[pandaboard] what linux distro's still support the pandaboard

[Kristoff]

Hi all,



I have a pandaboard (non-ES) laying on my desk for ... well ... very
long. Althou I have moved into microcontrollers (and now also FPGAs), I
do would like the thing to do something usefull, be it a NAS,
mediaplayer, or an android box or whatever.


Are there still linux distro's that officially support the pandaboard?


(I did try archlinux about a year ago which worked, ... untill you
pulled a lot of traffic over it and then the ethernet-interface froze. :-( )



Any advice on still using the pandaboard beside a (very expensive) paper
holder?



Cheerio! Kr. Bonne.

[H. Nikolaus Schaller]

Hi,

Post by Kristoff
Hi all,
I have a pandaboard (non-ES) laying on my desk for ... well ... very long. Althou I have moved into microcontrollers (and now also FPGAs), I do would like the thing to do something usefull, be it a NAS, mediaplayer, or an android box or whatever.
Are there still linux distro's that officially support the pandaboard?
(I did try archlinux about a year ago which worked, ... untill you pulled a lot of traffic over it and then the ethernet-interface froze. :-( )
Any advice on still using the pandaboard beside a (very expensive) paper holder?

we use Debian Jessie and this U-Boot / kernel (and maintain on low priority):
but on PandaBoard ES. If you can find out for us what has to be changed (especially
for booting) we can add Pandaboard (non-ES).

http://download.goldelico.com/letux-kernel/letux-4.7-rc7/
http://download.goldelico.com/letux-u-boot/PandaBoard-ES/
http://download.goldelico.com/letux-debian-rootfs/

A bootable SD card can be pulled together by http://projects.goldelico.com/p/gta04-makesd/

As soon as it works you can extend or replace Debian with something different.

-- hns

[Han Hartgers]

Dear Kristoff,

Happy new year!
Did you find a good working distribution for the Pandaboard? I have a
Pandaboard A1 and I am also looking for a good (modern) distribution for
my board to use it as scan and print server.

The things Robert Nelson is doing at:
https://eewiki.net/display/linuxonarm/PandaBoard is quite good. However
mainline u-boot (16.11) and kernel (4.9x) are not configuring "things"
correctly any more. Therefore is at the on-board Wifi interface not working
at this moment. I am currently wondering if I should try to get the things
or Robert Nelson working or if I should look for a different distribution.
Therefore I wanted to know if you have find a good working distribution
except for the old ubuntu 12.04.

Greetings,

Han

[Robert Nelson]

Post by Han Hartgers
Dear Kristoff,
Happy new year!
Did you find a good working distribution for the Pandaboard? I have a
Pandaboard A1 and I am also looking for a good (modern) distribution for my
board to use it as scan and print server.
https://eewiki.net/display/linuxonarm/PandaBoard is quite good. However
mainline u-boot (16.11) and kernel (4.9x) are not configuring "things"
correctly any more. Therefore is at the on-board Wifi interface not working
at this moment. I am currently wondering if I should try to get the things
or Robert Nelson working or if I should look for a different distribution.
Therefore I wanted to know if you have find a good working distribution
except for the old ubuntu 12.04.

Other then testing the display in v4.9.x. I haven't done really much
with the panda's anymore (they aren't the backbone of my build farm
anymore, wandboard's have replaced them).

For wifi on the A1, i thing we are missing a 32khz clock somewhere, as
back in the v3.7.x era (pre-device tree's), my A1's needed this:

https://github.com/RobertCNelson/stable-kernel/commit/2b19f1fd3b53f3a3519a2809cff59735361bbd80

Regards,

[Han Hartgers]

Post by Han Hartgers

Post by Han Hartgers
Dear Kristoff,
Happy new year!
Did you find a good working distribution for the Pandaboard? I have a
Pandaboard A1 and I am also looking for a good (modern) distribution

for my

Post by Han Hartgers
board to use it as scan and print server.
https://eewiki.net/display/linuxonarm/PandaBoard is quite good. However
mainline u-boot (16.11) and kernel (4.9x) are not configuring "things"
correctly any more. Therefore is at the on-board Wifi interface not

working

Post by Han Hartgers
at this moment. I am currently wondering if I should try to get the

things

Post by Han Hartgers
or Robert Nelson working or if I should look for a different

distribution.

Post by Han Hartgers
Therefore I wanted to know if you have find a good working distribution
except for the old ubuntu 12.04.

Other then testing the display in v4.9.x. I haven't done really much
with the panda's anymore (they aren't the backbone of my build farm
anymore, wandboard's have replaced them).
For wifi on the A1, i thing we are missing a 32khz clock somewhere, as
https://github.com/RobertCNelson/stable-kernel/commit/2b19f1fd3b53f3a3519a2809cff59735361bbd80
Regards,
--
Robert Nelson
https://rcn-ee.com/

Dear Robert,

Thanks for the pointer to the 3.7.x kernels. I have compiled and tried but
that was also not a big success.
(I did "get clone https://github.com/RobertCNelson/stable-kernel ; git
checkout origin origin/v3.7.x ; needed to adapt tools/host_det.sh to
support ubuntu 16.04 and ./build_kernel.sh)

The kernel boots but cannot find the root file system, it stops with "[
2.194671] Waiting for root device /dev/mmcblk0p1..."
It can be that I look for the wrong root device (did that change?) but I
see quite some errors relating with "mmc" on the serial terminal so I think
that this kernel cannot start the mmc system correctly.
[ 1.731323] omap_hsmmc 4809c000.mmc: unable to select pin group
[ 1.737640] omap_hsmmc 4809c000.mmc: Failed to get debounce clk
[ 1.743896] /ocp/***@4809c000: can't find DMA channel
[ 1.749206] omap_hsmmc 4809c000.mmc: unable to obtain RX DMA engine
channel 62
[ 1.756866] omap_hsmmc 480d5000.mmc: unable to select pin group
[ 1.763153] omap_hsmmc 480d5000.mmc: Failed to get debounce clk
[ 1.769378] /ocp/***@480d5000: can't find DMA channel
[ 1.774719] omap_hsmmc 480d5000.mmc: unable to obtain RX DMA engine
channel 60
Giving other errors in the attached log of the serial console it nearly
looks if the board is not set-up as described in the device tree. Did there
change something that I should "communicate" through u-boot?
Probably I have to look for an u-boot out of the same era and not using the
2016.11 one. Do you have in your git a pointer for that as well?

An other route, i am looking into, is to use the patches mentioned here
https://bugzilla.kernel.org/show_bug.cgi?id=119331 with some effort I get
this compiled for kernel 4.9 but no real success. I have the suspicion that
I have to adapt the dts file as well or some hook in the patch is not valid
and the init/activation of the CLK32GK is therefore not done.

It keeps me busy when the children are in bed.... (and I only wanted to
make the copy button on the scanner working for my wife. ;-))

It is a pity that TI had abandoned this OMAP4.... it was a very promising
board. The beagle-family and panda made the arm boards more accessible for
hobby developers like me.
I like the wandboard as well, but does your kernels support the VPU?

Kind regards,

Han

[Robert Nelson]

Post by Han Hartgers
Dear Robert,
Thanks for the pointer to the 3.7.x kernels. I have compiled and tried but
that was also not a big success.
(I did "get clone https://github.com/RobertCNelson/stable-kernel ; git
checkout origin origin/v3.7.x ; needed to adapt tools/host_det.sh to support
ubuntu 16.04 and ./build_kernel.sh)

yeah, it's been a few years for the branch, hence why i just linked to
the specific patch ;)

Post by Han Hartgers
The kernel boots but cannot find the root file system, it stops with "[
2.194671] Waiting for root device /dev/mmcblk0p1..."
It can be that I look for the wrong root device (did that change?) but I see
quite some errors relating with "mmc" on the serial terminal so I think that
this kernel cannot start the mmc system correctly.
[ 1.731323] omap_hsmmc 4809c000.mmc: unable to select pin group
[ 1.737640] omap_hsmmc 4809c000.mmc: Failed to get debounce clk
[ 1.749206] omap_hsmmc 4809c000.mmc: unable to obtain RX DMA engine
channel 62
[ 1.756866] omap_hsmmc 480d5000.mmc: unable to select pin group
[ 1.763153] omap_hsmmc 480d5000.mmc: Failed to get debounce clk
[ 1.774719] omap_hsmmc 480d5000.mmc: unable to obtain RX DMA engine
channel 60

well v3.7.x is "pre-dt" for omap4, so boot the board file... It was
really till 3.10/v3.11 did the omap4 becomes "useable" with dt..

Post by Han Hartgers
Giving other errors in the attached log of the serial console it nearly
looks if the board is not set-up as described in the device tree. Did there
change something that I should "communicate" through u-boot?
Probably I have to look for an u-boot out of the same era and not using the
2016.11 one. Do you have in your git a pointer for that as well?
An other route, i am looking into, is to use the patches mentioned here
https://bugzilla.kernel.org/show_bug.cgi?id=119331 with some effort I get
this compiled for kernel 4.9 but no real success. I have the suspicion that
I have to adapt the dts file as well or some hook in the patch is not valid
and the init/activation of the CLK32GK is therefore not done.
It keeps me busy when the children are in bed.... (and I only wanted to make
the copy button on the scanner working for my wife. ;-))
It is a pity that TI had abandoned this OMAP4.... it was a very promising
board.

Well it was abandoned in late 2012

http://www.eetimes.com/document.asp?doc_id=1262782

4 years later, most of it still works.. ;)

Post by Han Hartgers
The beagle-family and panda made the arm boards more accessible for
hobby developers like me.
I like the wandboard as well, but does your kernels support the VPU?

wandboard has real opensource 2d/3d acceleration with etnaviv

Just follow:
https://eewiki.net/display/linuxonarm/Wandboard

and use v4.9.x+

Regards.

[Han Hartgers]

Post by Kristoff

Post by Han Hartgers
Dear Robert,
Thanks for the pointer to the 3.7.x kernels. I have compiled and tried

but

Post by Han Hartgers
that was also not a big success.
(I did "get clone https://github.com/RobertCNelson/stable-kernel ; git
checkout origin origin/v3.7.x ; needed to adapt tools/host_det.sh to

support

Post by Han Hartgers
ubuntu 16.04 and ./build_kernel.sh)

yeah, it's been a few years for the branch, hence why i just linked to
the specific patch ;)

Post by Han Hartgers
The kernel boots but cannot find the root file system, it stops with "[
2.194671] Waiting for root device /dev/mmcblk0p1..."
It can be that I look for the wrong root device (did that change?) but I

see

Post by Han Hartgers
quite some errors relating with "mmc" on the serial terminal so I think

that

Post by Han Hartgers
this kernel cannot start the mmc system correctly.
[ 1.731323] omap_hsmmc 4809c000.mmc: unable to select pin group
[ 1.737640] omap_hsmmc 4809c000.mmc: Failed to get debounce clk
[ 1.749206] omap_hsmmc 4809c000.mmc: unable to obtain RX DMA engine
channel 62
[ 1.756866] omap_hsmmc 480d5000.mmc: unable to select pin group
[ 1.763153] omap_hsmmc 480d5000.mmc: Failed to get debounce clk
[ 1.774719] omap_hsmmc 480d5000.mmc: unable to obtain RX DMA engine
channel 60

It is a kind of a mix to me. I do have a omap4-panda.dtb file with this
kernel. Probably is the contents not so "complete" or is only u-boot using
this? (Loading the omap4-pandaES.dtb did not change the behaviour of my
kernel booting...)
Any thought on what this waiting for root device /dev/mmcblk0p1 can be?

Post by Kristoff
well v3.7.x is "pre-dt" for omap4, so boot the board file... It was
really till 3.10/v3.11 did the omap4 becomes "useable" with dt..

Post by Han Hartgers
Giving other errors in the attached log of the serial console it nearly
looks if the board is not set-up as described in the device tree. Did

there

Post by Han Hartgers
change something that I should "communicate" through u-boot?
Probably I have to look for an u-boot out of the same era and not using

the

Post by Han Hartgers
2016.11 one. Do you have in your git a pointer for that as well?
An other route, i am looking into, is to use the patches mentioned here
https://bugzilla.kernel.org/show_bug.cgi?id=119331 with some effort I

get

Post by Han Hartgers
this compiled for kernel 4.9 but no real success. I have the suspicion

that

Post by Han Hartgers
I have to adapt the dts file as well or some hook in the patch is not

valid

Post by Han Hartgers
and the init/activation of the CLK32GK is therefore not done.
It keeps me busy when the children are in bed.... (and I only wanted to

make

Post by Han Hartgers
the copy button on the scanner working for my wife. ;-))
It is a pity that TI had abandoned this OMAP4.... it was a very

promising

Post by Han Hartgers
board.

Well it was abandoned in late 2012
http://www.eetimes.com/document.asp?doc_id=1262782
4 years later, most of it still works.. ;)

Post by Han Hartgers
The beagle-family and panda made the arm boards more accessible for
hobby developers like me.
I like the wandboard as well, but does your kernels support the VPU?

I don't need the 2d/3d acceleration but I need the hardware h264
encoder/decoder as can used with gstreamer, that was a few years back a bit
of a hit or miss with which kernel I used.

Post by Kristoff
wandboard has real opensource 2d/3d acceleration with etnaviv
https://eewiki.net/display/linuxonarm/Wandboard
and use v4.9.x+
Regards.
--
Robert Nelson
https://rcn-ee.com/

[Robert Nelson]

Post by Han Hartgers
It is a kind of a mix to me. I do have a omap4-panda.dtb file with this
kernel. Probably is the contents not so "complete" or is only u-boot using
this? (Loading the omap4-pandaES.dtb did not change the behaviour of my
kernel booting...)
Any thought on what this waiting for root device /dev/mmcblk0p1 can be?

Yeah, like i said previously, v3.7.x is "pre-dt" for omap4, so STOP
using "omap4-panda.dtb/omap4-pandaES.dtb" and just use the classic
board file..

Post by Han Hartgers
I don't need the 2d/3d acceleration but I need the hardware h264
encoder/decoder as can used with gstreamer, that was a few years back a bit
of a hit or miss with which kernel I used.

Well someone needs to reverse-engineer that first..

Regards,

[Han Hartgers]

Post by Robert Nelson

Post by Han Hartgers
It is a kind of a mix to me. I do have a omap4-panda.dtb file with this
kernel. Probably is the contents not so "complete" or is only u-boot

using

Post by Han Hartgers
this? (Loading the omap4-pandaES.dtb did not change the behaviour of my
kernel booting...)
Any thought on what this waiting for root device /dev/mmcblk0p1 can be?

Yeah, like i said previously, v3.7.x is "pre-dt" for omap4, so STOP
using "omap4-panda.dtb/omap4-pandaES.dtb" and just use the classic
board file..
I do not fully understand what you mean here.

I (or actually your build script) is doing a make ARCH=arm
CROSS_COMPILE=omap2plus_defconfig is that the classic board file
way-of-work you point to?
To my surprise I do get as well: am335x-bone.dtb am335x-evm.dtb
omap3-beagle-xm.dtb omap3-tobi.dtb omap4-pandaES.dtb omap4-var_som.dtb
am335x-bonelt.dtb omap2420-h4.dtb omap3-evm.dtb omap4-panda.dtb
omap4-sdp.dtb omap5-evm.dtb (of which I am interested in only hte
omap4-pandaboard.dtb)
I had not expected these and I assumed that they are ignored by the
kernel??? Or is the kernel ignoring the board file data when it gets a
(wrong) device tree data as well?

If the last part is true then I have to look into u-boot not giving device
tree data to the 3.7.x kernel


Greetings,

Han

[Robert Nelson]

Post by Robert Nelson

Post by Robert Nelson

Post by Han Hartgers
It is a kind of a mix to me. I do have a omap4-panda.dtb file with this
kernel. Probably is the contents not so "complete" or is only u-boot using
this? (Loading the omap4-pandaES.dtb did not change the behaviour of my
kernel booting...)
Any thought on what this waiting for root device /dev/mmcblk0p1 can be?

Yeah, like i said previously, v3.7.x is "pre-dt" for omap4, so STOP
using "omap4-panda.dtb/omap4-pandaES.dtb" and just use the classic
board file..

I do not fully understand what you mean here.
I (or actually your build script) is doing a make ARCH=arm
CROSS_COMPILE=omap2plus_defconfig is that the classic board file
way-of-work you point to?
To my surprise I do get as well: am335x-bone.dtb am335x-evm.dtb
omap3-beagle-xm.dtb omap3-tobi.dtb omap4-pandaES.dtb omap4-var_som.dtb
am335x-bonelt.dtb omap2420-h4.dtb omap3-evm.dtb omap4-panda.dtb
omap4-sdp.dtb omap5-evm.dtb (of which I am interested in only hte
omap4-pandaboard.dtb)
I had not expected these and I assumed that they are ignored by the
kernel??? Or is the kernel ignoring the board file data when it gets a
(wrong) device tree data as well?

No, you don't get it.. Device Tree's where not created over night.
There was a long transition period, as each and every driver had to be
converted.

board file boot -> device tree boot.

v3.7.x is in the early/middle section of that..

Looking at your boot log:

******************
Running uname_boot ...
loading /boot/vmlinuz-3.7.10-x13 ...
3419792 bytes read in 209 ms (15.6 MiB/s)
loading /boot/dtbs/3.7.10-x13/omap4-panda.dtb ...
11810 bytes read in 38 ms (302.7 KiB/s)
debug: [console=ttyO2,115200n8 root=/dev/mmcblk0p1 ro rootfstype=ext4
rootwait] ...
debug: [bootz 0x82000000 - 0x88000000] ...
## Flattened Device Tree blob at 88000000
Booting using the fdt blob at 0x88000000
reserving fdt memory region: addr=9d000000 size=3000000
Loading Device Tree to 8fffa000, end 8ffffe21 ... OK

Starting kernel ...
******************

You want it to look like:

******************
Running uname_boot ...
loading /boot/vmlinuz-3.7.10-x13 ...
3419792 bytes read in 209 ms (15.6 MiB/s)
debug: [console=ttyO2,115200n8 root=/dev/mmcblk0p1 ro rootfstype=ext4
rootwait] ...
debug: [bootz 0x82000000] ...

Starting kernel ...
******************

See how the device tree is NOT used? That's the old board-file boot.

Regards,

[Han Hartgers]

Post by Robert Nelson
No, you don't get it.. Device Tree's where not created over night.
There was a long transition period, as each and every driver had to be
converted.
board file boot -> device tree boot.
v3.7.x is in the early/middle section of that..
******************
Running uname_boot ...
loading /boot/vmlinuz-3.7.10-x13 ...
3419792 bytes read in 209 ms (15.6 MiB/s)
loading /boot/dtbs/3.7.10-x13/omap4-panda.dtb ...
11810 bytes read in 38 ms (302.7 KiB/s)
debug: [console=ttyO2,115200n8 root=/dev/mmcblk0p1 ro rootfstype=ext4
rootwait] ...
debug: [bootz 0x82000000 - 0x88000000] ...
## Flattened Device Tree blob at 88000000
Booting using the fdt blob at 0x88000000
reserving fdt memory region: addr=9d000000 size=3000000
Loading Device Tree to 8fffa000, end 8ffffe21 ... OK
Starting kernel ...
******************
******************
Running uname_boot ...
loading /boot/vmlinuz-3.7.10-x13 ...
3419792 bytes read in 209 ms (15.6 MiB/s)
debug: [console=ttyO2,115200n8 root=/dev/mmcblk0p1 ro rootfstype=ext4
rootwait] ...
debug: [bootz 0x82000000] ...
Starting kernel ...
******************
See how the device tree is NOT used? That's the old board-file boot.
Regards,
--
Robert Nelson
https://rcn-ee.com/

OK now I understand. With the device tree available are the board-file
settings overwritten with a incomplete device tree data. I had expected
that they would be more additive construction so that there is more smooth
transition.

I deleted the device tree data from the file system and kernel 3.7.10 is
booting.

Wifi works also a sort of; CLK32KG is there and the WL12xx driver loads. I
can connect to an AP but I have problems with a "beacon loss" and that
makes the connection not so reliable. This was also what I now start to
remember from my experiences in 2011/2012...

I am probably better off by staying at kernel 4.9.x and soldering an
additional USB connector to add a USB wlan-stick. It is unsatisfying; I
will have to leave the on board module for what it is. (I can give it a try
if I understand why the old 2014 patches do not enable the CLK32KG signal
at a later time.)
By the way can it be that the cpu govenor is not working properly in kernel
4.9.x?
I see the following in my logs:
[ 15.748077] cpu cpu0: dev_pm_opp_set_rate: Regulator not registered with
OPP core
[ 15.755767] cpufreq: __target_index: Failed to change cpu frequency: -22
[ 1092.638519] cpu cpu0: dev_pm_opp_set_rate: Regulator not registered with
OPP core
[ 1092.646636] cpufreq: __target_index: Failed to change cpu frequency: -22

And when I remove the heatsink from the processor it crashes with a thermal
issue...
(I am not 100% certain about the crash because I was also measuring the
power rails and clks to the wifi module; so maybe I have loaded some signal
too much.)

Thanks for the support and much respect the overall very good work you are
presenting at the eewiki site! May I ask if you are employee of Digikey
so that you can do this or is this just your hobby?

[s***@gmail.com]

Post by Han Hartgers
Dear Kristoff,
Happy new year!
Did you find a good working distribution for the Pandaboard? I have a
Pandaboard A1 and I am also looking for a good (modern) distribution for
my board to use it as scan and print server.
https://eewiki.net/display/linuxonarm/PandaBoard is quite good. However
mainline u-boot (16.11) and kernel (4.9x) are not configuring "things"
correctly any more. Therefore is at the on-board Wifi interface not working
at this moment. I am currently wondering if I should try to get the things
or Robert Nelson working or if I should look for a different distribution.
Therefore I wanted to know if you have find a good working distribution
except for the old ubuntu 12.04.
Greetings,
Han

I'm using OpenWrt - Linux 4.4.14:

***@OpenWrt:/# uname -a
Linux OpenWrt 4.4.14 #84 SMP Sun Dec 18 19:29:30 UTC 2016 armv7l GNU/Linux

PandaBoard ES rev. B1

The only thing that's not working is audio...

Steve

[Robert Nelson]

Post by s***@gmail.com
Linux OpenWrt 4.4.14 #84 SMP Sun Dec 18 19:29:30 UTC 2016 armv7l GNU/Linux
PandaBoard ES rev. B1

Wrong board, wifi works fine on the ES's on mainline..

It's the older non-ES original panda that had wifi clock issues..

Regards,