On 01/21/2017 08:37 AM, Jan Kiszka wrote:
> On 2017-01-19 21:11, Ralf Ramsauer wrote:
>> Hi,
>>
>> I implemented a sysfs interface for the hypervisor console. This is useful 
>> for
>> debugging devices without UARTs. At least as long as the root-cell doesn't
>> crash...
>>
>> Additionally, I implemented continuous reading of the hypervisor console via
>> /dev/jailhouse, which is a pretty nice and easy way to follow the hypervisor
>> console from Linux.
>>
>> I use a page in the hypervisor's memory space as ringbuffer. If the root cell
>> has a certain permission enabled in its config, the hypervisor won't back 
>> that
>> page with empty_page. The Linux driver may then read from that page without 
>> any
>> interactions with the hypervisor. We do this without any signalling or 
>> locking.
>> Otherwise we would risk starvation of the hypervisor. The producer 
>> (hypervisor
>> debug write) only indicates ongoing writes by setting a flag.
>>
>> tree:
>> https://github.com/lfd/jailhouse/tree/hvconsoleV4
> 
> Started to play with it, looks good, but I need to check in more details
> again the next days or so.
Great, thanks.
> 
>>
>> PS: maybe we could think of enabling JAILHOUSE_CELL_DEBUG_CONSOLE as a 
>> default
>> flag for our demo inmate cell configuration. As Jan already pointed out
>> earlier, none of them is permissive.
> 
> I don't get this yet: which configs do you want to enable, and why? The
> linux inmates?
Yes, and other inmates as well. At the moment, inmates are not allowed
to use the debug console by default. The virtual jailhouse console is
kind of a debug feature as well. Currently we only get Jailhouse messages.

  Ralf
> 
> Jan
> 

-- 
You received this message because you are subscribed to the Google Groups 
"Jailhouse" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to