Re: [PATCH 23/42] KVM: Activate Virtualization On Demand

[Alexander Graf]

On 16.08.2010, at 15:49, Serge Belyshev wrote:

> Alexander Graf <agraf@xxxxxxx> writes:
> 
>> Hrm - try to use the following (probably whitespace broken and 100% untested) hacky patch:
>> 
>> diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
>> index 56c9b6b..bde9ee3 100644
>> --- a/arch/x86/kvm/svm.c
>> +++ b/arch/x86/kvm/svm.c
>> @@ -429,9 +429,11 @@ static int svm_hardware_enable(void *garbage)
>>        struct desc_struct *gdt;
>>        int me = raw_smp_processor_id();
>> 
>> +#if 0
>>        rdmsrl(MSR_EFER, efer);
>>        if (efer & EFER_SVME)
>>                return -EBUSY;
>> +#endif
>> 
> 
> I tested a bit different hack keeping rdmsrl but removing the if() and
> it works fine for me:
> 
> Index: linux/arch/x86/kvm/svm.c
> ===================================================================
> --- linux.orig/arch/x86/kvm/svm.c
> +++ linux/arch/x86/kvm/svm.c
> @@ -429,8 +429,7 @@ static int svm_hardware_enable(void *gar
> 	int me = raw_smp_processor_id();
> 
> 	rdmsrl(MSR_EFER, efer);
> -	if (efer & EFER_SVME)
> -		return -EBUSY;
> +	printk(KERN_DEBUG "svm_hardware_enable: efer %x on %d\n", efer, me);

Ouch, yes. Thanks for not taking my patch ;). The efer read is of course still mandatory.

> 
> 	if (!has_svm()) {
> 		printk(KERN_ERR "svm_hardware_enable: err EOPNOTSUPP on %d\n",
> 
> Here is an example of its output:
> 
> [231441.358111] svm_hardware_enable: efer 1d01 on 0
> [231441.358117] svm_hardware_enable: efer d01 on 3
> [231441.358124] svm_hardware_enable: efer d01 on 2
> [231441.358135] svm_hardware_enable: efer d01 on 1
> [231675.168967] svm_hardware_enable: efer d01 on 0
> [231675.168972] svm_hardware_enable: efer d01 on 1
> [231675.168983] svm_hardware_enable: efer d01 on 2
> [231675.168995] svm_hardware_enable: efer d01 on 3
> 
> 
> After reboot, the first kvm run always reports efer = 1d01 on the first cpu.
> In all subsequent runs before reboot efer stays d01 on all cpus.

Happy broken BIOS fun. Oh well, I guess you can live with the patched version and normal end-users will just update their BIOSes.


Alex

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/