linux-kernel - Re: [PATCH 0/6] KVM ARM64 pre_fault

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] [day] [month] [year] [list]

Message-ID: <aMMba1r3AVdC9cQw@linux.dev>
Date: Thu, 11 Sep 2025 11:56:43 -0700
From: Oliver Upton <oliver.upton@...ux.dev>
To: Jack Thomson <jackabt.amazon@...il.com>
Cc: maz@...nel.org, pbonzini@...hat.com, joey.gouly@....com,
	suzuki.poulose@....com, yuzenghui@...wei.com,
	catalin.marinas@....com, will@...nel.org, shuah@...nel.org,
	linux-arm-kernel@...ts.infradead.org, kvmarm@...ts.linux.dev,
	linux-kernel@...r.kernel.org, linux-kselftest@...r.kernel.org,
	isaku.yamahata@...el.com, roypat@...zon.co.uk,
	kalyazin@...zon.co.uk, jackabt@...zon.com
Subject: Re: [PATCH 0/6] KVM ARM64 pre_fault_memory

Hi Jack,

On Thu, Sep 11, 2025 at 02:46:42PM +0100, Jack Thomson wrote:
> From: Jack Thomson <jackabt@...zon.com>
> 
> Overview:
> 
> This patch series adds ARM64 support for the KVM_PRE_FAULT_MEMORY
> feature, which was previously only available on x86 [1]. This allows
> a reduction in the number of stage-2 faults during execution. This is
> beneficial in post-copy migration scenarios, particularly in memory
> intensive applications, where high latencies are experienced due to
> the stage-2 faults when pre-populating memory via UFFD / memcpy.

Thanks for posting the series. More of a general comment on the UAPI
documentation:

  "... However, KVM does not mark any newly created stage-2 PTE as Accessed."

This behavior is x86-specific since kvm_pgtable_stage2_map() will lay
down PTEs with the AF set. Probably shouldn't have documented the
internal state of the stage-2 in the first place but oh well, please
just update the UAPI description to make it clear this is x86-specific.

Thanks
Oliver