kotovalexarian-likes-github/ruby--ruby
1
0
Fork 0
mirror of https://github.com/ruby/ruby.git synced 2022-11-09 12:17:21 -05:00
Code Releases Activity

thread_pthread.c: fix deadlock on test_thread.rb::test_signal_at_join

Browse source 
Fixes: r64575 ("avoid lock ping-pong in do_gvl_timer & ubf_select")

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@64579 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This commit is contained in:
normal 2018-08-27 23:29:44 +00:00
parent 8500714adc
commit 1b1274f728
1 changed files with 6 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
thread_pthread.c
View file

@ -1320,6 +1320,7 @@ ubf_select(void *ptr)
{
rb_thread_t *th = (rb_thread_t *)ptr;
rb_vm_t *vm = th->vm;
const rb_thread_t *cur = ruby_thread_from_native(); /* may be 0 */
register_ubf_list(th);
@ -1328,9 +1329,12 @@ ubf_select(void *ptr)
* Therefore, we repeatedly call ubf_wakeup_thread() until a target thread
* exit from ubf function. We must have a timer to perform this operation.
* We use double-checked locking here because this function may be called
* while vm->gvl.lock is held in do_gvl_timer
* while vm->gvl.lock is held in do_gvl_timer.
* There is also no need to start a timer if we're the designated
* sigwait_th thread, otherwise we can deadlock with a thread
* in unblock_function_clear.
*/
if (vm->gvl.timer != ruby_thread_from_native()) {
if (cur != vm->gvl.timer && cur != sigwait_th) {
rb_native_mutex_lock(&vm->gvl.lock);
if (!vm->gvl.timer) {
rb_thread_wakeup_timer_thread(-1);