and define CONSTFUNC and PUREFUNC if available.
Note that I don't add those options as default because
it still shows many false-positive (it seems not to consider
longjmp).
* vm_eval.c (stack_check): get rb_thread_t* as an argument
to avoid duplicate call of GET_THREAD().
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@54952 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This reverts commit r54747.
!__clang__ is also essential.
Anyway clang inlines vm_getivar into both vm_call_ivar and
vm_getinstancevariable, which r54728 originally intended to.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@54751 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (INLINE): works with __NO_INLINE__ only,
__clang__ is not the point.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@54747 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
define as `static inline` when it is not optimized to keep
the symbol generated.
* vm_insnhelper.c (vm_getivar): use `INLINE` to force inline
so that a compiler inlines it into vm_getinstancevariable
and optimizes out is_attr and related branches.
* vm_insnhelper.c (vm_getivar): use `inline` to recommend inline.
Without this vm1_ivar_set is degraded.
benchmark results:
minimum results in each 5 measurements.
Execution time (sec)
name ruby 2.4.0dev (2016-04-23 trunk 54727) [x86_64-linux] ruby 2.4.0dev (2016-04-23 trunk 54733) [x86_64-linux] built-ruby
loop_whileloop 0.641 0.642 0.646
vm1_ivar* 1.002 0.999 0.831
vm1_ivar_set* 0.369 1.106 0.362
Speedup ratio: compare with the result of `ruby 2.4.0dev (2016-04-23
trunk 54727) [x86_64-linux]' (greater is better)
name ruby 2.4.0dev (2016-04-23 trunk 54733) [x86_64-linux]
built-ruby
loop_whileloop
0.998 0.991
vm1_ivar*
1.003 1.205
vm1_ivar_set*
0.334 1.018
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@54738 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (vm_getivar): add missing static to inline,
otherwise external symbol is referred and link fails when
optimization is disabled.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@54729 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
vm_getivar is called by vm_call_ivar and vm_getinstancevariable.
At least with GCC 4.8 and 5.3 on Linux, they are inlining it into
vm_call_ivar but not vm_getinstancevariable.
By `inline`, they correctly inline it and gains performance.
Speedup ratio: compare with the result of `ruby 2.4.0dev (2016-04-23
trunk 54727) [x86_64-linux]' (greater is better)
name built-ruby
loop_whileloop 1.001
vm1_ivar* 1.189
vm1_ivar_set* 1.024
Note tha `inline`'s meaning is different between old GCC
and C99. Old GCC's inline means C99's extern inline.
https://gcc.gnu.org/onlinedocs/gcc/Inline.html
Since Ruby specify -std=iso9899:1999, it works like C99.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@54728 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Take advantage of "%+" modifier in the format string instead of
explicitly calling rb_inspect to reduce object size.
On x86 32-bit:
text data bss dec hex filename
before: 2949572 12448 30680 2992700 2daa3c miniruby
after: 2949464 12448 30680 2992592 2da9d0 miniruby
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@53644 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_eval.c (make_no_method_exception): make format string
fstring.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@53525 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
functions) to the latter location.
This moving recovers performance a little.
[Bug #11829]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@53236 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
with keyword arguments for VM_METHOD_TYPE_ATTRSET type methods.
Normally, we can not use keyword arguments for this kind of methods,
(obj.foo = 1), but we can set alias names for them.
[Bug #11657]
* test/ruby/test_keyword.rb: add a test for this fix.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@53164 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (vm_call_method_missing): remove spaces just
before a tab to adjust indent.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@53062 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
and a pre-compilation/runtime loader sample.
[Feature #11788]
* iseq.c: add new methods:
* RubyVM::InstructionSequence#to_binary_format(extra_data = nil)
* RubyVM::InstructionSequence.from_binary_format(binary)
* RubyVM::InstructionSequence.from_binary_format_extra_data(binary)
* compile.c: implement body of this new feature.
* load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq):
call RubyVM::InstructionSequence.load_iseq(fname) with
loading script name if this method is defined.
We can return any ISeq object as a result value.
Otherwise loading will be continue as usual.
This interface is not matured and is not extensible.
So that we don't guarantee the future compatibility of this method.
Basically, you should'nt use this method.
* iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions)
from iseq.c.
* encoding.c (rb_data_is_encoding), internal.h: added.
* vm_core.h: add several supports for lazy load.
* add USE_LAZY_LOAD macro to specify enable or disable of
this feature.
* add several fields to rb_iseq_t.
* introduce new macro rb_iseq_check().
* insns.def: some check for lazy loading feature.
* vm_insnhelper.c: ditto.
* proc.c: ditto.
* vm.c: ditto.
* test/lib/iseq_loader_checker.rb: enabled iff suitable
environment variables are provided.
* test/runner.rb: enable lib/iseq_loader_checker.rb.
* sample/iseq_loader.rb: add sample compiler and loader.
$ ruby sample/iseq_loader.rb [dir]
will compile all ruby scripts in [dir].
With default setting, this compile creates *.rb.yarb files
in same directory of target .rb scripts.
$ ruby -r sample/iseq_loader.rb [app]
will run with enable to load compiled binary data.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
CREFs should not be shared by methods between `using'.
[Bug #11247]
* vm_insnhelper.c (vm_cref_replace_with_duplicated_cref): ditto.
* vm.c (vm_cref_dup): should copy refinements correctly.
* eval.c: use rb_vm_cref_replace_with_duplicated_cref().
* eval_intern.h: add a decl. of
rb_vm_cref_replace_with_duplicated_cref().
* vm_eval.c (eval_string_with_cref): do not need to pass
scope's CREF because VM can find out CREF from stack frames.
* test/ruby/test_refinement.rb: add a test.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52677 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* rb_method_definition_t::complemented_count to count shared method
entries because of complemented method entries and separate from
alias_count.
Shared `def' only by complemented method entries should not prevent
method re-definition warning.
* METHOD_ENTRY_COMPLEMENTED(me) to represent complemented method entry.
* METHOD_ENTRY_COMPLEMENTED_SET(me) to check it as complemented me.
* vm_insnhelper.c (aliased_callable_method_entry): should also
check me->def->complemented_count.
* vm_method.c (method_definition_addref_complement): add to count
complemented method entries number.
* vm_method.c (rb_method_definition_release): release `def' iff
alias_count == 0 and complemented_count == 0.
* test/ruby/test_module.rb: add a test.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52636 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (vm_call_method_missing): get rid of extra
garbage after argv.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52404 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
current cref only when cached CREF list includes singleton class.
Singleton classes have own namespaces, so that we need to check
cref as a key (#10943).
However, if current CREF list does not include singleton class,
no need to check CREF beacuse it should be same name space.
* vm_insnhelper.c (vm_get_const_key_cref): add a function returns
CREF only when it includes singleton class.
* vm_core.h: constify iseq_inline_cache_entry::ic_cref.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52371 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (VM_PROFILE_UP): use enum.
* vm_insnhelper.c (vm_profile_show_result): fix typo, "r->c" at
the last should be "c->c".
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52343 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
of arguments", distinguishing given and expected argument
numbers clearly. [Feature #9025]
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52264 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
vm_call_iseq_setup_normal_0start() is simple, however it has
some loops/conditions depends on ISeq::param.size and
ISeq::local_size (in vm_push_frame(), inlined into this function).
There are many simple methods which has a few parameters and local
variables. So that this patch introduces several special functions
generated in vm_call_iseq_optimized.inc by
tool/mk_call_iseq_optimized.rb.
This script makes
vm_call_iseq_setup_normal_0start_Xparams_Ylocals()
where X is 0 to 3 and Y is 1 to 6 (as current setting).
In this case, X * Y = 24 functions are created.
These functions creates fast method dispatch by inlining
vm_push_frame() with immediate params/locals sizes.
On my laptop, we can have the following results.
vm2_method* 1.083 (8.3% faster)
vm2_poly_method* 0.961 (3.4% slower)
It shows 8.3% faster for inner loop method dispatch (hit inline
cache), but 3.4% slower when inline cache miss because we need
to find a suitable call handler.
* common.mk: add a rule for vm_call_iseq_optimized.inc.
* tool/mk_call_iseq_optimized.rb: added.
* vm.c: include vm_call_iseq_optimized.inc.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52254 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c (vm_yield_with_cfunc): cast to suppress a
warning by VC.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52144 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
functions.
* vm_insnhelper.c (vm_yield_callee_setup_arg): remove this function
beacuse it is only delegation function.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52104 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Pointed out by Yukihiro Matsumoto <matz@ruby-lang.org>.
* gc.c (newobj_of_...): `of' is unnecessary.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52102 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
because $SAFE = 3 and 4 is not available.
Now, $SAFE is not checked on method dispatch at all.
* vm_eval.c, vm_insnhelper.c, vm_method.c: ditto.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52058 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Change all goto statement across blocks to tail call functions.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52049 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_args.c (vm_caller_setup_arg_block): store new ifunc for
symbol in control frame proc to be marked.
* proc.c (proc_new), vm_insnhelper.c (vm_yield_with_cfunc):
block->proc may be an ifunc now.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51995 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* rb_call_info (ci) has compiled fixed information.
* if ci->flag & VM_CALL_KWARG, then rb_call_info is
also rb_call_info_with_kwarg. This technique reduce one word
for major rb_call_info data.
* rb_calling_info has temporary data (argc, blockptr, recv).
for each method dispatch. This data is allocated only on
machine stack.
* rb_call_cache is for inline method cache.
Before this patch, only rb_call_info_t data is passed.
After this patch, above three structs are passed.
This patch improves:
* data locarity (rb_call_info is now read-only data).
* reduce memory consumption (rb_call_info_with_kwarg,
rb_calling_info).
* compile.c: use above data.
* insns.def: ditto.
* iseq.c: ditto.
* vm_args.c: ditto.
* vm_eval.c: ditto.
* vm_insnhelper.c: ditto.
* vm_insnhelper.h: ditto.
* iseq.h: add iseq_compile_data::ci_index and
iseq_compile_data::ci_kw_indx.
* tool/instruction.rb: introduce TS_CALLCACHE operand type.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51903 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
* vm_insnhelper.c: introduce shortcut functions for opt_pc == 0
because opt_pc is always 0 on shortcut function.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51827 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
tarui
naruse
nobu
ktsj
ko1
normal
hsbt
shugo
duerst