Re: [patch] KVM: do VMXOFF upon reboot

[Avi Kivity <avi@qumranet.com>]

Ingo Molnar wrote:
> * Ingo Molnar <mingo@elte.hu> wrote:
>
>   
>> So i think we should do the patch below - this makes reboot work even 
>> in atomic contexts. [...]
>>     
>
> hm, this causes problems if KVM is not active on a VT-capable CPU: even 
> on CPUs with VT supported, if a VT context is not actually activated, a 
> vmxoff causes an invalid opcode exception. So the updated patch below 
> uses a slightly more sophisticated approach to avoid that problem.
>
>   

There is already code to that effect.  Any idea why it is not called?

> static int kvm_reboot(struct notifier_block *notifier, unsigned long val,
>                        void *v)
> {
>     if (val == SYS_RESTART) {
>         /*
>          * Some (well, at least mine) BIOSes hang on reboot if
>          * in vmx root mode.
>          */
>         printk(KERN_INFO "kvm: exiting hardware virtualization\n");
>         on_each_cpu(kvm_arch_ops->hardware_disable, 0, 0, 1);
>     }
>     return NOTIFY_OK;
> }
>
> static struct notifier_block kvm_reboot_notifier = {
>     .notifier_call = kvm_reboot,
>     .priority = 0,
> };
>

Note that it performs the vmxoff on all cpus, not just one, and that it 
is svm friendly too.  Maybe it should check for values other than 
SYS_RESTART?


-- 
error compiling committee.c: too many arguments to function