On 11/07/2014 03:45 PM, Sebastian Reimers wrote: > $ /usr/bin/jackd -R -P89 -dalsa -d hw:0,0 -r48000 -p1024 -n3 > > With: kernel.sched_rt_runtime_us = -1 uhhh. Why do you need this? Once the ISO transfer completes, we get an interrupt. It is likely that the real transfer is not done yet so poll via a timer. Once the transfer is complete we invoke that completion handler which will schedule the tasklet which could lead to another delay if some interrupts fire up… There should be two back traces. > [ 79.236750] cppi41_irq() q 138 desc 00000000 One where we got an interrupt but the descriptor was gone. > [ 83.768813] ------------[ cut here ]------------ > [ 83.773711] WARNING: CPU: 0 PID: 278 at drivers/dma/cppi41.c:612 > cppi41_dma_control+0x2ac/0x2d8() > [ 83.828893] [<c02c9790>] (cppi41_dma_control) from [<c0391068>] and here we try to cancel a transfer but the descriptor is not coming out. And we had a complete transfer without a transfer… > [ 83.858292] [<c038ca5c>] (musb_urb_dequeue) from [<c0374ff0>] > [ 83.921365] [<c03ee79c>] (snd_pcm_do_stop) from [<c03eda30>] > [ 83.975034] [<c00ef134>] (SyS_ioctl) from [<c000e440>] and the transfer is canceled because jack assumed it took too long to complete? You didn't stop it on your own right? Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
