On 15/06/2022 22:36, Adrien Thierry wrote: > We've been encountering a BUG: scheduling while atomic while running the > 5.18.0-rt11 kernel on a Qualcomm SoC (see stacktrace below). > > It seems to occur because a spinlock is taken in the PSCI idle code path > in the idle loop. With the RT patchset applied and CONFIG_PREEMPT_RT > enabled, spinlocks can sleep, thus triggering the bug. > > In order to prevent this, replace the generic_pm_domain spinlock by a > raw spinlock. > > [ 2.994433] BUG: scheduling while atomic: swapper/6/0/0x00000002 > [ 2.994439] Modules linked in: > [ 2.994447] [<ffff80000810b0ec>] migrate_enable+0x3c/0x160 > [ 2.994461] CPU: 6 PID: 0 Comm: swapper/6 Not tainted 5.18.0-rt11+ #1 > [ 2.994464] Hardware name: Qualcomm SA8295P ADP (DT) > [ 2.994466] Call trace: > [ 2.994467] dump_backtrace+0xb0/0x120 > [ 2.994473] show_stack+0x1c/0x6c > [ 2.994477] dump_stack_lvl+0x64/0x7c > [ 2.994483] dump_stack+0x14/0x2c > [ 2.994487] __schedule_bug+0xa8/0xc0 > [ 2.994489] schedule_debug.constprop.0+0x154/0x170 > [ 2.994492] __schedule+0x58/0x520 > [ 2.994496] schedule_rtlock+0x20/0x44 > [ 2.994499] rtlock_slowlock_locked+0x110/0x260 > [ 2.994503] rt_spin_lock+0x74/0x94 > [ 2.994505] genpd_lock_nested_spin+0x20/0x30 Hi Adrian, I also hit it now on v6.1 RT kernel. I see no more discussions happened here. Do you have any progress on your side for this issue? Best regards, Krzysztof
