Re: [kvmarm] [RFC PATCH 0/3] KVM: ARM: Get rid of hardcoded VGIC addresses

[Jan Kiszka <jan.kiszka@siemens.com>]

On 2012-10-23 12:52, Peter Maydell wrote:
> On 23 October 2012 11:48, Jan Kiszka <jan.kiszka@siemens.com> wrote:
>> The current irqchip API is like this:
>>
>> KVM_CREATE_IRQCHIP (without any parameters)
>> ...
>> KVM_CREATE_VCPU
>> KVM_SET_IRQCHIP (or the other way around)
>> ...
>> KVM_RUN
>>
>> The arguments you cannot pass via KVM_CREATE_IRQCHIP - which is more
>> like a "Hey, there will be an IRQ chip!" - could be passed via
>> KVM_SET_IRQCHIP (it has 512 bytes space). Provided there are sane
>> configuration defaults, at least after KVM_CREATE_VCPU, KVM_SET_IRQCHIP
>> becomes optional. Then you don't need the a check on KVM_RUN.
> 
> KVM_SET_IRQCHIP is the state load ioctl, right (companion to
> KVM_GET_IRQCHIP)? It seems like a bit of an abuse to use that
> for configuration rather than for migration/sync of state
> with userspace...

Depends on how reconfigurable your chip is. x86 also sends the mapping
down this way, which happens to be guest configurable but is practically
static.

> 
> (For ARM we will just use the ONE_REG ABI for irqchip state
> save/load anyway, so KVM_SET_IRQCHIP isn't relevant.)

The less problems you should have using the SET interface to perform
one-time configuration.

BTW, I guess we will regret that one-reg ABI one day and have to
introduce a multi-reg version again for hot-standby, i.e. continuous
state migration. I know we also do this for c86 MSRs - that interface
has the same limitation.

Jan

-- 
Siemens AG, Corporate Technology, CT RTC ITP SDP-DE
Corporate Competence Center Embedded Linux