| |
| (1) following 'true' branch...
NN |
This is useful in followup patches e.g. when testing how interprocedural
paths are printed using multiline.exp, to avoid depending on precise line
numbers.
gcc/testsuite/ChangeLog:
* lib/gcc-dg.exp (cleanup-after-saved-dg-test): Reset global
nn_line_numbers_enabled.
* lib/multiline.exp (nn_line_numbers_enabled): New global.
(dg-enable-nn-line-numbers): New proc.
(maybe-handle-nn-line-numbers): New proc.
* lib/prune.exp (prune_gcc_output): Call maybe-handle-nn-line-numbers.
From-SVN: r280134
2020-01-10 Vladimir Makarov <vmakarov@redhat.com>
PR inline-asm/93207
* lra-constraints.c (match_reload): Permit input operands have the
same mode as output while other input operands have a different
mode.
2020-01-10 Vladimir Makarov <vmakarov@redhat.com>
PR inline-asm/93207
* gcc.target/i386/pr93207.c: New test.
From-SVN: r280133
Support common idioms for count trailing zeroes using an array lookup.
The canonical form is array[((x & -x) * C) >> SHIFT] where C is a magic
constant which when multiplied by a power of 2 creates a unique value
in the top 5 or 6 bits. This is then indexed into a table which maps it
to the number of trailing zeroes. When the table is valid, we emit a
sequence using the target defined value for ctz (0):
int ctz1 (unsigned x)
{
static const char table[32] =
{
0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9
};
return table[((unsigned)((x & -x) * 0x077CB531U)) >> 27];
}
Is optimized to:
rbit w0, w0
clz w0, w0
and w0, w0, 31
ret
gcc/
PR tree-optimization/90838
* tree-ssa-forwprop.c (check_ctz_array): Add new function.
(check_ctz_string): Likewise.
(optimize_count_trailing_zeroes): Likewise.
(simplify_count_trailing_zeroes): Likewise.
(pass_forwprop::execute): Try ctz simplification.
* match.pd: Add matching for ctz idioms.
testsuite/
PR tree-optimization/90838
* testsuite/gcc.target/aarch64/pr90838.c: New test.
From-SVN: r280132
gcc/ChangeLog:
2020-01-10 Stam Markianos-Wright <stam.markianos-wright@arm.com>
* config/aarch64/aarch64.c (aarch64_invalid_conversion): New function
for target hook.
(aarch64_invalid_unary_op): New function for target hook.
(aarch64_invalid_binary_op): New function for target hook.
gcc/testsuite/ChangeLog:
2020-01-10 Stam Markianos-Wright <stam.markianos-wright@arm.com>
* g++.target/aarch64/bfloat_cpp_typecheck.C: New test.
* gcc.target/aarch64/bfloat16_scalar_typecheck.c: New test.
* gcc.target/aarch64/bfloat16_vector_typecheck_1.c: New test.
* gcc.target/aarch64/bfloat16_vector_typecheck_2.c: New test.
From-SVN: r280130
Back in SVN r131862 richi removed this code to fix PR 34235, but didn't
remove the parallel code from the C front-end because the bug had previously
been fixed in r44080. This patch copies the code from C again.
* typeck.c (cp_build_binary_op): Restore short_shift code.
From-SVN: r280128
