Re: RCU stall

From: Bart Van Assche
Date: Thu Mar 24 2016 - 16:24:29 EST

On 03/22/2016 07:29 PM, Paul E. McKenney wrote:
Note that a soft lockup triggered at 10509.568010, well before the RCU
CPU stall warning.. And you have a second soft lockup at 10537.567212,
with the same funtion scsi_request_fn() at the top of the stack in both
stack traces. That function has a nice big "for (;;)" loop that does
not appear to have any iteration-limiting mechanism.

Hello Paul,

Your feedback is really appreciated. I have started with testing the four attached patches. With the tests I ran so far these patches were sufficient to avoid any soft lockup or RCU stall complaints. I will submit these patches to the appropriate maintainers once I have finished testing these patches.