Hi Just my 2 ct. Am Mittwoch, 8. Januar 2025, 10:48:56 CET schrieb Florian Paul Schmidt: > Hi! > > On 1/7/2025 3:39 PM, Sebastian Andrzej Siewior wrote: > > On 2024-12-20 14:18:43 [+0100], Florian Paul Schmidt wrote: > [...] > > > The backtrace appears long but everything up to folios_put_refs() > > appears preemtible. Then there is the interrupt handler. This looks > > okay. > > That was my impression as well. Nothing really surprising (at least for > my limited knowledge in these matters.) > > > Based on this trace, I don't see a section which disables preemption/ > > interrupts and be the reason for the interrupt happening. > > > > One thing that could delay your doing are small caches. > > Then there is looking within the system: An inaccurate/ low resolution > > clocksource and/ or timer device. It could also be the TLB infrastructure. I found that these events can also be very time consuming, probably combined with memory contention? I think this might be traced with performance counters but i never looked into this further than musing about the possibility. Best regards Tim
