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: <87wm69dvbu.fsf@redhat.com>
Date: Mon, 08 Sep 2025 14:42:29 +0300
From: Vitaly Kuznetsov <vkuznets@...hat.com>
To: Khushit Shah <khushit.shah@...anix.com>
Cc: "seanjc@...gle.com" <seanjc@...gle.com>, "pbonzini@...hat.com"
 <pbonzini@...hat.com>, "kvm@...r.kernel.org" <kvm@...r.kernel.org>,
 "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>, Shaju
 Abraham <shaju.abraham@...anix.com>
Subject: Re: [BUG] [KVM/VMX] Level triggered interrupts mishandled on
 Windows w/ nested virt(Credential Guard) when using split irqchip

Khushit Shah <khushit.shah@...anix.com> writes:

> Thanks you for the comments Vitaly!
>
>> On 8 Sep 2025, at 2:35 PM, Vitaly Kuznetsov <vkuznets@...hat.com> wrote:
>> 
>> Is there a specific reason to not enable any Hyper-V enlightenments for
>> your guest? For nested cases, features like Enightended VMCS
>> ('hv-evmcs'), 'hv-vapic', 'hv-apicv', ... can change Windows's behavior
>> a lot. I'd even suggest you start with 'hv-passthrough' to see if the
>> slowness goes away and if yes, then try to find the required set of
>> options you can use in your setup.
>
>
> Actually in production we use an extensive set of cpu features exposed to the guest, still the issue persists, 
> With the following hv-* options also the issue is present:
>        hypervisor=on,hv-time=on,hv-relaxed=on,hv-vapic=on,hv-spinlocks=0x2000,hv-vpindex=on,hv-runtime=on,hv-synic=on, 
>        hv-stimer=on,hv-tlbflush=on,hv-ipi=on,hv-evmcs=on
>

Try adding 'hv-apicv' (AKA 'hv-avic') to the list too (not to be
confused with 'hv-vapic').

>> 
>> On 8 Sep 2025, at 2:35 PM, Vitaly Kuznetsov <vkuznets@...hat.com> wrote:
>> 
>> Single CPU Windows guests are always very slow, doubly so when running nested.
>
> The bug was reproducible even with more cpus like (4,8), we use 1 to reduce noise in captured logs.
>
> I should also mention by slow boot we mean extremely slow (>3h).

Oh, this is next level) Do you know if the issue reproduces with newer
Windows, e.g. 2025?

Also, I've just recalled I fixed (well, 'workarounded') an issue similar
to yours a while ago in QEMU:

commit 958a01dab8e02fc49f4fd619fad8c82a1108afdb
Author: Vitaly Kuznetsov <vkuznets@...hat.com>
Date:   Tue Apr 2 10:02:15 2019 +0200

    ioapic: allow buggy guests mishandling level-triggered interrupts to make progress

maybe something has changed and it doesn't work anymore?

-- 
Vitaly


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ