Re: [Qemu-devel] [PATCH 0/6] Bypass tcg memory functions -v1.0-2009

[Anthony Liguori <aliguori@us.ibm.com>]

Paul Brook wrote:
> On Tuesday 20 January 2009, Glauber Costa wrote:
>   
>> This series is not very different from the last one that did it.
>> Just bringing it back from my vacations. Idea is to replace tcg
>> memory functions with kvm's, selectable at runtime. Right now
>> we use a conditional selection. In the future, we might use
>> function pointers to allow for easy coupling of any hypervisor.
>>     
>
> I dislike that you're introducing two different ways or doing the same thing.  
> Duplicating all the memory region tracking code seems like a bad way to solve 
> this problem.
>   

To me, this patch set is really about defining the memory API for QEMU's 
CPU emulation/virtualization core.  It can be made a lot prettier by 
introducing function pointers and such but doing the separation and 
having two implementations is the first step in that direction.

I believe this is the right level of abstraction.  Trying to make shared 
code with a lower level interface is much more complicated because TCG 
has some need for deep hooks in the memory code.  If TCG ever wants to 
use memory slots instead of a tree, I'd rather see common code 
abstracted into a set of libs than making the slot code the main code 
with hooks for TCG.

A big part of the reason for these patches is to make it possible to 
build QEMU without TCG in an architecturally clean way.

Regards,

Anthony Liguori

> Paul
>