[Qemu-devel] [PATCH 0/3] virtio support for QEMU

[Qemu-devel] [PATCH 0/3] virtio support for QEMU

[Anthony Liguori]

This patch series adds support for paravirtual device drivers that use 
virtio.  virtio is a frame work in Linux for abstracting the details of 
virtual IO so that a single device driver (like networking) can be used 
with multiple hypervisors using a small shim layer.  Currently, Linux 
supports a virtio network and a virtio block device and a shim layer for 
lguest.   I have a PCI shim layer for virtio that is suitable for use in 
QEMU.  This is currently in Rusty Russell's virtio tree awaiting the 
Linux merge window to open again.

Regards,

Anthony Liguori

[Qemu-devel] Re: [PATCH 0/3] virtio support for QEMU

[Dor Laor]

[-- Attachment #1: Type: text/plain, Size: 906 bytes --]

Anthony Liguori wrote:
>
> This patch series adds support for paravirtual device drivers that use
> virtio.  virtio is a frame work in Linux for abstracting the details of
> virtual IO so that a single device driver (like networking) can be used
> with multiple hypervisors using a small shim layer.  Currently, Linux
> supports a virtio network and a virtio block device and a shim layer for
> lguest.   I have a PCI shim layer for virtio that is suitable for use in
> QEMU.  This is currently in Rusty Russell's virtio tree awaiting the
> Linux merge window to open again.
>
> Regards,
>
> Anthony Liguori
>
Hi Anthony, Rusty,

Can you set up an uptodate git repository for the kernel part, other 
community members
will also benefit.
I got an oops for vring_kick immediately on the first pkt while testing 
this patchset.
Going to debug it tomorrow, hopefully it's not because of patching.
Thanks,
Dor.

[-- Attachment #2: Type: text/html, Size: 1544 bytes --]

[Qemu-devel] Re: [PATCH 0/3] virtio support for QEMU

[Anthony Liguori]

Dor Laor wrote:
> Anthony Liguori wrote:
>>
>> This patch series adds support for paravirtual device drivers that use
>> virtio.  virtio is a frame work in Linux for abstracting the details of
>> virtual IO so that a single device driver (like networking) can be used
>> with multiple hypervisors using a small shim layer.  Currently, Linux
>> supports a virtio network and a virtio block device and a shim layer for
>> lguest.   I have a PCI shim layer for virtio that is suitable for use in
>> QEMU.  This is currently in Rusty Russell's virtio tree awaiting the
>> Linux merge window to open again.
>>
>> Regards,
>>
>> Anthony Liguori
>>
> Hi Anthony, Rusty,
>
> Can you set up an uptodate git repository for the kernel part, other 
> community members
> will also benefit.
> I got an oops for vring_kick immediately on the first pkt while 
> testing this patchset.
> Going to debug it tomorrow, hopefully it's not because of patching.

Yeah, patches are required that aren't yet in Rusty's queue.  I'll 
update the list once we've gotten his queue straightened out.  The oops 
is most certainly from the use of unsigned int in vring_init instead of 
unsigned long.  I have a patch on 
virtualization@lists.linux-foundation.org for that but I don't think 
Rusty's pulled it in yet.

Regards,

Anthony Liguori

> Thanks,
> Dor.

[Qemu-devel] Re: [PATCH 0/3] virtio support for QEMU

[Rusty Russell]

On Wednesday 05 December 2007 11:10:46 Anthony Liguori wrote:
> Dor Laor wrote:
> > Can you set up an uptodate git repository for the kernel part, other
> Yeah, patches are required that aren't yet in Rusty's queue.  I'll
> update the list once we've gotten his queue straightened out.

I'm actually on vacation this week and next week, so my queue won't be updated 
for a while.  I'll put them all in once I'm back on deck...

Rusty.

[Qemu-devel] Re: [PATCH 0/3] virtio support for QEMU

[Avi Kivity]

Rusty Russell wrote:
> On Wednesday 05 December 2007 11:10:46 Anthony Liguori wrote:
>   
>> Dor Laor wrote:
>>     
>>> Can you set up an uptodate git repository for the kernel part, other
>>>       
>> Yeah, patches are required that aren't yet in Rusty's queue.  I'll
>> update the list once we've gotten his queue straightened out.
>>     
>
> I'm actually on vacation this week and next week, so my queue won't be updated 
> for a while.  I'll put them all in once I'm back on deck...
>
>   

I can host it in a branch on kvm.git, if I knew what patches to 
include.  Anthony, can you post pointers?

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

[Qemu-devel] Re: [PATCH 0/3] virtio support for QEMU

[Anthony Liguori]

Avi Kivity wrote:
> Rusty Russell wrote:
>> On Wednesday 05 December 2007 11:10:46 Anthony Liguori wrote:
>>  
>>> Dor Laor wrote:
>>>    
>>>> Can you set up an uptodate git repository for the kernel part, other
>>>>       
>>> Yeah, patches are required that aren't yet in Rusty's queue.  I'll
>>> update the list once we've gotten his queue straightened out.
>>>     
>>
>> I'm actually on vacation this week and next week, so my queue won't 
>> be updated for a while.  I'll put them all in once I'm back on deck...
>>
>>   
>
> I can host it in a branch on kvm.git, if I knew what patches to 
> include.  Anthony, can you post pointers?

http://hg.codemonkey.ws/linux-virtio

That's the required patches from rusty's queue plus my patches based off 
the kvm git tree as of 83f9425dc148dbe0d373b7e07f221affff9e91cb

Regards,

Anthony Liguori

[Qemu-devel] Re: [PATCH 0/3] virtio support for QEMU

[Avi Kivity]

Anthony Liguori wrote:
> This patch series adds support for paravirtual device drivers that use 
> virtio.  virtio is a frame work in Linux for abstracting the details 
> of virtual IO so that a single device driver (like networking) can be 
> used with multiple hypervisors using a small shim layer.  Currently, 
> Linux supports a virtio network and a virtio block device and a shim 
> layer for lguest.   I have a PCI shim layer for virtio that is 
> suitable for use in QEMU.  This is currently in Rusty Russell's virtio 
> tree awaiting the Linux merge window to open again.
>

Can you post this to kvm-devel as well next time?  I'm sure there is 
considerable interest in this work among kvm developers.

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