Wandboard kernel updated to 3.10.17-1.0.1

[Otavio Salvador]

Hello,

The Wandboard is one of the boards which are going to be used during
the Yocto Project Developer Day[1] which with be take place tomorrow,
after ELCE, in Düsseldorf, Germany.

1. https://www.yoctoproject.org/tools-resources/events/yocto-project-developer-day-elce-2014

During the final preparation of the board for the training, it has
been identified a bad behavior of 3.10.17-1.0.0 kernel regarding the
GPIO pins exposed[2] in the external pinheader JP-4. This has been
caused due a bad IOMUX configuration which is now fixed. As I were
touching the kernel I used the same opportunity to update[3] it to the
latest GA release which is provided by Freescale and update it to the
3.10.17-1.0.1 which fixes a GPU freeze issue.

2. http://wiki.wandboard.org/index.php/External_gpios
3. https://github.com/wandboard-org/linux/tree/wandboard_imx_3.10.17_1.0.1_ga

The fix is pushed to the github tree and has been also updated in the
Yocto Project BSP in Daisy[4] (1.6 release) and Dizzy[5] (upcoming 1.7
release); you can also use FSL Community BSP[6] to try this out.

4. https://github.com/Freescale/meta-fsl-arm-extra/tree/daisy
5. https://github.com/Freescale/meta-fsl-arm-extra/tree/dizzy
6. http://freescale.github.io

Good luck :-)

--
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

[Alfonso Tamés]

Awsome! thanks Otavio and good luck in the Developers Day.

[hansan]

Dear Otavio,

Thanks for the update to the kernel!!

I tried today this new kernel on my wandboard quad (based on a archlinux arm rootfs) and I did not get gpio75 gpio191 and gpio72 on J4 working as output.
echo 75 > /sys/class/gpio/export
works, however:
echo out > /sys/class/gpio/gpio75/direction
Does not make that the pin can be toggled by echoing a 0 or 1 to value.
Can you confirm that this should work / does not work?
I did not see directly a problem in the device tree, but I have to admit that I don't know so well where to start looking.

Greetings,

Han


On Wednesday, October 15, 2014 7:10:21 PM UTC+2, Otavio Salvador wrote:
snip

During the final preparation of the board for the training, it has
been identified a bad behavior of 3.10.17-1.0.0 kernel regarding the
GPIO pins exposed[2] in the external pinheader JP-4. This has been
caused due a bad IOMUX configuration which is now fixed. As I were
touching the kernel I used the same opportunity to update[3] it to the
latest GA release which is provided by Freescale and update it to the
3.10.17-1.0.1 which fixes a GPU freeze issue.

2. http://wiki.wandboard.org/index.php/External_gpios
3. https://github.com/wandboard-org/linux/tree/wandboard_imx_3.10.17_1.0.1_ga

Snip

[Otavio Salvador]

On Sun, Oct 26, 2014 at 6:24 PM, hansan <han.ha...@gmail.com> wrote:
> Thanks for the update to the kernel!!
>
> I tried today this new kernel on my wandboard quad (based on a archlinux arm
> rootfs) and I did not get gpio75 gpio191 and gpio72 on J4 working as output.
> echo 75 > /sys/class/gpio/export
> works, however:
> echo out > /sys/class/gpio/gpio75/direction
> Does not make that the pin can be toggled by echoing a 0 or 1 to value.
> Can you confirm that this should work / does not work?
> I did not see directly a problem in the device tree, but I have to admit
> that I don't know so well where to start looking.

It does; you may not have updated the dtb file.

[hansan]

On Sunday, October 26, 2014 9:42:56 PM UTC+1, Otavio Salvador wrote:

> Can you confirm that this should work / does not work?
> I did not see directly a problem in the device tree, but I have to admit
> that I don't know so well where to start looking.

It does; you may not have updated the dtb file.

I double checked and I have te new dtb files in my /boot/dtbs directories.
WIth the 3.0.35 kernel (yocto 1.5) are gpio pin 75,  191 and 72 working; it is not a hardware problem.....
Maybe there is happening more in the kernel which makes that these 3 gpio pin are blocked. It is for me difficult to debug, but I will learn something.

[John Weber]

I believe that for Yocto, the DTB is loaded from the fat partition, not /boot in the ext partition (rootfs).  Check Uboot to see where it loads from. 

--
You received this message because you are subscribed to the Google Groups "Wandboard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wandboard+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[hansan]

Thanks for the tip. I double checked and this is also working correctly, uboot mentions:
loading /boot/dtbs/imx6q-wandboard.dtb ...
 43322 bytes read in 82 ms (515.6 KiB/s)
[hansan@wandboard ~]$ ls -l /boot/dtbs/
total 88
-rw-r--r-- 1 root root 42669 Nov  1 21:35 imx6dl-wandboard.dtb
-rw-r--r-- 1 root root 43322 Nov  1 21:35 imx6q-wandboard.dtb

I believe they take the same.
Today I also noticed that I cannot read the value of gpio101.

I have used up to now archlinux. What I will try next is to compile the kernel (again  wandboard_imx_3.10.17_1.0.1_ga) and uboot in yocto to see if that version works for me instead of the versions made in the archlinux environment. (The kernel should be the same because I use the same source code but maybe it is a u-boot issue)

[hansan]

I found at last some time to continue with this.

I tried a yocto image with the right kernel:
root@wandboard-quad:~# uname -a
Linux wandboard-quad 3.10.17-1.0.1-wandboard+ga6e7fc5 #2 SMP PREEMPT Sat Dec 13 20:32:49 CET 2014 armv7l GNU/Linux

And with this kernel and image are gpio 75, 191 and 72 also not responding they stay at a constant level, while the other gpio's on JP4 can be toggled.  I have the same problem with the linaro 3.14.# based kernel. 
I can toggle all the GPIO's with a yocto image with the very old  3.0.35 kernel.

I see as major change between the kernels the introduction of the device tree. Gpio 75 and gpio 72 are also related to the boot mode of the imx6, maybe that has something to do with "blocking" these pins. Or the pinmux is not setup correctly..

Just a stupid question; is someone else able to get all the GPIO on JP4 working with the newer kernels?

[Otavio Salvador]

On Tue, Dec 23, 2014 at 7:59 PM, hansan <han.ha...@gmail.com> wrote:
> I found at last some time to continue with this.
>
> I tried a yocto image with the right kernel:
> root@wandboard-quad:~# uname -a
> Linux wandboard-quad 3.10.17-1.0.1-wandboard+ga6e7fc5 #2 SMP PREEMPT Sat Dec
> 13 20:32:49 CET 2014 armv7l GNU/Linux
>
> And with this kernel and image are gpio 75, 191 and 72 also not responding
> they stay at a constant level, while the other gpio's on JP4 can be toggled.
> I have the same problem with the linaro 3.14.# based kernel.
> I can toggle all the GPIO's with a yocto image with the very old 3.0.35
> kernel.
>
> I see as major change between the kernels the introduction of the device
> tree. Gpio 75 and gpio 72 are also related to the boot mode of the imx6,
> maybe that has something to do with "blocking" these pins. Or the pinmux is
> not setup correctly..
>
> Just a stupid question; is someone else able to get all the GPIO on JP4
> working with the newer kernels?

Yes. The kernel has been updated, by Alfonso, to 3.10.17-1.0.2.

This has been used in the Yocto Project Dev Day training, in Germany,
last October with succeed. So I think it is alright.

[hansan]

On Wednesday, December 24, 2014 12:13:55 AM UTC+1, Otavio Salvador wrote:

On Tue, Dec 23, 2014 at 7:59 PM, hansan <han.ha...@gmail.com> wrote:
> I found at last some time to continue with this.
>
>

> Just a stupid question; is someone else able to get all the GPIO on JP4
> working with the newer kernels?

Yes. The kernel has been updated, by Alfonso, to 3.10.17-1.0.2.

This has been used in the Yocto Project Dev Day training, in Germany,
last October with succeed. So I think it is alright.

 
Did they really check all the GPIO pins?  For me are 5 of them working as expected and without any problem. Three have a problem, 75 and 72 stay at 0.45V and 191 stays at 0V and cannot be toggled as output (or do not react).
I just checked the yocto 3.10.13 kernel (Linux wandboard-quad 3.10.31_1.1.0_beta2-wandboard) of yocto master and that version has the same problem.

I will try the 3.10.17-1.0.2 kernel (I did try already the 1.0.1 version but of the 1.0.2 I am not quite certain..) and will report.

have a nice holiday season!