Re: IRQ behaivour has been changed from v4.14 to v4.15-rc1

[Thomas Gleixner]

On Thu, 28 Dec 2017, Thomas Gleixner wrote:
> On Thu, 28 Dec 2017, Andy Shevchenko wrote:
> > On Thu, 2017-12-28 at 21:18 +0100, Thomas Gleixner wrote:
> > > Yes, you missed the typo in the command line. It should be:
> > > 
> > >  'trace_event=irq_vectors:* ftrace_dump_on_oops'
> > 
> > Indeed.
> > 
> > So, I had to disable LOCAL_TIMER_VECTOR, CALL_FUNCTION_VECTOR and
> > RESCHDULE_VECTOR tracing, otherwise I got a lot of spam and RCU stalls.
> 
> Fair enough.
> 
> > The result w/o above is (full log is available here https://pastebin.com
> > /J5yaTbM9):
> 
> Ok. Which irqs are related to that ISP thingy?
> 
> Are these interrupts MSI?

And looking at the log, I see that you can load the driver successfully and
the trouble starts afterwards when you actually use it.

Can you please enable CONFIG_GENERIC_IRQ_DEBUGFS and after login, check
which interrupt is assigned to that atomisp thingy and then provide the
output of

cat /sys/kernel/debug/irq/irqs/$ATOMISPIRQ

Thanks,

	tglx