linux-kernel - Re: [PATCH v3 3/3] KVM: SVM: Extend host physical APIC ID field to support more than 8-bit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <7108583e-d495-809f-9530-63dbd5e15179@amd.com>
Date:   Wed, 2 Feb 2022 11:07:17 +0700
From:   "Suthikulpanit, Suravee" <suravee.suthikulpanit@....com>
To:     Sean Christopherson <seanjc@...gle.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org, x86@...nel.org,
        pbonzini@...hat.com, joro@...tes.org, mlevitsk@...hat.com,
        tglx@...utronix.de, mingo@...hat.com, bp@...en8.de,
        peterz@...radead.org, hpa@...or.com, thomas.lendacky@....com,
        jon.grimm@....com
Subject: Re: [PATCH v3 3/3] KVM: SVM: Extend host physical APIC ID field to
 support more than 8-bit



On 2/1/2022 7:58 PM, Suthikulpanit, Suravee wrote:
>> Anyways, for the new larger mask, IMO dynamically computing the mask based on what
>> APIC IDs were enumerated to the kernel is pointless.  If the AVIC doesn't support
>> using bits 11:0 to address APIC IDs then KVM is silently hosed no matter what if
>> any APIC ID is >255.
> 
> The reason for dynamic mask is to protect the reserved bits, which varies between
> the current platform (i.e 11:8) vs. newer platform (i.e. 11:10), in which
> there is no good way to tell except to check the max_physical_apicid (see below).
> 
>> Ideally, there would be a feature flag enumerating the larger AVIC support so we
>> could do:
>>
>>     if (!x2apic_mode || !boot_cpu_has(X86_FEATURE_FANCY_NEW_AVIC))
>>         avic_host_physical_id_mask = GENMASK(7:0);
>>     else
>>         avic_host_physical_id_mask = GENMASK(11:0);
>>
>> but since it sounds like that's not the case, and presumably hardware is smart
>> enough not to assign APIC IDs it can't address, this can simply be
>>
>>     if (!x2apic_mode)
>>         avic_host_physical_id_mask = GENMASK(7:0);
>>     else
>>         avic_host_physical_id_mask = GENMASK(11:0);
>>
>> and patch 01 to add+export apic_get_max_phys_apicid() goes away.
> 
> Unfortunately, we do not have the "X86_FEATURE_FANCY_NEW_AVIC" CPUID bit :(
> 
> Also, based on the previous comment, we can't use the x2APIC mode in the host
> to determine such condition. Hence, the need for dynamic mask based on
> the max_physical_apicid.

I recheck this part, and it should be safe to assume that AVIC HW can support
upto 8-bit (old platform) vs. 12-bit (new platform) depending on the maximum
host physical APIC ID available on the system.

I'll simplify this in v4.

Regards,
Suravee