[PATCH v2] KVM: arm/arm64: Signal SIGBUS when stage2 discovers hwpoison memory

[gengdongjiu]

Hi James,

On 2017/6/21 17:53, James Morse wrote:
> I think we discussed this before[0], your CPU has a feature called 'hwpoison'
> that is uses to support RAS. Linux also has a feature called 'hwpoison' [1][2],
> which handles the offline-ing of memory pages when it receives a notification
> through APEI. I've tried to call this memory_failure() to avoid this confusion.
> 
> This patch is to handle stage2 faults when the page was removed from the stage2
> mapping by the memory_failure() code. v3 of this patch[3] does a much better job
> of describing this.
> 
> (... I don't think your question is related to this patch ...)

I know your meaning about the Linux 'hwpoison' feature.
Let see the code that how to get the "pfn"

///get the pfn
fault_ipa = kvm_vcpu_get_fault_ipa(vcpu);
gfn = fault_ipa >> PAGE_SHIFT;
pfn = gfn_to_pfn_prot(kvm, gfn, write_fault, &writable);


As shown in above code, when happen SEA, the fault_ipa is got from the HPFAR_EL2 register.
if the HPFAR_EL2 does not record the IPA. the fault_ipa is zero, then gfn is zero, so the pfn is unknown.

so below judgement always false although firmware notify the memory_failure through APEI, because we do not get the right fault memory page.
using this API "kvm_vcpu_get_fault_ipa" can not get the right fault memory page if cpu does not update the HPFAR_EL2.

+	if (pfn == KVM_PFN_ERR_HWPOISON) {
+		kvm_send_hwpoison_signal(hva, vma);
+		return 0;
+	}

so may be you need to double confirm that whether armv8.0/armv8.2  standard CPU can always update the  HPFAR_EL2 registers.