On Fri, Nov 27, 2020 at 10:44:20PM +0100, Manuel Bouyer wrote:
> On Fri, Nov 27, 2020 at 09:22:11PM +0100, Roger Pau Monné wrote:
> > > 
> > > But I can confirm that now, entering ^A^A^A gets interrupt going in again
> > 
> > I think there are some weird things with dpci interrupts that I'm
> > trying to understand. I have a patch now that will panic when the
> > buffer is full, so we will hopefully be able to see the whole trace of
> > events. There will be no need for you to press the 'T' key now, the
> > system will panic when the buffer is full.
> > 
> > Note this patch also removes the deassert done in pt_irq_time_out.
> 
> thanks
> the trace is at
> http://www-soc.lip6.fr/~bouyer/xen-log13.txt

Thanks! I think I've found the issue and I'm attaching a possible fix
(fix.patch) to this email. In any case I've also attached a further
debug patch, in case the fix turns out to be wrong. Please test the
fix first, as the debug patch will end up triggering a panic when the
buffer is full.

Roger.