signal_struct->count must die, initial changes
cover signal_struct->count must die, initial changes
signal_struct->count in its current form must die.
- it has no reasons to be atomic_t
- it looks like a reference counter, but it is not
- otoh, we really need to make task->signal refcountable, just look at the extremely ugly task_rq_unlock_wait() called from __exit_signals().
- we should change the lifetime rules for task->signal, it should be pinned to task_struct. We have a lot of code which can be simplified after that.
- it is not needed! while the code is correct, any usage of this counter is artificial, except fs/proc uses it correctly to show the number of threads.
This series removes the usage of sig->count from exit pathes.
Oleg.
Quoted 1 line
signal_struct->countin its current form must die.
Agreed. This series
Quoted 3 lines
- otoh, we really need to make
task->signalrefcountable, just look at the extremely uglytask_rq_unlock_wait()called from__exit_signals().
Another of many good reasons for...
Quoted 3 lines
- we should change the lifetime rules for
task->signal, it should be pinned totask_struct. We have a lot of code which can be simplified after that.
Yes, and sighand too. I think there will be quite a torrent of cleanup we can do after that.
Remember too that we should really change de_thread() altogether. It has long-standing signal-losing races that we should fix. Changing the lifetime rules may complicate that, or not. But they'll tie in together, and when we do that we'll replace the notify_count again with something else entirely.
Thanks,
Roland
Quoted 1 line · Oleg Nesterov
signal_struct->countin its current form must die.
Ack on this series.
--
Veaceslav