On 2019-10-24 14:57:16 [-0700], David Miller wrote: > From: Daniel Wagner <dwagner@xxxxxxx> > Date: Fri, 18 Oct 2019 10:28:17 +0200 > > > handle_simple_irq() expect interrupts to be disabled. The USB > > framework is using threaded interrupts, which implies that interrupts > > are re-enabled as soon as it has run. > ... > > Where are we with this patch? I'm tossing it. > > It seems Sebastian made a suggestion, someone else said his suggestion > should be tried, then everything died. > > Please follow up and post when something is ready. My suggestion with the nested-handler was nonsense. The suggestion with the local_irq_disable() before invoking generic_handle_irq() did avoid the warning but the NFS-root was not stable (and Daniel reported a lot of USB interrupts coming which indicates that the interrupt-handler is not acknowledging the interrupt properly). It works by reverting the IRQ-domain code as this patch does. "Works" as in "no warnings" and "NFS-root looks stable". You have two ACKs on that patch. I could ask Daniel to repost the patch with additional informations. My last information from Daniel was that the rpi tree works and I'm not sure if he is looking for the missing ingredient or preferring the removal of the non-working code. Sebastian
