Looking at using KVM for embedded product

Looking at using KVM for embedded product

[Tom Shoes]

Hi there,

    I am looking at using KVM for an embedded product. I am also new
to Virtualization so pardon if
I ask dumb questions. This is my first time posting to this forum.

   The embedded product that need to run KVM has:

         a. Intel processor with VT
         b. BIOS supports enabling VT
         c. Linux kernel 2.6.26 (from kernel.org)
         d. No VGA adapter
         e. Serial console
         f. BusyBox

  Need help to understand if KVM can be installed, configured and run
with c,d,e and f configuration
mentioned above?

  Also can I know
         a.  Is KVM code in Linux kernel 2.6.26 reliable?
         b.  Will latest KVM code from linux-kvm.org run on Linux kernel 2.6.26?
         c.  How can I find which version of KVM code is stable. Can I
find this using the version number
              of the KVM release?

    Thanks for any help on this.

Cheers
../Tom

Re: Looking at using KVM for embedded product

[Avi Kivity]

On 06/23/2010 04:22 AM, Tom Shoes wrote:
>    Need help to understand if KVM can be installed, configured and run
> with c,d,e and f configuration
> mentioned above?
>
>    Also can I know
>           a.  Is KVM code in Linux kernel 2.6.26 reliable?
>    

It may be reliable or not, but it's certainly not supported.  If you 
have problems or questions you're unlikely to receive support (on kvm or 
on Linux itself) on the mailing lists.

>           b.  Will latest KVM code from linux-kvm.org run on Linux kernel 2.6.26?
>    

You mean kvm-kmod?  It may, but I recommend at least 2.6.27 (when mmu 
notifiers were introduced).

>           c.  How can I find which version of KVM code is stable. Can I
> find this using the version number
>                of the KVM release?
>    

It depends on your needs.  Many users are satisfied with the lastest 
stable branch (2.6.32.y or 2.6.34.y).  Others want a distribution for 
longer term support.  The adventuous run kvm.git master branch.


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

Re: Looking at using KVM for embedded product

[Avi Kivity]

On 06/23/2010 05:29 PM, Tom Shoes wrote:
>>> On Wed, Jun 23, 2010 at 2:36 AM, Avi Kivity<avi@redhat.com>  wrote:
>>>        
> Thanks Avi for your response.
>    

Re-adding mailing list to cc list.

> Request for more clarification on the following:
>
> Is having no VGA adapter and using serial port as console an issue?
> When I mean by issue, I am refering to issue while configuring or
> running KVM and also its tools, library etc..
>    

Should not be a problem.

>> Many users are satisfied with the lastest stable branch (2.6.32.y or 2.6.34.y).
>>      
> Want to confirm if the KVM stable release go hand in hand with Linux kernel
> release?
>    

Yes - the kernel part.  In fact there are no independent kvm releases 
any more, they're just ordinary kernel releases.

The userspace part - qemu-kvm - is released independently (and has its 
own stable branch).

>> You mean kvm-kmod?  It may, but I recommend at least 2.6.27 (when mmu
>> notifiers were introduced).
>>      
> Thanks. Do you recommend 2.6.27 or any specific 2.6.27.y version.
>    

The latest version.  And I recommend 2.6.32.y or 2.6.34.y over 2.6.27.

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

Re: Looking at using KVM for embedded product

[Anjali Kulkarni]

Hi,

I am not sure if this is the right forum for my questions -
I am using Qemu+KVM on centOS to run my freebsd based virtual OS. The Qemu
emulates one of my nics - yukon -II 88E8022 chip. Now, my yukon driver
exists in kernel as well as user space(and I will use the one which I want
based on some configs). The kernel driver is able to read the bar0 from PCI
config and also read contents correctly. The user space driver gets the bar0
correctly, and mmaps it into it's process space, but when it reads contents
of bar0, it gets all 0s. This same code when written on a real device works.
Is it due to the fact Qemu is running my freebsd based OS in userland, so
does mmap require some tweaks?
Are there any documents which I could regarding how the Qemu/KVM traps the
read write to bar0 so that it can emulate the device reads/writes?
If this is not the right forum, can you point me to the right one?

Thanks
Anjali

Question about mmap from user space and accessing emulated device bar0 contents

[Anjali Kulkarni]

Changing subject,


Hi,

I am not sure if this is the right forum for my questions -
I am using Qemu+KVM on centOS to run my freebsd based virtual OS. The Qemu
emulates one of my nics - yukon -II 88E8022 chip. Now, my yukon driver
exists in kernel as well as user space(and I will use the one which I want
based on some configs). The kernel driver is able to read the bar0 from PCI
config and also read contents correctly. The user space driver gets the bar0
correctly, and mmaps it into it's process space, but when it reads contents
of bar0, it gets all 0s. This same code when written on a real device works.
Is it due to the fact Qemu is running my freebsd based OS in userland, so
does mmap require some tweaks?
Are there any documents which I could regarding how the Qemu/KVM traps the
read write to bar0 so that it can emulate the device reads/writes?
If this is not the right forum, can you point me to the right one?

Thanks
Anjali

Re: Looking at using KVM for embedded product

[Tom Shoes]

On Wed, Jun 23, 2010 at 7:33 AM, Avi Kivity <avi@redhat.com> wrote:

Thanks for your response Avi.

> The userspace part - qemu-kvm - is released independently (and has its own
> stable branch).

Does a certain version of qemu-kvm go hand in hand with a certain
version of KVM
code in the Linux kernel release?

Does the KVM community recommend or announce a version of qemu-kvm is
good to go after testing with a KVM code it releases?

Re: Looking at using KVM for embedded product

[Avi Kivity]

On 06/24/2010 03:21 PM, Tom Shoes wrote:
>> The userspace part - qemu-kvm - is released independently (and has its own
>> stable branch).
>>      
> Does a certain version of qemu-kvm go hand in hand with a certain
> version of KVM
> code in the Linux kernel release?
>
>    

No.

> Does the KVM community recommend or announce a version of qemu-kvm is
> good to go after testing with a KVM code it releases?
>    

No.  I generally test the latest qemu-kvm with the latest kvm.

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

Re: Looking at using KVM for embedded product

[Jan Kiszka]

Tom Shoes wrote:
> Hi there,
> 
>     I am looking at using KVM for an embedded product. I am also new
> to Virtualization so pardon if
> I ask dumb questions. This is my first time posting to this forum.
> 
>    The embedded product that need to run KVM has:
> 
>          a. Intel processor with VT
>          b. BIOS supports enabling VT
>          c. Linux kernel 2.6.26 (from kernel.org)

Why 2.6.26, specifically if you seem to have no customized bits?
Starting with a pre-historic kernel for a new project is hardly ever a
good choice.

>          d. No VGA adapter
>          e. Serial console
>          f. BusyBox
> 
>   Need help to understand if KVM can be installed, configured and run
> with c,d,e and f configuration
> mentioned above?
> 
>   Also can I know
>          a.  Is KVM code in Linux kernel 2.6.26 reliable?

That kernel is 2 years old, last version is 1.5 years old, and kvm was
still quite young into mainline at that time. There is a good chance you
may hit some bug that was fixed since then.

>          b.  Will latest KVM code from linux-kvm.org run on Linux kernel 2.6.26?

kvm-kmod builds down to 2.6.24, but it is likely only tested down to
2.6.27. And you will miss quite a few relevant performance optimization
due to limitations of 2.6.26.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

Re: Looking at using KVM for embedded product

[Jan Kiszka]

[ please use reply-all when posting to mailing lists ]

Tom Shoes wrote:
> On Wed, Jun 23, 2010 at 3:16 AM, Jan Kiszka <jan.kiszka@siemens.com> wrote:
> 
> Thanks Jan for the response.
> 
>> Why 2.6.26, specifically if you seem to have no customized bits?
>> Starting with a pre-historic kernel for a new project is hardly ever a
>> good choice.
> 
>   We had picked up 2.6.26 1 1/2 years back and now there is requirement
> to add virtualization. Many of the other bits has been ported to this kernel
> and 3/4th tested. So moving to newer kernel will be difficult unless there
> is a pressing requirement and this might be from the answer from you
> and Avi.

The effort you may save by sticking with 2.6.26 can quickly be consumed
by troubles and limitations you may face with this old kernel. Better
make sure your local bits are clean and easily portable to recent
kernels version - or even push them upstream. And if you already have a
test suite for your specific scenarios, switching the target shouldn't
be that problematic.

> 
>>>          b.  Will latest KVM code from linux-kvm.org run on Linux kernel 2.6.26?
>> kvm-kmod builds down to 2.6.24, but it is likely only tested down to
>> 2.6.27. And you will miss quite a few relevant performance optimization
>> due to limitations of 2.6.26.
>>
> 
> Do you mean the the KVM community test build every KVM releases down to
> 2.6.24

Speaking of kvm-kmod: http://buildbot.kiszka.org/kvm-kmod/waterfall

> and do testing (may be a cursory one) of KVM functionality down to
> 2.6.27?

Testing of kvm-kmod is a bit less ambitious. This compat package is not
that broadly used compared to the KVM modules that ship with standard or
distribution kernels. There are tests performed against 2.6.27 and
2.6.32 hosts ATM, but anything else depends on community feedback.
Still, we didn't face much kvm-kmod-specific issues in the recent past,
specifically after its stabilization phases.

However, if you decide to upgrade your kernel anyway, I would suggest to
go with the included KVM support first, only pick kvm-kmod if there is a
good reason.

Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux

Re: Looking at using KVM for embedded product

[Tom Shoes]

Thanks you for the suggestions and the link (a neat one).

Cheers
../Tom

Re: Looking at using KVM for embedded product

[john cooper]

Tom Shoes wrote:
> Hi there,
> 
>     I am looking at using KVM for an embedded product. I am also new
> to Virtualization so pardon if
> I ask dumb questions. This is my first time posting to this forum.
> 
>    The embedded product that need to run KVM has:
> 
>          a. Intel processor with VT
>          b. BIOS supports enabling VT
>          c. Linux kernel 2.6.26 (from kernel.org)
>          d. No VGA adapter
>          e. Serial console
>          f. BusyBox

Busybox is an interesting requirement in that context.  If
you are constrained with userland size and linking against
other than glibc, use of qemu could be interesting.  Can't
say I've built it other than linked against glibc and an
extensive list of runtime libraries.  Although I've never
tried to configure-down that dependency. 

-john

-- 
john.cooper@third-harmonic.com