Android howto

[Michael Trimarchi]

Hi all,

We recently buy a freerunner openmoko device and we want to try the
android platform on it.
Is there an howto to startup android on freerunner?

Regards Michael

[Sean McNeil]

There is a begginings of one that you can find at

http://wiki.openmoko.org/wiki/User:Seanmcneil3

The rootfs image is currently deleted as there was a licensing issue
with the codecs. I am actively working to provide media playback of at
least ogg files, but it might not be so easy to accomplish. Regardless,
there will be a new image posted on or before Monday morning.

Sean

michael wrote:
> Hi all,
>
> We recently buy a freerunner openmoko device and we want to try the
> android platform on it.
> Is there an howto to startup android on freerunner?
>
> Regards Michael

[Marco Trevisan (Treviño)]

Sean McNeil wrote:
> There is a begginings of one that you can find at
>
> http://wiki.openmoko.org/wiki/User:Seanmcneil3
>
> The rootfs image is currently deleted as there was a licensing issue
> with the codecs. I am actively working to provide media playback of at
> least ogg files, but it might not be so easy to accomplish. Regardless,
> there will be a new image posted on or before Monday morning.

Ehm... Not to annoy you, but, ehm... Isn't today Monday? :P

--
Treviño's World - Life and Linux
http://www.3v1n0.net/

[Yorick Moko]

On Mon, Dec 1, 2008 at 8:51 PM, Yorick Moko <[hidden email]> wrote:

> only three hours "monday" left in my TZ :)
>
> On Mon, Dec 1, 2008 at 6:39 PM, "Marco Trevisan (Treviño)"
> <[hidden email]> wrote:
>> Sean McNeil wrote:
>>> There is a begginings of one that you can find at
>>>
>>> http://wiki.openmoko.org/wiki/User:Seanmcneil3
>>>
>>> The rootfs image is currently deleted as there was a licensing issue
>>> with the codecs. I am actively working to provide media playback of at
>>> least ogg files, but it might not be so easy to accomplish. Regardless,
>>> there will be a new image posted on or before Monday morning.
>>
>> Ehm... Not to annoy you, but, ehm... Isn't today Monday? :P
>>
>> --
>> Treviño's World - Life and Linux
>> http://www.3v1n0.net/
>>
>>
>>
>

[Sean McNeil]

I am uploading a new rootfs.jffs as I type. It will take a few hours
with my slow connection :(

The issue with patents is causing great pain. Android is setup to either
build with all of Packet Videos opencore, or to eliminate it. If
eliminated, then the media playback and record cannot work as it uses
Packet Video in a JNI interface. Building a partial opencore library is
also quite difficult. I can't just eliminate particular codecs as they
are static libraries with lots of dependencies.

So, media handling is broken at the moment. You will see that it tells
you in logcat it is out of memory, but that is just a bogus error. I am
still working to fix it.

Sean

Yorick Moko wrote:

> On Mon, Dec 1, 2008 at 8:51 PM, Yorick Moko <[hidden email]> wrote:
>  
>> only three hours "monday" left in my TZ :)
>>
>> On Mon, Dec 1, 2008 at 6:39 PM, "Marco Trevisan (Treviño)"
>> <[hidden email]> wrote:
>>    
>>> Sean McNeil wrote:
>>>      
>>>> There is a begginings of one that you can find at
>>>>
>>>> http://wiki.openmoko.org/wiki/User:Seanmcneil3
>>>>
>>>> The rootfs image is currently deleted as there was a licensing issue
>>>> with the codecs. I am actively working to provide media playback of at
>>>> least ogg files, but it might not be so easy to accomplish. Regardless,
>>>> there will be a new image posted on or before Monday morning.
>>>>        
>>> Ehm... Not to annoy you, but, ehm... Isn't today Monday? :P
>>>
>>> --
>>> Treviño's World - Life and Linux
>>> http://www.3v1n0.net/
>>>
>>>
>>>
>>>      
>
>  

[Marco Trevisan (Treviño)]

Sean McNeil wrote:
> I am uploading a new rootfs.jffs as I type. It will take a few hours
> with my slow connection :(

Thanks so much...
What about the kernel for android? Do you have a newer version, maybe
with suspend/resume fixes? :P

--
Treviño's World - Life and Linux
http://www.3v1n0.net/

[Sean McNeil]

Marco Trevisan (Treviño) wrote:
> Sean McNeil wrote:
>  
>> I am uploading a new rootfs.jffs as I type. It will take a few hours
>> with my slow connection :(
>>    
>
> Thanks so much...
> What about the kernel for android? Do you have a newer version, maybe
> with suspend/resume fixes? :P

Not yet. I do have a newer kernel, but there are some problems with it.
I'll copy it over for you. It is at

http://people.openmoko.org/sean_mcneil/uImage-tracking

if you'd like to try it out.

Sean

[Marco Trevisan (Treviño)]

Sean McNeil wrote:

> Marco Trevisan (Treviño) wrote:
>> Thanks so much...
>> What about the kernel for android? Do you have a newer version, maybe
>> with suspend/resume fixes? :P
>
> Not yet. I do have a newer kernel, but there are some problems with it.
> I'll copy it over for you. It is at
>
> http://people.openmoko.org/sean_mcneil/uImage-tracking
>
> if you'd like to try it out.

Ok thanks... I'll try it.

Btw, do you have a small changelog for the android rootfs (I can't flash
it right now and I'm toooooo curious :P)?

--
Treviño's World - Life and Linux
http://www.3v1n0.net/

[Sean McNeil]

Marco Trevisan (Treviño) wrote:

> Sean McNeil wrote:
>  
>> Marco Trevisan (Treviño) wrote:
>>    
>>> Thanks so much...
>>> What about the kernel for android? Do you have a newer version, maybe
>>> with suspend/resume fixes? :P
>>>      
>> Not yet. I do have a newer kernel, but there are some problems with it.
>> I'll copy it over for you. It is at
>>
>> http://people.openmoko.org/sean_mcneil/uImage-tracking
>>
>> if you'd like to try it out.
>>    
>
> Ok thanks... I'll try it.
>
> Btw, do you have a small changelog for the android rootfs (I can't flash
> it right now and I'm toooooo curious :P)?
>  

This is an excellent idea that I haven't thought of doing. I'll try to
put something together.

[Andy Green]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:
| Marco Trevisan (Treviño) wrote:
|> Sean McNeil wrote:
|>
|>> I am uploading a new rootfs.jffs as I type. It will take a few hours
|>> with my slow connection :(
|>>
|>
|> Thanks so much...
|> What about the kernel for android? Do you have a newer version, maybe
|> with suspend/resume fixes? :P
|
| Not yet. I do have a newer kernel, but there are some problems with it.
| I'll copy it over for you. It is at
|
| http://people.openmoko.org/sean_mcneil/uImage-tracking
|
| if you'd like to try it out.

Just a thought, because we as a company don't "do" source packages, we
need to put a git head hash and the branch at least on the binary kernel
files to direct people to the sources matching any binaries we put out.
~ This was on my mind when doing the fair amount of work to get history
in tracking git branches.

Also, what's in this kernel that isn't in andy-tracking?  We need to
have whatever it is in our git so all our kernels do whatever it is
Android needs -- I thought we were already there.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkk0+JEACgkQOjLpvpq7dMp4UQCeKkm87Xbu76rJD5/isHp9Js0g
YzkAoIh/aX2i1crWrY5ntT/vfJ/wqrEX
=zZvH
-----END PGP SIGNATURE-----

[Sean McNeil]

Andy Green wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Somebody in the thread at some point said:
> | Marco Trevisan (Treviño) wrote:
> |> Sean McNeil wrote:
> |>
> |>> I am uploading a new rootfs.jffs as I type. It will take a few hours
> |>> with my slow connection :(
> |>>
> |>
> |> Thanks so much...
> |> What about the kernel for android? Do you have a newer version, maybe
> |> with suspend/resume fixes? :P
> |
> | Not yet. I do have a newer kernel, but there are some problems with it.
> | I'll copy it over for you. It is at
> |
> | http://people.openmoko.org/sean_mcneil/uImage-tracking
> |
> | if you'd like to try it out.
>
> Just a thought, because we as a company don't "do" source packages, we
> need to put a git head hash and the branch at least on the binary kernel
> files to direct people to the sources matching any binaries we put out.
> ~ This was on my mind when doing the fair amount of work to get history
> in tracking git branches.
>
> Also, what's in this kernel that isn't in andy-tracking?  We need to
> have whatever it is in our git so all our kernels do whatever it is
> Android needs -- I thought we were already there.

Yes, the hashes would be good to add. There are a few things for this
kernel:

1) The .config needs to be different. we can't, for instance, have the
console go out to the frame buffer. It just switches like mad whenever
there is log messages and perhaps gets confused.
2) I've included a number of drivers that are modules.
3) The recent patch to add ac power_supply is included.
4) I still have a hack to make the power button act as KEY_MENU when
pressed/released before the 1S key interrupt.

The plan is to still eliminate the need of #4 and hopefully #3 can make
it in. Not sure what to do about difference in .config.

[Cédric Berger]

On Tue, Dec 2, 2008 at 04:02, Sean McNeil <[hidden email]> wrote:

> Marco Trevisan (Treviño) wrote:
>>
>> Sean McNeil wrote:
>>
>> Btw, do you have a small changelog for the android rootfs (I can't flash
>> it right now and I'm toooooo curious :P)?
>>
>
> This is an excellent idea that I haven't thought of doing. I'll try to put
> something together.
>

Looks like the virtual keyboard that was proposed by... [oups I do not
remember who :-p ]... is not included yet.
Have you tried to integrate it ?

For now I have not yet seen what are the differences with previous
image (the broken one without video libs, that was quickly removed)

I will now try with your latest kernel.

Anyway, thanks for your work !

[Andy Green]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

| Yes, the hashes would be good to add. There are a few things for this
| kernel:

Great.  The ./build script takes care of it.

| 1) The .config needs to be different. we can't, for instance, have the
| console go out to the frame buffer. It just switches like mad whenever
| there is log messages and perhaps gets confused.

Huh.  Well is it OK to have tty0 as a console alright but low loglevel?
~ By default that is the situation in Qi now and it shouldn't print
anything unless it is KERN_ERR or worse (panic / OOPS).  For syslog
stuff this is a bootloader decision told by the kernel commandline really?

| 2) I've included a number of drivers that are modules.

Fine... shortly we will have to create a "recommended distro config"
like we have on stable, the moredrivers config we have now in the
tracking branches is not intended for that.

| 3) The recent patch to add ac power_supply is included.

OK I ask Balaji because I don't want to carry it as a local patch if
he's OK with the logic of including / selling it in the generic driver
for upstream.

| 4) I still have a hack to make the power button act as KEY_MENU when
| pressed/released before the 1S key interrupt.
|
| The plan is to still eliminate the need of #4 and hopefully #3 can make
| it in. Not sure what to do about difference in .config.

The .config thing we'll solve by integrating your module list into the
recommended packaging config... I really want to avoid any kind of
"android kernel" or "android config" and to do so will try to work with
the needs directly.

- -Andy

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkk1AsoACgkQOjLpvpq7dMoiSgCZARfc018wNy3v7zn0wggXBmcG
pAEAmgPOwkPe+L2uMfrTCbOo9zGIAO0M
=7hVk
-----END PGP SIGNATURE-----

[Sean McNeil]

Cédric Berger wrote:

> On Tue, Dec 2, 2008 at 04:02, Sean McNeil <[hidden email]> wrote:
>  
>> Marco Trevisan (Treviño) wrote:
>>    
>>> Sean McNeil wrote:
>>>
>>> Btw, do you have a small changelog for the android rootfs (I can't flash
>>> it right now and I'm toooooo curious :P)?
>>>
>>>      
>> This is an excellent idea that I haven't thought of doing. I'll try to put
>> something together.
>>
>>    
>
> Looks like the virtual keyboard that was proposed by... [oups I do not
> remember who :-p ]... is not included yet.
> Have you tried to integrate it ?
>  

Argg. Completely slipped my mind. I'm going back over your work right now.

> For now I have not yet seen what are the differences with previous
> image (the broken one without video libs, that was quickly removed)
>
> I will now try with your latest kernel.
>
> Anyway, thanks for your work !
>  

[Cédric Berger]

On Tue, Dec 2, 2008 at 10:52, Sean McNeil <[hidden email]> wrote:
> Cédric Berger wrote:

>> Looks like the virtual keyboard that was proposed by... [oups I do not
>> remember who :-p ]... is not included yet.
>> Have you tried to integrate it ?
>>
>
> Argg. Completely slipped my mind. I'm going back over your work right now.

Ok cool !

oh and it was Rui Castro I was talking about for the keyboard :
http://groups.google.com/group/android-framework/browse_thread/thread/c8ca666ee3fa9cca


>> For now I have not yet seen what are the differences with previous
>> image (the broken one without video libs, that was quickly removed)
>>
>> I will now try with your latest kernel.
>>

With latest kernel :
- when trying to shut down android, it is stuck showing "shutting down" ....
- backlight brigthness cannot be adjusted (it looks like it is at maximum).

And WIFI seems to still not work (with either kernel) : "impossible to
search networks"

[Sean McNeil]

Cédric Berger wrote:

> On Tue, Dec 2, 2008 at 10:52, Sean McNeil <[hidden email]> wrote:
>  
>> Cédric Berger wrote:
>>    
>
>  
>>> Looks like the virtual keyboard that was proposed by... [oups I do not
>>> remember who :-p ]... is not included yet.
>>> Have you tried to integrate it ?
>>>
>>>      
>> Argg. Completely slipped my mind. I'm going back over your work right now.
>>    
>
> Ok cool !
>
> oh and it was Rui Castro I was talking about for the keyboard :
> http://groups.google.com/group/android-framework/browse_thread/thread/c8ca666ee3fa9cca
>  

Yes, that's the one I will incorporate.

>
>  
>>> For now I have not yet seen what are the differences with previous
>>> image (the broken one without video libs, that was quickly removed)
>>>
>>> I will now try with your latest kernel.
>>>
>>>      
>
> With latest kernel :
> - when trying to shut down android, it is stuck showing "shutting down" ....
> - backlight brigthness cannot be adjusted (it looks like it is at maximum).
>  

Yes, I experience the same behavior here on shutdown. It is lower
priority to me than the other problems, though.
Backlight has moved in the new kernel. It's been a while since I've done
any verification on that.

You'll need to change /system/build.prop:

backlight.lcd=/sys/class/backlight/gta02-bl

I'm going to move everything forward to andy-tracking. Better to test
now than later.

> And WIFI seems to still not work (with either kernel) : "impossible to
> search networks"
>  

This has been broken for me a long time on various kernels. It must me
something I'm doing wrong. I'll need to get Werner to help me on this. I
just can't get anything on a scan at all.

[Werner Almesberger]

Sean McNeil wrote:
> This has been broken for me a long time on various kernels. It must me  
> something I'm doing wrong. I'll need to get Werner to help me on this. I  
> just can't get anything on a scan at all.

Does this happen with  iwlist scan  or with some other program ?

Also, if even iwlist scan doesn't work, does anything run before it
that touches WLAN ?

As far as I can tell, at least iwlist scan on a pristine system (i.e.,
one that doesn't do things with WLAN on its own) works for everyone.
The real problems only seem to kick in once wpa_supplicant had a
chance to run.

- Werner

[Jim Ancona]

Sean McNeil wrote:
> Cédric Berger wrote:

>> And WIFI seems to still not work (with either kernel) : "impossible to
>> search networks"
>>  
>
> This has been broken for me a long time on various kernels. It must me
> something I'm doing wrong. I'll need to get Werner to help me on this. I
> just can't get anything on a scan at all.

While scan doesn't work, I was able to connect to an open network by
using the Add a Wi-Fi network dialog in Settings. Unfortunately they
don't have a WPA2-Enterprise option, so I can't try it on out work network.

Jim

[Marco Trevisan (Treviño)]

Sean McNeil wrote:

> Cédric Berger wrote:
>> On Tue, Dec 2, 2008 at 10:52, Sean McNeil <[hidden email]> wrote:
>>> Cédric Berger wrote:
>>>> Looks like the virtual keyboard that was proposed by... [oups I do not
>>>> remember who :-p ]... is not included yet.
>>>> Have you tried to integrate it ?
>>>>      
>>> Argg. Completely slipped my mind. I'm going back over your work right
>>> now.  
>>
>> Ok cool !
>>
>> oh and it was Rui Castro I was talking about for the keyboard :
>> http://groups.google.com/group/android-framework/browse_thread/thread/c8ca666ee3fa9cca
>
> Yes, that's the one I will incorporate.

Thanks I hope you could complete the task soon since this is, right now,
the most-wanted feature :P

PS: Is it too hard/long to provide some build instructions? :P

--
Treviño's World - Life and Linux
http://www.3v1n0.net/

[Sean McNeil]

Werner Almesberger wrote:
> Sean McNeil wrote:
>  
>> This has been broken for me a long time on various kernels. It must me  
>> something I'm doing wrong. I'll need to get Werner to help me on this. I  
>> just can't get anything on a scan at all.
>>    
>
> Does this happen with  iwlist scan  or with some other program ?
>  

It is happening with wpa_cli. I basically get a message:

ar6000_ioctl_giwscan(): data length 0

I think it is an wext issue. with wpa_supplicant support.

> Also, if even iwlist scan doesn't work, does anything run before it
> that touches WLAN ?
>  

Yes, it would appear that iwlist is working. I'll need to see what it is
doing differently.

> As far as I can tell, at least iwlist scan on a pristine system (i.e.,
> one that doesn't do things with WLAN on its own) works for everyone.
> The real problems only seem to kick in once wpa_supplicant had a
> chance to run.

OK, so it is probably something in the wext layer?