summaryrefslogtreecommitdiff
path: root/vm_exec.c
AgeCommit message (Collapse)Author
2025-02-14Only count VM instructions in YJIT stats buildsAaron Patterson
The instruction counter is slowing multi-Ractor applications. I had changed it to use a thread local, but using a thread local is slowing single threaded applications. This commit only enables the instruction counter in YJIT stats builds until we can figure out a way to gather the information with lower overhead. Co-authored-by: Randy Stauner <[email protected]> Notes: Merged: https://github.com/ruby/ruby/pull/12670
2025-02-13[Feature #21116] Extract RJIT as a third-party gemNobuyoshi Nakada
Notes: Merged: https://github.com/ruby/ruby/pull/12740
2025-01-10Make rb_vm_insns_count a thread local variableAaron Patterson
`rb_vm_insns_count` is a global variable used for reporting YJIT statistics. It is a counter that tallies the number of interpreter instructions that have been executed, this way we can approximate how much time we're spending in YJIT compared to the interpreter. Unfortunately keeping this statistic means that every instruction executed in the interpreter loop must increment the counter. Normally this isn't a problem, but in multi-threaded situations (when Ractors are used), incrementing this counter can become quite costly due to page caching issues. Additionally, since there is no locking when incrementing this global, the count can't really make sense in a multi-threaded environment. This commit changes `rb_vm_insns_count` to a thread local. That way each Ractor has it's own copy of the counter and incrementing the counter becomes quite cheap. Of course this means that in multi-threaded situations, the value doesn't really make sense (but it didn't make sense before because of the lack of locking). The counter is used for YJIT statistics, and since YJIT is basically disabled when Ractors are in use, I don't think we care about inaccuracies (for the time being). We can revisit this counter when we give YJIT multi-threading support, but for the time being this commit restores multi-threaded performance. To test this, I used the benchmark in [Bug #20489]. Here is the performance on Ruby 3.2: ``` $ time RUBY_MAX_CPU=12 ./miniruby -v ../test.rb 8 8 ruby 3.2.0 (2022-12-25 revision a528908271) [x86_64-linux] [0...1, 1...2, 2...3, 3...4, 4...5, 5...6, 6...7, 7...8] ../test.rb:43: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. ________________________________________________________ Executed in 2.53 secs fish external usr time 19.86 secs 370.00 micros 19.86 secs sys time 0.02 secs 320.00 micros 0.02 secs ``` We can see the regression in performance on the master branch: ``` $ time RUBY_MAX_CPU=12 ./miniruby -v ../test.rb 8 8 ruby 3.5.0dev (2025-01-10T16:22:26Z master 4a2702dafb) +PRISM [x86_64-linux] [0...1, 1...2, 2...3, 3...4, 4...5, 5...6, 6...7, 7...8] ../test.rb:43: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. ________________________________________________________ Executed in 24.87 secs fish external usr time 195.55 secs 0.00 micros 195.55 secs sys time 0.00 secs 716.00 micros 0.00 secs ``` Here are the stats after this commit: ``` $ time RUBY_MAX_CPU=12 ./miniruby -v ../test.rb 8 8 ruby 3.5.0dev (2025-01-10T20:37:06Z tl 3ef0432779) +PRISM [x86_64-linux] [0...1, 1...2, 2...3, 3...4, 4...5, 5...6, 6...7, 7...8] ../test.rb:43: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. ________________________________________________________ Executed in 2.46 secs fish external usr time 19.34 secs 381.00 micros 19.34 secs sys time 0.01 secs 321.00 micros 0.01 secs ``` [Bug #20489] Notes: Merged: https://github.com/ruby/ruby/pull/12549
2023-12-19RJIT: Correct the macro that defines rb_vm_insns_countTakashi Kokubun
2023-07-27Remove an unused argument in vm_exec_coreTakashi Kokubun
2023-07-27Clean up OPT_STACK_CACHING (#8132)Takashi Kokubun
Notes: Merged-By: k0kubun <[email protected]>
2023-07-13YJIT: Make ratio_in_yjit always available (#8064)Takashi Kokubun
Notes: Merged-By: maximecb <[email protected]>
2023-07-12Remove an unused `#if 0` codeTakashi Kokubun
2023-07-11Remove obsoleted USE_INSNS_COUNTER macro for nowTakashi Kokubun
I don't use that these days. I'd like to remove this to ease a change I'm writing. I'll resurrect this when I need this again.
2022-11-16Using UNDEF_P macroS-H-GAMELINKS
Notes: Merged: https://github.com/ruby/ruby/pull/6721
2022-10-19Fix and improve coroutines for Darwin (macOS) ppc/ppc64. (#5975)Sergey Fedorov
Notes: Merged-By: ioquatix <[email protected]>
2022-07-21Expand tabs [ci skip]Takashi Kokubun
[Misc #18891] Notes: Merged: https://github.com/ruby/ruby/pull/6094
2021-10-20Cleanup diff against upstream. Add commentsAlan Wu
I did a `git diff --stat` against upstream and looked at all the files that are outside of YJIT to come up with these minor changes.
2021-10-20Try to convince get better code out of ClangAlan Wu
2021-08-09Suppress unused-function warning when OPT_THREADED_CODE != 1Nobuyoshi Nakada
Notes: Merged: https://github.com/ruby/ruby/pull/4720
2021-05-04Fix trivial -Wundef warningsBenoit Daloze
* See [Feature #17752] Co-authored-by: xtkoba (Tee KOBAYASHI) <[email protected]> Notes: Merged: https://github.com/ruby/ruby/pull/4428
2020-08-14Enable arm64 optimizations that exist for power/x86 (#3393)AGSaidi
* Enable unaligned accesses on arm64 64-bit Arm platforms support unaligned accesses. Running the string benchmarks this change improves performance by an average of 1.04x, min .96x, max 1.21x, median 1.01x * arm64 enable gc optimizations Similar to x86 and powerpc optimizations. | |compare-ruby|built-ruby| |:------|-----------:|---------:| |hash1 | 0.225| 0.237| | | -| 1.05x| |hash2 | 0.110| 0.110| | | 1.00x| -| * vm_exec.c: improve performance for arm64 | |compare-ruby|built-ruby| |:------------------------------|-----------:|---------:| |vm_array | 26.501M| 27.959M| | | -| 1.06x| |vm_attr_ivar | 21.606M| 31.429M| | | -| 1.45x| |vm_attr_ivar_set | 21.178M| 26.113M| | | -| 1.23x| |vm_backtrace | 6.621| 6.668| | | -| 1.01x| |vm_bigarray | 26.205M| 29.958M| | | -| 1.14x| |vm_bighash | 504.155k| 479.306k| | | 1.05x| -| |vm_block | 16.692M| 21.315M| | | -| 1.28x| |block_handler_type_iseq | 5.083| 7.004| | | -| 1.38x| Notes: Merged-By: nurse <[email protected]>
2020-07-10RUBY_CONST_ASSERT: use STATIC_ASSERT instead卜部昌平
Static assertions shall be done using STATIC_ASSERT these days. Notes: Merged: https://github.com/ruby/ruby/pull/3296
2019-12-28Add VM insns counter like debug_counter (#2789)Takashi Kokubun
Notes: Merged-By: k0kubun <[email protected]>
2018-03-04vm_exec.c: reduced repeated same callsnobu
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@62648 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-10-27catch up recent changes for call threaded code VM.ko1
Fix compile errors for OPT_CALL_THREADED_CODE (in vm_opts.h). git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60493 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-10-27vm_exec_core() accepts `ec` instead of `th`.ko1
* vm_exec.c (vm_exec_core): accepts `ec` instead of `th`. * vm_args.c (vm_caller_setup_arg_block): also accepts `ec`. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60477 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-10-26Use rb_execution_context_t instead of rb_thread_tko1
to represent execution context [Feature #14038] * vm_core.h (rb_thread_t): rb_thread_t::ec is now a pointer. There are many code using `th` to represent execution context (such as cfp, VM stack and so on). To access `ec`, they need to use `th->ec->...` (adding one indirection) so that we need to replace them by passing `ec` instead of `th`. * vm_core.h (GET_EC()): introduced to access current ec. Also remove `ruby_current_thread` global variable. * cont.c (rb_context_t): introduce rb_context_t::thread_ptr instead of rb_context_t::thread_value. * cont.c (ec_set_vm_stack): added to update vm_stack explicitly. * cont.c (ec_switch): added to switch ec explicitly. * cont.c (rb_fiber_close): added to terminate fibers explicitly. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60440 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-10-23Drop to support NaCl platform.hsbt
Because NaCl and PNaCl are already sunset status. see https://bugs.chromium.org/p/chromium/issues/detail?id=239656#c160 configure.ac: Patch for this file was provided by @nobu. [Feature #14041][ruby-core:83497][fix GH-1726] git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@60374 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-06-28move fields to ec.ko1
* vm_core.h (rb_thread.h): move errinfo and trace_arg to rb_execution_context_t. * cont.c (fiber_switch, rb_cont_call): do not restore "trace_arg" here. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@59199 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2017-05-09rb_execution_context_t: move stack, stack_size and cfp from rb_thread_tnormal
The goal is to reduce rb_context_t and rb_fiber_t size by removing the need to store the entire rb_thread_t in there. [ruby-core:81045] Work-in-progress: soon, we will move more fields here. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@58614 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2016-11-05* vm_insnhelper.h: rename REG_(CFP|PC|SP|EP) to VM_REG_....ko1
[Bug #12527] * vm_exec.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@56609 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-07-23* vm_core.h: constify rb_iseq_constant_body::iseq_encoded andko1
rb_control_frame_t::pc. * compile.c (rb_iseq_translate_threaded_code): catch up this fix. * iseq.c: ditto. * vm_exec.c (vm_exec_core): ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@51356 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-08-12* vm_exec.c: improve performance in ppc64 arch.hsbt
[ruby-core:63437] [Feature #9997] git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@47149 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-01-14* vm_exec.c (cfp): Fixes a SEGV issue in r44554.yugui
r11 can be broken by subroutine and sometimes causes SEGV at runtime. Use r13 instead. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@44590 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-01-10* vm_exec.c (cfp): Avoid generating invalid binary foryugui
NativeClient. r15 on x86_64 is reserved by NativeClient. So r15 to cfp used to generate invalid binary under some combinations of compiler optimization flags. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@44554 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-12-18* vm_exec.h (VM_DEBUG_STACKOVERFLOW): added.ko1
disable stack overflow check for every stack pushing as default. * vm_exec.c (vm_stack_overflow_for_insn): ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@44275 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-11-18vm_core.h: extract VM_STACK_OVERFLOWED_Pnobu
* vm_core.h (VM_STACK_OVERFLOWED_P, WHEN_VM_STACK_OVERFLOWED): extract condition from CHECK_VM_STACK_OVERFLOW. * vm_exec.c (vm_stack_overflow_for_insn): move rb_bug call. * vm_exec.h (CHECK_VM_STACK_OVERFLOW_FOR_INSN): share the condition with CHECK_VM_STACK_OVERFLOW. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@43702 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2013-03-18* vm_exec.c: Correct predefined macro name. This typo is introduced bynaruse
r36534 and should be backported to ruby_2_0_0. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@39806 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-12-03* vm_exec.c: check VM_COLLECT_USAGE_DETAILS.ko1
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@38164 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-12-03* vm_exec.c: vm_analysis_insn should be statkc.ko1
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@38160 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-12* probes.d: add DTrace probe declarations. [ruby-core:27448]tenderlove
* array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <[email protected]> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <[email protected]> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-08-07* vm_exec.c, insns.def (leave): solve problems onko1
OPT_CALL_THREADED_CODE. Catch up finish frame structure on OPT_CALL_THREADED_CODE. * vm_core.h: add rb_thread_t#retval for temporary space on OPT_CALL_THREADED_CODE. * vm.c (th_init): clear rb_thread_t#retval as Qundef. * vm_dump.c (rb_vmdebug_debug_print_pre): fix debug print format. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@36652 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-07-25Suppress warnings.naruse
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@36534 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-06-15* vm_core.h: remove VM_FRAME_MAGIC_FINISH (finish frame type).ko1
Before this commit: `finish frame' was place holder which indicates that VM loop needs to return function. If a C method calls a Ruby methods (a method written by Ruby), then VM loop will be (re-)invoked. When the Ruby method returns, then also VM loop should be escaped. `finish frame' has only one instruction `finish', which returns VM loop function. VM loop function executes `finish' instruction, then VM loop function returns itself. With such mechanism, `leave' instruction (which returns one frame from current scope) doesn't need to check that this `leave' should also return from VM loop function. Strictly, one branch can be removed from `leave' instructon. Consideration: However, pushing the `finish frame' needs costs because it needs several memory accesses. The number of pushing `finish frame' is greater than I had assumed. Of course, pushing `finish frame' consumes additional control frame. Moreover, recent processors has good branch prediction, with which we can ignore such trivial checking. After this commit: Finally, I decide to remove `finish frame' and `finish' instruction. Some parts of VM depend on `finish frame', so the new frame flag VM_FRAME_FLAG_FINISH is introduced. If this frame should escape from VM function loop, then the result of VM_FRAME_TYPE_FINISH_P(cfp) is true. `leave' instruction checks this flag every time. I measured performance on it. However on my environments, it improves some benchmarks and slows some benchmarks down. Maybe it is because of C compiler optimization parameters. I'll re-visit here if this cause problems. * insns.def (leave, finish): remove finish instruction. * vm.c, vm_eval.c, vm_exec.c, vm_backtrace.c, vm_dump.c: apply above changes. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@36099 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2011-05-31* vm_exec.c: remove conditions for clangnaruse
because clang version 3.0 (trunk 132165) doesn't need them. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@31846 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2011-04-23* include/ruby/intern.h: pcc can't use __builtin_constant_p.naruse
* vm_exec.c: change condition. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@31327 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2010-10-07* vm_exec.c (vm_exec_core): Treat clang as non gcc on thisnaruse
context: It has __asm__ but doesn't works well. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@29419 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2009-10-27* vm_exec.c (DECL_SC_REG): use __clang__.naruse
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@25504 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2009-10-26* vm_exec.c (DECL_SC_REG): check defined(__asm__) before use it.naruse
LLVM/clang defines __GNUC__ but doesn't have __asm__.. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@25493 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2009-01-19* vm.c: add a prefix "rb_" to exposed functionsko1
vm_get_ruby_level_next_cfp(), rb_vm_make_env_object(), vm_stack_to_heap(), vm_make_proc(), vm_invoke_proc(), vm_get_sourceline(), vm_cref(), vm_localjump_error(), vm_make_jump_tag_but_local_jump(), vm_jump_tag_but_local_jump(). This changes may affect only core because most of renamed functions require a pointer of not-exposed struct such as rb_thread_t or NODE. In short, they are core functions. * cont.c, eval.c, eval_intern.h, load.c, proc.c, thread.c, vm_core.h, vm_dump.c, vm_eval.c, vm_exec.c, vm_insnhelper.c: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@21659 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-11-14* README.EXT, README.EXT.ja, vm_exec.c, vm_insnhelper.c: updatemame
filenames. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@20233 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-28* vm_exec.c (vm_exec_core): add an UNLIKELY() hint.ko1
* vm_insnhelper.h (BASIC_OP_UNREDEFINED_P): add a LIKELY() hint. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19605 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2008-09-23* common.mk: clean upko1
- remove blockinlining.$(OBJEXT) to built - make ENCODING_H_INCLDUES variable (include/ruby/encoding.h) - make VM_CORE_H_INCLUDES variable (vm_core.h) - simplify rules. - make depends rule to output depend status using gcc -MM. * include/ruby/mvm.h, include/ruby/vm.h: rename mvm.h to vm.h. * include/ruby.h: ditto. * load.c: add inclusion explicitly. * enumerator.c, object.c, parse.y, thread.c, vm_dump.c: remove useless inclusion. * eval_intern.h: cleanup inclusion. * vm_core.h: rb_thread_t should be defined in this file. * vm_evalbody.c, vm_exec.c: rename vm_evalbody.c to vm_exec.c. * vm.h, vm_exec.h: rename vm.h to vm_exec.h. * insnhelper.h, vm_insnhelper.h: rename insnhelper.h to vm_insnhelper.h. * vm.c, vm_insnhelper.c, vm_insnhelper.h: - rename vm_eval() to vm_exec_core(). - rename vm_eval_body() to vm_exec(). - cleanup include order. * vm_method.c: fix comment. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@19466 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
generated by cgit v1.2.3 (git 2.25.1) at 2025-06-19 19:33:56 +0000