[pull] master from ruby:master#891
Merged
pull[bot] merged 22 commits intoturkdevops:masterfrom Mar 27, 2026
Merged
Conversation
`tstring_beg` in particular is needed by `yard`. Before: > 1980 examples, 606 failures, 15 pending After: > 1980 examples, 582 failures, 15 pending Thought it would be more, but oh well. It needs `on_sp` which I guess is why there are not many new passes ruby/prism@e1b18cb582
In `ripper`, both go through the same converion logic. Needed for rspec, no other failures in their own tests ruby/prism@510258aa2b
This checks that the value returned from the function registered with rb_define_alloc_func is of the correct class. When this was first introduced in 1fe40b7 (by Matz on 2001-10-03), allocation was done via user-defined Object#allocate, so it made sense to have a runtime check in release builds. Now that it's defined via rb_define_alloc_func in the C extension API, I don't think it's necessary. The check is surprisingly expensive. Removing it makes Object.new about 10% faster. It allows the C compiler to optimize the call to the function pointer as a tail-call. Removing this also allows ZJIT/YJIT to call the function directly (ZJIT already does this by having a list of known safe allocation functions). There's no way for users to ever have seen this check, other than by writing a misbehaving C extension, which returns objects with the wrong class. [Feature #21966]
…6581) For sends with ARGS_BLOCKARG (e.g. `foo(&block)`), the block arg sits on the stack above the receiver and regular arguments. The profiling (both interpreter and exit profiling) only records types for the receiver and regular args (argc + 1 values), but profile_stack was mapping those types onto stack positions starting from the top, which incorrectly mapped them onto the block arg and the wrong operands. Fix by detecting ARGS_BLOCKARG at the profile_stack call site and passing a stack_offset of 1 to skip the block arg. This allows resolve_receiver_type_from_profile to find the correct receiver type, enabling method dispatch optimization for these sends after recompilation via exit profiling.
* ZJIT: Check native stack before compiling ISEQ When the native/machine stack is nearly exhausted, don't compile and enter ZJIT code. JIT-compiled code uses more native stack per call frame than the interpreter, so falling back to the interpreter avoids SystemStackError in cases where the interpreter would still have room. This matches what YJIT does in rb_yjit_iseq_gen_entry_point(). * ZJIT: Add skipped_native_stack_full counter
When an ISEQ has already reached MAX_ISEQ_VERSIONS, converting no-profile sends to SideExits is counterproductive: the exit fires every time but can never trigger recompilation. Keep them as Send fallbacks so the interpreter handles them directly without the overhead of a SideExit + no_profile_send_recompile call.
Instead of storing is_final_version as a field on Function, compute it dynamically in convert_no_profile_sends by checking the payload's version count against MAX_ISEQ_VERSIONS.
Keep the original SideExit recompile test alongside the new final-version test since they cover different behaviors. Remove the intermediate HIR snapshot since hir_string() now sees the auto-compiled version as final.
…d,Proc}#source_location" This reverts commit c161328.
This reverts commit 1d0cc4c.
…#source_location" This reverts commit 31c3bcb.
…ition" This reverts commit 8f5e0d8.
This reverts commit edff523.
…ler> Introduce a BlockHandler enum to represent how a block is passed to a send-like instruction. For now it has only the BlockIseq(IseqPtr) variant, making this a pure mechanical rename with no behavior change.
Add a BlockArg variant to BlockHandler for sends that pass a block via &proc (ARGS_BLOCKARG). Previously these were represented as Some(BlockIseq(null)), requiring a normalization hack in reduce_send_to_ccall. Now the three cases are explicit: None (no block), Some(BlockIseq) (literal block), Some(BlockArg) (&proc). This also fixes a latent bug: with specialized_instruction disabled (as power_assert does), regular calls use `send` instead of opt_send_without_block. A null blockiseq was misinterpreted as having a block, causing gen_block_handler_specval to write a captured block handler with a null code pointer.
Extend convert_no_profile_sends to handle SendNoProfiles (sends with literal blocks), not just SendWithoutBlockNoProfiles. The match guards against BlockArg sends since &proc forwarding can't be exit-profiled.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )