684e84df7c
* Use rb_ivar_get() for general case of getivar Pretty straight forward. Buys about 1% coverage on railsbench. * Update yjit_codegen.c Co-authored-by: Maxime Chevalier-Boisvert <maximechevalierb@gmail.com> |
||
---|---|---|
.github | ||
basictest | ||
benchmark | ||
bin | ||
bootstraptest | ||
ccan | ||
coroutine | ||
coverage | ||
cygwin | ||
defs | ||
doc | ||
enc | ||
ext | ||
gems | ||
include | ||
internal | ||
lib | ||
libexec | ||
man | ||
misc | ||
missing | ||
sample | ||
spec | ||
template | ||
test | ||
tool | ||
win32 | ||
.appveyor.yml | ||
.cirrus.yml | ||
.dir-locals.el | ||
.document | ||
.editorconfig | ||
.gdbinit | ||
.gitattributes | ||
.gitignore | ||
.rspec_parallel | ||
.travis.yml | ||
BSDL | ||
CONTRIBUTING.md | ||
COPYING | ||
COPYING.ja | ||
GPL | ||
KNOWNBUGS.rb | ||
LEGAL | ||
NEWS.md | ||
README.EXT | ||
README.EXT.ja | ||
README.ja.md | ||
README.md | ||
README_ruby.md | ||
aclocal.m4 | ||
addr2line.c | ||
addr2line.h | ||
array.c | ||
array.rb | ||
ast.c | ||
ast.rb | ||
autogen.sh | ||
bignum.c | ||
builtin.c | ||
builtin.h | ||
class.c | ||
common.mk | ||
compar.c | ||
compile.c | ||
complex.c | ||
configure.ac | ||
constant.h | ||
cont.c | ||
darray.h | ||
debug.c | ||
debug_counter.c | ||
debug_counter.h | ||
dir.c | ||
dir.rb | ||
dln.c | ||
dln.h | ||
dln_find.c | ||
dmydln.c | ||
dmyenc.c | ||
dmyext.c | ||
encindex.h | ||
encoding.c | ||
enum.c | ||
enumerator.c | ||
error.c | ||
eval.c | ||
eval_error.c | ||
eval_intern.h | ||
eval_jump.c | ||
file.c | ||
gc.c | ||
gc.h | ||
gc.rb | ||
gem_prelude.rb | ||
golf_prelude.rb | ||
goruby.c | ||
hash.c | ||
hrtime.h | ||
id_table.c | ||
id_table.h | ||
inits.c | ||
insns.def | ||
internal.h | ||
io.c | ||
io.rb | ||
iseq.c | ||
iseq.h | ||
kernel.rb | ||
lex.c.blt | ||
load.c | ||
loadpath.c | ||
localeinit.c | ||
main.c | ||
marshal.c | ||
marshal.rb | ||
math.c | ||
memory_view.c | ||
method.h | ||
mini_builtin.c | ||
miniinit.c | ||
mjit.c | ||
mjit.h | ||
mjit_compile.c | ||
mjit_worker.c | ||
nilclass.rb | ||
node.c | ||
node.h | ||
numeric.c | ||
numeric.rb | ||
object.c | ||
pack.c | ||
pack.rb | ||
parse.y | ||
prelude.rb | ||
probes.d | ||
probes_helper.h | ||
proc.c | ||
process.c | ||
ractor.c | ||
ractor.rb | ||
ractor_core.h | ||
random.c | ||
range.c | ||
rational.c | ||
re.c | ||
regcomp.c | ||
regenc.c | ||
regenc.h | ||
regerror.c | ||
regexec.c | ||
regint.h | ||
regparse.c | ||
regparse.h | ||
regsyntax.c | ||
ruby-runner.c | ||
ruby.c | ||
ruby_assert.h | ||
ruby_atomic.h | ||
rubystub.c | ||
scheduler.c | ||
signal.c | ||
siphash.c | ||
siphash.h | ||
sparc.c | ||
sprintf.c | ||
st.c | ||
strftime.c | ||
string.c | ||
struct.c | ||
symbol.c | ||
symbol.h | ||
test_asm.sh | ||
thread.c | ||
thread_pthread.c | ||
thread_pthread.h | ||
thread_sync.c | ||
thread_win32.c | ||
thread_win32.h | ||
time.c | ||
timev.h | ||
timev.rb | ||
trace_point.rb | ||
transcode.c | ||
transcode_data.h | ||
transient_heap.c | ||
transient_heap.h | ||
util.c | ||
variable.c | ||
variable.h | ||
version.c | ||
version.h | ||
vm.c | ||
vm_args.c | ||
vm_backtrace.c | ||
vm_callinfo.h | ||
vm_core.h | ||
vm_debug.h | ||
vm_dump.c | ||
vm_eval.c | ||
vm_exec.c | ||
vm_exec.h | ||
vm_insnhelper.c | ||
vm_insnhelper.h | ||
vm_method.c | ||
vm_opts.h | ||
vm_sync.c | ||
vm_sync.h | ||
vm_trace.c | ||
vsnprintf.c | ||
warning.rb | ||
yjit.h | ||
yjit.rb | ||
yjit_asm.c | ||
yjit_asm.h | ||
yjit_asm_tests.c | ||
yjit_codegen.c | ||
yjit_codegen.h | ||
yjit_core.c | ||
yjit_core.h | ||
yjit_iface.c | ||
yjit_iface.h | ||
yjit_utils.c | ||
yjit_utils.h |
README.md
YJIT - Yet Another Ruby JIT
DISCLAIMER: Please note that this project is in early stages of development. It is very much a work in progress, it may cause your software to crash, and current performance results are likely to leave you feeling underwhelmed.
YJIT is a lightweight, minimalistic Ruby JIT built inside the CRuby/MRI binary. It lazily compiles code using a Basic Block Versioning (BBV) architecture. The target use case is that of servers running Ruby on Rails, an area where CRuby's MJIT has not yet managed to deliver speedups. To simplify development, we currently support only MacOS and Linux on x86-64, but an ARM64 backend is part of future plans. This project is open source and falls under the same license as CRuby.
If you wish to learn more about the architecture, there 3 recorded conference talks and two published papers:
- YJIT: Building a New JIT Compiler Inside CRuby (MoreVMs 2021)
- Simple and Effective Type Check Removal through Lazy Basic Block Versioning (ECOOP 2015 talk)
- Interprocedural Type Specialization of JavaScript Programs Without Type Analysis (ECOOP 2016 talk)
To cite this repository in your publications, please use this bibtex snippet:
@misc{yjit_ruby_jit,
author = {Chevalier-Boisvert, Maxime and Wu, Alan and Patterson, Aaron},
title = {YJIT - Yet Another Ruby JIT},
year = {2021},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/Shopify/ruby/tree/yjit}},
}
Installation
Start by cloning the yjit
branch of the Shopify/ruby
repository:
git clone https://github.com/Shopify/ruby.git yjit
cd yjit
The YJIT ruby
binary can be built with either GCC or Clang. We recommend enabling debug symbols so that assertions are enabled during development as this makes debugging easier. Enabling debug mode will also make it possible for you to disassemble code generated by YJIT, and get access to stat counters. More detailed build instructions are provided in the Ruby README.
./autogen.sh
./configure cppflags=-DRUBY_DEBUG --prefix=$HOME/.rubies/ruby-yjit
make -j16 install
You can test that YJIT works correctly by running:
# Quick tests found in /bootstraptest
make btest
# Complete set of tests
make -j16 test-all
Usage
Once YJIT is built, you can either use ./miniruby
from within your build directory, or switch to the YJIT version of ruby
by using the chruby
tool:
chruby ruby-yjit
ruby myscript.rb
You can dump statistics about compilation and execution by running YJIT with the --yjit-stats
command-line option:
./miniruby --yjit-stats myscript.rb
The machine code generated for a given method can be printed by adding puts YJIT.disasm(method(:method_name))
to a Ruby script. Note that no code will be generated if the method is not compiled.
Benchmarking
We have collected a set of benchmarks and implemented a simple benchmarking harness in the yjit-bench repository. This benchmarking harness is designed to disable CPU frequency scaling, set process affinity and disable address space randomization so that the variance between benchmarking runs will be as small as possible. Please kindly note that we are at an early stage in this project.
Source Code Organization
The YJIT source code is divided between:
yjit_asm.c
: x86 in-memory assembler we use to generate machine codeyjit_asm_tests.c
: tests for the in-memory assembleryjit_codegen.c
: logic for translating Ruby bytecode to machine codeyjit_core.c
: basic block versioning logic, core structure of YJITyjit_iface.c
: code YJIT uses to interface with the rest of CRubyyjit.h
: C definitions YJIT exposes to the rest of the CRubyyjit.rb
:YJIT
Ruby module that is exposed to Rubytest_asm.sh
: script to compile and run the in-memory assembler teststool/ruby_vm/views/vm.inc.erb
: template instruction handler used to hook into the interpreter
The core of CRuby's interpreter logic is found in:
insns.def
: defines Ruby's bytecode instructions (gets compiled intovm.inc
)vm_insnshelper.c
: logic used by Ruby's bytecode instructionsvm_exec.c
: Ruby interpreter loop
Contributing
We welcome open source contributors. You should feel free to open new issues to report bugs or just to ask questions. Suggestions on how to make this readme file more helpful for new contributors are most welcome.
Bug fixes and bug reports are very valuable to us. If you find bugs in YJIT, it's very possible be that nobody has reported this bug before, or that we don't have a good reproduction for it, so please open an issue and provide some information about your configuration and a description of how you encountered the problem. If you are able to produce a small reproduction to help us track down the bug, that is very much appreciated as well.
If you would like to contribute a large patch to YJIT, we suggest opening an issue or a discussion on this repository so that we can have an active discussion. A common problem is that sometimes people submit large pull requests to open source projects without prior communication, and we have to reject them because the work they implemented does not fit within the design of the project. We want to save you time and frustration, so please reach out and we can have a productive discussion as to how you can contribute things we will want to merge into YJIT.