[meta-fsl-arm][PATCH v3 0/8] iMX6 Yocto Support based on 12.09.01 BSP


Otavio Salvador
 

Hello,

This is the third version of the patchset. The segfault seems fixed
and I'd like to ask you to test and comment on them.

If we don't find new problems from this I'd like to merge this
patchset soon so we reduce the amount of not-merged patches being
handled. This is also going to offer a good base for work in next
integration steps as for example Qt support.

The the GStreamer codec support is being finished and we'll send it as
another patchset soon.

Changes for v3:

* Drop merged patches (firmware-imx, imx-lib, imx-test);
* Rework GPU packaging to include test applications;
* Fix Linux kernel configuration to have GPU module built-in;

Changes for v2:

* Fix Vivant Xorg driver to rdepends on Xorg EXA module;
* Minor cleanup in Xorg driver packaging;

Adrian Alonso (2):
xf86-dri-vivante: Add Vivant DRI driver for Xorg
xf86-video-imxfb-vivante: Add Vivant Xorg driver for iMX SoCs

Daiane Angolini (1):
linux-imx (3.0.35): Configure GPU module to be builtin

Otavio Salvador (5):
linux-imx (3.0.35): Move defconfig to mx6 directory
gpu-viv-bin-mx6q: Add Vivant GPU libraries
mesa-dri: Change provided set of libraries when building for iMX6
xserver-xf86-config: Use Vivant accelerated driver for iMX6
imx-base: Use xf86-video-imxfb-vivant for iMX6 targets

conf/machine/include/imx-base.inc | 1 +
.../gpu-viv-bin-mx6q/gpu-viv-bin-mx6q.inc | 144 ++++++++++++++++++++
.../gpu-viv-bin-mx6q/gpu-viv-bin-mx6q_12.09.01.bb | 9 ++
recipes-graphics/mesa/mesa-dri_8.0.4.bbappend | 4 +-
recipes-graphics/mesa/mesa-dri_git.bbappend | 4 +-
.../xorg-driver/xf86-dri-vivante_12.09.01.bb | 31 +++++
.../Makefile.am-remove-prefixed-include-path.patch | 29 ++++
.../xf86-video-imxfb-vivante_12.09.01.bb | 27 ++++
.../xorg-xserver/xserver-xf86-config/mx6/xorg.conf | 8 +-
.../xorg-xserver/xserver-xf86-config_0.1.bbappend | 2 +-
.../linux/linux-imx-3.0.35/{ => mx6}/defconfig | 2 +-
recipes-kernel/linux/linux-imx_3.0.35.bb | 2 +-
12 files changed, 255 insertions(+), 8 deletions(-)
create mode 100644 recipes-graphics/gpu-viv-bin-mx6q/gpu-viv-bin-mx6q.inc
create mode 100644 recipes-graphics/gpu-viv-bin-mx6q/gpu-viv-bin-mx6q_12.09.01.bb
create mode 100644 recipes-graphics/xorg-driver/xf86-dri-vivante_12.09.01.bb
create mode 100644 recipes-graphics/xorg-driver/xf86-video-imxfb-vivante/Makefile.am-remove-prefixed-include-path.patch
create mode 100644 recipes-graphics/xorg-driver/xf86-video-imxfb-vivante_12.09.01.bb
rename recipes-kernel/linux/linux-imx-3.0.35/{ => mx6}/defconfig (99%)

--
1.7.10.4


Daiane
 

Overall GPU support is up and running (with and without X11)

Please, make sure commit log and subject was fixed, and
COMPATIBLE_MACHINE is set for any needed package.


Daiane


Otavio Salvador
 

On Fri, Nov 23, 2012 at 9:18 AM, Daiane Angolini <daiane.list@...> wrote:
Overall GPU support is up and running (with and without X11)

Please, make sure commit log and subject was fixed, and
COMPATIBLE_MACHINE is set for any needed package.
I did it; this patchset is now merged! Yey!

:-D

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Thomas Senyk <thomas.senyk@...>
 

I've just tested a 'bitbake core-image-x11' and it builds without any problem.

I got a couple of warnings such as:
WARNING: File '/opt/viv_samples/vdk/tutorial2' from gpu-viv-bin-mx6q was
already stripped, this will prevent future debugging!

... I don't mind them, but "silencing" them would be helpful/good enough
(no clue if this is possible/feasible)


I would like to tell you that I've also tested this on the hardware, but it my
imx6 got "stolen" by colleagues ;)


Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?
(so far I've only tested it with 12.04.01)
- How is the gstreamer work progressing?

On Thu, November 22, 2012 16:49:33 Otavio Salvador wrote:
Hello,

This is the third version of the patchset. The segfault seems fixed
and I'd like to ask you to test and comment on them.

If we don't find new problems from this I'd like to merge this
patchset soon so we reduce the amount of not-merged patches being
handled. This is also going to offer a good base for work in next
integration steps as for example Qt support.

The the GStreamer codec support is being finished and we'll send it as
another patchset soon.

Changes for v3:

* Drop merged patches (firmware-imx, imx-lib, imx-test);
* Rework GPU packaging to include test applications;
* Fix Linux kernel configuration to have GPU module built-in;

Changes for v2:

* Fix Vivant Xorg driver to rdepends on Xorg EXA module;
* Minor cleanup in Xorg driver packaging;

Adrian Alonso (2):
xf86-dri-vivante: Add Vivant DRI driver for Xorg
xf86-video-imxfb-vivante: Add Vivant Xorg driver for iMX SoCs

Daiane Angolini (1):
linux-imx (3.0.35): Configure GPU module to be builtin

Otavio Salvador (5):
linux-imx (3.0.35): Move defconfig to mx6 directory
gpu-viv-bin-mx6q: Add Vivant GPU libraries
mesa-dri: Change provided set of libraries when building for iMX6
xserver-xf86-config: Use Vivant accelerated driver for iMX6
imx-base: Use xf86-video-imxfb-vivant for iMX6 targets

conf/machine/include/imx-base.inc | 1 +
.../gpu-viv-bin-mx6q/gpu-viv-bin-mx6q.inc | 144
++++++++++++++++++++ .../gpu-viv-bin-mx6q/gpu-viv-bin-mx6q_12.09.01.bb |
9 ++
recipes-graphics/mesa/mesa-dri_8.0.4.bbappend | 4 +-
recipes-graphics/mesa/mesa-dri_git.bbappend | 4 +-
.../xorg-driver/xf86-dri-vivante_12.09.01.bb | 31 +++++
.../Makefile.am-remove-prefixed-include-path.patch | 29 ++++
.../xf86-video-imxfb-vivante_12.09.01.bb | 27 ++++
.../xorg-xserver/xserver-xf86-config/mx6/xorg.conf | 8 +-
.../xorg-xserver/xserver-xf86-config_0.1.bbappend | 2 +-
.../linux/linux-imx-3.0.35/{ => mx6}/defconfig | 2 +-
recipes-kernel/linux/linux-imx_3.0.35.bb | 2 +-
12 files changed, 255 insertions(+), 8 deletions(-)
create mode 100644 recipes-graphics/gpu-viv-bin-mx6q/gpu-viv-bin-mx6q.inc
create mode 100644
recipes-graphics/gpu-viv-bin-mx6q/gpu-viv-bin-mx6q_12.09.01.bb create mode
100644 recipes-graphics/xorg-driver/xf86-dri-vivante_12.09.01.bb create
mode 100644
recipes-graphics/xorg-driver/xf86-video-imxfb-vivante/Makefile.am-remove-pr
efixed-include-path.patch create mode 100644
recipes-graphics/xorg-driver/xf86-video-imxfb-vivante_12.09.01.bb rename
recipes-kernel/linux/linux-imx-3.0.35/{ => mx6}/defconfig (99%)


Otavio Salvador
 

On Mon, Nov 26, 2012 at 2:23 PM, Thomas Senyk
<thomas.senyk@...> wrote:
I've just tested a 'bitbake core-image-x11' and it builds without any problem.

I got a couple of warnings such as:
WARNING: File '/opt/viv_samples/vdk/tutorial2' from gpu-viv-bin-mx6q was
already stripped, this will prevent future debugging!

... I don't mind them, but "silencing" them would be helpful/good enough
(no clue if this is possible/feasible)
I did a brief check how to quiet them but didn't find a way so I let
them on for now.

I would like to tell you that I've also tested this on the hardware, but it my
imx6 got "stolen" by colleagues ;)
Please do! :)

Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?
(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?

- How is the gstreamer work progressing?
We expect to have it done soon...

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Thomas Senyk <thomas.senyk@...>
 

On Mon, November 26, 2012 14:00:42 Daiane Angolini wrote:
On Mon, Nov 26, 2012 at 2:45 PM, Otavio Salvador

<otavio@...> wrote:
Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?

(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?
I tested tiger (under /opt/viv_samples) (and cover flow) and it worked
very fine. I tested it on X11 image and with fb image.

Do you think this test is enough? What would be other good tests?
Yes @ "enough" .. at least for me to do the next bigger test: Qt5 and eglfs
plugin.
I'll do this test as soon as I get my board back! :)


Daiane


Daiane
 

On Mon, Nov 26, 2012 at 2:45 PM, Otavio Salvador
<otavio@...> wrote:
Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?
(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?
I tested tiger (under /opt/viv_samples) (and cover flow) and it worked
very fine. I tested it on X11 image and with fb image.

Do you think this test is enough? What would be other good tests?

Daiane


Otavio Salvador
 

On Mon, Nov 26, 2012 at 3:00 PM, Daiane Angolini <daiane.list@...> wrote:
On Mon, Nov 26, 2012 at 2:45 PM, Otavio Salvador
<otavio@...> wrote:
Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?
(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?
I tested tiger (under /opt/viv_samples) (and cover flow) and it worked
very fine. I tested it on X11 image and with fb image.

Do you think this test is enough? What would be other good tests?
Later Qt will prove if it works or not ;-)

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Otavio Salvador
 

On Mon, Nov 26, 2012 at 2:53 PM, Thomas Senyk
<thomas.senyk@...> wrote:
On Mon, November 26, 2012 14:00:42 Daiane Angolini wrote:
On Mon, Nov 26, 2012 at 2:45 PM, Otavio Salvador

<otavio@...> wrote:
Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?

(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?
I tested tiger (under /opt/viv_samples) (and cover flow) and it worked
very fine. I tested it on X11 image and with fb image.

Do you think this test is enough? What would be other good tests?
Yes @ "enough" .. at least for me to do the next bigger test: Qt5 and eglfs
plugin.
I'll do this test as soon as I get my board back! :)
By the way, HP people, Martin and I are working in Qt5 packaging for
OE. We're doing our work at https://github.com/meta-qt5/meta-qt5

It is not yet working but you might want to help ;-)

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Thomas Senyk <thomas.senyk@...>
 

On Mon, November 26, 2012 15:12:08 Otavio Salvador wrote:
On Mon, Nov 26, 2012 at 2:53 PM, Thomas Senyk

<thomas.senyk@...> wrote:
On Mon, November 26, 2012 14:00:42 Daiane Angolini wrote:
On Mon, Nov 26, 2012 at 2:45 PM, Otavio Salvador

<otavio@...> wrote:
Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?

(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?
I tested tiger (under /opt/viv_samples) (and cover flow) and it worked
very fine. I tested it on X11 image and with fb image.

Do you think this test is enough? What would be other good tests?
Yes @ "enough" .. at least for me to do the next bigger test: Qt5 and
eglfs
plugin.
I'll do this test as soon as I get my board back! :)
By the way, HP people, Martin and I are working in Qt5 packaging for
OE. We're doing our work at https://github.com/meta-qt5/meta-qt5
Nice!! I took a quick look at the files ...and as with the qt4 recipes, it's
... interesting ... how complex things can get when writing a recipe ;)

I'll test it tomorrow ... I guess the proper channel to talk about this is
openembedded-devel@... ?

Greets
Thomas


It is not yet working but you might want to help ;-)


Gary Thomas <samoht.yrag@...>
 

On 2012-11-26 09:45, Otavio Salvador wrote:
On Mon, Nov 26, 2012 at 2:23 PM, Thomas Senyk
<thomas.senyk@...> wrote:
I've just tested a 'bitbake core-image-x11' and it builds without any problem.

I got a couple of warnings such as:
WARNING: File '/opt/viv_samples/vdk/tutorial2' from gpu-viv-bin-mx6q was
already stripped, this will prevent future debugging!

... I don't mind them, but "silencing" them would be helpful/good enough
(no clue if this is possible/feasible)
I did a brief check how to quiet them but didn't find a way so I let
them on for now.

I would like to tell you that I've also tested this on the hardware, but it my
imx6 got "stolen" by colleagues ;)
Please do! :)
I've tried these new changes on my imx6qsabrelite board and X now comes up (no
more segmentation violation), but the touch screen isn't working. I don't see
a module available for it (it used to be tsc2004).

How do I get the touch screen going?

Thanks


Daiane
 

I'm using LVDS and it is not working as well.

I don't know tsc2004, where do you connect this display?

You could try to use egalaxy driver, may you do that? (and let me know
the results =P)


Daiane


Otavio Salvador
 

On Mon, Nov 26, 2012 at 3:48 PM, Thomas Senyk
<thomas.senyk@...> wrote:
On Mon, November 26, 2012 15:12:08 Otavio Salvador wrote:
On Mon, Nov 26, 2012 at 2:53 PM, Thomas Senyk

<thomas.senyk@...> wrote:
On Mon, November 26, 2012 14:00:42 Daiane Angolini wrote:
On Mon, Nov 26, 2012 at 2:45 PM, Otavio Salvador

<otavio@...> wrote:
Two additional questions:
- Has anyone tested "egl on fb" (without x11) with this new drivers?

(so far I've only tested it with 12.04.01)
Daiane has tested it I think. Daiane?
I tested tiger (under /opt/viv_samples) (and cover flow) and it worked
very fine. I tested it on X11 image and with fb image.

Do you think this test is enough? What would be other good tests?
Yes @ "enough" .. at least for me to do the next bigger test: Qt5 and
eglfs
plugin.
I'll do this test as soon as I get my board back! :)
By the way, HP people, Martin and I are working in Qt5 packaging for
OE. We're doing our work at https://github.com/meta-qt5/meta-qt5
Nice!! I took a quick look at the files ...and as with the qt4 recipes, it's
... interesting ... how complex things can get when writing a recipe ;)

I'll test it tomorrow ... I guess the proper channel to talk about this is
openembedded-devel@... ?
In fact we've been discussing it at oe@...

The recipes are not working for non x86 targets as it is still trying
to run native tools during target builds. It is still work in progress
but any help is good.

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Gary Thomas <samoht.yrag@...>
 

On 2012-11-26 11:04, Daiane Angolini wrote:
I'm using LVDS and it is not working as well.

I don't know tsc2004, where do you connect this display?

You could try to use egalaxy driver, may you do that? (and let me know
the results =P)
That driver is compiled in and it's trying to use it, but failing.
I get this error when I boot:
egalax_ts 2-0004: egalax_ts: failed to read firmware version
egalax_ts: probe of 2-0004 failed with error -5

I had this working with a previous kernel, using these modules:
egalax_ts
tsc2004
galcore

n.b. when you reply, please quote what you are replying to. If you
read these messages via email (as I do), all context is lost and it's
really hard to figure out what you are saying. Not everyone uses
Google Group's swank interface!


Otavio Salvador
 

On Mon, Nov 26, 2012 at 4:27 PM, Gary Thomas <samoht.yrag@...> wrote:
On 2012-11-26 11:04, Daiane Angolini wrote:

I'm using LVDS and it is not working as well.

I don't know tsc2004, where do you connect this display?

You could try to use egalaxy driver, may you do that? (and let me know
the results =P)

That driver is compiled in and it's trying to use it, but failing.
I get this error when I boot:
egalax_ts 2-0004: egalax_ts: failed to read firmware version
egalax_ts: probe of 2-0004 failed with error -5

I had this working with a previous kernel, using these modules:
egalax_ts
tsc2004
Right.

galcore
This is now built-in.

Since you had it working previously please compare the dmesg output
between both kernels so we can have a light where to look for it?

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Gary Thomas <samoht.yrag@...>
 

On 2012-11-26 11:40, Otavio Salvador wrote:
On Mon, Nov 26, 2012 at 4:27 PM, Gary Thomas <samoht.yrag@...> wrote:
On 2012-11-26 11:04, Daiane Angolini wrote:

I'm using LVDS and it is not working as well.

I don't know tsc2004, where do you connect this display?

You could try to use egalaxy driver, may you do that? (and let me know
the results =P)

That driver is compiled in and it's trying to use it, but failing.
I get this error when I boot:
egalax_ts 2-0004: egalax_ts: failed to read firmware version
egalax_ts: probe of 2-0004 failed with error -5

I had this working with a previous kernel, using these modules:
egalax_ts
tsc2004
Right.

galcore
This is now built-in.

Since you had it working previously please compare the dmesg output
between both kernels so we can have a light where to look for it?
I'll see what I can do. I built it using meta-fsl-arm in October, so
I'll have to rewind meta-fsl-arm to that time period since the linux
kernel recipes have been updated (and older versions deleted).

Does someone have a working image for the sabrelite that uses this
touch screen (it was the only one available as I recall when I bought
the board in October)? I'd really like to try a known image so as to
eliminate any errors I might have added when making the build.


Otavio Salvador
 

On Mon, Nov 26, 2012 at 4:54 PM, Gary Thomas <samoht.yrag@...> wrote:
On 2012-11-26 11:40, Otavio Salvador wrote:

On Mon, Nov 26, 2012 at 4:27 PM, Gary Thomas <samoht.yrag@...>
wrote:

On 2012-11-26 11:04, Daiane Angolini wrote:


I'm using LVDS and it is not working as well.

I don't know tsc2004, where do you connect this display?

You could try to use egalaxy driver, may you do that? (and let me know
the results =P)


That driver is compiled in and it's trying to use it, but failing.
I get this error when I boot:
egalax_ts 2-0004: egalax_ts: failed to read firmware version
egalax_ts: probe of 2-0004 failed with error -5

I had this working with a previous kernel, using these modules:
egalax_ts
tsc2004

Right.

galcore

This is now built-in.

Since you had it working previously please compare the dmesg output
between both kernels so we can have a light where to look for it?

I'll see what I can do. I built it using meta-fsl-arm in October, so
I'll have to rewind meta-fsl-arm to that time period since the linux
kernel recipes have been updated (and older versions deleted).

Does someone have a working image for the sabrelite that uses this
touch screen (it was the only one available as I recall when I bought
the board in October)? I'd really like to try a known image so as to
eliminate any errors I might have added when making the build.
In any case this seems to be kernel related and not a BSP issue.
Fabio, could you take a look?

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Fabio Estevam
 

On Mon, Nov 26, 2012 at 4:57 PM, Otavio Salvador
<otavio@...> wrote:

In any case this seems to be kernel related and not a BSP issue.
Fabio, could you take a look?
The "failed to read firmware version" seems to be an I2C issue. The
I2C controllers send some commands to the panel, and it is failing.

I will take try to reproduce this when I have a chance.

Regards,

Fabio Estevam


Otavio Salvador
 

On Fri, Nov 23, 2012 at 9:35 AM, Otavio Salvador
<otavio@...> wrote:
On Fri, Nov 23, 2012 at 9:18 AM, Daiane Angolini <daiane.list@...> wrote:
Overall GPU support is up and running (with and without X11)

Please, make sure commit log and subject was fixed, and
COMPATIBLE_MACHINE is set for any needed package.
I did it; this patchset is now merged! Yey!
I just merged the VPU patches as well :-)

--
Otavio Salvador O.S. Systems
E-mail: otavio@... http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br


Fabio Estevam
 

On Tue, Nov 27, 2012 at 1:13 AM, Fabio Estevam <festevam@...> wrote:
On Mon, Nov 26, 2012 at 4:57 PM, Otavio Salvador
<otavio@...> wrote:

In any case this seems to be kernel related and not a BSP issue.
Fabio, could you take a look?
The "failed to read firmware version" seems to be an I2C issue. The
I2C controllers send some commands to the panel, and it is failing.
I think the I2C read failure is normal, as the sabrelite panel is not egalax.

You should use the Hannstar panel.

Regards,

Fabio Estevam