linux-kernel - Re: [PATCH] KVM: X86: Make fpu allocation a common function

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]

Date:   Tue, 15 Oct 2019 08:48:02 +0800
From:   Xiaoyao Li <xiaoyao.li@...el.com>
To:     Sean Christopherson <sean.j.christopherson@...el.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>
Cc:     kvm@...r.kernel.org, linux-kernel@...r.kernel.org,
        Paolo Bonzini <pbonzini@...hat.com>,
        Radim Krčmář <rkrcmar@...hat.com>,
        Jim Mattson <jmattson@...gle.com>
Subject: Re: [PATCH] KVM: X86: Make fpu allocation a common function

On 10/15/2019 2:37 AM, Sean Christopherson wrote:
> On Mon, Oct 14, 2019 at 06:58:49PM +0200, Vitaly Kuznetsov wrote:
>> Xiaoyao Li <xiaoyao.li@...el.com> writes:
>>
>>> They are duplicated codes to create vcpu.arch.{user,guest}_fpu in VMX
>>> and SVM. Make them common functions.
>>>
>>> No functional change intended.
>>
>> Would it rather make sense to move this code to
>> kvm_arch_vcpu_create()/kvm_arch_vcpu_destroy() instead?
> 
> Does it make sense?  Yes.  Would it actually work?  No.  Well, not without
> other shenanigans.
> 
> FPU allocation can't be placed after the call to .create_vcpu() becuase
> it's consumed in kvm_arch_vcpu_init().   FPU allocation can't come before
> .create_vcpu() because the vCPU struct itself hasn't been allocated.  The
> latter could be solved by passed the FPU pointer into .create_vcpu(), but
> that's a bit ugly and is not a precedent we want to set.
> 

That's exactly what I found.

> At a glance, FPU allocation can be moved to kvm_arch_vcpu_init(), maybe
> right before the call to fx_init().
> 

Yeah, putting here is better.

I'm wondering the semantic of create, init, setup. There are 
vcpu_{create,init,setup}, and IIUC, vcpu_create is mainly for data 
structure allocation and vcpu_{init,setup} should be for data structure 
initialization/setup (and maybe they could/should merge into one)

But I feel the current codes for vcpu creation a bit messed, especially 
of vmx.