Re: [PATCH v3] xen/arm: vgic to ignore GICD ICPENDRn registers access

[Ian Jackson <iwj@xxxxxxxxxxxxxx>]

~Bertrand Marquis writes ("Re: [PATCH v3] xen/arm: vgic to ignore GICD ICPENDRn 
registers access"):
> [+Ian]
> > On 20 Oct 2021, at 11:10, Hongda Deng <Hongda.Deng@xxxxxxx> wrote:
> > 
> > Currently, Xen will return IO unhandled when guests access GICD ICPENRn
> > registers. This will raise a data abort inside guest. For Linux Guest,
> > these virtual registers will not be accessed. But for Zephyr, in its
> > GIC initialization code, these virtual registers will be accessed. And
> > zephyr guest will get an IO data abort in initilization stage and enter
> > fatal error. Emulating ICPENDR is not easy with the existing vGIC, so
> > we currently ignore these virtual registers access and print a message
> > about whether they are already pending instead of returning unhandled.
> > More details can be found at [1].
> > 
> > [1] https://github.com/zephyrproject-rtos/zephyr/blob/eaf6cf745df3807e6e
> > cc941c3a30de6c179ae359/drivers/interrupt_controller/intc_gicv3.c#L274
> > 
> > Signed-off-by: Hongda Deng <hongda.deng@xxxxxxx>
> Reviewed-by: Bertrand Marquis <bertrand.marquis@xxxxxxx>
> 
> Ian this is fixing a bug in the vgic implementation which is preventing to run
> Zephyr as a guest on top of Xen. Xen support has now been merged in Zephyr
> so this is required to use it.
> 
> Could we consider adding this patch into the 4.16 release ?

Hi.  I'm definitely open to the idea, especially if it goes in soon.

I think this needs an ARM maintainer review, still ?

It doesn't seem entirely straightforward.  I'd like to hear from the
maintainer, to confirm that they agree it's a bugfix, and to get an
idea of the risks vs benefits of this patch.

Thanks,
Ian.