linux-kernel - Re: [PATCH] x86: KVM: Add feature flag for AMD's FsGsKernelGsBaseNonSerializing

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: <01009a2a-929e-ce16-6f44-1d314e6bcba5@intel.com>
Date:   Tue, 3 Oct 2023 17:57:23 -0700
From:   Dave Hansen <dave.hansen@...el.com>
To:     Jim Mattson <jmattson@...gle.com>, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org,
        Pawan Gupta <pawan.kumar.gupta@...ux.intel.com>,
        Jiaxi Chen <jiaxi.chen@...ux.intel.com>,
        Kim Phillips <kim.phillips@....com>,
        Paolo Bonzini <pbonzini@...hat.com>,
        Sean Christopherson <seanjc@...gle.com>,
        "H. Peter Anvin" <hpa@...or.com>, x86@...nel.org,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Borislav Petkov <bp@...en8.de>, Ingo Molnar <mingo@...hat.com>,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH] x86: KVM: Add feature flag for AMD's
 FsGsKernelGsBaseNonSerializing

On 10/3/23 17:20, Jim Mattson wrote:
> Define an X86_FEATURE_* flag for
> CPUID.80000021H:EAX.FsGsKernelGsBaseNonSerializing[bit 1], and
> advertise the feature to userspace via KVM_GET_SUPPORTED_CPUID.
...
> +#define X86_FEATURE_BASES_NON_SERIAL	(20*32+ 1) /* "" FSBASE, GSBASE, and KERNELGSBASE are non-serializing */

This is failing to differentiate two *VERY* different things.

FSBASE, GSBASE, and KERNELGSBASE themselves are registers.  They have
*NOTHING* to do with serialization.  WRFSBASE, for instance is not
serializing.  Reading (with RDMSR) or using any of those three registers
is not serializing.

The *ONLY* thing that relates them to serialization is the WRMSR
instruction which itself is (mostly) architecturally serializing and the
fact that WRMSR has historically been the main way to write those three
registers.

The AMD docs call this out, which helps.  But the changelog, comments
and probably the feature naming need some work.

Why does this matter, btw?  Why do guests need this bit passed through?