Message-ID: <453D2FFA.3040506@qumranet.com>
Date: Mon, 23 Oct 2006 23:11:22 +0200
From: Avi Kivity <avi@qumranet.com>
User-Agent: Thunderbird 1.5.0.7 (X11/20061008)
MIME-Version: 1.0
To: Antonio Vargas <windenntw@gmail.com>
CC: Arnd Bergmann <arnd@arndb.de>, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 8/13] KVM: vcpu execution loop
References: <453CC390.9080508@qumranet.com> <200610232141.45802.arnd@arndb.de>	 <453D230D.7070403@qumranet.com> <200610232229.41934.arnd@arndb.de>	 <453D27F8.8020509@qumranet.com> <69304d110610231402k7913df63l7e493f95b5d92911@mail.gmail.com>
In-Reply-To: <69304d110610231402k7913df63l7e493f95b5d92911@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1065
Lines: 26

Antonio Vargas wrote:
>>
>> I could do that, but I feel that's more brittle.  I might need more (or
>> other) fields later on.  It will also cost me more  pushes on the stack
>> (no real performance or space impact, just C64-era frugality).
>
> maybe thats the mindsent needed to make these virtual cpu patches
> without eating away all the cpu power with more than needed
> abstractions ;)
>

Unfortunately not.  Saving a cycle or two doesn't help when a vm exit 
costs thousands of cycles, and worse, kills your tlb.

The key is eliminating unnecessary exits.  I have plans for massively 
optimizing the mmu virtualization, and the next AMD core will do that in 
hardware (look for a "nested page tables" sticker before you buy).

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/