Re: cr3 OOS optimisation breaks 32-bit GNU/kFreeBSD guest

From: Avi Kivity <avi@redhat.com>
To: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>, kvm@vger.kernel.org
Subject: Re: cr3 OOS optimisation breaks 32-bit GNU/kFreeBSD guest
Date: Sun, 05 Apr 2009 14:41:02 +0300	[thread overview]
Message-ID: <49D898CE.2050308@redhat.com> (raw)
In-Reply-To: <20090405112917.GA4105@amt.cnet>

Marcelo Tosatti wrote:
>> Here's one way to make this work:
>>
>>  - add a hash of global pagetables, indexed by virtual address instead  
>> of the pagetable's gfn
>>  - invlpg checks this hash in addition to the recursive walk
>>
>> We'd need to make the virtual address part of sp->role to avoid needing  
>> to link the same page multiple times in the virtual address hash.
>>     
>
> Humpf, yes. It seems its too expensive/complex to handle this, for such
> small gain (~= 2% on AIM7 with RHEL3 guest).
>
> Are you okay with just disabling the global pages optimization?
>   

Definitely to plug the hole; and probably for later as well, unless 
people cry out due to regressions.

Please send it in two patches:  one a trivial one to disable global page 
detection which can be sent to -stable as well, and a follow on which 
rips out the global page machinery until (and if) we decide to 
reimplement it correctly.

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