| Age | Commit message (Collapse) | Author |
|---|
|
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
|
|
Notes:
Merged: https://github.com/ruby/ruby/pull/12743
|
|
Notes:
Merged: https://github.com/ruby/ruby/pull/12739
|
|
Bumps [capstone](https://github.com/capstone-rust/capstone-rs) from 0.12.0 to 0.13.0.
- [Release notes](https://github.com/capstone-rust/capstone-rs/releases)
- [Changelog](https://github.com/capstone-rust/capstone-rs/blob/master/CHANGELOG.md)
- [Commits](https://github.com/capstone-rust/capstone-rs/compare/capstone-v0.12.0...capstone-v0.13.0)
---
updated-dependencies:
- dependency-name: capstone
dependency-type: direct:production
update-type: version-update:semver-minor
...
Signed-off-by: dependabot[bot] <[email protected]>
Notes:
Merged: https://github.com/ruby/ruby/pull/12697
|
|
The Cargo feature was removed in 2de8b5b8054f311c4cee112dcab5208b66cc62a4
and it's available in all build configs now.
[ci skip]
|
|
|
|
|
|
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
|
|
* 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 <[email protected]>
|
|
It's in gen_send_general(), so nothing specifically to do with iseqs.
Notes:
Merged: https://github.com/ruby/ruby/pull/12550
|
|
`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
|
|
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
|
|
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
|
|
Notes:
Merged-By: k0kubun <[email protected]>
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
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 <[email protected]>
Co-authored-by: Maxime Chevalier-Boisvert <[email protected]>
Co-authored-by: Aaron Patterson <[email protected]>
Co-authored-by: Kevin Menard <[email protected]>
Co-authored-by: Randy Stauner <[email protected]>
Notes:
Merged-By: maximecb <[email protected]>
|
|
|
|
```
warning: fields `blue_begin` and `blue_end` are never read
```
Notes:
Merged: https://github.com/ruby/ruby/pull/12310
|
|
Notes:
Merged: https://github.com/ruby/ruby/pull/12310
|
|
Notes:
Merged: https://github.com/ruby/ruby/pull/12310
|
|
jit_prepare_lazy_frame_call is a complicated trick and comes with memory
overhead. Every use of the function should come with justification.
|
|
* YJIT: Spill/load argument registers to reuse blocks
* Mention the immediate function name
* Explain the context behind spill/load operations
Notes:
Merged-By: k0kubun <[email protected]>
|
|
* YJIT: Use fully-qualified name for OPTIONS in get_options!
* YJIT: Only enable disassembly colors for tty
Notes:
Merged-By: maximecb <[email protected]>
|
|
* YJIT: Generate specialized code for Symbol for objtostring
Co-authored-by: John Hawthorn <[email protected]>
* Update yjit/src/codegen.rs
---------
Co-authored-by: John Hawthorn <[email protected]>
Co-authored-by: Maxime Chevalier-Boisvert <[email protected]>
Notes:
Merged-By: maximecb <[email protected]>
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
Fewer allocations on boot, too.
Suggested-by: https://github.com/ruby/ruby/pull/12217
Notes:
Merged: https://github.com/ruby/ruby/pull/12220
|
|
Notes:
Merged: https://github.com/ruby/ruby/pull/12202
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
* 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 <[email protected]>
* YJIT: Implement opt_duparray_send include_p
Co-authored-by: Andrew Novoselac <[email protected]>
* 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 <[email protected]>
Notes:
Merged-By: maximecb <[email protected]>
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
It's good to monitor compilation failures.
Notes:
Merged-By: maximecb <[email protected]>
|
|
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
|
|
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
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
Notes:
Merged-By: maximecb <[email protected]>
|
|
* YJIT: Specialize `String#[]` (`String#slice`) with fixnum arguments
String#[] is in the top few C calls of several YJIT benchmarks:
liquid-compile rubocop mail sudoku
This speeds up these benchmarks by 1-2%.
* YJIT: Try harder to get type info for `String#[]`
In the large generated code of the mail gem the context doesn't have
the type info. In that case if we peek at the stack and add a guard
we can still apply the specialization
and it speeds up the mail benchmark by 5%.
Co-authored-by: Maxime Chevalier-Boisvert <[email protected]>
Co-authored-by: Takashi Kokubun (k0kubun) <[email protected]>
---------
Co-authored-by: Maxime Chevalier-Boisvert <[email protected]>
Co-authored-by: Takashi Kokubun (k0kubun) <[email protected]>
Notes:
Merged-By: maximecb <[email protected]>
|
|
* Use FL_USER0 for ELTS_SHARED
This makes space in RString for two bits for chilled strings.
* Mark strings returned by `Symbol#to_s` as chilled
[Feature #20350]
`STR_CHILLED` now spans on two user flags. If one bit is set it
marks a chilled string literal, if it's the other it marks a
`Symbol#to_s` chilled string.
Since it's not possible, and doesn't make much sense to include
debug info when `--debug-frozen-string-literal` is set, we can't
include allocation source, but we can safely include the symbol
name in the warning message, making it much easier to find the source
of the issue.
Co-Authored-By: Étienne Barrié <[email protected]>
---------
Co-authored-by: Étienne Barrié <[email protected]>
Co-authored-by: Jean Boussier <[email protected]>
|
|
When we run with RUBY_FREE_AT_EXIT, there's a false-positive memory leak
reported in YJIT because the METHOD_CODEGEN_TABLE is never freed. This
commit adds rb_yjit_free_at_exit that is called at shutdown when
RUBY_FREE_AT_EXIT is set.
Reported memory leak:
==699816== 1,104 bytes in 1 blocks are possibly lost in loss record 1 of 1
==699816== at 0x484680F: malloc (vg_replace_malloc.c:446)
==699816== by 0x155B3E: UnknownInlinedFun (unix.rs:14)
==699816== by 0x155B3E: UnknownInlinedFun (stats.rs:36)
==699816== by 0x155B3E: UnknownInlinedFun (stats.rs:27)
==699816== by 0x155B3E: alloc (alloc.rs:98)
==699816== by 0x155B3E: alloc_impl (alloc.rs:181)
==699816== by 0x155B3E: allocate (alloc.rs:241)
==699816== by 0x155B3E: do_alloc<alloc::alloc::Global> (alloc.rs:15)
==699816== by 0x155B3E: new_uninitialized<alloc::alloc::Global> (mod.rs:1750)
==699816== by 0x155B3E: fallible_with_capacity<alloc::alloc::Global> (mod.rs:1788)
==699816== by 0x155B3E: prepare_resize<alloc::alloc::Global> (mod.rs:2864)
==699816== by 0x155B3E: resize_inner<alloc::alloc::Global> (mod.rs:3060)
==699816== by 0x155B3E: reserve_rehash_inner<alloc::alloc::Global> (mod.rs:2950)
==699816== by 0x155B3E: hashbrown::raw::RawTable<T,A>::reserve_rehash (mod.rs:1231)
==699816== by 0x5BC39F: UnknownInlinedFun (mod.rs:1179)
==699816== by 0x5BC39F: find_or_find_insert_slot<(usize, fn(&mut yjit::codegen::JITState, &mut yjit::backend::ir::Assembler, *const yjit::cruby::autogened::rb_callinfo, *const yjit::cruby::autogened::rb_callable_method_entry_struct, core::option::Option<yjit::codegen::BlockHandler>, i32, core::option::Option<yjit::cruby::VALUE>) -> bool), alloc::alloc::Global, hashbrown::map::equivalent_key::{closure_env#0}<usize, usize, fn(&mut yjit::codegen::JITState, &mut yjit::backend::ir::Assembler, *const yjit::cruby::autogened::rb_callinfo, *const yjit::cruby::autogened::rb_callable_method_entry_struct, core::option::Option<yjit::codegen::BlockHandler>, i32, core::option::Option<yjit::cruby::VALUE>) -> bool>, hashbrown::map::make_hasher::{closure_env#0}<usize, fn(&mut yjit::codegen::JITState, &mut yjit::backend::ir::Assembler, *const yjit::cruby::autogened::rb_callinfo, *const yjit::cruby::autogened::rb_callable_method_entry_struct, core::option::Option<yjit::codegen::BlockHandler>, i32, core::option::Option<yjit::cruby::VALUE>) -> bool, std::hash::random::RandomState>> (mod.rs:1413)
==699816== by 0x5BC39F: hashbrown::map::HashMap<K,V,S,A>::insert (map.rs:1754)
==699816== by 0x57C5C6: insert<usize, fn(&mut yjit::codegen::JITState, &mut yjit::backend::ir::Assembler, *const yjit::cruby::autogened::rb_callinfo, *const yjit::cruby::autogened::rb_callable_method_entry_struct, core::option::Option<yjit::codegen::BlockHandler>, i32, core::option::Option<yjit::cruby::VALUE>) -> bool, std::hash::random::RandomState> (map.rs:1104)
==699816== by 0x57C5C6: yjit::codegen::reg_method_codegen (codegen.rs:10521)
==699816== by 0x57C295: yjit::codegen::yjit_reg_method_codegen_fns (codegen.rs:10464)
==699816== by 0x5C6B07: rb_yjit_init (yjit.rs:40)
==699816== by 0x393723: ruby_opt_init (ruby.c:1820)
==699816== by 0x393723: ruby_opt_init (ruby.c:1767)
==699816== by 0x3957D4: prism_script (ruby.c:2215)
==699816== by 0x3957D4: process_options (ruby.c:2538)
==699816== by 0x396065: ruby_process_options (ruby.c:3166)
==699816== by 0x236E56: ruby_options (eval.c:117)
==699816== by 0x15BAED: rb_main (main.c:43)
==699816== by 0x15BAED: main (main.c:62)
After this patch, there are no more memory leaks reported when running
RUBY_FREE_AT_EXIT with Valgrind on an empty Ruby script:
$ RUBY_FREE_AT_EXIT=1 valgrind --leak-check=full ruby -e ""
...
==700357== HEAP SUMMARY:
==700357== in use at exit: 0 bytes in 0 blocks
==700357== total heap usage: 36,559 allocs, 36,559 frees, 6,064,783 bytes allocated
==700357==
==700357== All heap blocks were freed -- no leaks are possible
Notes:
Merged-By: maximecb <[email protected]>
|
|
In [1], we started checking for gen_branch failures, but I made two
crucial mistakes. One, defer_compilation() had the same issue as
gen_branch() but wasn't checked. Two, returning None from a codegen
function does not throw away the block. Checking how gen_single_block()
handles codegen functions, you can see that None terminates the block
with an exit, but does not overall return an Err. This handling is fine
for unimplemented instructions, for example, but incorrect in case
gen_branch() fails. The missing branch essentially corrupts the
block; adding more code after a missing branch doesn't correct the code.
Always abandon the block when defer_compilation() or gen_branch() fails.
[1]: cb661d7d82984cdb54485ea3f4af01ac21960882
Fixup: [1]
Notes:
Merged: https://github.com/ruby/ruby/pull/12035
Merged-By: XrXr
|
|
So that the Rust panic message is forwarded to the RUBY_CRASH_REPORT
system, instead of only the static "YJIT panicked" message done so
previously. This helps with triaging crashes since it's easier than
trying to parse stderr output.
Sample:
<internal:yjit_hook>:2: [BUG] YJIT: panicked at src/codegen.rs:1197:5:
explicit panic
...
Notes:
Merged: https://github.com/ruby/ruby/pull/12027
|
|
Fix as the compiler orders:
```
warning: unused return value of `into_raw_fd` that must be used
--> ../src/yjit/src/disasm.rs:123:21
|
123 | file.into_raw_fd(); // keep the fd open
| ^^^^^^^^^^^^^^^^^^
|
= note: losing the raw file descriptor may leak resources
= note: `#[warn(unused_must_use)]` on by default
help: use `let _ = ...` to ignore the resulting value
|
123 | let _ = file.into_raw_fd(); // keep the fd open
| +++++++
warning: unused return value of `into_raw_fd` that must be used
--> ../src/yjit/src/log.rs:84:21
|
84 | file.into_raw_fd(); // keep the fd open
| ^^^^^^^^^^^^^^^^^^
|
= note: losing the raw file descriptor may leak resources
help: use `let _ = ...` to ignore the resulting value
|
84 | let _ = file.into_raw_fd(); // keep the fd open
| +++++++
```
Notes:
Merged: https://github.com/ruby/ruby/pull/12009
|
|
* YJIT: Replace Array#each only when YJIT is enabled
* Add comments about BUILTIN_ATTR_C_TRACE
* Make Ruby Array#each available with --yjit as well
* Fix all paths that expect a C location
* Use method_basic_definition_p to detect patches
* Copy a comment about C_TRACE flag to compilers
* Rephrase a comment about add_yjit_hook
* Give METHOD_ENTRY_BASIC flag to Array#each
* Add --yjit-c-builtin option
* Allow inconsistent source_location in test-spec
* Refactor a check of BUILTIN_ATTR_C_TRACE
* Set METHOD_ENTRY_BASIC without touching vm->running
Notes:
Merged-By: maximecb <[email protected]>
|
|
|
|
We got some core dumps in the wild where a PendingBranch had everything
as None, leading to a panic unwrapping in PendingBranch::into_branch().
This happened while compiling a `branchif`.
It seems that the only way this can happen is when core::gen_branch()
fails, but not due to OOM. We wouldn't have reach into_branch() when
OOM, and the only way to not leave markers that would've set the
branch's start_addr to some value in gen_branch() is for set_target() to
fail, causing an early return.
Unfortunately, it's hard to tell the exact sequence of events that led
to this situation, but regardless, the dumps show us that we should
check for errors in gen_branch().
Because gen_branch() is used deep in the stack during compilation (e.g.
guard_known_class() -> jit_chain_guard() -> gen_branch()), it'd be bad
for compile speed to propagate the error everywhere, not to mention the
massive patch required. Opt for a flag checked near the end of
compilation.
Notes:
Merged: https://github.com/ruby/ruby/pull/11938
Merged-By: XrXr
|
|
|
|
Notes:
Merged-By: k0kubun <[email protected]>
|
|
Type information in the context for no additional work!
This is the `if (special_object_p(obj)) return obj;` path in
rb_obj_dup() and for Numeric#dup, it's always the identity function.
Notes:
Merged: https://github.com/ruby/ruby/pull/11926
|
|
* YJIT: Fill in commented-out assertion
* YJIT: Rename yjit_reg_method() and add links in docs
Notes:
Merged-By: maximecb <[email protected]>
|
|
* Update yjit-bindgen deps
* YJIT: Allow shareable consts in multi-ractor mode
* Update yjit/src/codegen.rs
Co-authored-by: Alan Wu <[email protected]>
---------
Co-authored-by: Alan Wu <[email protected]>
Notes:
Merged-By: maximecb <[email protected]>
|
