Non PCI passthrough on xen-unstable tree

Non PCI passthrough on xen-unstable tree

[manish jaggi]

[-- Attachment #1.1: Type: text/plain, Size: 279 bytes --]

Hi julien,

I am trying to evaluate your changes (
http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
support for PCI passthrough for xen on cavium thunder. Can you please
provide steps how you tested non-pci passthrough with your git.


-Regards
Manish

[-- Attachment #1.2: Type: text/html, Size: 627 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Non PCI passthrough on xen-unstable tree

[manish jaggi]

Hi Julien,

Waiting for your reply

On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com> wrote:
> Hi julien,
>
> I am trying to evaluate your changes
> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
> support for PCI passthrough for xen on cavium thunder. Can you please
> provide steps how you tested non-pci passthrough with your git.
>
>
> -Regards
> Manish
>

Re: Non PCI passthrough on xen-unstable tree

[Julien Grall]

On 08/12/2014 05:40 AM, manish jaggi wrote:
> Hi Julien,

Hello,

> Waiting for your reply

It's not necessary to ping after just 2 days, one was a day off here...

> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com> wrote:
>> Hi julien,
>>
>> I am trying to evaluate your changes
>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
>> support for PCI passthrough for xen on cavium thunder. Can you please
>> provide steps how you tested non-pci passthrough with your git.

Everything should be written in the cover letter. I won't copy again the
steps here.

If there is a missing step please ask.

Regards,

-- 
Julien Grall

Re: Non PCI passthrough on xen-unstable tree

[manish jaggi]

On 13 August 2014 20:00, Julien Grall <julien.grall@linaro.org> wrote:
> On 08/12/2014 05:40 AM, manish jaggi wrote:
>> Hi Julien,
>
> Hello,
>
>> Waiting for your reply
>
> It's not necessary to ping after just 2 days, one was a day off here..
Oops sorry.
>
>> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com> wrote:
>>> Hi julien,
>>>
>>> I am trying to evaluate your changes
>>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
>>> support for PCI passthrough for xen on cavium thunder. Can you please
>>> provide steps how you tested non-pci passthrough with your git.
>
> Everything should be written in the cover letter. I won't copy again the
> steps here.
>
Can you please help us with the following so that we can test your
patch and use it in our development
Which linux version  was used, defconfig and domU.cfg used ?
Was it only on midway or can work on foundation model

> If there is a missing step please ask.
>
> Regards,
>
> --
> Julien Grall

Re: Non PCI passthrough on xen-unstable tree

[Julien Grall]

On 08/13/2014 06:29 PM, manish jaggi wrote:
>>> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com> wrote:
>>>> I am trying to evaluate your changes
>>>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
>>>> support for PCI passthrough for xen on cavium thunder. Can you please
>>>> provide steps how you tested non-pci passthrough with your git.
>>
>> Everything should be written in the cover letter. I won't copy again the
>> steps here.
>>
> Can you please help us with the following so that we can test your
> patch and use it in our development
> Which linux version  was used, defconfig and domU.cfg used ?

Why are you talking about linux version/defconfig? As said on my cover
letter nothing as been modified in Linux. The only requirements is the
kernel must have a driver for your device.

> Was it only on midway or can work on foundation model

Even though I did my development on midway, it should work on every
platform with devices under an SMMU.

Regards,

-- 
Julien Grall

Re: Non PCI passthrough on xen-unstable tree

[manish jaggi]

On 14 August 2014 14:18, Julien Grall <julien.grall@linaro.org> wrote:
> On 08/13/2014 06:29 PM, manish jaggi wrote:
>>>> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com> wrote:
>>>>> I am trying to evaluate your changes
>>>>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
>>>>> support for PCI passthrough for xen on cavium thunder. Can you please
>>>>> provide steps how you tested non-pci passthrough with your git.
>>>
>>> Everything should be written in the cover letter. I won't copy again the
>>> steps here.
>>>
>> Can you please help us with the following so that we can test your
>> patch and use it in our development
>> Which linux version  was used, defconfig and domU.cfg used ?
>
> Why are you talking about linux version/defconfig? As said on my cover
> letter nothing as been modified in Linux. The only requirements is the
> kernel must have a driver for your device.
>
I was not able to make out from the cover letter.

>> Was it only on midway or can work on foundation model
>
> Even though I did my development on midway, it should work on every
> platform with devices under an SMMU.
>
> Regards,
>
It would be very helpful if you could please provide steps to test this feature.
I have to integrate your work into my tree for testing PCI
pass-through on cavium thunder platform.

> --
> Julien Grall

Re: Non PCI passthrough on xen-unstable tree

[Julien Grall]

On 14/08/14 10:20, manish jaggi wrote:
> On 14 August 2014 14:18, Julien Grall <julien.grall@linaro.org> wrote:
>> On 08/13/2014 06:29 PM, manish jaggi wrote:
>>>>> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com> wrote:
>>>>>> I am trying to evaluate your changes
>>>>>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am adding
>>>>>> support for PCI passthrough for xen on cavium thunder. Can you please
>>>>>> provide steps how you tested non-pci passthrough with your git.
>>>>
>>>> Everything should be written in the cover letter. I won't copy again the
>>>> steps here.
>>>>
>>> Can you please help us with the following so that we can test your
>>> patch and use it in our development
>>> Which linux version  was used, defconfig and domU.cfg used ?
>>
>> Why are you talking about linux version/defconfig? As said on my cover
>> letter nothing as been modified in Linux. The only requirements is the
>> kernel must have a driver for your device.
>>
> I was not able to make out from the cover letter.

"To avoid adding code in DOM0 to manage platform device deassignment, 
the user has to mark the device disabled in the device tree."

>>> Was it only on midway or can work on foundation model
>>
>> Even though I did my development on midway, it should work on every
>> platform with devices under an SMMU.
>>
>> Regards,
>>
> It would be very helpful if you could please provide steps to test this feature.
> I have to integrate your work into my tree for testing PCI
> pass-through on cavium thunder platform.

Below, the steps to use on midway. You will have to adapt to your platform:

1) In the u-boot script disable the device you plan to passthrough:
   fdt set /soc/ethernet@fff51000 status disabled
2) Ask the toolstack to passthrough the device by adding in the vm 
config file:
   dtdev= [ "/soc/ethernet@fff51000" ]

There is some minor error in my branch passthrough-v2. I've pushed a new 
one here:

git://xenbits.xen.org/people/julieng/xen-unstable.git passthrough-v2.1

Regards,

-- 
Julien Grall

Re: Non PCI passthrough on xen-unstable tree

[manish jaggi]

On 15 August 2014 03:43, Julien Grall <julien.grall@linaro.org> wrote:
>
>
> On 14/08/14 10:20, manish jaggi wrote:
>>
>> On 14 August 2014 14:18, Julien Grall <julien.grall@linaro.org> wrote:
>>>
>>> On 08/13/2014 06:29 PM, manish jaggi wrote:
>>>>>>
>>>>>> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com>
>>>>>> wrote:
>>>>>>>
>>>>>>> I am trying to evaluate your changes
>>>>>>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am
>>>>>>> adding
>>>>>>> support for PCI passthrough for xen on cavium thunder. Can you please
>>>>>>> provide steps how you tested non-pci passthrough with your git.
>>>>>
>>>>>
>>>>> Everything should be written in the cover letter. I won't copy again
>>>>> the
>>>>> steps here.
>>>>>
>>>> Can you please help us with the following so that we can test your
>>>> patch and use it in our development
>>>> Which linux version  was used, defconfig and domU.cfg used ?
>>>
>>>
>>> Why are you talking about linux version/defconfig? As said on my cover
>>> letter nothing as been modified in Linux. The only requirements is the
>>> kernel must have a driver for your device.
>>>
>> I was not able to make out from the cover letter.
>
>
> "To avoid adding code in DOM0 to manage platform device deassignment, the
> user has to mark the device disabled in the device tree."
>
>
>>>> Was it only on midway or can work on foundation model
>>>
>>>
>>> Even though I did my development on midway, it should work on every
>>> platform with devices under an SMMU.
>>>
>>> Regards,
>>>
>> It would be very helpful if you could please provide steps to test this
>> feature.
>> I have to integrate your work into my tree for testing PCI
>> pass-through on cavium thunder platform.
>
>
> Below, the steps to use on midway. You will have to adapt to your platform:
>
> 1) In the u-boot script disable the device you plan to passthrough:
>   fdt set /soc/ethernet@fff51000 status disabled
> 2) Ask the toolstack to passthrough the device by adding in the vm config
> file:
>   dtdev= [ "/soc/ethernet@fff51000" ]
>
> There is some minor error in my branch passthrough-v2. I've pushed a new one
> here:
>
> git://xenbits.xen.org/people/julieng/xen-unstable.git passthrough-v2.1
>
Thanks Julien for replying. I am trying to understand your code and
map it with the documentation (devicetree/bindings/iommu/arm,smmu.txt
and SMMU v2 Arch).

While I was trying to understand the drivers/passthrough/arm/smmu.c I
was not able to find a device tree that contains the compatible nodes
in 3.16 kernel (grep smmuvx / mmu-masters returned nothing).
Can you please help with this.

> Regards,
>
> --
> Julien Grall

Re: Non PCI passthrough on xen-unstable tree

[Stefano Stabellini]

On Tue, 2 Sep 2014, manish jaggi wrote:
> On 15 August 2014 03:43, Julien Grall <julien.grall@linaro.org> wrote:
> >
> >
> > On 14/08/14 10:20, manish jaggi wrote:
> >>
> >> On 14 August 2014 14:18, Julien Grall <julien.grall@linaro.org> wrote:
> >>>
> >>> On 08/13/2014 06:29 PM, manish jaggi wrote:
> >>>>>>
> >>>>>> On 10 August 2014 07:53, manish jaggi <manishjaggi.oss@gmail.com>
> >>>>>> wrote:
> >>>>>>>
> >>>>>>> I am trying to evaluate your changes
> >>>>>>> (http://www.gossamer-threads.com/lists/xen/devel/342003)  while  I am
> >>>>>>> adding
> >>>>>>> support for PCI passthrough for xen on cavium thunder. Can you please
> >>>>>>> provide steps how you tested non-pci passthrough with your git.
> >>>>>
> >>>>>
> >>>>> Everything should be written in the cover letter. I won't copy again
> >>>>> the
> >>>>> steps here.
> >>>>>
> >>>> Can you please help us with the following so that we can test your
> >>>> patch and use it in our development
> >>>> Which linux version  was used, defconfig and domU.cfg used ?
> >>>
> >>>
> >>> Why are you talking about linux version/defconfig? As said on my cover
> >>> letter nothing as been modified in Linux. The only requirements is the
> >>> kernel must have a driver for your device.
> >>>
> >> I was not able to make out from the cover letter.
> >
> >
> > "To avoid adding code in DOM0 to manage platform device deassignment, the
> > user has to mark the device disabled in the device tree."
> >
> >
> >>>> Was it only on midway or can work on foundation model
> >>>
> >>>
> >>> Even though I did my development on midway, it should work on every
> >>> platform with devices under an SMMU.
> >>>
> >>> Regards,
> >>>
> >> It would be very helpful if you could please provide steps to test this
> >> feature.
> >> I have to integrate your work into my tree for testing PCI
> >> pass-through on cavium thunder platform.
> >
> >
> > Below, the steps to use on midway. You will have to adapt to your platform:
> >
> > 1) In the u-boot script disable the device you plan to passthrough:
> >   fdt set /soc/ethernet@fff51000 status disabled
> > 2) Ask the toolstack to passthrough the device by adding in the vm config
> > file:
> >   dtdev= [ "/soc/ethernet@fff51000" ]
> >
> > There is some minor error in my branch passthrough-v2. I've pushed a new one
> > here:
> >
> > git://xenbits.xen.org/people/julieng/xen-unstable.git passthrough-v2.1
> >
> Thanks Julien for replying. I am trying to understand your code and
> map it with the documentation (devicetree/bindings/iommu/arm,smmu.txt
> and SMMU v2 Arch).
> 
> While I was trying to understand the drivers/passthrough/arm/smmu.c I
> was not able to find a device tree that contains the compatible nodes
> in 3.16 kernel (grep smmuvx / mmu-masters returned nothing).
> Can you please help with this.

Calxeda had some in the ECX-2000 DTS, but the changes never made it
upstream:

http://lists.infradead.org/pipermail/linux-arm-kernel/2013-October/202966.html