March 2026 - ruby-core - ml.ruby-lang.org

[ruby-core:125083] [Ruby Misc#21960] Improve #backtrace to not confuse terminals
by svoop (Sven Schwyn) 25 May '26

25 May '26

Issue #21960 has been reported by svoop (Sven Schwyn). ---------------------------------------- Misc #21960: Improve #backtrace to not confuse terminals https://bugs.ruby-lang.org/issues/21960 * Author: svoop (Sven Schwyn) * Status: Open ---------------------------------------- The `#backtrace` of exceptions are currently printed like so: ``` /path/to/whatever.rb:8:in `/' /path/to/whatever.rb:8:in `show' (...) ``` Most terminals recognize `/path/to/whatever.rb:8` as a local file and make it clickable as "open in editor", however, some (e.g. Ghostty) include everything up to the next whitespace, in this case `/path/to/whatever.rb:8:in`. This is then used as an argument to `open` (on macOS) which will not work due to the `:in` part. There are two ways to improve this: ## Replace the colon with a space Given the existence of `#backtrace_locations` which is better suited for programmatically working with backtrace locations, it should not break things to slightly modify the `#backtrace` to look like this: ``` /path/to/whatever.rb:8 in `/' /path/to/whatever.rb:8 in `show' (...) ^^^ space instead of colon ``` ## Add OCS 8 sequences [OCS 8](https://github.com/Alhadis/OSC8-Adoption) are the `<a>` tag for terminal output. While many terminals support it, it might cause visual noise in logs and therefore would have to be an opt-in feature e.g. via a Ruby interpreter argument. -- https://bugs.ruby-lang.org/

7 16

[ruby-core:125045] [Ruby Bug#21955] `Fiber#transfer`: machine stack not released when fiber terminates, causing `FiberError: can't set a guard page`
by ioquatix (Samuel Williams) 23 May '26

23 May '26

Issue #21955 has been reported by ioquatix (Samuel Williams). ---------------------------------------- Bug #21955: `Fiber#transfer`: machine stack not released when fiber terminates, causing `FiberError: can't set a guard page` https://bugs.ruby-lang.org/issues/21955 * Author: ioquatix (Samuel Williams) * Status: Open * Assignee: ioquatix (Samuel Williams) * Backport: 3.2: REQUIRED, 3.3: REQUIRED, 3.4: REQUIRED, 4.0: REQUIRED ---------------------------------------- When a fiber terminates (falls off the end of its block, or raises an unhandled exception) after being reached via `Fiber#transfer`, its machine stack is **not** returned to the fiber pool. The stack is only freed when the `Fiber` object is eventually garbage collected. In production, where GC does not run continuously, terminated fibers accumulate unreleased stacks. Each stack allocation contains a guard page set with `mprotect(PROT_NONE)`, which splits a kernel VMA. On Linux this exhausts the per-process `vm.max_map_count` limit and raises: ``` FiberError: can't set a guard page: Cannot allocate memory ``` The symptom is confusing: the fibers are all dead (`alive? == false`), new fibers cannot be created, but GC has not run. ## Root cause In `cont.c`, `fiber_switch()`, the eager stack release after `fiber_store` returns is guarded by `resuming_fiber`: ```c // cont.c (affected versions) #ifndef COROUTINE_PTHREAD_CONTEXT if (resuming_fiber && FIBER_TERMINATED_P(fiber)) { RB_VM_LOCKING() { fiber_stack_release(fiber); } } #endif ``` `resuming_fiber` is only non-`NULL` when the switch was initiated by `Fiber#resume` (which passes `resuming_fiber = fiber`). `Fiber#transfer` passes `resuming_fiber = NULL`, so the condition is never true and the stack is silently leaked until GC. Additionally, `fiber_raise` on a suspended non-yielding (transferred) fiber calls `fiber_transfer_kw`, also passing `resuming_fiber = NULL`, so the same leak occurs when a transferred fiber is terminated by a raised exception. ## Reproduction Set a low map count limit, then run the attached script: ``` sudo bash -c "echo 10000 > /proc/sys/vm/max_map_count" ruby test_fiber_transfer_leak.rb ``` Restore afterwards (`sysctl vm.max_map_count` shows the default, typically 65530). ```ruby GC.disable leaked = [] count = 0 begin 10_000.times do f = Fiber.new { } # terminates immediately leaked << f # hold reference so Fiber object is not GC'd f.transfer # transfer, not resume => stack NOT released (bug) count += 1 puts "[#{count} fibers] all dead: #{leaked.none?(&:alive?)}" if count % 1000 == 0 end puts "No error — fix is applied." rescue FiberError => e puts "FiberError after #{count} fibers: #{e.message}" puts "All terminated (alive?=false): #{leaked.none?(&:alive?)}" ensure GC.enable leaked.clear GC.start end ``` Expected output on **unpatched** Ruby (vm.max_map_count=10000): ``` [1000 fibers] all dead: true [2000 fibers] all dead: true [3000 fibers] all dead: true [4000 fibers] all dead: true FiberError after 4096 fibers: can't set a guard page: Cannot allocate memory All terminated (alive?=false): false ``` Expected output on **patched** Ruby: ``` [1000 fibers] all dead: true ... [10000 fibers] all dead: true No error — fix is applied. ``` ## Fix Drop the `resuming_fiber &&` guard. After `fiber_store` returns we are executing in the caller's context — we are never on `fiber`'s stack — so releasing it is unconditionally safe. `fiber_stack_release` is already idempotent (guarded by `fiber->stack.base == NULL`), so the resume path is unaffected. ```c #ifndef COROUTINE_PTHREAD_CONTEXT if (FIBER_TERMINATED_P(fiber)) { RB_VM_LOCKING() { fiber_stack_release(fiber); } } #endif ``` Pull request with fix: https://github.com/ruby/ruby/pull/16416 -- https://bugs.ruby-lang.org/

3 2

[ruby-core:125099] [Ruby Bug#21961] Marshal.load freeze option fail to free linked strings
by byroot (Jean Boussier) 23 May '26

23 May '26

Issue #21961 has been reported by byroot (Jean Boussier). ---------------------------------------- Bug #21961: Marshal.load freeze option fail to free linked strings https://bugs.ruby-lang.org/issues/21961 * Author: byroot (Jean Boussier) * Status: Open * Backport: 3.2: WONTFIX, 3.3: REQUIRED, 3.4: REQUIRED, 4.0: REQUIRED ---------------------------------------- ```ruby str = "test" arr = [str, str] strings = Marshal.load(Marshal.dump(arr), freeze: true) p strings.map(&:frozen?) # => [true, false] ``` Expected: `[true, true]` -- https://bugs.ruby-lang.org/

4 3

[ruby-core:125078] [Ruby Bug#21959] rb_internal_thread_event_hooks_rw_lock is not reinitialized after fork causing deadlocks
by anmarchenko_datadog (Andrey Marchenko) 23 May '26

23 May '26

Issue #21959 has been reported by anmarchenko_datadog (Andrey Marchenko). ---------------------------------------- Bug #21959: rb_internal_thread_event_hooks_rw_lock is not reinitialized after fork causing deadlocks https://bugs.ruby-lang.org/issues/21959 * Author: anmarchenko_datadog (Andrey Marchenko) * Status: Open * ruby -v: 4.0.1 * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN, 4.0: UNKNOWN ---------------------------------------- Ruby's GVL Instrumentation API uses a read-write lock (rb_internal_thread_event_hooks_rw_lock) to protect the list of thread event hooks: - Read lock — acquired on every GVL transition to iterate and call hook callbacks (rb_thread_execute_hooks) - Write lock — acquired when adding/removing hooks (rb_internal_thread_add_event_hook, rb_internal_thread_remove_event_hook) After fork(), Ruby reinitializes several internal locks (e.g. vm->ractor.sched.lock, timer_th.waiting_lock), but not rb_internal_thread_event_hooks_rw_lock. This wasn't added with the GVL Instrumentation API. The full reproducer is available here: https://github.com/anmarchenko/ruby-locks-fork-bug ## Deadlock sequence 1. Parent process has thread event hooks registered (e.g. by a profiler like dd-trace-rb) 2. Multiple threads run concurrently, causing GVL transitions — each transition acquires a read lock on the rwlock 3. fork() happens while a thread holds the read lock 4. In the child, only the forking thread survives — the thread that held the lock is gone, but the lock state is copied as-is 5. Child tries to add or remove a hook → needs write lock → blocks forever on a lock that will never be released 6. Deadlock ## Impact This affects any Ruby C extension using the GVL Instrumentation API in combination with fork-based servers (Resque, Unicorn, Passenger, etc.). The original report comes from dd-trace-rb's profiler deadlocking Resque workers on Alpine Linux (musl libc): https://github.com/DataDog/dd-trace-rb/issues/4967 -- https://bugs.ruby-lang.org/

4 5

[ruby-core:124887] [Ruby Bug#21926] Thread#value on popen3 wait thread hangs in finalizer
by stevecrozz (Stephen Crosby) 23 May '26

23 May '26

Issue #21926 has been reported by stevecrozz (Stephen Crosby). ---------------------------------------- Bug #21926: Thread#value on popen3 wait thread hangs in finalizer https://bugs.ruby-lang.org/issues/21926 * Author: stevecrozz (Stephen Crosby) * Status: Open * ruby -v: 3.3.7 * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN, 4.0: UNKNOWN ---------------------------------------- Calling Thread#value on an Open3.popen3 wait thread from a finalizer completes in Ruby 3.2 but hangs in Ruby 3.3+. See repro.rb below. When the Ruby process hangs in these conditions, it no longer responds to signals and it seems to be unable to run any other threads. This affects the schmooze gem (and potentially other code using Open3.popen3 with finalizers), causing test suites to hang intermittently. ``` ruby # repro.rb require 'open3' class ProcessWrapper def initialize @stdin, @stdout, @stderr, @wait_thread = Open3.popen3("cat") ObjectSpace.define_finalizer(self, self.class.make_finalizer(@stdin, @stdout, @stderr, @wait_thread)) end def self.make_finalizer(stdin, stdout, stderr, wait_thread) proc do stdin.close rescue nil stdout.close rescue nil stderr.close rescue nil wait_thread.value # Hangs here in Ruby 3.3+ end end end 100.times { ProcessWrapper.new } GC.stress = true 1000.times { Object.new } puts "done" ``` ## Environment - Linux x86_64 - Tested on Ruby 3.2.7, 3.3.7, 3.4.8 -- https://bugs.ruby-lang.org/

5 5

[ruby-core:124594] [Ruby Bug#21844] Inconsistent ArgumentError message for Data::define.new
by jnchito (Junichi Ito) 20 May '26

20 May '26

Issue #21844 has been reported by jnchito (Junichi Ito). ---------------------------------------- Bug #21844: Inconsistent ArgumentError message for Data::define.new https://bugs.ruby-lang.org/issues/21844 * Author: jnchito (Junichi Ito) * Status: Open * ruby -v: ruby 4.0.1 (2026-01-13 revision e04267a14b) +PRISM [arm64-darwin25] * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN, 4.0: UNKNOWN ---------------------------------------- The code below shows `Data::define.new` treats symbol and string keys equivalently: ``` ruby C = Data.define(:a, :b) C.new(a: 1, b: 1) #=> #<data C a=1, b=1> C.new('a' => 1, 'b' => 1) #=> #<data C a=1, b=1> ``` But it acts differently when detecting missing keywords: ```ruby C.new(a: 1) #=> 'Data#initialize': missing keyword: :b (ArgumentError) C.new('a' => 1) #=> 'Data#initialize': missing keywords: :a, :b (ArgumentError) ``` I feel it should work like this: ```ruby C.new('a' => 1) #=> 'Data#initialize': missing keyword: :b (ArgumentError) ``` I created a PR to fix it: https://github.com/ruby/ruby/pull/15910 -- https://bugs.ruby-lang.org/

6 14

[ruby-core:124286] [Ruby Bug#21789] `IO#flush` doesn't invoke fiber scheduler io_write hook
by noteflakes (Sharon Rosner) 16 May '26

16 May '26

Issue #21789 has been reported by noteflakes (Sharon Rosner). ---------------------------------------- Bug #21789: `IO#flush` doesn't invoke fiber scheduler io_write hook https://bugs.ruby-lang.org/issues/21789 * Author: noteflakes (Sharon Rosner) * Status: Open * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN ---------------------------------------- When calling `IO#flush` or closing an IO with unflushed writes, the flushing is done by calling `rb_io_blocking_region_wait` and eventually doing a `write` system call. When done in a non-blocking fiber with an active fiber scheduler, this will invoke the `#blocking_operation_wait` fiber scheduler hook, which will run the system call in a worker thread, and therefore the flushing bypasses the async I/O capabilities of the fiber scheduler. The proposed fix is to test for the presence of a fiber scheduler in `io_flush_buffer_async` and invoke the `#io_write` fiber scheduler hook with the unflushed part of the IO's write buffer. Associated PR here: https://github.com/ruby/ruby/pull/15609 -- https://bugs.ruby-lang.org/

2 1

[ruby-core:111146] [Ruby master Bug#19172] `ruby_thread_has_gvl_p` is innacurate sometimes -- document or change?
by ivoanjo (Ivo Anjo) 15 May '26

15 May '26

Issue #19172 has been reported by ivoanjo (Ivo Anjo). ---------------------------------------- Bug #19172: `ruby_thread_has_gvl_p` is innacurate sometimes -- document or change? https://bugs.ruby-lang.org/issues/19172 * Author: ivoanjo (Ivo Anjo) * Status: Open * Priority: Normal * ruby -v: (All Ruby versions) * Backport: 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN ---------------------------------------- Howdy 👋! I work for Datadog [on the ddtrace gem](https://github.com/DataDog/dd-trace-rb) and I found a... sharp edge on the internal `ruby_thread_has_gvl_p` API. I am aware that `ruby_thread_has_gvl_p` is documented an experimental API that is exported as a symbol but not present on the VM include files. ### Background In the ddtrace profiling component, we setup a signal handler and then periodically send SIGPROF signals to try to interrupt the running Ruby thread (e.g. the thread that is holding the global VM lock or equivalent). In the signal handler, we need to perform some API calls which are not safe to do without the GVL. So we need to check if the signal handler got called in the thread that has the GVL. ### The issue ```ruby int ruby_thread_has_gvl_p(void) { rb_thread_t *th = ruby_thread_from_native(); if (th && th->blocking_region_buffer == 0) { return 1; } else { return 0; } } ``` In its current implementation, `ruby_thread_has_gvl_p` only checks if the thread has a `blocking_region_buffer` or not. Unfortunately, this means that when called from a thread that lost the GVL but not due to blocking (e.g. via `rb_thread_schedule()`), it can still claim that a thread is holding the GVL when that is not the case. I ran into this issue in https://github.com/DataDog/dd-trace-rb/pull/2415, and needed to find a workaround. ### Next steps Since this is an internal VM API, I'm not sure you'd want to change the current behavior, so I was thinking of perhaps two options: * Is it worth changing `ruby_thread_has_gvl_p` to be accurate in the case I've listed? * If not, would you accept a PR to document its current limitations, so that others don't run into the same issue I did? -- https://bugs.ruby-lang.org/

3 3

[ruby-core:111678] [Ruby master Feature#19315] Lazy substrings in CRuby
by Eregon (Benoit Daloze) 13 May '26

13 May '26

Issue #19315 has been reported by Eregon (Benoit Daloze). ---------------------------------------- Feature #19315: Lazy substrings in CRuby https://bugs.ruby-lang.org/issues/19315 * Author: Eregon (Benoit Daloze) * Status: Open * Priority: Normal ---------------------------------------- CRuby should implement lazy substrings, i.e., "abcdef"[1..3] must not copy bytes. Currently CRuby only reuse the char* if the substring is until the end of the buffer. But it should also work wherever the substring starts and ends. Yes, it means RSTRING_PTR() might need to allocate to \0-terminate, so be it, it's worth it. There is already code for this (`SHARABLE_MIDDLE_SUBSTRING`), but it's disabled by default and `RSTRING_PTR()` needs to be changed to deal with this. It seems a good idea to introduce a variant of `RSTRING_PTR` which doesn't guarantee \0-termination, so such callers can then use the existing bytes always without copy. There are countless workarounds for this missing optimization, all not worth and all less readable: * https://bugs.ruby-lang.org/issues/19314 * https://github.com/ruby/net-protocol/pull/14 * Manual lazy substrings which track string + index + length * More but I don't remember all now, feel free to comment or link more urls/tickets. -- https://bugs.ruby-lang.org/

8 13

[ruby-core:123484] [Ruby Bug#21640] Core Pathname is missing 3 methods / is partially-defined
by Eregon (Benoit Daloze) 13 May '26

13 May '26

Issue #21640 has been reported by Eregon (Benoit Daloze). ---------------------------------------- Bug #21640: Core Pathname is missing 3 methods / is partially-defined https://bugs.ruby-lang.org/issues/21640 * Author: Eregon (Benoit Daloze) * Status: Open * ruby -v: ruby 3.5.0dev (2025-10-09T08:06:20Z master a29c90c3b0) +PRISM [x86_64-linux] * Backport: 3.2: UNKNOWN, 3.3: UNKNOWN, 3.4: UNKNOWN ---------------------------------------- ``` $ ruby -e 'puts Pathname.instance_methods(false).sort; puts Pathname.singleton_methods.sort' > core_pathname_methods.txt $ ruby -rpathname -e 'puts Pathname.instance_methods(false).sort; puts Pathname.singleton_methods.sort' > require_pathname_methods.txt $ diff core_pathname_methods.txt require_pathname_methods.txt 36a37 > find 72a74 > rmtree 98a101 > mktmpdir ``` So `#find`, `#rmtree` and `.mktmpdir` are missing from core Pathname. And indeed, they give NoMethodError, e.g. ``` $ ruby -ve 'Pathname.new("doesnotexist").rmtree' ruby 3.5.0dev (2025-10-09T08:06:20Z master a29c90c3b0) +PRISM [x86_64-linux] -e:1:in '<main>': undefined method 'rmtree' for an instance of Pathname (NoMethodError) ``` I think this is confusing and unexpected for most users. Either Pathname should be fully defined or not at all, having a partially-defined Pathname seems particularly confusing (a bit like an autoload that failed in the middle or so). AFAIK the only core class which used to have this was Fiber with `alive?` and `transfer`, that certainly caused its share of confusion, and Fiber has since been fixed, now even without `require "fiber"` Fiber has all methods. Furthermore these 3 methods are not documented as needing `require 'pathname'` in their documentation: https://github.com/ruby/ruby/blob/master/lib/pathname.rb The reason these 3 methods are not in core pathname seems to be (from [this comment](https://github.com/ruby/pathname/issues/64#issuecomment-3388907046)) > That's consensus with akr and me. We should avoid loading other libraries by simply calling methods from the embedded core classes. I understand that concern and I share it, but I think the situation of having a partially-defined Pathname in core is quite problematic. I think there are 2 solutions: * Define all Pathname methods in core. The `require` are done lazily inside the methods so AFAIK there is no technical issue blocking that, but it's not ideal design-wise that a core class can load default gems (i.e. give up that concern for this particular case). * Do not define Pathname in core and let it be a default gem as it was in Ruby 3.4. I discuss the second in more details, because I found more problems than I would have imagined with core Pathname: ## Should Pathname be in core? #17473 proposed to make Pathname core. That sounded great to me at first thought, but as I realize the practical problems with it I'm thinking it's actually better to keep Pathname non-core and as a default gems. Here are concerns with having Pathname in core: 1. Pathname inherently depends on other default gems/stdlib like `find`, `fileutils` and `tmpdir`. This is important to make Pathname useful. The docs even phrase it like this: `ri Pathname` > ... > All functionality from File, FileTest, and some from Dir and > FileUtils is included, in an unsurprising way. It is essentially a > facade for all of these, and more. 2. core Pathname is missing 3 methods compared to gem Pathname (shown above) 3. the pathname gem is likely to gain extra features, methods and bug fixes. For example I see [4 PRs](https://github.com/ruby/pathname/pulls) adding new methods. It means core Pathname will always be outdated and potentially missing some new methods. 4. One can still use the pathname gem even if it's in core, but surprisingly just `require 'pathname'` is not enough: ``` $ gem install pathname Building native extensions. This could take a while... Successfully installed pathname-0.4.0 1 gem installed $ ruby -e 'require "pathname"; puts $".grep(/pathname/)' pathname.so /home/eregon/prefix/ruby-master/lib/ruby/3.5.0+4/pathname.rb ^ wrong that's the stdlib pathname $ ruby -e 'gem "pathname"; require "pathname"; puts $".grep(/pathname/)' # some warnings, reported in https://github.com/ruby/pathname/issues/66 pathname.so /home/eregon/prefix/ruby-master/lib/ruby/gems/3.5.0+4/gems/pathname-0.4.0/lib/pathname.rb ^ correct ``` This is another source of confusion caused by core pathname. It even happens with Bundler! ``` $ cat Gemfile source "https://rubygems.org" gem "pathname" $ bundle install $ ruby -e 'require "bundler/setup"; require "pathname"; puts $".grep(/pathname/); puts; puts $:' pathname.so /home/eregon/prefix/ruby-master/lib/ruby/3.5.0+4/pathname.rb /home/eregon/tmp/bundler-pathname/vendor/bundle/ruby/3.5.0+4/gems/pathname-0.4.0/lib /home/eregon/tmp/bundler-pathname/vendor/bundle/ruby/3.5.0+4/extensions/x86_64-linux/3.5.0+4-static/pathname-0.4.0 /home/eregon/prefix/ruby-master/lib/ruby/gems/3.5.0+4/gems/bundler-2.8.0.dev/lib /home/eregon/prefix/ruby-master/lib/ruby/site_ruby/3.5.0+4 /home/eregon/prefix/ruby-master/lib/ruby/site_ruby/3.5.0+4/x86_64-linux /home/eregon/prefix/ruby-master/lib/ruby/site_ruby /home/eregon/prefix/ruby-master/lib/ruby/vendor_ruby/3.5.0+4 /home/eregon/prefix/ruby-master/lib/ruby/vendor_ruby/3.5.0+4/x86_64-linux /home/eregon/prefix/ruby-master/lib/ruby/vendor_ruby /home/eregon/prefix/ruby-master/lib/ruby/3.5.0+4 /home/eregon/prefix/ruby-master/lib/ruby/3.5.0+4/x86_64-linux $ bundle exec ruby -e 'require "pathname"; puts $".grep(/pathname/); puts; puts $:' same as above ``` I'm not sure why, but that seems a serious bug that the `pathname` gem apparently can't be used at all with Bundler. 4. due to this difference in terms of methods for core & gem pathname it seems complicated to keep core Pathname and gem Pathname in sync, since e.g. everything is naturally in one `.rb` file in the gem, but in two `.rb` files in core 5. the pathname gem has to `remove_const :Pathname` to avoid conflicts and warnings in https://github.com/ruby/pathname/blob/4689b0b78d081ae855f325e086d95803fa5bd…. This is bad for Ruby JITs, especially if core Pathname methods might have been used, as it will invalidate JITs compilations, invalidate some inline caches (also bad for the interpreter), caused the JIT to do more compilations which means slower warmup, makes it much harder to persist JITed code across process executions, etc. Based on all of these I think it would actually be better to not have Pathname in core, and let it be a default gem. Having to `require "pathname"` as one always needed to do so far seems far better than a partially-defined core Pathname with rough edge cases (those concerns). -- https://bugs.ruby-lang.org/

7 24