public inbox for kvm@vger.kernel.org
 help / color / mirror / Atom feed
From: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
To: Ingo Molnar <mingo-X9Un+BFzKDI@public.gmane.org>
Cc: kvm-devel <kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org>
Subject: Re: [patch] KVM: paravirtual guest support
Date: Tue, 09 Jan 2007 12:13:56 +0200	[thread overview]
Message-ID: <45A36AE4.6000904@qumranet.com> (raw)
In-Reply-To: <20070109093354.GA10318-X9Un+BFzKDI@public.gmane.org>

Ingo Molnar wrote:
> Subject: [patch] KVM: paravirtual guest support
> From: Ingo Molnar <mingo-X9Un+BFzKDI@public.gmane.org>
>
> this enables a CONFIG_PARAVIRT Linux guest kernel to establish a 
> hypercall API to a KVM host. If successfully registered, then the Linux 
> guest will optimize a few things like its interrupt controller, io-delay 
> and it also registers its cr3-cache structures with the host. (but the 
> host will not touch those, just yet)
>
> (this is fully backwards compatible - if the WRMSR fails then the Linux
> guest continues to execute as a native kernel.)
>
> Signed-off-by: Ingo Molnar <mingo-X9Un+BFzKDI@public.gmane.org>
> ---
> +			/*
> +			 * Cache-hit: we load the cached host-CR3 value.
> +			 * This never causes any VM exit. (if it does then the
> +			 * hypervisor could do nothing with this instruction
> +			 * and the guest OS would be aborted)
> +			 */
> +			asm volatile("movl %0, %%cr3"
> +				: : "r" (cache->entry[idx].host_cr3));
> +			goto out;
> +		}
> +	}
> +
> +	/*
> +	 * Cache-miss. Load the guest-cr3 value into cr3, which will
> +	 * cause a VM exit to the hypervisor, which then loads the
> +	 * host cr3 value and updates the cr3_cache.
> +	 */
> +	asm volatile("movl %0, %%cr3" : : "r" (guest_cr3));
> +out:
> +	put_cpu_var(para_state);
> +}
> +
>   

There's a problem here.  A cache-hit cr3 is an hpa, while a cache-miss 
cr3 is a gpa.  The two could alias:

/* pseudoassembler */

guest:
  mov $0x1234, %cr3 /* cache miss */

host:
  creates shadow for 0x1234 at 0x5678
  set cr3 cache entry: $0x1234 -> 0x5678

guest:
  mov $0x5678, %cr3 /* cache miss */
  vmx accepts that as a cached cr3 equivalent to the guest's 0x1234, 
whereas a 0x5678 gpa cr3 was intended.

The only solution I see is to use the hypercall API on a cache miss.  
Once the guest indicates it wants to use the cr3 cache, set_cr3() should 
fault, and cr3 should only be accepted from the hypercall.

-- 
error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

  parent reply	other threads:[~2007-01-09 10:13 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-01-09  9:33 [patch] KVM: paravirtual guest support Ingo Molnar
     [not found] ` <20070109093354.GA10318-X9Un+BFzKDI@public.gmane.org>
2007-01-09  9:41   ` Ingo Molnar
2007-01-09 10:13   ` Avi Kivity [this message]
     [not found]     ` <45A36AE4.6000904-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
2007-01-09 10:46       ` Ingo Molnar

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=45A36AE4.6000904@qumranet.com \
    --to=avi-atkuwr5tajbwk0htik3j/w@public.gmane.org \
    --cc=kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org \
    --cc=mingo-X9Un+BFzKDI@public.gmane.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox