Subject: Re: [PATCH v2] KVM: x86: avoid redundant REQ_EVENT
To: =?UTF-8?B?UmFkaW0gS3LEjW3DocWZ?= <rkrcmar@redhat.com>,
        Roman Kagan <rkagan@virtuozzo.com>,
        Denis Plotnikov <dplotnikov@virtuozzo.com>, den@virtuozzo.com,
        kvm@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20161214105908.322638-1-dplotnikov@virtuozzo.com>
 <b3da88d7-709b-3228-cf3b-3694fd00a97d@redhat.com>
 <20161215071840.GB7704@rkaganb.sw.ru> <20161215143054.GC6667@potion>
From: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <4876bc4a-b72a-897e-84a7-c34c3f934705@redhat.com>
Date: Thu, 15 Dec 2016 15:32:45 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101
 Thunderbird/45.5.1
MIME-Version: 1.0
In-Reply-To: <20161215143054.GC6667@potion>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 651
Lines: 17


On 15/12/2016 15:30, Radim Krčmář wrote:
> 
> One useless round of KVM_REQ_EVENT is not going change nested
> performance by much and it is not the only thing we could improve wrt.
> TPR ... I would just leave it for now and take care of it when we
>  * don't to update PPR at all with APICv -- it is already correct
>  * drop the KVM_REQ_EVENT with flex priority, because lower TPR cannot
>    unmask an interrupt

I agree.  I still don't like the patch very much, because I feel like an
explicit state machine ("can KVM_REQ_EVENT do anything?") would be more
maintainable.  But if I don't come up with anything we'll go with this
patch.

Paolo