Message-ID: <58DA41B0.8080203@arm.com>
Date: Tue, 28 Mar 2017 11:57:52 +0100
From: James Morse <james.morse@arm.com>
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Icedove/31.6.0
MIME-Version: 1.0
To: Tyler Baicar <tbaicar@codeaurora.org>
CC: christoffer.dall@linaro.org, marc.zyngier@arm.com, pbonzini@redhat.com,
        rkrcmar@redhat.com, linux@armlinux.org.uk, catalin.marinas@arm.com,
        will.deacon@arm.com, rjw@rjwysocki.net, lenb@kernel.org,
        matt@codeblueprint.co.uk, robert.moore@intel.com, lv.zheng@intel.com,
        nkaje@codeaurora.org, zjzhang@codeaurora.org, mark.rutland@arm.com,
        akpm@linux-foundation.org, eun.taik.lee@samsung.com,
        sandeepa.s.prabhu@gmail.com, labbott@redhat.com, shijie.huang@arm.com,
        rruigrok@codeaurora.org, paul.gortmaker@windriver.com, tn@semihalf.com,
        fu.wei@linaro.org, rostedt@goodmis.org, bristot@redhat.com,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu,
        kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-acpi@vger.kernel.org, linux-efi@vger.kernel.org,
        devel@acpica.org, Suzuki.Poulose@arm.com, punit.agrawal@arm.com,
        astone@redhat.com, harba@codeaurora.org, hanjun.guo@linaro.org,
        john.garry@huawei.com, shiju.jose@huawei.com, joe@perches.com
Subject: Re: [PATCH V13 10/10] arm/arm64: KVM: add guest SEA support
References: <1490136425-4324-1-git-send-email-tbaicar@codeaurora.org> <1490136425-4324-11-git-send-email-tbaicar@codeaurora.org>
In-Reply-To: <1490136425-4324-11-git-send-email-tbaicar@codeaurora.org>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2708
Lines: 83

Hi Tyler,

On 21/03/17 22:47, Tyler Baicar wrote:
> Currently external aborts are unsupported by the guest abort
> handling. Add handling for SEAs so that the host kernel reports
> SEAs which occur in the guest kernel.

Looks good,

Can we squash the APEI changes into the patch that added them? This would be one
fewer patches that then need the ACPI maintainer to review.


> diff --git a/arch/arm/kvm/mmu.c b/arch/arm/kvm/mmu.c
> index 962616f..105b6ab 100644
> --- a/arch/arm/kvm/mmu.c
> +++ b/arch/arm/kvm/mmu.c
> @@ -1406,6 +1407,24 @@ static void handle_access_fault(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa)
>  		kvm_set_pfn_accessed(pfn);
>  }
>  
> +static bool is_abort_synchronous(unsigned long fault_status) {

I missed kvm_vcpu_dabt_isextabt() when I suggested we would need a helper (my
fault). Can we use that instead?

(my argument that the unused encodings are reserved doesn't hold if KVM is
already doing this... )


> @@ -1426,23 +1445,31 @@ int kvm_handle_guest_abort(struct kvm_vcpu *vcpu, struct kvm_run *run)
>  	unsigned long hva;
>  	bool is_iabt, write_fault, writable;
>  	gfn_t gfn;
> -	int ret, idx;
> +	int ret, idx, sea_status = 1;
> +
> +	/* Check the stage-2 fault is trans. fault or write fault */
> +	fault_status = kvm_vcpu_trap_get_fault_type(vcpu);
> +
> +	fault_ipa = kvm_vcpu_get_fault_ipa(vcpu);
> +
> +	/* The host kernel will handle the synchronous external abort. There
> +	 * is no need to pass the error into the guest.
> +	 */
> +	if (is_abort_synchronous(fault_status))
> +		sea_status = handle_guest_sea((unsigned long)fault_ipa,
> +				    kvm_vcpu_get_hsr(vcpu));


Why not return from here if the error has been handled?

You use sea_status to skip the next two things that KVM might do, but it goes on
to try and process this, possibly calling user_mem_abort(), surely all this is
unnecessary?


>  
>  	is_iabt = kvm_vcpu_trap_is_iabt(vcpu);
> -	if (unlikely(!is_iabt && kvm_vcpu_dabt_isextabt(vcpu))) {
> +	if (unlikely(!is_iabt && kvm_vcpu_dabt_isextabt(vcpu)) && sea_status) {
>  		kvm_inject_vabt(vcpu);
>  		return 1;
>  	}
>  
> -	fault_ipa = kvm_vcpu_get_fault_ipa(vcpu);
> -
>  	trace_kvm_guest_fault(*vcpu_pc(vcpu), kvm_vcpu_get_hsr(vcpu),
>  			      kvm_vcpu_get_hfar(vcpu), fault_ipa);
>  
> -	/* Check the stage-2 fault is trans. fault or write fault */
> -	fault_status = kvm_vcpu_trap_get_fault_type(vcpu);
>  	if (fault_status != FSC_FAULT && fault_status != FSC_PERM &&
> -	    fault_status != FSC_ACCESS) {
> +	    fault_status != FSC_ACCESS && sea_status) {
>  		kvm_err("Unsupported FSC: EC=%#x xFSC=%#lx ESR_EL2=%#lx\n",
>  			kvm_vcpu_trap_get_class(vcpu),
>  			(unsigned long)kvm_vcpu_trap_get_fault(vcpu),


Thanks,

James