summaryrefslogtreecommitdiff
path: root/yjit/src
AgeCommit message (Collapse)Author
2025-05-09Rename `rb_shape_get_shape_by_id` -> `RSHAPE`Jean Boussier
Notes: Merged: https://github.com/ruby/ruby/pull/13283
2025-05-08Move `object_id` in object fields.Jean Boussier
And get rid of the `obj_to_id_tbl` It's no longer needed, the `object_id` is now stored inline in the object alongside instance variables. We still need the inverse table in case `_id2ref` is invoked, but we lazily build it by walking the heap if that happens. The `object_id` concern is also no longer a GC implementation concern, but a generic implementation. Co-Authored-By: Matt Valentine-House <matt@eightbitraptor.com> Notes: Merged: https://github.com/ruby/ruby/pull/13159
2025-05-08Refactor OBJ_TOO_COMPLEX_SHAPE_ID to not be referenced outside shape.hJean Boussier
Also refactor checks for `->type == SHAPE_OBJ_TOO_COMPLEX`. Notes: Merged: https://github.com/ruby/ruby/pull/13159
2025-05-08Rename `ivptr` -> `fields`, `next_iv_index` -> `next_field_index`Jean Boussier
Ivars will longer be the only thing stored inline via shapes, so keeping the `iv_index` and `ivptr` names would be confusing. Instance variables won't be the only thing stored inline via shapes, so keeping the `ivptr` name would be confusing. `field` encompass anything that can be stored in a VALUE array. Similarly, `gen_ivtbl` becomes `gen_fields_tbl`. Notes: Merged: https://github.com/ruby/ruby/pull/13159
2025-05-05YJIT: End the block after OPTIMIZE_METHOD_TYPE_CALL (#13245)Takashi Kokubun
Notes: Merged-By: k0kubun <takashikkbn@gmail.com>
2025-05-05Make rb_shape.capacity an `attr_index_t`Jean Boussier
Notes: Merged: https://github.com/ruby/ruby/pull/13257
2025-05-02YJIT: ZJIT: Share identical glue functionsAlan Wu
Working towards having YJIT and ZJIT in the same build, we need to deduplicate some glue code that would otherwise cause name collision. Add jit.c for this and build it for YJIT and ZJIT builds. Update bindgen to look at jit.c; some shuffling of functions in the output, but the set of functions shouldn't have changed. Notes: Merged: https://github.com/ruby/ruby/pull/13229
2025-04-29ZJIT: Disable ZJIT instructions when USE_ZJIT is 0 (#13199)Takashi Kokubun
* ZJIT: Disable ZJIT instructions when USE_ZJIT is 0 * Test the order of ZJIT instructions * Add more jobs that disable JITs * Show instruction names in the message Notes: Merged-By: k0kubun <takashikkbn@gmail.com>
2025-04-28ZJIT: Drop trace_zjit_* instructions (#13189)Takashi Kokubun
Notes: Merged-By: k0kubun <takashikkbn@gmail.com>
2025-04-28YJIT: Fix potential infinite loop when OOM (GH-13186)Rian McGuire
Avoid generating an infinite loop in the case where: 1. Block `first` is adjacent to block `second`, and the branch from `first` to `second` is a fallthrough, and 2. Block `second` immediately exits to the interpreter, and 3. Block `second` is invalidated and YJIT is OOM While pondering how to fix this, I think I've stumbled on another related edge case: 1. Block `incoming_one` and `incoming_two` both branch to block `second`. Block `incoming_one` has a fallthrough 2. Block `second` immediately exits to the interpreter (so it starts with its exit) 3. When Block `second` is invalidated, the incoming fallthrough branch from `incoming_one` might be rewritten first, which overwrites the start of block `second` with a jump to a new branch stub. 4. YJIT runs of out memory 5. The incoming branch from `incoming_two` is then rewritten, but because we're OOM we can't generate a new stub, so we use `second`'s exit as the branch target. However `second`'s exit was already overwritten with a jump to the branch stub for `incoming_one`, so `incoming_two` will end up jumping to `incoming_one`'s branch stub. Fixes [Bug #21257] Notes: Merged: https://github.com/ruby/ruby/pull/13186 Merged-By: XrXr
2025-04-25Inline Class#new.Aaron Patterson
This commit inlines instructions for Class#new. To make this work, we added a new YARV instructions, `opt_new`. `opt_new` checks whether or not the `new` method is the default allocator method. If it is, it allocates the object, and pushes the instance on the stack. If not, the instruction jumps to the "slow path" method call instructions. Old instructions: ``` > ruby --dump=insns -e'Object.new' == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,10)> 0000 opt_getconstant_path <ic:0 Object> ( 1)[Li] 0002 opt_send_without_block <calldata!mid:new, argc:0, ARGS_SIMPLE> 0004 leave ``` New instructions: ``` > ./miniruby --dump=insns -e'Object.new' == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,10)> 0000 opt_getconstant_path <ic:0 Object> ( 1)[Li] 0002 putnil 0003 swap 0004 opt_new <calldata!mid:new, argc:0, ARGS_SIMPLE>, 11 0007 opt_send_without_block <calldata!mid:initialize, argc:0, FCALL|ARGS_SIMPLE> 0009 jump 14 0011 opt_send_without_block <calldata!mid:new, argc:0, ARGS_SIMPLE> 0013 swap 0014 pop 0015 leave ``` This commit speeds up basic object allocation (`Foo.new`) by 60%, but classes that take keyword parameters see an even bigger benefit because no hash is allocated when instantiating the object (3x to 6x faster). Here is an example that uses `Hash.new(capacity: 0)`: ``` > hyperfine "ruby --disable-gems -e'i = 0; while i < 10_000_000; Hash.new(capacity: 0); i += 1; end'" "./ruby --disable-gems -e'i = 0; while i < 10_000_000; Hash.new(capacity: 0); i += 1; end'" Benchmark 1: ruby --disable-gems -e'i = 0; while i < 10_000_000; Hash.new(capacity: 0); i += 1; end' Time (mean ± σ): 1.082 s ± 0.004 s [User: 1.074 s, System: 0.008 s] Range (min … max): 1.076 s … 1.088 s 10 runs Benchmark 2: ./ruby --disable-gems -e'i = 0; while i < 10_000_000; Hash.new(capacity: 0); i += 1; end' Time (mean ± σ): 627.9 ms ± 3.5 ms [User: 622.7 ms, System: 4.8 ms] Range (min … max): 622.7 ms … 633.2 ms 10 runs Summary ./ruby --disable-gems -e'i = 0; while i < 10_000_000; Hash.new(capacity: 0); i += 1; end' ran 1.72 ± 0.01 times faster than ruby --disable-gems -e'i = 0; while i < 10_000_000; Hash.new(capacity: 0); i += 1; end' ``` This commit changes the backtrace for `initialize`: ``` aaron@tc ~/g/ruby (inline-new)> cat test.rb class Foo def initialize puts caller end end def hello Foo.new end hello aaron@tc ~/g/ruby (inline-new)> ruby -v test.rb ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin24] test.rb:8:in 'Class#new' test.rb:8:in 'Object#hello' test.rb:11:in '<main>' aaron@tc ~/g/ruby (inline-new)> ./miniruby -v test.rb ruby 3.5.0dev (2025-03-28T23:59:40Z inline-new c4157884e4) +PRISM [arm64-darwin24] test.rb:8:in 'Object#hello' test.rb:11:in '<main>' ``` It also increases memory usage for calls to `new` by 122 bytes: ``` aaron@tc ~/g/ruby (inline-new)> cat test.rb require "objspace" class Foo def initialize puts caller end end def hello Foo.new end puts ObjectSpace.memsize_of(RubyVM::InstructionSequence.of(method(:hello))) aaron@tc ~/g/ruby (inline-new)> make runruby RUBY_ON_BUG='gdb -x ./.gdbinit -p' ./miniruby -I./lib -I. -I.ext/common ./tool/runruby.rb --extout=.ext -- --disable-gems ./test.rb 656 aaron@tc ~/g/ruby (inline-new)> ruby -v test.rb ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [arm64-darwin24] 544 ``` Thanks to @ko1 for coming up with this idea! Co-Authored-By: John Hawthorn <john@hawthorn.email>
2025-04-18Fix yjit-bindgenTakashi Kokubun
Notes: Merged: https://github.com/ruby/ruby/pull/13131
2025-04-18Fix bindgenTakashi Kokubun
Notes: Merged: https://github.com/ruby/ruby/pull/13131
2025-03-07YJIT: Add Counter::invalidate_everythingAlan Wu
When YJIT is forced to discard all the code, that's bad for performance, so there should be an easy way to know about it. Notes: Merged: https://github.com/ruby/ruby/pull/12882
2025-03-06YJIT: Rename get_temp_regs2() back to get_temp_regs() (#12866)Takashi Kokubun
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2025-03-03Allow YJIT `mem-size` and `call-threshold` to be set at runtime via ↵annichai-stripe
`YJIT.enable()` (#12505) * first commit * yjit.rb change * revert formatting * rename mem-size to exec-mem-size for correctness * wip, move setting into rb_yjit_enable directly * remove unused helper functions * add in call threshold * input validation with extensive eprintln * delete test script * exec-mem-size -> mem-size * handle input validation with asserts * add test cases related to input validation * modify test cases * move validation out of rs, into rb * add comments * remove trailing spaces * remove logging Co-authored-by: Takashi Kokubun <takashikkbn@gmail.com> * remove helper fn * Update test/ruby/test_yjit.rb Co-authored-by: Takashi Kokubun <takashikkbn@gmail.com> * trailing white space --------- Co-authored-by: Alan Wu <XrXr@users.noreply.github.com> Co-authored-by: Takashi Kokubun <takashikkbn@gmail.com> Co-authored-by: Maxime Chevalier-Boisvert <maxime.chevalierboisvert@shopify.com> Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
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 <randy.stauner@shopify.com> Notes: Merged: https://github.com/ruby/ruby/pull/12670
2025-02-12Remove dead iv_index_tbl field in RObjectPeter Zhu
Notes: Merged: https://github.com/ruby/ruby/pull/12739
2025-01-30YJIT: Remove comments that refer to the removed "stats" featureAlan Wu
The Cargo feature was removed in 2de8b5b8054f311c4cee112dcab5208b66cc62a4 and it's available in all build configs now. [ci skip]
2025-01-30YJIT: Turn on dead code lint for the stats moduleAlan Wu
2025-01-30YJIT: Explicitly specify C ABI to fix a nightly Rust warningAlan Wu
2025-01-29YJIT: A64: Remove assert that trips when OOM at page boundaryAlan Wu
With a well-timed OOM around a page switch in the backend, it can return RetryOnNextPage twice and crash due to the assert. (More places can signal OOM now since VirtualMem tracks Rust malloc heap size for --yjit-mem-size.) Return error in these cases instead of crashing. Fixes: https://github.com/Shopify/ruby/issues/566 Notes: Merged: https://github.com/ruby/ruby/pull/12668
2025-01-28YJIT: Initialize locals in ISeqs defined with `...` (#12660)Alan Wu
* YJIT: Fix indentation [ci skip] Fixes: cdf33ed5f37f9649c482c3ba1d245f0d80ac01ce * YJIT: Initialize locals in ISeqs defined with `...` Previously, callers of forwardable ISeqs moved the stack pointer up without writing to the stack. If there happens to be a stale value in the area skipped over, it could crash due to "try to mark T_NONE". Also, the uninitialized local variables were observable through `binding`. Initialize the locals to nil. [Bug #21021] Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2025-01-10YJIT: Rename send_iseq_forwarding->send_forwardingAlan Wu
It's in gen_send_general(), so nothing specifically to do with iseqs. Notes: Merged: https://github.com/ruby/ruby/pull/12550
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
2025-01-08YJIT: Filter `&` calls from specialized C method codegenAlan Wu
Evident with the crash reported in [Bug #20997], the C replacement codegen functions aren't authored to handle block arguments (nor should they because the extra code from the complexity defeats optimization). Filter sites with VM_CALL_ARGS_BLOCKARG. Notes: Merged: https://github.com/ruby/ruby/pull/12536
2025-01-04YJIT: Fix crash when yielding keyword argumentsAlan Wu
Previously, the code for dropping surplus arguments when yielding into blocks erroneously attempted to drop keyword arguments when there is in fact no surplus arguments. Fix the condition and test that supplying the exact number of keyword arguments as require compiles without fallback. Notes: Merged: https://github.com/ruby/ruby/pull/12499
2024-12-23YJIT: Return None if entry block compilation fails (#12445)Takashi Kokubun
Notes: Merged-By: k0kubun <takashikkbn@gmail.com>
2024-12-17YJIT: Load registers on JIT entry to reuse blocks (#12355)Takashi Kokubun
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-12-13YJIT: Speculate block arg for `c_func_method(&nil)` calls (#12326)Alan Wu
A good amount of call sites always pass nil as block argument, but the nil doesn't show up in the context. Put a runtime guard for those cases to handle it. Particular relevant for the `ruby-lsp` benchmark in `yjit-bench`. Up to a 2% speedup across headline benchmarks. Co-authored-by: Takashi Kokubun <takashikkbn@gmail.com> Co-authored-by: Maxime Chevalier-Boisvert <maxime.chevalierboisvert@shopify.com> Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org> Co-authored-by: Kevin Menard <kevin@nirvdrum.com> Co-authored-by: Randy Stauner <randy.stauner@shopify.com> Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-12-12YJIT: Allow then-unknown `static_mut_refs` on older Rusts [ci skip]Alan Wu
2024-12-11YJIT: Fix unread field lint in release buildsAlan Wu
``` warning: fields `blue_begin` and `blue_end` are never read ``` Notes: Merged: https://github.com/ruby/ruby/pull/12310
2024-12-11YJIT: Disable static_mut_refs for nowAlan Wu
Notes: Merged: https://github.com/ruby/ruby/pull/12310
2024-12-11YJIT: Use the correct size constantAlan Wu
Notes: Merged: https://github.com/ruby/ruby/pull/12310
2024-12-09YJIT: Add a comment about a lazy frame callTakashi Kokubun
jit_prepare_lazy_frame_call is a complicated trick and comes with memory overhead. Every use of the function should come with justification.
2024-12-09YJIT: Spill/load argument registers to reuse blocks (#12287)Takashi Kokubun
* YJIT: Spill/load argument registers to reuse blocks * Mention the immediate function name * Explain the context behind spill/load operations Notes: Merged-By: k0kubun <takashikkbn@gmail.com>
2024-12-09YJIT: Only enable disassembly colors for tty (#12283)Max Bernstein
* YJIT: Use fully-qualified name for OPTIONS in get_options! * YJIT: Only enable disassembly colors for tty Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-12-04YJIT: Generate specialized code for Symbol for objtostring (#12247)Maximillian Polhill
* YJIT: Generate specialized code for Symbol for objtostring Co-authored-by: John Hawthorn <john@hawthorn.email> * Update yjit/src/codegen.rs --------- Co-authored-by: John Hawthorn <john@hawthorn.email> Co-authored-by: Maxime Chevalier-Boisvert <maximechevalierb@gmail.com> Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-12-04YJIT: track time since initialization (#12263)Maxime Chevalier-Boisvert
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-29YJIT: Avoid std::ffi::CString with rb_intern2() during bootAlan Wu
Fewer allocations on boot, too. Suggested-by: https://github.com/ruby/ruby/pull/12217 Notes: Merged: https://github.com/ruby/ruby/pull/12220
2024-11-28YJIT: Add missing prepare before calling str_dupJohn Hawthorn
Notes: Merged: https://github.com/ruby/ruby/pull/12202
2024-11-26YJIT: Implement opt_reverse insn (#12175)Randy Stauner
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-26Optimize instructions when creating an array just to call `include?` (#12123)Randy Stauner
* Add opt_duparray_send insn to skip the allocation on `#include?` If the method isn't going to modify the array we don't need to copy it. This avoids the allocation / array copy for things like `[:a, :b].include?(x)`. This adds a BOP for include? and tracks redefinition for it on Array. Co-authored-by: Andrew Novoselac <andrew.novoselac@shopify.com> * YJIT: Implement opt_duparray_send include_p Co-authored-by: Andrew Novoselac <andrew.novoselac@shopify.com> * Update opt_newarray_send to support simple forms of include?(arg) Similar to opt_duparray_send but for non-static arrays. * YJIT: Implement opt_newarray_send include_p --------- Co-authored-by: Andrew Novoselac <andrew.novoselac@shopify.com> Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-25YJIT: fix small typo in command line options help (#12167)Maxime Chevalier-Boisvert
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-20YJIT: Make compilation_failure a default stat (#12128)Alan Wu
It's good to monitor compilation failures. Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-20YJIT: Refactor to forward jump_to_next_insn() return valueAlan Wu
It's more concise this way and since `return Some(EndBlock)` is the only correct answer, no point repeating it everywhere. Notes: Merged: https://github.com/ruby/ruby/pull/12124
2024-11-20YJIT: Abandon block when gen_outlined_exit() failsAlan Wu
When CodeBlock::set_page fails (part of next_page(), see their docs for exact conditions), it can cause gen_outlined_exit() to fail while there is still plenty of memory available. Previously, this can have YJIT running incomplete code due to taking the early return in end_block_with_jump() that manifested as crashes with SIGILL. Add and use a wrapper with error handling. Notes: Merged: https://github.com/ruby/ruby/pull/12124
2024-11-14YJIT: Specialize String#dup (#12090)Takashi Kokubun
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-14YJIT: Specialize Integer#pred (#12082)Takashi Kokubun
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
2024-11-13YJIT: Add inline_block_count stat (#12081)Takashi Kokubun
Notes: Merged-By: maximecb <maximecb@ruby-lang.org>
generated by cgit v1.2.3 (git 2.25.1) at 2026-05-14 21:46:56 +0000