linux-kernel - Re: [PATCH 4/4] KVM: X86: Only get rflags when needed in permission

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: <Ya/Y1+6BR4exkTKK@google.com>
Date:   Tue, 7 Dec 2021 21:57:43 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Lai Jiangshan <jiangshanlai@...il.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        Paolo Bonzini <pbonzini@...hat.com>,
        Lai Jiangshan <laijs@...ux.alibaba.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        Dave Hansen <dave.hansen@...ux.intel.com>, x86@...nel.org,
        "H. Peter Anvin" <hpa@...or.com>
Subject: Re: [PATCH 4/4] KVM: X86: Only get rflags when needed in
 permission_fault()

On Tue, Dec 07, 2021, Lai Jiangshan wrote:
> From: Lai Jiangshan <laijs@...ux.alibaba.com>
> 
> In same cases, it doesn't need to get rflags for SMAP checks.
> 
> For example: it is user mode access, it could have contained other
> permission fault, SMAP is not enabled, it is implicit supervisor
> access, or it is nested TDP pagetable.

I don't disagree that reading RFLAGS is silly and _may_ have worse performance,
but I'd prefer any change have actual numbers to justify that it's an improvement
or at least a wash / in the noise.

Too much of the MMU code (and KVM in general) has optimizations like this that
have bitrotted horribly over the years.  And in many/most cases, the original commit
didn't provide performance numbers, so it's not even clear that the "optimizations"
were _ever_ a net win.