On 01/10/21 18:47, Valentin Schneider wrote: > On 30/09/21 00:10, Frederic Weisbecker wrote: >> PREEMPT_RT has made rcu_core() preemptible, making it unsafe against >> concurrent NOCB (de-)offloading. >> >> Thomas suggested to drop the local_lock() based solution and simply >> check the offloaded state while context looks safe but that's not >> enough. Here is a bit of rework. >> >> git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git >> rcu/rt >> >> HEAD: aac1c58961446c731f2e989bd822ca1fd2659bad >> >> Thanks, >> Frederic > > FWIW I've had RCU torture with NOCB toggling running for half a day on my > Arm Juno and nothing to report. I still need to rebase this on an -rt tree > and give it a spin with CONFIG_PREEMPT_RT. Rebased on top of v5.15-rc4-rt7-rebase with Thomas' patch reverted and ran the same thing under CONFIG_PREEMPT_RT, nothing seems to catch on fire, so: Tested-by: Valentin Schneider <valentin.schneider@xxxxxxx>
