On Tue, 2019-10-08 at 15:08 +0200, Petr Mladek wrote: > On Mon 2019-10-07 10:59:10, Qian Cai wrote: > > It is almost impossible to eliminate all the indirect call chains from > > console_sem/console_owner_lock to zone->lock because it is too normal that > > something later needs to allocate some memory dynamically, so as long as it > > directly call printk() with zone->lock held, it will be in trouble. > > It is not normal the something needs to allocate memory under > console_sem. Console drivers are supposed to get the message > out even when the system is in really bad state and it is not > possible to allocate memory. I consider this a bug in the console > driver.
Again, it is not directly under console_sem. It is *indirect*. console_sem --> lockA lockA --> lockB lockB --> lockC Anything allocating memory with lockB or lockC held will form the problematic lock chain to trigger the lockdep splat with memory offline.