Vivante GPU on i.MX6solo

Vivante GPU on i.MX6solo

[Gary Thomas]

I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
to run anything using the accelerated graphics, e.g. X server using
the Vivante libraries, it crashes hard.

I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
Vivante GC2000.

Is there any way to get Vivante GPU support for my board/SOC?

Thanks

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: Vivante GPU on i.MX6solo

[Otavio Salvador]

Hello Gary,

On Mon, Mar 24, 2014 at 12:13 PM, Gary Thomas <gary@mlbassoc.com> wrote:
> I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
> a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
> to run anything using the accelerated graphics, e.g. X server using
> the Vivante libraries, it crashes hard.
>
> I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
> Vivante GC2000.
>
> Is there any way to get Vivante GPU support for my board/SOC?

Well, it is lacking a lot of information so it is quite hard to guess.
My guess is that you have an old BSP kernel and without the 4.6.9p13
patches. When those things happen you see a segfault in Xorg
initialization using the Vivante drivers.

Good luck debugging.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

Re: Vivante GPU on i.MX6solo

[Gary Thomas]

On 2014-03-24 20:07, Otavio Salvador wrote:
> Hello Gary,
> 
> On Mon, Mar 24, 2014 at 12:13 PM, Gary Thomas <gary@mlbassoc.com> wrote:
>> I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
>> a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
>> to run anything using the accelerated graphics, e.g. X server using
>> the Vivante libraries, it crashes hard.
>>
>> I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
>> Vivante GC2000.
>>
>> Is there any way to get Vivante GPU support for my board/SOC?
> 
> Well, it is lacking a lot of information so it is quite hard to guess.
> My guess is that you have an old BSP kernel and without the 4.6.9p13
> patches. When those things happen you see a segfault in Xorg
> initialization using the Vivante drivers.

This well could be - I use the linux-boundary recipe for 3.0.35+
(my board is nearly identical to a SabreLite from Boundary Devices)

I'm a bit confused by the '4.6.9p13' reference.  Is that part of
a GPU recipe version?  or a kernel version?

Can you point me to any discussion of this, or the kernel (which tree)
revision that made this change, etc?

Thanks

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: Vivante GPU on i.MX6solo

[Otavio Salvador]

On Tue, Mar 25, 2014 at 8:03 AM, Gary Thomas <gary@mlbassoc.com> wrote:
> On 2014-03-24 20:07, Otavio Salvador wrote:
>> Hello Gary,
>>
>> On Mon, Mar 24, 2014 at 12:13 PM, Gary Thomas <gary@mlbassoc.com> wrote:
>>> I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
>>> a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
>>> to run anything using the accelerated graphics, e.g. X server using
>>> the Vivante libraries, it crashes hard.
>>>
>>> I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
>>> Vivante GC2000.
>>>
>>> Is there any way to get Vivante GPU support for my board/SOC?
>>
>> Well, it is lacking a lot of information so it is quite hard to guess.
>> My guess is that you have an old BSP kernel and without the 4.6.9p13
>> patches. When those things happen you see a segfault in Xorg
>> initialization using the Vivante drivers.
>
> This well could be - I use the linux-boundary recipe for 3.0.35+
> (my board is nearly identical to a SabreLite from Boundary Devices)

Currently linux-boundary has it right. It depends if you have dropped
the patches applied in our recipe or not.

> I'm a bit confused by the '4.6.9p13' reference.  Is that part of
> a GPU recipe version?  or a kernel version?

Yes, GPU driver version.

> Can you point me to any discussion of this, or the kernel (which tree)
> revision that made this change, etc?

https://github.com/Freescale/meta-fsl-arm-extra/commit/37c793bea49a1140a414660c262d62c3e2b130c9

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

Re: Vivante GPU on i.MX6solo

[Gary Thomas]

On 2014-03-25 05:40, Otavio Salvador wrote:
> On Tue, Mar 25, 2014 at 8:03 AM, Gary Thomas <gary@mlbassoc.com> wrote:
>> On 2014-03-24 20:07, Otavio Salvador wrote:
>>> Hello Gary,
>>>
>>> On Mon, Mar 24, 2014 at 12:13 PM, Gary Thomas <gary@mlbassoc.com> wrote:
>>>> I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
>>>> a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
>>>> to run anything using the accelerated graphics, e.g. X server using
>>>> the Vivante libraries, it crashes hard.
>>>>
>>>> I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
>>>> Vivante GC2000.
>>>>
>>>> Is there any way to get Vivante GPU support for my board/SOC?
>>>
>>> Well, it is lacking a lot of information so it is quite hard to guess.
>>> My guess is that you have an old BSP kernel and without the 4.6.9p13
>>> patches. When those things happen you see a segfault in Xorg
>>> initialization using the Vivante drivers.
>>
>> This well could be - I use the linux-boundary recipe for 3.0.35+
>> (my board is nearly identical to a SabreLite from Boundary Devices)
> 
> Currently linux-boundary has it right. It depends if you have dropped
> the patches applied in our recipe or not.
> 
>> I'm a bit confused by the '4.6.9p13' reference.  Is that part of
>> a GPU recipe version?  or a kernel version?
> 
> Yes, GPU driver version.
> 
>> Can you point me to any discussion of this, or the kernel (which tree)
>> revision that made this change, etc?
> 
> https://github.com/Freescale/meta-fsl-arm-extra/commit/37c793bea49a1140a414660c262d62c3e2b130c9
> 

Yes, it seems that I had missed this patch which seems to have only
been on master-next until 2014-03-20.

I'll give it another go now.

Thanks

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: Vivante GPU on i.MX6solo

[Otavio Salvador]

On Tue, Mar 25, 2014 at 9:03 AM, Gary Thomas <gary@mlbassoc.com> wrote:
> On 2014-03-25 05:40, Otavio Salvador wrote:
>> On Tue, Mar 25, 2014 at 8:03 AM, Gary Thomas <gary@mlbassoc.com> wrote:
>>> On 2014-03-24 20:07, Otavio Salvador wrote:
>>>> Hello Gary,
>>>>
>>>> On Mon, Mar 24, 2014 at 12:13 PM, Gary Thomas <gary@mlbassoc.com> wrote:
>>>>> I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
>>>>> a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
>>>>> to run anything using the accelerated graphics, e.g. X server using
>>>>> the Vivante libraries, it crashes hard.
>>>>>
>>>>> I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
>>>>> Vivante GC2000.
>>>>>
>>>>> Is there any way to get Vivante GPU support for my board/SOC?
>>>>
>>>> Well, it is lacking a lot of information so it is quite hard to guess.
>>>> My guess is that you have an old BSP kernel and without the 4.6.9p13
>>>> patches. When those things happen you see a segfault in Xorg
>>>> initialization using the Vivante drivers.
>>>
>>> This well could be - I use the linux-boundary recipe for 3.0.35+
>>> (my board is nearly identical to a SabreLite from Boundary Devices)
>>
>> Currently linux-boundary has it right. It depends if you have dropped
>> the patches applied in our recipe or not.
>>
>>> I'm a bit confused by the '4.6.9p13' reference.  Is that part of
>>> a GPU recipe version?  or a kernel version?
>>
>> Yes, GPU driver version.
>>
>>> Can you point me to any discussion of this, or the kernel (which tree)
>>> revision that made this change, etc?
>>
>> https://github.com/Freescale/meta-fsl-arm-extra/commit/37c793bea49a1140a414660c262d62c3e2b130c9
>>
>
> Yes, it seems that I had missed this patch which seems to have only
> been on master-next until 2014-03-20.
>
> I'll give it another go now.

Great; this should get it working.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

Re: Vivante GPU on i.MX6solo

[Gary Thomas]

On 2014-03-25 06:08, Otavio Salvador wrote:
> On Tue, Mar 25, 2014 at 9:03 AM, Gary Thomas <gary@mlbassoc.com> wrote:
>> On 2014-03-25 05:40, Otavio Salvador wrote:
>>> On Tue, Mar 25, 2014 at 8:03 AM, Gary Thomas <gary@mlbassoc.com> wrote:
>>>> On 2014-03-24 20:07, Otavio Salvador wrote:
>>>>> Hello Gary,
>>>>>
>>>>> On Mon, Mar 24, 2014 at 12:13 PM, Gary Thomas <gary@mlbassoc.com> wrote:
>>>>>> I have a board which uses the i.MX6solo chip.  This CPU (SOC) has
>>>>>> a different GPU than its cousins i.MX6Q and i.MX6dual.  When I try
>>>>>> to run anything using the accelerated graphics, e.g. X server using
>>>>>> the Vivante libraries, it crashes hard.
>>>>>>
>>>>>> I know that my SOC has a Vivante GC880 whereas the i.MX6Q uses the
>>>>>> Vivante GC2000.
>>>>>>
>>>>>> Is there any way to get Vivante GPU support for my board/SOC?
>>>>>
>>>>> Well, it is lacking a lot of information so it is quite hard to guess.
>>>>> My guess is that you have an old BSP kernel and without the 4.6.9p13
>>>>> patches. When those things happen you see a segfault in Xorg
>>>>> initialization using the Vivante drivers.
>>>>
>>>> This well could be - I use the linux-boundary recipe for 3.0.35+
>>>> (my board is nearly identical to a SabreLite from Boundary Devices)
>>>
>>> Currently linux-boundary has it right. It depends if you have dropped
>>> the patches applied in our recipe or not.
>>>
>>>> I'm a bit confused by the '4.6.9p13' reference.  Is that part of
>>>> a GPU recipe version?  or a kernel version?
>>>
>>> Yes, GPU driver version.
>>>
>>>> Can you point me to any discussion of this, or the kernel (which tree)
>>>> revision that made this change, etc?
>>>
>>> https://github.com/Freescale/meta-fsl-arm-extra/commit/37c793bea49a1140a414660c262d62c3e2b130c9
>>>
>>
>> Yes, it seems that I had missed this patch which seems to have only
>> been on master-next until 2014-03-20.
>>
>> I'll give it another go now.
> 
> Great; this should get it working.
> 

I've merged up to the latest boundary branch+rev and now the
GPU is working on my i.MX6solo :-)

Thanks

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------