ruby-changes:51848
From: normal <ko1@a...>
Date: Thu, 26 Jul 2018 17:30:15 +0900 (JST)
Subject: [ruby-changes:51848] normal:r64062 (trunk): cont.c (ec_switch): prevent delayed/missed trap interrupt race
normal 2018-07-26 17:30:10 +0900 (Thu, 26 Jul 2018) New Revision: 64062 https://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=revision&revision=64062 Log: cont.c (ec_switch): prevent delayed/missed trap interrupt race timer-thread may set trap interrupt with rb_threadptr_check_signal at any time independent of GVL. This means timer-thread may set the trap interrupt flag on the previous execution context; causing the flag to be unnoticed until a future ec switch (or lost completely if the ec is done). Note: I avoid relying on th->interrupt_lock here and use atomics because we won't be able to rely on it for proposed lazy timer-thread [Misc #14937]. This regression affects Ruby 2.5 as it was introduced by moving interrupt_flag to `ec' which is an unstable pointer. Ruby <= 2.4 was unaffected because vm->main_thread->interrupt_flag never changed. [ruby-core:88119] [Bug #14939] Modified files: trunk/cont.c Index: cont.c =================================================================== --- cont.c (revision 64061) +++ cont.c (revision 64062) @@ -262,7 +262,17 @@ static inline void https://github.com/ruby/ruby/blob/trunk/cont.c#L262 ec_switch(rb_thread_t *th, rb_fiber_t *fib) { rb_execution_context_t *ec = &fib->cont.saved_ec; + ruby_current_execution_context_ptr = th->ec = ec; + + /* + * timer-thread may set trap interrupt on previous th->ec at any time; + * ensure we do not delay (or lose) the trap interrupt handling. + */ + if (th->vm->main_thread == th && rb_signal_buff_size() > 0) { + RUBY_VM_SET_TRAP_INTERRUPT(ec); + } + VM_ASSERT(ec->fiber_ptr->cont.self == 0 || ec->vm_stack != NULL); } -- ML: ruby-changes@q... Info: http://www.atdot.net/~ko1/quickml/