the wake_up from a non threaded isr in RT will end up to try to get the spin_lock of of that wait_queue, but this kind of locking in not allowed in non threaded isr and therefore the use of wake_up_locked does not touches the spin_locks of the queue... and it seems to work On Fri, Jan 27, 2017 at 7:24 PM, Andy Ng <andreas2025@xxxxxxxxx> wrote: > i have replaced wake_up call with wake_up_locked ... and it seems better. > > On Fri, Jan 27, 2017 at 2:09 PM, Andy Ng <andreas2025@xxxxxxxxx> wrote: >> I have activated kprobes and it seems that timer tick and schedule >> don't get called. I have a gpio line to monitor the kprobe entry and >> exit handlers >> >> >> On Fri, Jan 27, 2017 at 2:04 PM, Andy Ng <andreas2025@xxxxxxxxx> wrote: >>> Thank you. However when I turn debug on the problem goes away :( >>> >>> On Thu, Jan 26, 2017 at 4:57 PM, Sebastian Andrzej Siewior >>> <bigeasy@xxxxxxxxxxxxx> wrote: >>>> On 2017-01-25 13:54:15 [+0000], Andy Ng wrote: >>>>> Dear List, >>>>> >>>>> I have an interesting problem on 3.0.35 with the rt56 patch on ARM. >>>>> >>>>> I have some external events that drive my ISR, which I have set it up >>>>> as non_threaded one. >>>>> When the interrupt rate is low, 1.8msec things are going fine, but >>>>> when the rate is higher, the wake_up call from the isr does not >>>>> return. The system is frozen. No Magic SysRq. Nothing. >>>>> >>>>> All will work fine once I change my ISR to be a threaded one. >>>>> >>>>> Any ideas where to search? >>>> >>>> If you you enable lockdep and sleeping-while-atomic warning you should >>>> see quite a few warnings. That would be a good starting point to search. >>>> >>>>> Kind regards, >>>>> Andy >>>> >>>> Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
