Re: [Xen-devel] [Very RFC PATCH] Livepatch - initial ARM64/32 support.

[Julien Grall <julien.grall@xxxxxxx>]

Hi

On 11/08/2016 21:05, Stefano Stabellini wrote:
> On Thu, 11 Aug 2016, Julien Grall wrote:
> > On 09/08/2016 06:18, Konrad Rzeszutek Wilk wrote:
> > > Hey!
> > Hi Konrad,
> >
> > > Over the last couple of months in my spare time I was playing
> > > with making livepatch work with ARM64 (using the FoundationModel
> > > simulator) and I finally got it working tonight.
> > Congrats!
> Indeed, congratulations! I hope it was fun :-)
>
>
> > > Sending out the patches just in case they don't work tomorrow :-)
> > >
> > > The ARM32 part is going slowly - as I don't have a simulator
> > > and using a real board takes time.
> > Whilst I can see usage of livepatch for Xen ARM64 (e.g server), I am less
> > convinced for ARM32. I am bit worry to check-in code that will get rotten in
> > the long term. So do you see any usage on ARM32?
> Actually I think it is useful for embedded use cases too: think about
> security updates for any embedded devices which cannot easily be
> rebooted. Or even if they can be rebooted, it might create a
> competitive advantage to build a car that can be updated for critical
> security fixes without requesting the driver to stop.
I see your point, although I think updating Xen while driving sounds 
really risky :). There are still a small window (few ms/ns) where 
nothing can run. That might be an issue with real-time operating system.
This leads to few questions as I am not very familiar with livepatching. 
Who decide when to patch? Is it the scheduler or the user requesting the 
patching?
Also, could this be aborted because a task with higher priority is incoming?

Regards,

--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel