Issue #15599 has been updated by wanabe (_ wanabe). Status changed from Open to Closed This issue was fixed between preview 2 and preview 3 of Ruby 3.3.0. ``` $ cat a.rb class A def a1() end end $ cat base.rb 1000.times do fork do autoload :A, './a' t1 = Thread.new { raise if A.instance_methods(false).size != 1 } t2 = Thread.new { require './a' } begin t1.join t2.join rescue Exception => e exit 1 end end end p Process.waitall.map {|_pid, status| status.success? ? :success : :failure }.group_by(&:itself).transform_values(&:size).to_a puts :fin $ docker run -it --rm -v.:/work -w/work --env ALL_RUBY_SINCE=3.2.10 rubylang/all-ruby /all-ruby/all-ruby base.rb ruby-3.2.10 #<Thread:0x00007119bd008e88 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x00007119bd008e88 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x00007119bd008e88 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x00007119bd008e88 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception [[:success, 921], [:failure, 79]] fin ruby-3.2.11 #<Thread:0x00007eb67a448f68 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception [[:success, 922], [:failure, 78]] fin ruby-3.3.0-preview1 #<Thread:0x00007429d36e67e0 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception [[:success, 931], [:failure, 69]] fin ruby-3.3.0-preview2 #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception #<Thread:0x0000751915115f78 base.rb:5 run> terminated with exception (report_on_exception is true): base.rb:5:in `block (3 levels) in <main>': unhandled exception [[:success, 904], [:failure, 96]] fin ruby-3.3.0-preview3 [[:success, 1000]] fin ... ruby-4.0.6 [[:success, 1000]] fin ``` I found the commit is be1bbd5b7d40ad863ab35097765d3754726bbd54 == https://bugs.ruby-lang.org/issues/19842. I don't know why this is the case, but I suspect that the refactoring done to achieve the original goal (M:N threads) had a positive effect. ``` $ git checkout be1bbd5b7d40ad863ab35097765d3754726bbd54~ && make -j miniruby >/dev/null 2>&1 && ./miniruby base.rb HEAD is now at 096ee0648e2 Bump ruby/setup-ruby from 1.155.0 to 1.156.0 [[:success, 945], [:failure, 55]] fin $ git checkout be1bbd5b7d40ad863ab35097765d3754726bbd54 && make -j miniruby >/dev/null 2>&1 && ./miniruby base.rb Previous HEAD position was 096ee0648e2 Bump ruby/setup-ruby from 1.155.0 to 1.156.0 HEAD is now at be1bbd5b7d4 M:N thread scheduler for Ractors [[:success, 1000]] fin ``` ---------------------------------------- Bug #15599: Mixing autoload and require causes deadlock and incomplete definition. https://bugs.ruby-lang.org/issues/15599#change-118480 * Author: akr (Akira Tanaka) * Status: Closed * ruby -v: ruby 2.7.0dev (2019-02-11 trunk 67049) [x86_64-linux] * Backport: 2.4: UNKNOWN, 2.5: UNKNOWN, 2.6: UNKNOWN ---------------------------------------- I found that mixing autoload and require causes deadlock and incomplete definition. ``` % cat a.rb class A def a1() end end % cat base.rb autoload :A, './a' t1 = Thread.new { p A.instance_methods(false) } t2 = Thread.new { require './a' } t1.join t2.join % ruby base.rb Traceback (most recent call last): 1: from base.rb:6:in `<main>' base.rb:6:in `join': No live threads left. Deadlock? (fatal) 3 threads, 3 sleeps current:0x000055cc6943fde0 main thread:0x000055cc6935f4b0 * #<Thread:0x000055cc6938f190 sleep_forever> rb_thread_t:0x000055cc6935f4b0 native:0x00007f58d256eb40 int:0 base.rb:6:in `join' base.rb:6:in `<main>' * #<Thread:0x000055cc6968cfc8@base.rb:3 sleep_forever> rb_thread_t:0x000055cc69736180 native:0x00007f58ce7b3700 int:0 mutex:0x000055cc6943fde0 cond:1 depended by: tb_thread_id:0x000055cc6935f4b0 /home/akr/ruby/o0/lib/ruby/2.7.0/rubygems/core_ext/kernel_require.rb:54:in `require' /home/akr/ruby/o0/lib/ruby/2.7.0/rubygems/core_ext/kernel_require.rb:54:in `require' base.rb:3:in `block in <main>' * #<Thread:0x000055cc6968ccd0@base.rb:4 sleep_forever> rb_thread_t:0x000055cc6943fde0 native:0x00007f58ce5b1700 int:0 /tmp/a/a.rb:1:in `<top (required)>' /home/akr/ruby/o0/lib/ruby/2.7.0/rubygems/core_ext/kernel_require.rb:54:in `require' /home/akr/ruby/o0/lib/ruby/2.7.0/rubygems/core_ext/kernel_require.rb:54:in `require' base.rb:4:in `block in <main>' % ruby base.rb [:a1] % ruby base.rb [] ``` The last run which prints [] means incomplete definition of A which a1 method is not defined. -- https://bugs.ruby-lang.org/