[PRJ][OS]Linux and Android on HTC Diamond "NOT YET WORKING"

[dabs]

I am pretty certain my card is formatted as a vfat single storage media.

Not near a linux box atm, but diskutil under os x gave me:
/dev/disk4
#: TYPE NAME SIZE IDENTIFIER
0: FDisk_partition_scheme *5.7 Gi disk4
1: DOS_FAT_32 SDHC6GB 5.7 Gi disk4s5


I have booted up an old fedora live image i had lying around in a vm, but it dog-slow, and I'm not entirely sure I trust it to format my card, seems like asking for trouble formatting a disk mounted inside a vm running a live image and using the phone in mass storage mode..

fdisk -l gives:
$ fdisk -l
Cannot open /dev/sda
Cannot open /dev/sdb
Which doesn't look promising...

Am stuck at home sick for New Years :(, so at least I have something to mess around with...

I have no particular aversion to repartitioning for an extra ext2 partition, I am only using 2gb on the 6gb card..

I will try to replicate what jobo did, but first il poke around and try to find why this happens on the raph100

#mount -t vfat /dev/mmcblk0p1 /mmc
mount: mounting /dev/mmcblk0p1 on /mmc failed: Invalid argument

[Willis111]

hi willis111, have you tried what i sent you in the private message ? we can chat by PM or in the irc. dont wish to turn this thread into support ... ;)

- lt

Not really trying to turn it into a support thread but it would be nice if I saw one reply saying "I think you're an idiot cause it boots fine on my cdma touch diamond". I read and replied to your pm though, thank you very much. Figured you'd be busy working on something else which was why I was looking for input from other cdma owners.

[touch of jobo]

(disclaimer: if this turns out to be an embarrassing drunk post, I shall edit it into something more appropriate.. ..later.)

wow ! first confirmation of the sd working in raph ... cool and thanks jobo !
yes it is meant to be the external sd for raphael. from netripper's info raph and diam uses (mostly) the same board design -- even the sd part. the difference is that diam built the sd inside.
i was wondering why the raph phones had no luck getting the clock right ... great to hear that at least one model works. :)
- lt
When you first posted your diff after you first mounted your mmc (http://forum.xda-developers.com/showthread.php?p=3060104), it didn't work for me. I got some CRC errors and finally error -110 (iirc). Then, when I applied the mmc2 diff, the one you said had nothing new, just some cleanup and the kernel boot option thing (http://forum.xda-developers.com/showthread.php?p=3071440), I could see my SD card.

After that, I was able to replicate what others had done.


[...] @jobo,
What are the pros/cons of using vfat imgs vs. an ext2 partition? [...]
Well .. when I first read about how android ran on vogue, I partitioned my SD card to have a vfat partition large enough to hold the system and data images and a large ext2 (or yaffs2?) partition to be the external storage. Then I found out it was useless but it's been sitting there ever since. Now, 2 months (or so) later, it finally gets mounted. :D
It could be a pro that its more suitable to put a complete root fs on (im not sure how well vfat behaves when it comes to symlinks, permissions, case sensitivity, whatnot). It could be a con that wince doesn't see it.

Could you post the procedure you followed (or even better zip/gz up the whole lot and up it someplace, but I know thats a lot to ask) for running from an ext2 partition.
I assume you extract the imgs to their relevent folders on the root of the ext2 partition then hack the initrd to mount everything where it should be, but a little tut would be very nice.

In order to create the ext2 partition and fs, I mounted it directly and used fdisk and mkfs.ext2 respectively. That was long ago.
In order to get this android setup running, I only created a directory named 'tmp' on that fs (/dev/mmcblk0p2 in msm linux on raph, /dev/sdc2 in ubuntu in a vbox with the sd card in a card reader in a usb port) and put all images in there. That was in the final couple of hours of last year.
I changed init to mount /dev/mmcblk0p2 as ext2 in stead of /dev/mmcblk0p1 as vfat. That's it. Other than that I just downloaded lavenders file and followed the recipe.
..and put the root fs next to haret.
Oh yeah .. and I changed init to exec /init.android in stead of /(system/?)bin/sh, so I don't have to type.
ps. Whats the deal with it crashing immediately before loading busybox when usb is not connected? Odd...
One of those things where the linux kernel just takes over where windows left it. No (complete) init code or init code that assumes some state.

* Typing on the vkeyb now works and no longer has the side effect of back/cancel.

:( This is a bit of a bummer ..
In the android 1.0 setup that runs off SD card, typing works, but end and f4 kill it, and there is no way (afaik) to go 'back/cancel'. When I boot into a single ram fs (like we've done with android 0.8 for a while now...) I can not type properly because any key (other than the 2 that kill it) have the (side?-)effect of going 'back'. This means you can really 'do' more stuff (if you can call it that) or at least fiddle around more and not get to a dead end...

Happy new year. |)

-Jobo

[sfxc13]

Not really trying to turn it into a support thread but it would be nice if I saw one reply saying "I think you're an idiot cause it boots fine on my cdma touch diamond". I read and replied to your pm though, thank you very much. Figured you'd be busy working on something else which was why I was looking for input from other cdma owners.

I was able to succesfully boot Android 1.0 on a Sprint Touch Diamond (CDMA)
what I did:
Follow the steps on here:
http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux
After it says "can't access tty; job control turned off" you are supposed to type /init.android the problem is the screen calibration is flip flopped so when you touch the top of the screen it is trying to press the bottom. To get it to work properly I pressed and held where I though the buttons would be at the bottom of the screen (the top buttons are closest to the bottom and the bottom buttons are closer to the middle) the helpful part is if you press and hold it while highlight the key you are pressing until you let go. So you can hold the stylus until the correct keyboard letter is highlighted and then let go.

For the record I still got the funky colors that existed on .8

Great works guys!!! Everybody I know REALLY appreciates all your hard work.

[sfxc13]

doh! figured out how to get the screen calibration to come up
delete msmts_calib=<xmin>.<ymin>.<Xmax>.<Ymax> from your default.txt file

[touch of jobo]

I am pretty certain my card is formatted as a vfat single storage media.

Not near a linux box atm, but diskutil under os x gave me:
/dev/disk4
#: TYPE NAME SIZE IDENTIFIER
0: FDisk_partition_scheme *5.7 Gi disk4
1: DOS_FAT_32 SDHC6GB 5.7 Gi disk4s5


Is that an extended partition?
Try
mmcdevp1=/dev/mmcblk0p5 #Add this line
mknod ${mmcdevp1} b 179 5 #Add this line too

(change those 1's to 5's. I changed them to 2's to use the 2nd partition.)

-Jobo

[dabs]

DOH!! You are spot on jobo! Somewhat of a schoolboy error... That will teach me to assume cards are formatted normally out of the box..

No need to change the initrd, I just formatted the damn thing properly and bingo - we have android.

Many, many thanks (and a happy new year!!)

/me leaves muttering about extended partitions....

ps. To use android/busybox without usb connected do a soft-reset immediately before running HaRET.

[Arktronic]

Has anyone else noticed in 1.0 that Android's Ringer Volume screen pops up momentarily when the reset button is pressed? I guess the button press is processed by Android before the lower-level stuff takes over and reboots, so that button is mapped...

Oh, and happy new year!

[slag02]

Thanks, that got it booting for me.
Unfortunately now that there's no screen calibration, Android seems to think the digitizer is on upside down so I can't type the command to start the gui. Is there any way to fix it?

EDIT: Actually now that I look at it, regardless of which initrd (original or kiwi's) I use it fails to mount the mmc using either device. I'm on a sprint raph 800. Maybe the sd card has a different name? Would it be possible to change the init file to mount the filesystems from internal storage (as in program memory, not the diamond's internal storage)?


It doesn't work for me, I get no such file or directory for all 3 imgs



I have been messing with it since I saw it in my RSS reader

From what I can see the issue is the MMC is not mounting in the right location, perhaps an easy fix for someone?

Here is an Image of it loaded and I went to the /dev directory and ran a dmesg then an ls

http://farm4.static.flickr.com/3093/3155731556_6f3abbef4d_o.jpg

AWESOME WORK!

I wish I could help out more, maybe I will start pick up a Java book?
HAPPY NEW YEAR

[mdrobnak]

Wow great job everyone!

I was on IRC just a day or two ago, it seems like a lot of work has happened with respect to RAPHX00 devices...

I will have to test this out tomorrow.. (later)

Next thing to get working will be audio and modem stuff for phone calls & data, then it should be a very usable OS while other features are worked on...

And here I estimated 3 months to my coworkers as to when I'd see Android on my Fuze..it's great to see that open source collaborative development works so well - everyone working on this kinda blows me away in the amount of knowledge that they have...

I will definitely stop on IRC to see what I can do with the phone stuff, but I'm very rusty, so we'll see...lol

Happy New Year everyone!

-Matt

[ermin15]

Hi, there
I like many others have my diamond and would really love to get android running on it, I found this thread.. and whatta long list of questions.. whithout answers...

this thread is just a list of 120 pages of disrespect and the wiki is useless...

Im a developper myself and have been working on many projects and seen a lot of them die because of structure and focus problems.
lists grow longer and longer with nothing but complains. no really progress..
this is looking like one of these "projects"

so before giving up, bying an iphone, a mac go to a fancy cafe drinking laté and look like im important........

first,
is there a "real" project somewhere, or is there only this meaningless talk ??
whatabout taking an agile aproach, create a real project somewhere ??
where is the developers listed, who is doing what, what order should we fix what ??
why are there focus on the colors when the phone is useless. without the mic and speakers working?
i would love to see any of you guys go to work in a car without the engine, and even without the tires...
if this is fixed, a lot would start using it, interest will grow and maybe a community will appear..

second,
what is it people want ? is it android on the diamond or is it the penguin logo when booting ??
anyone know a project with android running on windows mobile ?
the diamond is born with windows and orig drivers.. maybe it is there we should focus.



Maybe at this point this thread should fork, as the differences between diam and raph seem to be becoming more of an issue...


I think we have reached a point where many here think that this thread is not enough for this project. We could migrate to forum, but I think I have found a better solution: How about a social network like the one I created here (http://htclinux.ning.com/)?

Why a whole social network? First of all, it's quite unique. I personally didn't encounter a project with it's own network. It has nice features besides the forum. For example: chat. Wouldn't it be nice to write your "thank you, good job..."-posts in the chat, where it doesn't bother anyone? It's also convenient for other less important chit-chat.
But most important of all - i think it would be fun.

If this solution is accepted, the devs can pm me with their usernames and i will give them admin-status. Of course I would also forfeit my own.

[Gavin_H]

I was interested to notice that it remembers your settings when you reboot. I tested the long-press by adding shortcuts to the home screen and moving the clock and it remembers these changes everytime I re-boot.

I cannot wait until you guys get this working fully. :D

[holydhaliwal]

Hi good job everyone, when booting thoughi get errors saying[QUOTE][init: cannot find /system/bin/servicemanager , disabling servicemanager/QUOTE]

And then similar errors for rild, app_process, mediaserver, playmp3, dbus-daemon and installd, then nothing happens...

Any ideas? thanks.

[Bimboy]

Hi good job everyone, when booting thoughi get errors saying init: cannot find /system/bin/servicemanager , disabling servicemanager

And then similar errors for rild, app_process, mediaserver, playmp3, dbus-daemon and installd, then nothing happens...

Any ideas? thanks.

Are you typing in /init.android after "nothing happens"?

I got most of those errors, and Android does boot after issuing the command for it. However, on my Fuze (RAPH110), it boots to safe mode with no networking.

[abc12345]

Are you typing in /init.android after "nothing happens"?

I got most of those errors, and Android does boot after issuing the command for it. However, on my Fuze (RAPH101), it boots to safe mode with no networking.

You have a RAPH101? my fuze says 110. Are there different versions of the fuze?

[dabs]

Hi good job everyone, when booting thoughi get errors saying
[init: cannot find /system/bin/servicemanager , disabling servicemanager

And then similar errors for rild, app_process, mediaserver, playmp3, dbus-daemon and installd, then nothing happens...

Any ideas? thanks.

This is what mine displayed when the mmc was not mounted.

type "ls /mmc" to check the card has been successfully mounted. If it hasn't that will be empty and you should see lines this right before the busybox shell:
"mount: mounting /dev/mmcblk0p1 on /mmc failed: Invalid argument"
followed by three lines saying:
"mount: Unable to mount '/mmc/foo' No such file or directory"

^ That's from memory so may not be exactly right...

How is your SD card partitioned... Read jobo and lavender.t's posts over the last couple pages for clues.


@Devs:

What work needs to be done to map the front buttons? As I see it the biggest problem we face is currently inability to navigate menus due to no back button. Some buttons do appear to be mapped however, for example closing the slide on the raph types 14, and as someone mentioned above, the reset button is mapped to volume :)

The left d-pad button seems to be mapped to something, would be very nice if we could just xmodmap that to Back!

[Bimboy]

You have a RAPH101? my fuze says 110. Are there different versions of the fuze?

Sorry.. Finger malfunction... RAPH110 here. The rest is true. Can get to safe mode only. There is only one known version of the Fuze to date.

[slag02]

Here is a Screenshot of the Errors that Sprint RAPH800 Users are seeing*

http://farm4.static.flickr.com/3226/3157347414_f4d74df910_o.jpg


*footnote:
I have used both initrd files and the error is the same barring the small change between them - on this boot I also had removed the calibration flag from the default.txt

[keerttttt]

Is this version of android 1.0 we are booting now also the newest version around? There was a new version released a few days ago right?

[abc12345]

Sorry.. Finger malfunction... RAPH110 here. The rest is true. Can get to safe mode only. There is only one known version of the Fuze to date.

I can only boot up in safe mode also

[Bimboy]

I can only boot up in safe mode also

First, not looking for help. Thought I would pass this along in case it helps.


I can see BusyBox shell starts and the next thing I see before the command prompt is "/bin/sh: can't access tty; control turned off"

I then init android, it boots in safe mode, and I immediately get a pop up:

"Sorry!
Application Settings (in process
com.android.settings) is not responding"

with two buttons: [Force Exit] and [Wait]

[keerttttt]

First, not looking for help. Thought I would pass this along in case it helps.


I can see BusyBox shell starts and the next thing I see before the command prompt is "/bin/sh: can't access tty; control turned off"

I then init android, it boots in safe mode, and I immediately get a pop up:

"Sorry!
Application Settings (in process
com.android.settings) is not responding"

with two buttons: [Force Exit] and [Wait]
i guess everyone get's that forcing error. Some programs hang most times. If you try to dial a number it hangs also.

Maybe this is because it cannot reach things like battery status, network, bluetooth e.t.c.

[butterdori]

i guess everyone get's that forcing error. Some programs hang most times. If you try to dial a number it hangs also.

Maybe this is because it cannot reach things like battery status, network, bluetooth e.t.c.

Yep, i also get that..
But when I dial a number, it goes back to home screen

[keerttttt]

Yep, i also get that..
But when I dial a number, it goes back to home screen
Oh, i thought i got a error there too, but it could be that it just returns to the home screen yes.

[fredfredburger123]

i believe that it returns to the home screen because it has no mobile signal

things need mapping to enable such things

has anyone thought about how to migrate it as the main os?

sorry if its a daft question

[touch of jobo]

Is this version of android 1.0 we are booting now also the newest version around? There was a new version released a few days ago right?

Yes. Well.. we can.

# begin build properties
# autogenerated by buildinfo.sh
ro.build.id=MAIN
ro.build.version.incremental=eng.job.20090101.2251 27
ro.build.version.sdk=3
ro.build.version.release=1.5
ro.build.date=Fri Jan 2 08:17:06 CET 2009
ro.build.date.utc=1230880626
ro.build.user=job
ro.build.host=jobuntu
ro.build.tags=test-keys

This recipe gives you a bootable set of the latest android:
* Check out the latest (current) android,
* Build it (..over night (or over work) if you build on a vbox with little ram :rolleyes: as it takes a looong time),
* Combine the root, system and data folders from ./out/target/product/generic into a single tree,
* Change init.rc to disable some lines that create the /system and /data mount points and mount them,
* Put this tree in the root of an ext2 partition on the SD card,
* Create a default.txt with no INITRD, and a CMDLINE that says: "root=/dev/mmcblk0p2 rootdelay=10 init=/init console=tty0 mem=64M msmsdcc_id=2"
* Boot the kernel. (The linuxtogo htc-msm kernel you know and love, not the kernel that comes with the SDK. One that will mount your internal storage or SD card.)

I have no idea why mem=64M but it does appear to be necessary. :o


When dragging the tab up or down, I would think that the lower half should be darkened (behind a dark semi transparent layer) but it isn't. The background only gets dark when the tab is completely up.
In this latest build, when dragging the tab thing up or down, it looks as I would expect. :)
It no longer says 'Safe mode'.
It still dies horribly when left idle for 1 minute or when hitting END or F4. :(

-Jobo

[plbelanger]

I believe the new version you are speaking of is the cupcake perhaps? I don't know if it's a release in the traditional sense just addons e.g. onscreen keyboard..something diamond users need =)

[keerttttt]

Yes. Well.. we can.

# begin build properties
# autogenerated by buildinfo.sh
ro.build.id=MAIN
ro.build.version.incremental=eng.job.20090101.2251 27
ro.build.version.sdk=3
ro.build.version.release=1.5
ro.build.date=Fri Jan 2 08:17:06 CET 2009
ro.build.date.utc=1230880626
ro.build.user=job
ro.build.host=jobuntu
ro.build.tags=test-keys

This recipe gives you a bootable set of the latest android:
* Check out the latest (current) android,
* Build it (..over night (or over work) if you build on a vbox with little ram :rolleyes: as it takes a looong time),
* Combine the root, system and data folders from ./out/target/product/generic into a single tree,
* Change init.rc to disable some lines that create the /system and /data mount points and mount them,
* Put this tree in the root of an ext2 partition on the SD card,
* Create a default.txt with no INITRD, and a CMDLINE that says: "root=/dev/mmcblk0p2 rootdelay=10 init=/init console=tty0 mem=64M msmsdcc_id=2"
* Boot the kernel. (The linuxtogo htc-msm kernel you know and love, not the kernel that comes with the SDK. One that will mount your internal storage or SD card.)

I have no idea why mem=64M but it does appear to be necessary. :o


In this latest build, when dragging the tab thing up or down, it looks as I would expect.
It no longer says 'Safe mode'.
It still dies horribly when left idle for 1 minute or when hitting END or F4. :(

-Jobo
I understand that you have done this yourself and have the files of the new build?
Can you please place that build online? i would like to test that one on my phone also ;)

I'm not that good with linux.

[balsat]

I just compiled Android 1.0 and now the "bks" is assign as the back button!
A can dump the hole ext2 FS if anyone want it?

[NoBody20013]

hi,

what is the "bks" button? and how did you map it.

-yes please upload your build so we can test it or see your changes....

[fredfredburger123]

yes please - i would like to test also

[ermin15]

Android on Notebooks!?!?!? Find out more here (http://venturebeat.com/2009/01/01/android-netbooks-on-their-way-likely-by-2010/).

[balsat]

hi,

what is the "bks" button? and how did you map it.

-yes please upload your build so we can test it or see your changes....


Bks=backspace on the soft keyboard... I didn't map the key, it is automatic maped in Android build eng.root.20090102.152519

Someone with about 70-80 mb free webspace?

[dejant_9]

Someone with about 70-80 mb free webspace?

yes. PM me if you want.

[Untouchab1e]

yes. PM me if you want.


I can set you up with a free 250MB Web hosting account if you want.. Just let me know ;)

[balsat]

yes. PM me if you want.


I send you a PM... The dump is 40 mb

[c4software]

hi,

its possible to make a dump like the previous release aka : 3 files cache system and userdata ?

[balsat]

The files are now uploaded http://217.128.108.212/master/ AND
http://87.72.13.87/

The android_fs.tar.gz is a dump of my /dev/mmcblk0p2 ext2 partition

Thanks to dejant_9 for hosting

[c4software]

same use as before ?

Edit: WoW web Server seems really slow, if ou need an other dedicated webserver pm me.

[balsat]

same use as before ?

Edit: WoW web Server seems really slow, if ou need an other dedicated webserver pm me.

No you should unpack the contents of android_fs.tar.gz to a ext2 partition on your sdcard.

but i'am uploading userdata.img and system.img now, but thees img'es i NOT in ext2 format, but yaffs2

[fredfredburger123]

hey sorry for being numb!!

am trying to make an ext2 partition of my 4gb internal storage on diamond but gparted sees it all as free space?

will it kill all my current data if i give say 2gb to ext2?

Thanks

[c4software]

but i'am uploading userdata.img and system.img now, but thees img'es i NOT in ext2 format, but yaffs2

With this image's i just have to replace old image with yours ?

[balsat]

hey sorry for being numb!!

am trying to make an ext2 partition of my 4gb internal storage on diamond but gparted sees it all as free space?

will it kill all my current data if i give say 2gb to ext2?

Thanks

First of all... MAKE A BACKUP of your fat partition.
I use fdisk under Ubuntu and it see the fat 4gb part just fine!

Delete the partition and create 2 new, the first a fat16 LBA, the second ext2

[c4software]

what is the advantage to split the internal storage?

[fredfredburger123]

First of all... MAKE A BACKUP of your fat partition.
I use fdisk under Ubuntu and it see the fat 4gb part just fine!

Delete the partition and create 2 new, the first a fat16 LBA, the second ext2

thanks for that

[touch of jobo]

I understand that you have done this yourself and have the files of the new build?
Can you please place that build online? i would like to test that one on my phone also ;)


I have. I did. I can.
By now, balsat has done the same, and dumped the file system (http://forum.xda-developers.com/showthread.php?p=3113378).
Mine is only 30M (link) (http://members.home.nl/j0b0/android-singlefs-jobo.cpio.gz), not 40. I guess balsat had already booted it, so it comes with a pre-populated dalvik-cache. (This is why the 1st boot is waaay slower than subsequent boots.)
My ISP seems to have trouble with web space, so if that other link doesn't work, you can get it from here (http://tibook.jb.b4m.com/j0b0/android-singlefs-jobo.cpio.gz) for now. (I'm not going to leave it there if it saturates my pipe too much. Please try the other link (http://members.home.nl/j0b0/android-singlefs-jobo.cpio.gz) first.)

The reason I posted that recipe (which I think is pretty clean) was so that we don't need to fling multi-meg images around. (30 or 40M is a lot for not being able to do anything really...) The fact that the process is almost completely automated and involves very little tinkering is imho A Good Thing, because it means there are only few steps that need to be repeated (and that could potentially break) when updates happen.

-Jobo

EDIT: This is _not_ a file system image that gets mounted from an init script in a ramdisk. You have to unpack it into the root of a partition of your SD card and boot into it using a command line like I posted before (http://forum.xda-developers.com/showthread.php?p=3112480).

EDIT: Can anyone confirm they get the on screen keyboard (see attached picture) when you click in the google search widget (on the right most screen)?

EDIT: Oh yeah .. to get wallpaper pictures fit nicely, they must be 720x640 ;)

[balsat]

I have mounted a fat partition at /sdcard but Android can't see any sdcard, anybody know why?

[iRiKi]

what is the advantage to split the internal storage?

none, it's actually a disadvantage. it's like in the desktop windows world, two partitions because of incompatibilities in the m$ OS (again) :>

[abc12345]

Can someone reupload the files somewhere else because, the link is incredibly slow

[keerttttt]

I have. I did. I can.
By now, balsat has done the same, and dumped the file system (http://forum.xda-developers.com/showthread.php?p=3113378).
Mine is only 30M (link) (http://members.home.nl/j0b0/android-singlefs-jobo.cpio.gz), not 40. I guess balsat had already booted it, so it comes with a pre-populated dalvik-cache. (This is why the 1st boot is waaay slower than subsequent boots.)
My ISP seems to have trouble with web space, so if that other link doesn't work, you can get it from here (http://tibook.jb.b4m.com/j0b0/android-singlefs-jobo.cpio.gz) for now. (I'm not going to leave it there if it saturates my pipe too much. Please try the other link (http://members.home.nl/j0b0/android-singlefs-jobo.cpio.gz) first.)

The reason I posted that recipe (which I think is pretty clean) was so that we don't need to fling multi-meg images around. (30 or 40M is a lot for not being able to do anything really...) The fact that the process is almost completely automated and involves very little tinkering is imho A Good Thing, because it means there are only few steps that need to be repeated (and that could potentially break) when updates happen.

-Jobo

EDIT: This is _not_ a file system image that gets mounted from an init script in a ramdisk. You have to unpack it into the root of a partition of your SD card and boot into it using a command line like I posted before (http://forum.xda-developers.com/showthread.php?p=3112480).

EDIT: Can anyone confirm they get the on screen keyboard (see attached picture) when you click in the google search widget (on the right most screen)?

EDIT: Oh yeah .. to get wallpaper pictures fit nicely, they must be 720x640 ;)

I don't want to be an asshole ;), but is it possible to put it in a .ext2.img format so that i/we don't have to mess around with partitions and stuff?

[balsat]

Can anyone confirm they get the on screen keyboard (see attached picture) when you click in the google search widget (on the right most screen)?


Yes i got the same nice keyboard when i press on the google search within the android i compilede!

[Gavin_H]

I don't want to be an asshole ;), but is it possible to put it in a .ext2.img format so that i/we don't have to mess around with partitions and stuff?

I would be really grateful for this also, I'm not technically gifted and I don't really want to partition my drive.

[lavender.t]

got some requests for the source diffs so i'm attaching the patch file here. the only change from my previous diff attachment is the kernel parameters of msmts_calib.

this is the source that i built my zImage from.

- lt

[abc12345]

I dont want to sound like a noob but where do I place the other two img's

[c4software]

@balsat

Can you make system.img and userdata.img image in ext2 format ?

[touch of jobo]

I don't want to be an asshole ;), but is it possible to put it in a .ext2.img format so that i/we don't have to mess around with partitions and stuff?
That wouldn't help you because it is the entire file hierarchy on a single fs: root, system, data, cache. You would still have to roll a ram disk to boot from that mounts the image and chroots into it. ("This is _not_ a file system image that gets mounted from an init script in a ramdisk.") At the risk of being an asshole myself: If you have a setup that lets you do that, then I'm sure you can pump this cpio archive into any flavor file system. If not, an ext2 image wouldn't help either, and it might be more practical to use the method where you have an initrd mount a vfat partition and then some image files (http://forum.xda-developers.com/showthread.php?p=3105104).

Yes i got the same nice keyboard when i press on the google search within the android i compilede!
Cool. I guess it's time to make the kernel vkeyb toggleable again.. :)

got some requests for the source diffs so i'm attaching the patch file here. the only change from my previous diff attachment is the kernel parameters of msmts_calib.
this is the source that i built my zImage from.
- lt
Thanks! My screen is getting thin in those 2 corners...

-Jobo

UPDATE: This kernel (link) (http://members.home.nl/j0b0/zImage-jobo) has the (latest?) clock and mmc code (it sees the SD card in my raph100) as well as the msmts_calib boot option from lavender.t (so you don't have to calibrate every time)..
..and a toggleable vkeyb (so it doesn't get in the way when you don't need it). To the left of the vkeyb, there is a green square. You can drag the vkeyb around the screen with this. The square goed red while dragging. If you tap on it without dragging (which is really hard, so there is a 3 pixel threshold) it turns the vkeyb off (it wont undraw until the that part of the screen updates) and the square turns blue, so you can tell it's turned off even when its not (yet) undrawn. Again, you can drag it around, or tap on it to turn the vkeyb back on (which of course draws immediately).

BTW: Should you encounter unexpected behavior: This kernel does not init halibut-keypad or trout-keypad. Also, the htcraphael defconfig has CONFIG_TOUCHSCREEN_SYNAPTICS_I2C_RMI enabled. Disabling it didn't seem to make a difference so I left it disabled.

[Gavin_H]

Could someone not just update the wiki with the latest version of Android 1 in place of the link there currently?

[sibbor]

Cool. I guess it's time to make the kernel vkeyb toggleable again.. :)Getting so excited about the nice progress here!

Too bad, my MMC seem to have been damaged. I can read in high speeds from the card, but writing data to the card is VERY slow (50~100 kb/s)

Also my device seems to drop USB connection after a few seconds... sigh, can't test these stuff out anymore. Will try to get ActiveSync up and flash my device again.

[Toug]

Could someone not just update the wiki with the latest version of Android 1 in place of the link there currently?

Yes I would know too if Android is working and what are bugs still present, so If somebody can update the wiki?

I left Windows for Ubuntu and I would leave Windows mobile too.

Thanks

[kowen]

It's so nice to see progress here, with android 1.0
Great job you guys !!
Unfortunately I cant be of help, but looking forward to any progress.
I know how much work these things take (software engineer)

[michbeck100]

got some requests for the source diffs so i'm attaching the patch file here. the only change from my previous diff attachment is the kernel parameters of msmts_calib.

this is the source that i built my zImage from.

- lt

Great work lavender.t!

I guess you've used the sources from linuxtogo.com (http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=shortlog;h=refs/heads/htc-msm-2.6.25)? Will you also commit your changes? It's easier to keep track of the changes that way.

Cheers

[balsat]

I got a GSM signal on my diamond, and it connect to my GSM provider "3", but when i touch the softdialpad it crashes.

In /system/build.prop i changed the "rild.libargs=-d /dev/ttyS0" to "rild.libargs=-d /dev/smd0"

I can receive SMS, but i can't send yet

[abc12345]

can someone please explain better what to do

[keerttttt]

can someone please explain better what to do

Yeah, for me too. Can't seem to find out a way to:
1. make a ext2 partition on my diamond (after i have resized it of course)
2. Read the cpio file...
3. copy all the files needed to there.

I'm on windows, and with me a lot of other diamond users i would guess (quite often linux users prefer a unix based phone over a windows based phone.

For me the solution with the 3 .img files was perfect... so i secretly hope that someone has somekind of same solution :)

On the other hand, i don't wanna ruin anyone's drive or developing, it's going great the last days. I'm really proud of what a community like this can accomplish :)

[lavender.t]

so i secretly hope that someone has somekind of same solution :) ...

your secret wish has been secretly granted ... :) check out the cupcake images at willis111's server: http://xda.willboost.com/cupcake-20090103-2.tar.gz.

the zImage there has jobo's draggable vkeyb patch.

here i also incorporated the smd issues that mdrobnak and netripper figured out. so like the 0.8 image you had before, it is now able to place phone calls. still, no voice no rings, etc.

the change to use smd was as balsat suggested in the irc to modify build.prop in /system:

-rild.libargs=-d /dev/ttyS0
+rild.libargs=-d /dev/smd0

thanks guys !

- lt

[keerttttt]

your secret wish has been secretly granted ... :) check out the cupcake images at willis111's server: http://xda.willboost.com/cupcake-20090103-2.tar.gz.

the zImage there has jobo's draggable vkeyb patch.

here i also incorporated the smd issues that mdrobnak and netripper figured out. so like the 0.8 image you had before, it is now able to place phone calls. still, no voice no rings, etc.

the change to use smd was as balsat suggested in the irc to modify build.prop in /system:

-rild.libargs=-d /dev/ttyS0
+rild.libargs=-d /dev/smd0

thanks guys !

- lt
Really cool dude, booting now :D

[keerttttt]

Really cool dude, booting now :D
Allright, so i booted, it works great. the slide problem is now gone.
It automatically found my network and logged on and i could make a call. Of course, you can't hear anything on both sides, but it works.

When i tried to put roaming on (that's needed for my network's data connection) the settings dialog freezed and forced quitted. Really nice work though, beginning to get a lot better and better!

Too bad that while in the call, the screen got lines and all like before and went to dark... weird that that happens... power management issues?
Edit: same thing as when you type F4. What's even weirder. i can see the windows/touchflo 3d a little bit when the screen goes weird. That's even weirder!

[touch of jobo]

[...] the zImage there has jobo's draggable vkeyb patch.

thanks guys !
- lt
no no no no .. thank _you_! :)

Does it also have the toggle thing and the non-flicker thing? Or should I try and produce a clean diff?

-Jobo

[rl2000]

Isn't it about time to remove the "drawing" yellow dots thing from the framebuffer?

[keerttttt]

Isn't it about time to remove the "drawing" yellow dots thing from the framebuffer?

Yeah, i was thinking the same. Also, how can you disable the system keyboard again? i can drag it, but can't seem to disable it.

[rl2000]

I just tried the cupcake-20090103-2.tar.gz that was posted and it asks me for a PIN code? And the right one is "incorrect". Anybody else seeing this? I'll try to boot it a couple more times. Also, I can't seem to disable the keyboard either, only drag it.

UPDATE: Disabling PIN code skips this but as soon as I get the desktop it tells me I've lost connection due to roaming is disabled. When I try to enable it gives an error message and I'm stuck in that menu. Also as someone else mentioned F4 is a killswitch :P

The other Fkeys seem to enable some color chart like the ones shown on screenshots back when there was talk about the screwed colors of the CDMA version.

Booting a couple more times and selecting to ignore that I don't have a gsm connection I can use the system okay but the "bks" backspace thing doesn't go bakc in the menus though.

[keerttttt]

I just tried the cupcake-20090103-2.tar.gz that was posted and it asks me for a PIN code? And the right one is "incorrect". Anybody else seeing this? I'll try to boot it a couple more times. Also, I can't seem to disable the keyboard either, only drag it.

I would advise you to not try to input a pin for more then 2 times (maybe it will block...). You can try to disable your sim pin code and see if the message dissapears. Good to know thougs that iw probably doesn't communicate good in pin checking.

On the other hand, it could be (i'm not so lectured on sim cards) that because you are booting from windows mobile, the sim is already in ready mode, but android sees that there is a pin code on the sim... something like that..

Anyhow, minor issue for now i would say.

[touch of jobo]

[...] Also, how can you disable the system keyboard again? i can drag it, but can't seem to disable it.
[...] Also, I can't seem to disable the keyboard either, only drag it.
Tap the green square. It should turn blue. The vkeyb is now off. It doesn't go away until the screen updates that section, but in android thats quite often.
If it works as in those attached images with the kernel I posted earlier (in the update) (http://forum.xda-developers.com/showthread.php?p=3115937), then that code is just not in here and I shall produce a diff. If that one doesn't work either, you're doing something wrong. ;) (Tapping without dragging, even with a 3 pixel treshold is not always easy, depending on the time of day..)

-Jobo

EDIT: I just tried the kernel in the package lavender.t posted (http://forum.xda-developers.com/showthread.php?p=3117891). It does not have the toggle feature. (So it's not you ;) ) Also, that kernel makes android run in safe mode for me.

EDIT: A diff of the entire tree turned out a bit messed up, but this (http://members.home.nl/j0b0/msm_vkeyb-jobo20090103.diff.txt) is a diff for msm_vkeyb.c that has the toggle behavior.

[c4software]

Hi,

In my Diamond. the Bks button on Vkeyboard seems not map because when i press it i'm not return in the home screen.

maybe mapping keys would be the major todo point.

PS: My diamond gone sleep, and i can't turn on. (reboot :D)
PS': Call Works :D

[plbelanger]

As soon as wifi and making/receiving calls work I'll jump on...till then, keep up the great work! At this rate, perhaps the touch diamond will be the 2nd Android device :)

[keerttttt]

Yeah, i think mapping at least a hardware key to get back to the home screen would be a major todo point.

[holydhaliwal]

Hey guys im trying to get this working but have had no success so far, from reading through the last 4-5 pages i think its a problem with my sd card not being mounted, as probably a few others have, i just want to be sure before i start reformating it,

1- i should have one vfat partition and one ext2 partion?
2- which one should have the /tmp in its root directory?
3- Also will windows mobile be able to read the "storage" partition,
4- Another thing is how big should the partition that is holding the /tmp folder be?

Thanks a bunch guys! good progress in the last week or so!:)

[Bimboy]

EDIT: I just tried the kernel in the package lavender.t posted (http://forum.xda-developers.com/showthread.php?p=3117891). It does not have the toggle feature. (So it's not you ;) ) Also, that kernel makes android run in safe mode for me.

I used lavender.t's image, along with your zImage-Jobo, and I did get the toggle. I'm not in safe mode (for the first time!!!), and I can make calls, but do not have data capabilities. I also get the new pop-up keyboard when using google search.

If I go into the Settings app, I can see that my SDHC is not mounted in Android, as it says 'not found'. Also, just after the SDHC card is found, I get the following:

[ 3.406379] EXT2-fs warning: mounting unchecked fs, runni
ng e2fsck is recommended

I am assuming this is due to me not having created a partition on my SD card, and merely have it formatted FAT32 with my img files in a tmp dir?

[keerttttt]

Hey guys im trying to get this working but have had no success so far, from reading through the last 4-5 pages i think its a problem with my sd card not being mounted, as probably a few others have, i just want to be sure before i start reformating it,

1- i should have one vfat partition and one ext2 partion?
2- which one should have the /tmp in its root directory?
3- Also will windows mobile be able to read the "storage" partition,
4- Another thing is how big should the partition that is holding the /tmp folder be?

Thanks a bunch guys! good progress in the last week or so!:)

Try using the lavender.t images, that is the most simple solution.

[plbelanger]

[ 3.406379] EXT2-fs warning: mounting unchecked fs, runni
ng e2fsck is recommended


I have used linux exclusively for 10 years and you can disregard that...It's just a warning because the fs hasn't been checked...you can check it if you like using fsck. However, I'm not sure exactly all the internals of android and this port work as of now.

[iAmNotorious]

your secret wish has been secretly granted ... :) check out the cupcake images at willis111's server: http://xda.willboost.com/cupcake-20090103-2.tar.gz.

the zImage there has jobo's draggable vkeyb patch.

here i also incorporated the smd issues that mdrobnak and netripper figured out. so like the 0.8 image you had before, it is now able to place phone calls. still, no voice no rings, etc.

the change to use smd was as balsat suggested in the irc to modify build.prop in /system:

-rild.libargs=-d /dev/ttyS0
+rild.libargs=-d /dev/smd0

thanks guys !

- lt

I hope you don't mind but I'm impatient and that server was a bit slow so I uploaded it to my place to mirror..

http://dirtyhippie.org/files/cupcake-20090103-2.tar.gz

[holydhaliwal]

Try using the lavender.t images, that is the most simple solution.

I am using lavender's images however they aren't working for me, i have a cdma diamond btw...so in what way should i re-partition?

[Gavin_H]

Ok, I've been messing with lavender.t's upload and in case it's helpful the following works:

I can receive but not send SMS.
I cannot access the internet.
I can make and receive calls, but they are silent.
I can play snake. (It's crap)

If you go to the 'Spare Parts' App and change the "End Button Behaviour' to "Go To Home" that will change the "End" key on the soft keyboard to go back to the home page.

F1 on the soft keyboard appears to work as the 'Menu' hard button.

[ChiefmasterB]

Hm... I´m not a developer.... but I wonder why some guys get android running fully on an ASUS Eee PC?

http://www.chip.de/bildergalerie/Android-Google-OS-auf-Asus-Eee-PC-1000H-Galerie_34300064.html?show=2

The review says that the guy needed just 4 hours to get android running on the PC.....

any help for diamond-porting?

[NielDK]

Hm... I´m not a developer.... but I wonder why some guys get android running fully on an ASUS Eee PC?

http://www.chip.de/bildergalerie/Android-Google-OS-auf-Asus-Eee-PC-1000H-Galerie_34300064.html?show=2

The review says that the guy needed just 4 hours to get android running on the PC.....

any help for diamond-porting?

I think the guys working on this here has come just about as far as the EEE port. Just, remember, the EEE is NOT a phone, so the radio part (phone, SMS and so) is not applicable for that project, as it is for a phone.

[sandmen]

Hi guys,
nice to hear the progress with linux + android.
Is it possible to make a small howto in the wiki.
1. Bake a cupcake like lavender.t
1.1 howto make kernel
1.2 make userdata + system
1.3 make android

thanks

[Krad213]

And what about G1's rom leaked out is it completly useless ? Cuz G1 has almost the same hardware ...

[keerttttt]

Ok, I've been messing with lavender.t's upload and in case it's helpful the following works:

If you go to the 'Spare Parts' App and change the "End Button Behaviour' to "Go To Home" that will change the "End" key on the soft keyboard to go back to the home page.

Hey, that's a nice discovery. makes checking android out much easyer! thanx

[touch of jobo]

Hi guys,
nice to hear the progress with linux + android.
Is it possible to make a small howto in the wiki.
1. Bake a cupcake like lavender.t
1.1 howto make kernel
1.2 make userdata + system
1.3 make android
thanks

I put a 'build your own (http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux#Buildx20.Yourx20.O wn)' section underneath 'useful links', with how to build haret, how to build kernel, how to build android.
(I shall add a bit more on how to copy it to your phone and boot into it.)


I just tried the cupcake-20090103-2.tar.gz that was posted and it asks me for a PIN code? And the right one is "incorrect". Anybody else seeing this?

[...] the change to use smd was as balsat suggested in the irc to modify build.prop in /system:
-rild.libargs=-d /dev/ttyS0
+rild.libargs=-d /dev/smd0
thanks guys !
- lt

After I applied that change, it started prompting me for a PIN at startup too. It doesn't matter whether or not I had unlocked the SIM prior to booting linux. It doesn't matter whether I use 'a kernel that makes android run in safe mode' (whatever that means) or not. Disabling SIM PIN gets rid of this. Phone and SMS works.

-Jobo

[sandmen]

@touch of jobo
i want to try.
thanks.

I think this patch is not included or klick (http://forum.xda-developers.com/showpost.php?p=3115412&postcount=1303)

[katmandu421]

Aha!! The stylus must be out before booting in order to get the onscreen keyboard!

...but how in the world do you type /init.android on it :eek:

EDIT: Wow, my screen is mirrored, down is up, up is down. I think I remember seeing this in .8 at some point...and there was a fix, Any ideas?

[keerttttt]

Aha!! The stylus must be out before booting in order to get the onscreen keyboard!

...but how in the world do you type /init.android on it :eek:

EDIT: Wow, my screen is mirrored, down is up, up is down. I think I remember seeing this in .8 at some point...and there was a fix, Any ideas?

What kind of phone do you have, and what kind of provider/brand.

[keerttttt]

By the way, maybe we can look how the ppl at the vogue android topic mapped hardware buttons. Don't they commit their changes?

[dcordes]

can somebody with gsm diamond please build the current htc-msm-2.6.25 revision with htcdiamond_defconfig and check if everything works as it does with the old code plus lavender.t patch?

[katmandu421]

What kind of phone do you have, and what kind of provider/brand.

CDMA HTC Touch Doamond on Sprint, Juicy's ROM

[NetRipper]

can somebody with gsm diamond please build the current htc-msm-2.6.25 revision with htcdiamond_defconfig and check if everything works as it does with the old code plus lavender.t patch?
Please note that I split up Raphael and Diamond in the git tree (http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=shortlog;h=refs/heads/htc-msm-2.6.25).

The Diamond and the Raphael have their own board files now, due to differences with SD card detect and probably more differences to come.

When booting this kernel keep in mind that you need to configure the correct mtype in haret's default.txt:
1805 for Diamond
1910 for Raphael

I have tested this for Raphael to work properly, but not yet for Diamond. Would be nice if anyone could.

edit: just fixed an error in git that prevented diamond board from compiling into the kernel... also, you can use this zImage (http://netripper.com/raphael/20090401_zImage_git) for the test, just cleanly compiled from a clean defconfig.

[sandmen]

I try it, just downloading.

i need a little bit more time. (old debian and no JDK5 update 12)
i set up a new virtual machine :(

[petay]

i have been playing with the image on the wikki and added the files from the cupcake branch and i must say its fantastic!!! i eagerly await the day when its the only os on my diamond.

one thing i will sat is that i dont see the need for the on screen floaty keyboard anymore. once android is booted it is just in the way, i only need it to get back to the home screen. once the hardware buttons are properly mapped i will have no use for the floaty keyboard.

Also when the screen fades (if you leave the fone for long enough) is there any way to get it back on??

[balsat]

Please note that I split up Raphael and Diamond in the git tree (http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=shortlog;h=refs/heads/htc-msm-2.6.25).

The Diamond and the Raphael have their own board files now, due to differences with SD card detect and probably more differences to come.

When booting this kernel keep in mind that you need to configure the correct mtype in haret's default.txt:
1805 for Diamond
1910 for Raphael

I have tested this for Raphael to work properly, but not yet for Diamond. Would be nice if anyone could.

edit: just fixed an error in git that prevented diamond board from compiling into the kernel... also, you can use this zImage (http://netripper.com/raphael/20090401_zImage_git) for the test, just cleanly compiled from a clean defconfig.



I just compiled it, and it do not find my /dev/mmcblk0p2.

[dejant_9]

edit: just fixed an error in git that prevented diamond board from compiling into the kernel... also, you can use this zImage (http://netripper.com/raphael/20090401_zImage_git) for the test, just cleanly compiled from a clean defconfig.

I've used the haret.exe and cache.ext2 of the wiki, the 3 images of the cupcake-20090103-2 and rename this "20090401_zImage_git.txt" to "zImage" and put all of this in a tmp folder on my phone. No problem android boot in normal mode. I've the draggable keyboard, i can make a call etc.. i juste have some error messages with "force close" and "wait" buttons.

[NetRipper]

I've used the haret.exe and cache.ext2 of the wiki, the 3 images of the cupcake-20090103-2 and rename this "20090401_zImage_git.txt" to "zImage" and put all of this in a tmp folder on my phone. No problem android boot in normal mode. I've the draggable keyboard, i can make a call etc.. i juste have some error messages with "force close" and "wait" buttons.
Erm, the kernel (20090401_zImage_git) does not have the draggable keyboard. I think you must have done something wrong. :)

I just compiled it, and it do not find my /dev/mmcblk0p2
I see it as well. Let me find the problem and fix it.

[dejant_9]

woops! sorry,i made a mistake.. i test now this new kernel.

edit: (your right, no draggable keyboard ..) android boot with the default.txt of the wiki. but,


When booting this kernel keep in mind that you need to configure the correct mtype in haret's default.txt:
1805 for Diamond
1910 for Raphael



My diamond boot fine with mtype set to 1910. No problem, same functions available.
But with mtype set to 1805, init stop to " [21.700061] trout_pwrsink_set:STUB!"

With the other kernel(cupcake) and mtype set to 1805, haret crash

[NetRipper]

Fixed the Diamond mmc problem in git. Latest git should supports mmc for Raphael and Diamond (GSM versions) without needing to specify msmsdcc_id. Here's a zImage (http://netripper.com/raphael/20090401_zImage_git-01), if someone with a Diamond could verify to make sure, that'd be great.

But with mtype set to 1805, init stop to " [21.700061] trout_pwrsink_set:STUB!"

Should be fixed now.

@jobo
Did you mention having a cleaner .diff for the draggable and toggable vkeyb than the one you already provided? I still got the one from a few weeks back. If so, let me know. :)

And anyone know any other patches that are already available and should be added to git?

[touch of jobo]

Yeah, i think mapping at least a hardware key to get back to the home screen would be a major todo point.

I've been trying to get some hard buttons working, but not really successfully. :( It looks like we don't get all GPIs, maybe the hardware keyboard on the Raphael is not even GPI driven, and then there is the thing where monitoring one GPI will inhibit another, just to make things difficult (and slow)...

Without changing the GPIO drivers in arm/mach-msm (except for disabling halibut/trout keypads in board_htcraphael.c:halibut_init()), the following GPIs work work: (Tested on raph100)
38 - keyboard slider open
40 - left 1/3 of the navi pad: home/left/phone
41 - volume up / volume down
42 - right 2/3 of the navi pad: up/down/center/right/back/end
83 - power button
94 - navi pad touched (not even pressed) anywhere

This is the test driver code I wrote to investigate (http://members.home.nl/j0b0/board_htcraphael-gpiotest.c.txt), in case someone can use it to get further.

In this kernel (http://members.home.nl/j0b0/zImage-jobo) (yes, I keep replacing that one), the volume up and down buttons do 'end'.

UPDATE: The gpio test driver apparently makes diamonds hang. I updated the kernel with one that only monitors the GPI for the volume buttons (http://members.home.nl/j0b0/zImage-jobo) (which should act as the 'end' key) and no others. Maybe that helps. (Let me know.. :)) I also built one without the gpio driver (http://members.home.nl/j0b0/zImage-nogpio-jobo), so that gives you nothing new (except the vkeyb toggle?).
[...]
@jobo
Did you mention having a cleaner .diff for the draggable and toggable vkeyb than the one you already provided? I still got the one from a few weeks back. If so, let me know. :)
I attached the diff for msm_vkeyb.c only to this post (http://forum.xda-developers.com/showthread.php?p=3118229). If that conflicts or doesn't work, I can generate a complete diff of the entire tree, but that's quite messy...

-Jobo

[keerttttt]

I've been trying to get some hard buttons working, but not really successfully. :( It looks like we don't get all GPIs, maybe the hardware keyboard on the Raphael is not even GPI driven, and then there is the thing where monitoring one GPI will inhibit another, just to make things difficult (and slow)...

Without changing the GPIO drivers in arm/mach-msm (except for disabling halibut/trout keypads in board_htcraphael.c:halibut_init()), the following GPIs work work: (Tested on raph100)
38 - keyboard slider open
40 - left 1/3 of the navi pad: home/left/phone
41 - volume up / volume down
42 - right 2/3 of the navi pad: up/down/center/right/back/hangup
83 - power button
94 - navi pad touched (not even pressed) anywhere

This is the test driver code I wrote to investigate (http://members.home.nl/j0b0/board_htcraphael-gpiotest.c.txt), in case someone can use it to get further.

In this kernel (http://members.home.nl/j0b0/zImage-jobo) (yes, I keep replacing that one), the volume up and down buttons do 'end'.


I attached the diff for msm_vkeyb.c only to this post (http://forum.xda-developers.com/showthread.php?p=3118229). If that conflicts or doesn't work, I can generate a complete diff of the entire tree, but that's quite messy...

-Jobo
With your new zImage i get the following error:
Console handover: boot [htc_fb0] -> real [tty0]

At least, that's the last line it shows, then it stops.

I have a HTC Touch Diamond unbranded (DIAM100)

[keerttttt]

Fixed the Diamond mmc problem in git. Latest git should supports mmc for Raphael and Diamond (GSM versions) without needing to specify msmsdcc_id. Here's a zImage (http://netripper.com/raphael/20090401_zImage_git-01), if someone with a Diamond could verify to make sure, that'd be great.



Should be fixed now.

@jobo
Did you mention having a cleaner .diff for the draggable and toggable vkeyb than the one you already provided? I still got the one from a few weeks back. If so, let me know. :)

And anyone know any other patches that are already available and should be added to git?

Tested it, get the same message a couple of times , but it boots through though. seems working fine...

[NetRipper]

With your new zImage i get the following error:
Console handover: boot [htc_fb0] -> real [tty0]

At least, that's the last line it shows, then it stops.

I have a HTC Touch Diamond unbranded (DIAM100)
Jobo has a Raphael. Could be that some gpio's that are raphael-specific that jobo is trying to access is causing a hang on the Diamond. I.e. the "keyboard-slider" is something the Diamond doesn't have.

[dejant_9]

With your new zImage i get the following error:
Console handover: boot [htc_fb0] -> real [tty0]

At least, that's the last line it shows, then it stops.

I have a HTC Touch Diamond unbranded (DIAM100)

same here on diamond unbranded.

Originally Posted by NetRipper :
Fixed the Diamond mmc problem in git. Latest git should supports mmc for Raphael and Diamond (GSM versions) without needing to specify msmsdcc_id. Here's a zImage, if someone with a Diamond could verify to make sure, that'd be great.
Should be fixed now.

no, same problem..

[NetRipper]

same here on diamond unbranded.

Originally Posted by NetRipper :
Fixed the Diamond mmc problem in git. Latest git should supports mmc for Raphael and Diamond (GSM versions) without needing to specify msmsdcc_id. Here's a zImage, if someone with a Diamond could verify to make sure, that'd be great.
Should be fixed now.

no, same problem..
Are you using mtype 1805? And USB must be in Activesync mode (not disk drive). Other than that it should work on GSM Diamond. Multiple people have tested and verified as ok. You're sure you use the 01 zImage? :)

[touch of jobo]

With your new zImage i get the following error:
Console handover: boot [htc_fb0] -> real [tty0]
At least, that's the last line it shows, then it stops.
I have a HTC Touch Diamond unbranded (DIAM100)
The gpio test driver apparently makes diamonds hang. I updated the kernel with one that only monitors the GPI for the volume buttons (http://members.home.nl/j0b0/zImage-jobo) (which should act as the 'end' key) and no others. Maybe that helps. (Let me know.. :)) I also built one without the gpio driver (http://members.home.nl/j0b0/zImage-nogpio-jobo), so that gives you nothing new (except the vkeyb toggle?).
[...] You're sure you use the 01 zImage? :)
I think that was @me. Another case of my gpio test thingy breaking diamonds... (@dejant_9: was it? and if so: does the update fix it?)

-Jobo

[DfenZ]

can someone post the latest progress or update the wiki? i just get confused reading the thread : P

[holydhaliwal]

I can confirm that i have gotten it working on my cdma telus diamond (finnally) of course colors are still messed up, messed up in a different way then with android 0.8 though i noticed, its a lot more blue lol. Also performence seems to be a little laggy, anyone else notice this? Not sure if this has to due with early code or anything...

[Gavin_H]

I noticed that when going through the F buttons on the "floaty" keyboard that a bunch of different sets of colour bars come up when you go between F8-F10. Maybe if you messed with those you could sort out the colours? Unlikely I know, but maybe a clue to the actual solution?

[lavender.t]

Fixed the Diamond mmc problem in git....

it worked like a charm in my DIAM110 :)

regarding the code, netripper, should we, at a certain point, separate MACH_HTCRAPHAEL and MACH_HTCDIAMOND totally ? now they are built together only to be switched by the MTYPE.

it means some more code changes like quite some #ifdef CONFIG_MACH_HTCRAPHAEL need to be changed to #if defined(CONFIG_MACH_HTCRAPHAEL) || defined(CONFIG_MACH_HTCDIAMOND).

- lt

[mdrobnak]

Wow everyone's on a roll..

I'll have to check out these later kernels to see how they are on the Fuze..

-Matt

[dejant_9]

(@dejant_9: was it? and if so: does the update fix it?)

-Jobo
kernel with gpio work for my diamond. vol down = go to home :D


@NetRipper yes it was the good zimage.. :rolleyes:
i don't know, maybe Activesync problem. today it work fine..

so, Netripper fixed the Diamond mmc problem,and now diamond boot fine with the good mtype (1805). Jobo update kernel with volume down patch and draggable keyboard but diamond don't boot with the good mtype. Maybe everyone have to start with a unified kernel?

And after that, maybe someone can try to update the wiki with latest kernel and all latest ext2.img, I want to make that but my english is so bad :(

[timmy1980]

I can confirm that i have gotten it working on my cdma telus diamond (finnally) of course colors are still messed up, messed up in a different way then with android 0.8 though i noticed, its a lot more blue lol. Also performence seems to be a little laggy, anyone else notice this? Not sure if this has to due with early code or anything...

I did notice performance was a little slow too (diam100) but assume that this is down to it still being in the early stages?

I'm really impressed with the progress here, seems like Lavender getting 1.0 to boot was the break through everyone was waiting for.. since then there has been some awesome work from all involved! (netripper, jobo etc.. sorry if i've forgotten to credit anyone!) Very exciting stuff!

On a side note, I noticed that the ambient light sensor works in android... screen brightness is adjusted automatically. Is this expected?

[touch of jobo]

I have been looking into hardware buttons and the navi pad a bit more.
The 'home', 'send', and 'left' button are the same (pressing one while olding down the other doesn't give another click). Same for 'end', 'back', 'right'. Up and down can be pressed at the same time, and both give a click, but might be wired together, maybe even along with the center button. To tell where you press, the surface is a touch pad. (The input device driver for the navi pad would be a combination of multiple (2) platform devices?)
Looking at this touch pad, I found the following:

In HaRET, when you look at physical address 0xA990000C, the first byte of that word changes a lot when touching the navi pad (the thing that fires GPI 94 in my earlier tests). There is a little bit of a pattern in values versus finger movement, but not entirely clear or consistent; maybe have to only read at the right moment (in the irq handler?). 0xA9900004 contains 0x072D.

In Linux, When I watch 0xA990000C, it contains some value that doesn't change, but it's different every time I boot, so I guess it just doesn't get updated. 0xA9900004 contains 0x035D.

All these addresses are in the I2C region according to the memory map on the wiki. In i2c_msm.c/msm_i2c_probe(), it says:


enum {
I2C_WRITE_DATA = 0x00,
I2C_CLK_CTL = 0x04,
I2C_STATUS = 0x08,
I2C_READ_DATA = 0x0c,
I2C_INTERFACE_SELECT = 0x10,

[... skipping about 350 lines to msm_i2c_probe() ...]

/* I2C_HS_CLK = I2C_CLK/(3*(HS_DIVIDER_VALUE+1) */
/* I2C_FS_CLK = I2C_CLK/(2*(FS_DIVIDER_VALUE+3) */
/* FS_DIVIDER_VALUE = ((I2C_CLK / I2C_FS_CLK) / 2) - 3 */
i2c_clk = 19200000; /* input clock */
target_clk = 100000;
/* target_clk = 200000; */
fs_div = ((i2c_clk / target_clk) / 2) - 3;
hs_div = 3;
clk_ctl = ((hs_div & 0x7) << 8) | (fs_div & 0xff);
writel(/*clk_ctl*/ 0x72d, dev->base + I2C_CLK_CTL);
printk(KERN_INFO "msm_i2c_probe: clk_ctl %x, %d Hz\n",
clk_ctl, i2c_clk / (2 * ((clk_ctl & 0xff) + 3)));

dmesg says:
[ 0.430000] msm_i2c_probe: clk_ctl 35d, 100000 Hz
So the driver calculates a (wrong?) clock control value for the i2c bus. Merely writing 0x072D in stead of clk_ctl is not enough to make it work. (If only we knew someone who is good with clock divider values and stuff :p )

About the not updating thing: In msm_ts.c, we apparently need to clear some bits in some registers for the position to get updated again:
/* Without next 2 lines, position won't get updated */
writel(readl(MSM_TS_BASE + TSSC_CTL) & ~0x400, MSM_TS_BASE + TSSC_CTL);
writel(readl(MSM_TS_BASE + TSSC_CTL) & ~0xc00, MSM_TS_BASE + TSSC_CTL);

.. maybe we need something similar for the i2c data (after all .. it _is_ another touch panel). I have no on clue how to figure out which bit in which address if any. (There are so many...)

Another thing I could tinker with (unless someone with actual knowledge tells me its a waste of time) is the 'interface select' register. In HaRET, that address contains 0x0300 (which is _SELECT_SDL | _SELECT_SDA but not _SELECT_INTF_SELECT). Let's see...

-Jobo

[balsat]

kernel with gpio work for my diamond. vol down = go to home :D


And if you hold the vol up for a while = Phone power off

[zycho]

which files do you use for diam300?
can anyone pls post current working links to the files?

[Blazeitup123]

Android now running on HP Mini-Note 2133

http://www.engadget.com/2009/01/05/move-over-eee-android-now-running-on-hp-mini-note-2133/

[jbelman]

Am using DIAM100 running WM 6.1 (ROM VER: 1.93.161.2 WWE, RADIO VER: 1.00.25.03, Protcol Version: 52.26a.25.09U)

Connected to PC using ActiveSync and moved the following to a folder on internal storage.:


http://members.home.nl/j0b0/zImage-nogpio-jobo
http://members.home.nl/j0b0/zImage-jobo
Haret.exe (0.5.2)
default.txt:

set MTYPE 1805
set KERNEL zImage-nogpio-jobo
set RAMSIZE 0x7800000
set RAMADDR 0x10000000
bootlinux


Launch haret.exe shows: Found machine Generic MSM7xxxA

Run then I just get "HaRET: Booting Linux ... " (left this for 5 minutes no joy)

So swapped default.txt to:
set MTYPE 1805
set KERNEL zImage-jobo
set RAMSIZE 0x7800000
set RAMADDR 0x10000000
bootlinux


Run then again I get "HaRET: Booting Linux ... " (left this for 5 minutes no joy)

What am I missing?

[lupine_85]

I've been looking into the slide-out keyboard as well, taking the approach of disassembling the Windows driver: KbdNopAll.dll, keypad.dll, OEM_Misc.dll and MicroP.dll - nothing particularly amazing just yet though, i'm afraid. gpio33 is our friend, somehow... i just don't have a good handle on how windows does it all yet. once i do, hopefully i should be able to extract the i2c / gpio information that gets the hardware keyboard working. jobo: maybe we can collaborate somewhat on this? the slide-out keyboard doesn't seem to be a traditional row/cols one...

[dejant_9]

What am I missing?
all :D

read the wikki: http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux
section progress: and create the tmp folder on the root of internal storage,
then replace the wikki's ext2.img by the ext2.img in the cupcakes (http://forum.xda-developers.com/showthread.php?t=402002&page=132 , post #1315)
and replace the zimage by the latest zimage (by NetRipper or Jobo)
and use the default.txt in the wikki with jobo's zimage. for NetRipper's zimage, and if you have a diamond, just change mtype 1910 by mtype 1805. Nothing more.

[zycho]

when i enter my pin in android, it always told me 'incorrect pin' .. but its the right one.
thats the problem?

[dejant_9]

already see.. read the forum.. and try to desactivate the pin in WM

[keerttttt]

Yeah, when i'm using Jobo's latest zImage it also hangs on the "Haret Booting Linux" box. No terminal screen or what so-ever.

[touch of jobo]

And if you hold the vol up for a while = Phone power off
Are you sure? The vol-up and vol-down buttons are really the same. And even if they're not, I made only 1 GPI 'do' something. Nice to know anyway that it work someone else than myself ..and on a Diamond ;)

[...] jobo: maybe we can collaborate somewhat on this? the slide-out keyboard doesn't seem to be a traditional row/cols one...
Of course :) .. anything I can do other than investigate and report, please tell me. I am also in the irc channel (not always paying attention tho).

already see.. read the forum.. and try to deactivate the pin in WM
I added this one to the FAQ (http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux#FAQ) on the wiki (http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux). :rolleyes:

-Jobo

[NetRipper]

Run then again I get "HaRET: Booting Linux ... " (left this for 5 minutes no joy)
The kernels that jobo produce don't have raphael/diamond split yet I think, so you'll need to use mtype 1910 (also for diamond, on his kernels).

regarding the code, netripper, should we, at a certain point, separate MACH_HTCRAPHAEL and MACH_HTCDIAMOND totally ? now they are built together only to be switched by the MTYPE.

it means some more code changes like quite some #ifdef CONFIG_MACH_HTCRAPHAEL need to be changed to #if defined(CONFIG_MACH_HTCRAPHAEL) || defined(CONFIG_MACH_HTCDIAMOND).

- lt
Yes, still need to do that. I'm still wondering how we will proceed in the future though. Now we just have HTCRAPHAEL and HTCDIAMOND. However, we will also need to get a CDMA version for both, such as HTCRAPHAEL800 and HTCDIAMOND800. I also heard the raph500 and diam500 may differ from eachother as well, so we also need HTCRAPAHEL500 and HTCDIAMOND500.

Do we really want a:
#if defined(CONFIG_MACH_HTCRAPHAEL) || defined(CONFIG_MACH_HTCDIAMOND) || defined(CONFIG_MACH_HTCRAPHAEL500) || defined(CONFIG_MACH_HTCDIAMOND500) || defined(CONFIG_MACH_HTCRAPHAEL800)|| defined(CONFIG_MACH_HTCDIAMOND800)
? :)

One of the spots this will be is the irq.c under mach-msm, pretty low-level.

I guess we're not the first to run into something similar, how is it normally solved in a kernel?

[balsat]

Are you sure? The vol-up and vol-down buttons are really the same. And even if they're not, I made only 1 GPI 'do' something. Nice to know anyway that it work someone else than myself ;)


No i make a mistake, both the up and down button= Power off if hold >1 sec.
nice work with the "Back-button" is safe me for a lot of "ps" and "kill".

[touch of jobo]

The kernels that jobo produce don't have raphael/diamond split yet I think, so you'll need to use mtype 1910 (also for diamond, on his kernels).
Correct.

Do we really want a:
#if defined(CONFIG_MACH_HTCRAPHAEL) || defined(CONFIG_MACH_HTCDIAMOND) || defined(CONFIG_MACH_HTCRAPHAEL500) || defined(CONFIG_MACH_HTCDIAMOND500) || defined(CONFIG_MACH_HTCRAPHAEL800)|| defined(CONFIG_MACH_HTCDIAMOND800)
? :)
Wouldn't it be much nicer to keep the machine/board as generic as possible and distinguish in probe functions? (Otherwise the number of machines doubles every time a variable is encountered.)

-Jobo

[lupine_85]

:Of course :) .. anything I can do other than investigate and report, please tell me. I am also in the irc channel (not always paying attention tho)

Heh. well, i'm at the gym just now (i LOVE mobile internet!) - i'll try to catch you when i get back. mostly it's just trying different stuff, see what works, document everything - includig what doesn't ;). I'm mostly on the lookout for constants like the gpios, memory addresses, etc - let me know if there's aything yu think i should look for in particular!

[xianthax]

I have been looking into hardware buttons and the navi pad a bit more.
The 'home', 'send', and 'left' button are the same (pressing one while olding down the other doesn't give another click). Same for 'end', 'back', 'right'. Up and down can be pressed at the same time, and both give a click, but might be wired together, maybe even along with the center button. To tell where you press, the surface is a touch pad. (The input device driver for the navi pad would be a combination of multiple (2) platform devices?)

*snip*

-Jobo

hey guys, looking at the hardware blast here: http://www.phonewreck.com/wiki/index.php?title=HTC_Touch_Pro

there is a CY8C20434 and its mounted on the PCB behind the home/back/send/hangup buttons. This is a little 8bit uC with an analog cap sense blocks, its probably running the touch sense aspects of the panel. It most likely communicates with the qualcomm chip via i2c. I would venture that the windows driver gets a GPIO when a button is pushed, they it makes a call to the cypress part to figure out where on the pad a finger is present.

Unfortunately since its not a one function unit and has a programmable uC on board, the interface to talk to it is not documented but something the HTC designers cooked up and burned into the cypress chip, it may be hard to crack. I have an O-scope with I2C bus reading capabilities and surface mount rework equipment at home, worse come to worse i can attempt to put a scope on the bus and see whats happening, that could be very tough in the limited space inside the unit.

There is also a ATMEGA86V listed on that hardware page, a part i can't find, however the ATMEGA series are simple, general purpose 8bit uC's the only thing i can think of that its doing would be managing some sort of key GPIOs? It also appears to be mounted to the same PCB behind the back/home/etc buttons. Doesn't make much sense as the keyboard connector doesn't go to this board so, i dun know.

cheers,

x

[pauldgroot]

When pressing the volume buttons or the end key on the Virtual Keyboard my diamond hangs.

I'm using the last images and tried using the latest kernels from jobo.

I'd love to get back to the homescreen if possible.

Btw. sending and recieving sms messages works fine!

[keerttttt]

Correct.


Wouldn't it be much nicer to keep the machine/board as generic as possible and distinguish in probe functions? (Otherwise the number of machines doubles every time a variable is encountered.)

-Jobo
Yes, changing the Mtype worked, booted fine. And the volume button worked.
Also, indeed holding the button gives me the option to shutdown. However, it sorta 'hangs' while shutting down. But who cares :P

[Arktronic]

The 'home', 'send', and 'left' button are the same (pressing one while olding down the other doesn't give another click). Same for 'end', 'back', 'right'. Up and down can be pressed at the same time, and both give a click, but might be wired together, maybe even along with the center button. To tell where you press, the surface is a touch pad. (The input device driver for the navi pad would be a combination of multiple (2) platform devices?)

The entire area below the screen is supposed to be capacitive, so it shouldn't be able to detect which buttons are pressed when using gloves, for example. However, WinMo DOES see a difference between the four corners and the Enter button when using cloth to press them. So apparently there is a physical difference when pressing the buttons. A problem will probably arise when trying to get the arrows to work because it's unlikely that there are enough physical buttons to recognize arrows versus other keys without the help of capacitance.

[c4software]

Hi everyone with the lastest kernel and my HTC diamond when i push the down level sound button. my screen make a fade to black, and i don't find anyway to bring back android.

PS: Thaks everyone for your work.

[lavender.t]

... there is a CY8C20434 and its mounted on the PCB behind the home/back/send/hangup buttons. ...

didnt realize the keypad was so hitech ... hope i dont have to take apart my diamond physically to write code ... :)

also @jobo. i played with your gpio-test.c (very helpful) and haret last night and found out the same except that there was an I2C touchpad involved. put it in the irc before i went to sleep. just to summarize here, including your comments in the code which i also verified in DIAM110:

GPI 37 -- stylus detection.
GPI 40, 41, 42 -- keypad columns: 0, 1, 2.
GPI 83 -- power button.
GPI 94 -- touchpad.

GPO 32, 33, 34, 35 -- keypad rows: 0, 1, 2, 3.

for DIAM110, got the following key maps: <row, col>

<2, 1> -- volumn-up
<3, 1> -- volumn-down

<1, 2> -- down
<2, 2> -- up
<3, 2> -- enter (the round button)

<1, 0> -- left-half of the touchpad: (home, send, left)
<0, 2> -- right-half of the touchpad: (back, hangup, right)

- lt

[Akhara]

nice work

i will drink for your success hahahaa :D

[BGH80]

This is all great news. Looks like we will have a 100% working Android by this time tomorrow.

[touch of jobo]

Hi everyone with the lastest kernel and my HTC diamond when i push the down level sound button. my screen make a fade to black, and i don't find anyway to bring back android.
PS: Thaks everyone for your work.
You need to set the 'end button behavior' (in spare parts app) to 'go home', and not sleep. (I tested it myself just now.) I merely made the volume buttons 'hit' the end key. I don't influence what the end key key does.

[...]
GPI 40, 41, 42 -- keypad columns: 0, 1, 2.
GPO 32, 33, 34, 35 -- keypad rows: 0, 1, 2, 3.
[...]
- lt
Awesome! Gonna play with that a bit more. I don't think I could get those rows to fire, only the columns..
In linux that is. In HaRET everything fires.
Thanks
-Jobo

[lavender.t]

I don't think I could get those rows to fire, only the columns..
In linux that is. In HaRET everything fires.
Thanks
-Jobo

that's correct ... only the GPIs (columns, etc) register for IRQs. when the IRQs are handles, the gpio matrix code outputs the scan patterns to the the GPO rows to determine which <row, col> has a key press:

1000
0100
0010
0001

i'm also attaching the keypad code that i modified from halibut-keypad.c if it could save you some time :). for now it just maps the keys to 0 - 9, A and B just to find out the GPIO key mapping.

- lt

[mdrobnak]

This is all great news. Looks like we will have a 100% working Android by this time tomorrow.

I'm not sure I'd go that far ;-)

But everyone is working hard to get things working as soon as possible.

-Matt

[vitaminjay1982]

so i was reading through the stuff when you go to load android. and i noticed that it said "ept max 512mb"

would this have anything to do with the mms slot not reading a 1 gig or 2 gig card?

just a shot in the dark...

[touch of jobo]

that's correct ...
[...]
i'm also attaching the keypad code that i modified from halibut-keypad.c if it could save you some time :). for now it just maps the keys to 0 - 9, A and B just to find out the GPIO key mapping.
- lt

Thanks, but I just finished putting your pin mappings in a keypad driver... :)

board-htcraphael.c (http://members.home.nl/j0b0/board-htcraphael-20090106-jobo.c)
board-htcraphael-keypad.c (http://members.home.nl/j0b0/board-htcraphael-keypad-20090106-jobo.c)

It maps: 'Send' for left half, 'End' for right half, vol up/down, up/down/enter as you would think.
This kernel (http://members.home.nl/j0b0/zImage-keypad-jobo) has those key assignments. (It might be more useful to map 'menu'/'back' for 'up'/'down', since you would scroll and select in lists by touch.)

-Jobo

[Bimboy]

that's correct ... only the GPIs (columns, etc) register for IRQs. when the IRQs are handles, the gpio matrix code outputs the scan patterns to the the GPO rows to determine which <row, col> has a key press:

1000
0100
0010
0001

i'm also attaching the keypad code that i modified from halibut-keypad.c if it could save you some time :). for now it just maps the keys to 0 - 9, A and B just to find out the GPIO key mapping.

- lt

Do you know if it just goes left to right, row 1-4 (for Raph100), and row 1-5 (for Raph110 - Fuze) for physical keys? If so, then we need (again) another configurations for the different keypad layouts.

I updated the board-htcdiamond-keypad.c driver with the remainder of the alphabet, but realized that it might be the physical key location.


Edit: Nevermind... I see Jobo posted. the question does still remain about the Raph110 vs Raph100 though

[zEar]

The entire area below the screen is supposed to be capacitive, so it shouldn't be able to detect which buttons are pressed when using gloves, for example. However, WinMo DOES see a difference between the four corners and the Enter button when using cloth to press them. So apparently there is a physical difference when pressing the buttons. A problem will probably arise when trying to get the arrows to work because it's unlikely that there are enough physical buttons to recognize arrows versus other keys without the help of capacitance.

Hi all, maybe Trancecoder theory is not so dumb, try to press the buttons with your stylus, nothing happens !

[freeemet]

Hi all, maybe Trancecoder theory is not so dumb, try to press the buttons with your stylus, nothing happens !

Definitely!

Further, if someone has Core Player, load a movie and pause it. Now press the "call end" button with the stylus. The video goes forward one step, which means that this thing is mapped to some key after all.

[xianthax]

Hi all, maybe Trancecoder theory is not so dumb, try to press the buttons with your stylus, nothing happens !

Its a bit of a misconception that all capacitive touch panels can't be activated with gloves on. Current technology is sensitive enough to allow them to work through thin non conductive materials and its also very dependant on the thickness of the dielectric overlay. My fuze works through a scarf for instance even using touch only actions like zoom in opera

For an object to be detected by a capacitive panel it must hold charge (like your finger). The stylus is rubber tipped and probably too small even if it was a conductive metal, won't work. Try the zoom control with the negative terminal of a AAA battery, works great...

anyway it looks like there is no problem accessing the mechanical key presses. Do we know have a chart of i2c addresses for the components in the phone? If we can figure out the address of the cypress IC I could probably write a test program to check its response to commands, most i2c comms protocols are pretty simple hopefully its just a memory mapped/register style system. i'm also hoping the atmega controller isn't involved. i think the hardware site had a typo and meant the atmega88v which is interesting in itself as its the middle device in a pin compatible family, would indicate it may be doing something non trivial.

-x

[mdrobnak]

Thanks, but I just finished putting your pin mappings in a keypad driver... :)

board-htcraphael.c (http://members.home.nl/j0b0/board-htcraphael-20090106-jobo.c)
board-htcraphael-keypad.c (http://members.home.nl/j0b0/board-htcraphael-keypad-20090106-jobo.c)

It maps: 'Send' for left half, 'End' for right half, vol up/down, up/down/enter as you would think.
This kernel (http://members.home.nl/j0b0/zImage-keypad-jobo) has those key assignments. (It might be more useful to map 'menu'/'back' for 'up'/'down', since you would scroll and select in lists by touch.)

-Jobo

Great work man!
I tested it on the Fuze, works pretty well. Only thing missing is a Menu button. Did we find any GPIO related to the PTT button on the Fuze?

The other thing is, I'm not sure what the difference in configuration between your kernel & NetRipper's kernel is, but yours does not produce the Safe Mode message in the lower left corner, where his does...

lupine_85 started doing some disassembling of oem_misc.dll, I'm going to try and see if I can do some as well, but I don't know if I'll have time tomorrow.

-Matt
-

[c4software]

Nice a most of buttons works. but i have a question (maybe dumb...).

Why we need to put on usb the diamond to boot correctly on linux ??

And always no way to wake up the diamond when android put it in sleep ? (at this time, i just disable timeout screen on android to fix the problem)

[lavender.t]

Hi all, maybe Trancecoder theory is not so dumb, try to press the buttons with your stylus, nothing happens !

true. both jobo's and my tests confirmed this and jobo was already a step further to look into the I2C issues.

this is indeed not a good news which means more dev work for us to figure out the I2C interface and decode the data to map keys -- which could be significant amount of new kernel code.

xianthax's input is very valuable too. for the time being we can simply map the left and right halves into SEND/END keys so at least some features may work.

- lt

[petay]

there is a test tool for the bottom buttons linked in this post
http://forum.xda-developers.com/showthread.php?p=2509753#post2509753

its navdbgtool.eve in the zip it can show where your finger is on the buttons.

just had a play withthe new zImage and its lovely!!! i can really see this doing a lot of good for the diamond/pro.

there are a few errors popping up about settings not responding and i get a few errors about the softkeyboard (im assuming its the cupcake added softkeyboard as there android errors)

also, would mapping the power key allow us to recover from the screen timeout???

[dejant_9]

there is a test tool for the bottom buttons linked in this post
http://forum.xda-developers.com/showthread.php?p=2509753#post2509753

its navdbgtool.eve in the zip it can show where your finger is on the buttons.


there is a log function in this tools. te log is create in \my documents

[jbelman]

Got my DIAM100 on WM6.1 (VF-UK) up and running using:

default:

set mytype 1910
set kernel zImage-jobo
set initrd initrd.droid.cpio.gz
set ramsize 0x6000000
set ramaddr 0x10000000
set cmdline "mem=64M
msmsdcc_id=2 msmts_calib=0x7a.0x5e.0x35a.0x37f"
boot


With:


haret18
zImage-jobo

[From touch of jobo (http://members.home.nl/j0b0/)]


cache.ext2
system.ext2
userdata.ext2

[From cupcakes package (http://forum.xda-developers.com/showthread.php?t=402002&page=132)]


initrd.droid.cpio

[From Wiki package (http://xda.willboost.com/diam-20081229.tar.gz)]

Although it seem's to only load 50% of the time, it will hang and not load keyboard then if reset it will load ok, if reset again it will not etc.

Thanks for everyone's help so far

[dejant_9]

check this files (http://217.128.108.212/master/android/latest/) and remember that you have to get active sync connected before executing haret.exe, and stylus have to be out

[Gluck]

SMS is fully working. I can send and recieve SMS even with Russian text(only recieve)!

But why Dialer is crashed if i press button? This is same AT commands to modem, which sends SMS manager, as i know.

I testet it on RAPH100 with zimage-keyboard-jobo and partitioned SD card.

[touch of jobo]

[...] I'm not sure what the difference in configuration between your kernel & NetRipper's kernel is, but yours does not produce the Safe Mode message in the lower left corner, where his does...
-Matt
My best guess is that NetRippers kernel has the halibut or trout keypad driver enabled, monitoring a GPI that not wired to the keypad but to something else. This would cause android to boot while a key (menu button??) is pressed, making it run in 'Safe mode'.

[...] Why we need to put on usb the diamond to boot correctly on linux ??
From what I see, if you ever had USB connected since last reset, it must be connected when staring linux from HaRET or it hangs. But if you haven't had USB connected since last reset, it will boot up just fine. (Which may give the impression that it works 'every other time'.)

[...] this is indeed not a good news which means more dev work for us to figure out the I2C interface and decode the data to map keys -- which could be significant amount of new kernel code.
- lt
Wouldn't we have to figure out the i2c anyway, for things like light sensor, accelerometer, LEDs, maybe camera controls, .. .. oh yeah .. and the slide-out querty keyboard :)? (..of course that doesn't make it any better news..)

[...] also, would mapping the power key allow us to recover from the screen timeout???
No. Right now, the machine dies when it goes to sleep, no matter _how_ it goes to sleep (timeout, button, menu, ...).
EDIT: Correction: It does not die, it goes blind. As mdrobnak points out in the post below this one, it's still running.

SMS is fully working. I can send and recieve SMS even with Russian text(only recieve)!
Cool! Thats the 3rd or 4th report of SMS working.

-Jobo

[mdrobnak]

My best guess is that NetRippers kernel has the halibut or trout keypad driver enabled, monitoring a GPI that not wired to the keypad but to something else. This would cause android to boot while a key (menu button??) is pressed, making it run in 'Safe mode'.

No. Right now, the machine dies when it goes to sleep, no matter _how_ it goes to sleep (timeout, button, menu, ...).

-Jobo

Wrt point one, that makes sense.

Wrt the power - I'm not convinced it's "dead", as I left it in a call, and the screen died, but my terminal was very much alive...It's just a power management issue with the LCD I think. (I'm being nit-picky here - to the average user, yes, it's "dead.")

-Matt

[c4software]

From what I see, if you ever had USB connected since last reset, it must be connected when staring linux from HaRET or it hangs. But if you haven't had USB connected since last reset, it will boot up just fine. (Which may give the impression that it works 'every other time'.)


Yep same for me i try like you say and its boot.

[Gavin_H]

check this files (http://217.128.108.212/master/android/latest/) and remember that you have to get active sync connected before executing haret.exe, and stylus have to be out

Which of these needs to be replaced if I'm already running the TMP contents from the wiki, except with the cupcake ext2s uploaded by lavender.t?

'Cause your webspace is so, so slow it would take me all night to download everything.

[dejant_9]

Haret.exe and cache.ext2.img are the same than wikki's. If you have the lavender.t cupcake, you juste need the zImage, and default.txt
This webspace is my own old PC in server mode. It was just for test. Sorry i don't have another webspace with more than 130mo for sharing datas..

[esbenr]

I've just got a Diamond in spare so i thought that i'll try to put some Android on to it.

Where can i find a guide and the nescary files?

Did you guys get to fix the minor issues?

[chemical1979]

I've just got a Diamond in spare so i thought that i'll try to put some Android on to it.

Where can i find a guide and the nescary files?

Did you guys get to fix the minor issues?

In the first post in this thread is a link to a wiki where you can find the information you need.

[Gavin_H]

Haret.exe and cache.ext2.img are the same than wikki's. If you have the lavender.t cupcake, you juste need the zImage, and default.txt
This webspace is my own old PC in server mode. It was just for test. Sorry i don't have another webspace with more than 130mo for sharing datas..

That's ok, thanks.

[balsat]

I got the files to on my home server, it might be a little bit faster!
http://87.72.13.87/

[c4software]

if you need a didcated server to upload/share files. i can share my dedicated server if you want. i upload the lastest file tonight. i post the link when its done.

[Bimboy]

Tip: Some of you may not know this, but you can go into the Settings application and change the sleep timeout value, under the display app. This will keep the OS active for longer than a minute, unattended.

Nice to see the Vol+ and Vol- buttons being mapped to actual vol up and down, center, up, and down as well as power button! Good progress.

[keerttttt]

Wrt point one, that makes sense.

Wrt the power - I'm not convinced it's "dead", as I left it in a call, and the screen died, but my terminal was very much alive...It's just a power management issue with the LCD I think. (I'm being nit-picky here - to the average user, yes, it's "dead.")

-Matt
I was in a test call on my phone to my home phone. and the screen went dead.
I pushed the end button after a few seconds... and the phone hanged up. so yeah, i would say that it's only a screen fail :)

Oh yeah, buttons work nice, volume and end/call buttons :)

Other thing, when i tried to call to my phone, it ringed 2 times... but nothing to see on my phone itself. after 2 rings it went to voicemail.

Then, when i called to home from my mobile phone, that worked. Strange...

[timmy1980]

It maps: 'Send' for left half, 'End' for right half, vol up/down, up/down/enter as you would think.
This kernel (http://members.home.nl/j0b0/zImage-keypad-jobo) has those key assignments. (It might be more useful to map 'menu'/'back' for 'up'/'down', since you would scroll and select in lists by touch.)

-Jobo

Tried this zimage and it works great on my diam100, buttons work as described making it feel more and more like it should!

Not sure if this is related or not (not even sure it could be!) I have a problem in WM where by whilst connected to my laptop in either activesync or disk mode, after a few seconds my diamond screen turns off and I lose the USB connection. The only reason that I mention it is that it seems to have been doing it since I started playing around with Android??? More than likely a coincidence I realise but thought I would mention it anyway.

[TheLord-Akira]

tested on diam100

sms works
making a call works (without sound of course)
buttons OK
getting a call not works (caller get busy signal)

this is my second android test. first one (first android 1.0 image from ehmm i dont know) was very fast and smooth. the last one with radio OK is very slow like wm ^^.

i hope it dont depand on radio.

great work guys.

[c4software]

Strange, on my diamond i can make/receive call.

[ripnetuk]

Haret.exe and cache.ext2.img are the same than wikki's. If you have the lavender.t cupcake, you juste need the zImage, and default.txt
This webspace is my own old PC in server mode. It was just for test. Sorry i don't have another webspace with more than 130mo for sharing datas..

If there is really a shortage of bandwidth / space to distribute these rootfs's, why done we just open a DropBox account, and stick it on there? it allows up to 2gb of free storage, and can generate a public http:// link to let others download the file. It also comes with client software to keep the online storage in line with a local folder, so to 'release' a file, you just copy it to the folder, and right-click and do 'copy public url'. It has none of the annoyingness associated with RapidShare et al...

Alternatively, I have a unlimited bandwidth and unlimited storage account with DreamHost, so I can host anything that needs hosting...

Really excited with all the recent progress... thanks to everyone working on this :)

george

[Arktronic]

Other thing, when i tried to call to my phone, it ringed 2 times... but nothing to see on my phone itself. after 2 rings it went to voicemail.

Then, when i called to home from my mobile phone, that worked. Strange...

Same thing happens on my Fuze. Which device do you have? (BTW, it'd be nice if everyone put their device info in their signatures ;))

[gunar_adrian]

I tried to install Android on a Diamond with 7201 chipset and this is the log.
Any idea?

[frysee]

another way for hosting the necessary files: Wuala (http://www.wuala.com/en/)

Once uploaded, the files are (encrypted, splitted and) spread all over the Wuala network - you get 1GB space to start.

This would solve the bandwith and storage problems and its easier to follow all the progress :)

A little example: Linux/Android on HTC Diamond (http://www.wuala.com/Linux+and+Android+on+HTC+Diamond?lang=en)

The files can be accessed by using your browser or downloading the optional java-client for more options (e.g commenting, adding your own files, filesystem integration etc.) ;)

greetz

[camdenb]

Hi folks, just writing to report that I have successfully booted android on an HTC Fuze, using the directions in the wiki and LordKiwi's initrd.cpio. Radio does not appear to work, sound does not appear to work, hardware buttons don't work, and I cannot find this spare parts app!

but it boots!

now that I've been through the fires of initial setup, I am willing to do further testing if anybody needs me to.

EDIT: Also, thanks a ton for all your hard work, guys!

[lattiS710]

Which files from this site do I need to run android with (almost) working radio and hardware buttons?

http://87.72.13.87/

sorry for this noob question :o

[jringo718]

I am having trouble with mounting my 1GB Sandisk Microsd. I don't have a linux system, only Vista, so I am not sure how I stop the /mmc/tmp/system.ext2 not found, etc etc errors from showing up. I have transferred the files through a memory card reader as well as through activesync and i've tried it with activesync on and off.

I have tried a combination of the files from last week and now I am using the Haret18 and zimage-jobo from his site, the system.ext2.img and userdata.img from the cupcake gz, and the cache.ext2.img and intrid.droid.gz from the wiki(the intrid.droid.gz is from Lordkiwi's modified Touch Pro intrid) I put everything in tmp on the root of memory card.

I have a Sprint CDMA Touch Pro btw. It shows that mmc error then I calibrate the screen and type /init.android and press enter. It says loading ANDROID and a bunch of stuff and then just hangs.

Thanks guys.

[nuke1]

Which files from this site do I need to run android with (almost) working radio and hardware buttons?

http://87.72.13.87/

sorry for this noob question :o
Please state your device: Raph or Diam?

[lattiS710]

nuke1, u r really helping me a lot, thank you :)

i have a diamond

[counterfitninja]

Quote:
Originally Posted by lattiS710 View Post
Which files from this site do I need to run android with (almost) working radio and hardware buttons?

http://87.72.13.87/

sorry for this noob question
Please state your device: Raph or Diam?

Im trying Diamond as well, not much luck. Got it work earlier this week, but wanted to try new keyboards

[xianthax]

more on the touch part of the navipad, forgive me if this is already known

fired up HaRET (very good work whoever created this diog tool btw)

i noticed that just touching the navi pad fires these IRQs:

001936: mem IRQS(1) 95300084=00000001 (00000001)
004332: mem IRQS(0) 95300080=00080000 (00080000)
004332: mem IRQS(0) 95300080=00010000 (00010000)

this is interesting, in normal embedded design practices with i2c you want to avoid polling on the bus because its a slow bus and designed to be shared with multiple devices. In this case the cypress part only supports i2c slave operation and therefore can never initiate a transfer itself. the presence of this IRQ would indicate to me that the cypress touch sense IC has a GPIO pin or 2 tied back to the qualcomm chip to trigger this IRQ in the qualcomm chip effectively saying "hey i have some information you may want"

I would try just sending a read request to the cypress IC after catching this IRQ (i'm assuming you can find the trigger for this IRQ in the linux kernel too?) you may find that the cypress IC will just dump some information to you, probably just the finger location(s) on the touchpad.

If this is true then there is no need to decipher the i2c protocol but just decipher the data that comes back from the cypress IC which should be pretty easy...

I'd try the atmel controller also, as it may have something to do with this?

so the questions are:
1) does simply touching the navipad fire an IRQ or event in linux?
2) what are the i2c addresses of the cypress and atmel parts?

if someone is familiar with the i2c driver in the linux kernel, would you be willing to scan the bus? Unfortunately i work in mostly custom embedded OS's or bare metal so i'm not familiar with the procedure in linux

By scan i mean for each possible address do this:

Start Condition
7bit Address with R/~w = 0 (shouldn't matter usually)
Stop

and record if the address transmission is ack'd.

addresses that are reserved and don't need checking, assume lsb is R/W:
0x00 through 0x0F
0xF0 through 0xFF

that would give us the list of all I2C addresses that are active in the device, and may be faster than trying to disassemble the windows dll to find them

cheers,

x

[iservealot]

In an effort to not sound dumb here, am I reading correctly that there has been an "end" button mapped correctly?

When I try to the new zImages posted here, all I get it a faded grey screen whenever a button is pressed. ...am I missing something here?

[xvx45]

In an effort to not sound dumb here, am I reading correctly that there has been an "end" button mapped correctly?

When I try to the new zImages posted here, all I get it a faded grey screen whenever a button is pressed. ...am I missing something here?
With files found here : http://forum.xda-developers.com/showpost.php?p=3130207&postcount=1426 work great with my DIAM100, for the moment I only test sound with calling my voice mail but no sound of course, keymap work grea too.

[allofthepeople]

I'm running the Sprint HTC Touch Pro, any one ever fix the frezing after the pwrsink_set: STUB! messages yet? I also am not seeing a keyboard.
Thanks

[mdrobnak]

In an effort to not sound dumb here, am I reading correctly that there has been an "end" button mapped correctly?

When I try to the new zImages posted here, all I get it a faded grey screen whenever a button is pressed. ...am I missing something here?

You need to run the "Spare Parts" application in Android, and map the end key to go back home. The default is to go to sleep, which causes the issue you're experiencing.

-Matt

[touch of jobo]

Do you know if it just goes left to right, row 1-4 (for Raph100), and row 1-5 (for Raph110 - Fuze) for physical keys? If so, then we need (again) another configurations for the different keypad layouts.
We can make any row the top row (or bottom row for that matter) by putting the gpi number for that row first (or last for that matter) in the row_gpios array. same for columns.

[...] the presence of this IRQ would indicate to me that the cypress touch sense IC has a GPIO pin or 2 tied back to the qualcomm chip to trigger this IRQ in the qualcomm chip effectively saying "hey i have some information you may want" [...]
Yes, if you request an irq for gpio #94, that will fire when you touch (not even press) the navi pad.

[...] If this is true then there is no need to decipher the i2c protocol but just decipher the data that comes back from the cypress IC which should be pretty easy...
It seems that there are more devices on the bus than just the navi touch pad. So we would have to be able to tell from the data which device sent it.

You need to run the "Spare Parts" application in Android, and map the end key to go back home. The default is to go to sleep, which causes the issue you're experiencing.
Added this one to FAQ on the wiki :) (2x or 3x is frequently)




Following the example of lavender.t, I added a new kernel boot option for the virtual keyboard: msmvkeyb_toggle=hide makes the vkeyb initially hidden/collapsed (you see the blue square only) and msmvkeyb_toggle=off makes it not even init at all. (Now that we dont need the vkeyb for all the time for everything anymore..)


static int __init msmvkeyb_toggle_setup(char *str)
{
if (!strcmp(str,"off"))
vkeyb_toggle = -1;
else if (!strcmp(str,"hide"))
vkeyb_toggle = 0;
else
vkeyb_toggle = 1;
return 0;
}
__setup("msmvkeyb_toggle=", msmvkeyb_toggle_setup);


Also, I connected the 'keyboard slide open'-gpio, so that android turns landscape when you open slide open the keyboard on a raphael, see attached pictures. Here is kernel with this boot option and the rotation (http://members.home.nl/j0b0/zImage-rotate-jobo). I shall post code/diffs shortly..

-Jobo

Hmmm ... that wallpaper image got shrunk by the attachment system .. it should be 720x640 (http://members.home.nl/j0b0/wallpapers/wallpaper_trees.jpg).

[sibbor]

another way for hosting the necessary files: Wuala (http://www.wuala.com/en/)

Once uploaded, the files are (encrypted, splitted and) spread all over the Wuala network - you get 1GB space to start.

This would solve the bandwith and storage problems and its easier to follow all the progress :)

A little example: Linux/Android on HTC Diamond (http://www.wuala.com/Linux+and+Android+on+HTC+Diamond?lang=en)

The files can be accessed by using your browser or downloading the optional java-client for more options (e.g commenting, adding your own files, filesystem integration etc.) ;)

greetzWuala has fast downloads and it's easy to use... overall, seems great! I became a member on there and hope that updates will be uploaded there.

Downloaded all the latest files for Android 1.0 for my Diamond. Tested jobo's image that was on there (no clue if it was the latest one around!?). It had dragable & hideable onscreen keyboard either way.

First try = failure. Got stuck after some mmcblk0 stuff. I read it might be with USB connection from WM6 - if the device had been connected to a PC and was disconnected when trying to start Andriod it would failed. In my case I've justed transferred the files to my device and disconnected it from the PC - so USB connection was NOT active.
After a reboot of my device I managed to get past the prior stage and boot into Android from the shell using /init.android.

Using a DIAM*00 - how do I find out exact version?
Got nothing new to add, but atleast I tested it out ;)!

init: cannot find '/system/bin/playmp3' disabling bootsound or something similar in the shell right after /init.andriod
The com.android.settings error
LCD display screen issues during sleep state - same as everyone else
I was able to see my carrier and the network in use in the "status screen" of Andriod


Following the example of lavender.t, I added a new kernel boot option for the virtual keyboard: msmvkeyb_toggle=hide makes the vkeyb initially hidden/collapsed (you see the blue square only) and msmvkeyb_toggle=off makes it not even init at all. (Now that we dont need the vkeyb for all the time for everything anymore..)

Also, I connected the 'keyboard slide open'-gpio, so that android turns landscape when you open slide open the keyboard on a raphael, see attached pictures. Here is kernel with this boot option and the rotation (http://members.home.nl/j0b0/zImage-rotate-jobo). I shall post code/diffs shortly..

-JoboSince you just released this new zImage I was able to test it out on my Diamond. Added the command to default.txt:set cmdline "mem=64M msmsdcc_id=2 msmts_calib=0x7a.0x5e.0x35a.0x37f msmvkeyb_toggle=hide", and it works. My Diamond booted into Android in landscape mode, and I wasn't able to change it. I understand this is a image for Raph, but reporting either way. Also the virtual keyboard didn't rotate (and I don't know if it's supposed to - probably not...). So typing on the Diamond was a pain - of course.

Another thing I noticed: Android perhaps thought there's a hardware keyboard slided out? Because I can't seem to get the official onscreen keyboard up in the Google searchbox.

[NetRipper]

My Diamond booted into Android in landscape mode, and I wasn't able to change it. I understand this is a image for Raph, but reporting either way. Also the virtual keyboard didn't rotate (and I don't know if it's supposed to - probably not...). So typing on the Diamond was a pain - of course.
The GPIO that indicates whether the keyboard is out (1) or in (0) always reports 1 on the Diamond, which is why you get the landscape on diamond.

[frysee]

Wuala has fast downloads and it's easy to use... overall, seems great! I became a member on there and hope that updates will be uploaded there.

Thats what I said :D
Would be nice to concentrate the new files there as it works with good speed for everyone.

Yesterday I tested the new files by myself, android wont accept my PIN - any ideas? :(

greetz

[NoBody20013]

as far as i know: try disabling PIN in Win Mobile

[mdrobnak]

as far as i know: try disabling PIN in Win Mobile

What he said. :-)

It's in the FAQ on the Wiki

http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux

-Matt

[frysee]

What he said. :-)

It's in the FAQ on the Wiki

http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux

-Matt

Wow, I'm sure this wasnt in the wiki yesterday :rolleyes:

greetz

[sibbor]

The GPIO that indicates whether the keyboard is out (1) or in (0) always reports 1 on the Diamond, which is why you get the landscape on diamond.Alright I see. Quick question: I wasn't able to use the official on-screen keyboard with that image. Was it because Android inactivates it if GPIO for the keyboard reports 1, or was it because it doesn't support landscape mode, any clue?
Wow, I'm sure this wasnt in the wiki yesterday :rolleyes:It was added recently ;)

[NetRipper]

Alright I see. Quick question: I wasn't able to use the official on-screen keyboard with that image. Was it because Android inactivates it if GPIO for the keyboard reports 1, or was it because it doesn't support landscape mode, any clue?
I honestly don't know. Best guess would be that it doesn't use the android-keyboard because it is in landscape, and on G1, that means the keyboard is out. Android probably assume a keyboard when it is in landscape mode. The kernel just sends a EV_SW (swap) event, not a specific event that indicates the keyboard is actually out (as far as I know, jobo could confirm this perhaps).

[c4software]

I make a video of Android on Diamond.

http://www.youtube.com/watch?v=Ws7lSl8UfuY


For those who are afraid to test.

[allofthepeople]

I finally got it to show the keyboard and last prompt but the visual keyboards still not working for me?

I'm running the Sprint HTC Touch Pro, any one ever fix the frezing after the pwrsink_set: STUB! messages yet? I also am not seeing a keyboard.
Thanks

[touch of jobo]

The GPIO that indicates whether the keyboard is out (1) or in (0) always reports 1 on the Diamond, which is why you get the landscape on diamond.
Correct. It would be nice it it were always low on diamonds, but what can we do..

[...] Was it because Android inactivates it if GPIO for the keyboard reports 1, or was it because it doesn't support landscape mode, any clue?
Best guess would be that it doesn't use the android-keyboard because it is in landscape, and on G1, that means the keyboard is out. Android probably assume a keyboard when it is in landscape mode. [...]
I don't get the soft keyboard either when switched to landscape mode, and NetRippers best guess is mine as well: Android doesn't do soft keyboard (or soft keyboard doesn't do itself) when the keyboard is open. (The GPI doesn't request landscape mode, it indicates hardware keyboard open.)
[...] The kernel just sends a EV_SW (swap) event, not a specific event that indicates the keyboard is actually out (as far as I know, jobo could confirm this perhaps).
The kernet sends and EV_SW (switch, not swap) and does provide whether the switch is on or off. But in Diamonds, the gpi in question indicates that the keyboard is open.

I put this code:
void htcraphael_init_keypad(void)
{
/* no slider switch for diamonds */
if (!machine_is_htcraphael())
htcraphael_keypad_data.info_count = 1;

..in board_htcraphael_keypad.c (http://members.home.nl/j0b0/board_htcraphael_keypad-20080107.c.txt), as a quickie to inhibit the switch on non-raphaels (thanks for that tip last night NetRipper), but I guess I have to build separate kernels for raph and diam (or update to the latest git ;) ? ..or both? :confused:) for that to work?

Wow, I'm sure this wasnt in the wiki yesterday
Right. :) Please add frequently asked questions to the FAQ, preferably along with the answers.

-Jobo

EDIT: Oh, and by the way, I changed the keypad mapping so that up='menu' and down='back'. Having a back button is nice, so you don't have to go all the way back to the home screen every time. I can't seem to get the menu button to perform a menu like function.. I tried KEY_MENU, 229 (softkey1) and 230 (softkey2) but neither work.
Moreover, apps or activities seem to crash a lot when tapping the wrong key at the wrong time. I see this message a lot:
W/KeyCharacterMap( 198): No keyboard for id 0

[NetRipper]

..in board_htcraphael_keypad.c, as a quickie to inhibit the switch on non-raphaels (thanks for that tip last night NetRipper), but I guess I have to build separate kernels for raph and diam (or update to the latest git ? ..or both? ) for that to work?
The fun thing about machine_is_htcraphael() is that it checks the mtype at runtime. Any #if or #ifdef's are compiletime. With the machine_is_xxx() style we can build a single kernel with support for both diamond and raphael. However, diamond users must set mtype 1805 and raphael users must set mtype 1910 (we now all use 1910). This is already present in git, which is why I suggest you also update your local copy to the latest git. Currently the kernels you build cant be used with mtype 1805 (it will simply hang at the haret screen, as the kernel bails out directly when it doesn't recognize the mtype).

Also, when you update your git, you will have two board files. One for raphael and one for diamond. That way you can also just enable the GPI thingy in raphael, and leave it disabled by default for any other mach's. It's also done this way with the SD slot card detection.

Obviously, for the kernel to support both raphael and diamond, you need to have both CONFIG_MACH_HTCDIAMOND and CONFIG_MACH_HTCRAPHAEL enabled in your .config file, so that it builds both mach's.

[poxo]

when I'm in the android htc my screen gradually fades diam 2 seconds, then nothing blank screen.
happen after a few dozen seconds or when I press the pad "arrow" or when I press the button "end", "top down", "page up "....
Apart from that everything is fine, as-OK, sms-OK, OK-pad, ....

you have a solution, thank you in advance Paschal.

[jlavoie]

change screen timeout to never and also change the end button behaviour to go home and not sleep.

[poxo]

when I'm in the android htc my screen gradually fades diam 2 seconds, then nothing blank screen.
happen after a few dozen seconds or when I press the pad "arrow" or when I press the button "end", "top down", "page up "....
Apart from that everything is fine, as-OK, sms-OK, OK-pad, ....

you have a solution, thank you in advance Pascal.

[butterdori]

I don't get the soft keyboard either when switched to landscape mode, and NetRippers best guess is mine as well: Android doesn't do soft keyboard (or soft keyboard doesn't do itself) when the keyboard is open. (The GPI doesn't request landscape mode, it indicates hardware keyboard open.)

The kernet sends and EV_SW (switch, not swap) and does provide whether the switch is on or off. But in Diamonds, the gpi in question indicates that the keyboard is open.


Instead of using the keyboard out GPI, is there something that can trigger just the landscape view (using the g-sensor to switch between them for diamond, perhaps)? This leads to the question, is there a (cupcake) keyboard for landscape as well?

Also, when you update your git, you will have two board files. One for raphael and one for diamond. That way you can also just enable the GPI thingy in raphael, and leave it disabled by default for any other mach's. It's also done this way with the SD slot card detection.
are compiletime. With the machine_is_xxx() style we can build a single kernel with support for bot
Obviously, for the kernel to support both raphael and diamond, you need to have both CONFIG_MACH_HTCDIAMOND and CONFIG_MACH_HTCRAPHAEL enabled in your .config file, so that it builds both mach's.


Would making a kernel that supports both devices increase its size significantly? Or would this more depend on what the differences are between the two..

[Arktronic]

The fun thing about machine_is_htcraphael() is that it checks the mtype at runtime. Any #if or #ifdef's are compiletime. With the machine_is_xxx() style we can build a single kernel with support for both diamond and raphael.

While a single kernel is probably a good idea at this early stage of development, once the kernel is more or less stabilized, we should switch to compile-time conditionals and have separate builds for Diamond and Raphael as that would make both kernels smaller, easier to debug, and (theoretically) faster.

EDIT: butterdori beat me to it :)

[allofthepeople]

nevermind... another site says delete calb line in default.txt

I finally got it to show the keyboard and last prompt but the visual keyboards still not working for me?

[poxo]

change screen timeout to never and also change the end button behaviour to go home and not sleep.

OK, "dispaly screen" = timeout never

but when I press on keypad "<---" the screen goes blank !@]@!!@°°@!!!:mad:

[curiousblue]

OK, "dispaly screen" = timeout never

but when I press on keypad "<---" the screen goes blank !@]@!!@°°@!!!:mad:

this is a problem thats been discussed earlier in the thread. take a look at the wiki and FAQ. (http://wiki.xda-developers.com/index.php?pagename=RaphaelLinux)

[jlavoie]

OK, "dispaly screen" = timeout never

but when I press on keypad "<---" the screen goes blank !@]@!!@°°@!!!:mad:

from wiki:

When I boot android 1.0, the screen fades to black / gray / weird liney pattern when I hit some key (home, end, F4) or even when I do nothing

This happens when the phone goes to sleep. There are 2 things to prevent this: For the timeout, go to settings->sound and display->display timeout->never timeout. (default is 1 minute) For the button/key thing, go to spare parts->end key behavior->go home. (default is sleep or home+sleep)

[poxo]

thank you :)

[Dasken]

if the development is going forward, i´m realy hoping to juse adroid in my HTC Tuch Diamond:)

[allofthepeople]

mines doing the cannot find /system/bin/servicemanager , disabling servicemanager thing I tried "is/mmc" like suggested it says "/bin/sh: is:/mmc: not found"

[Arktronic]

FYI, it's not "is/mmc", it's "ls /mmc" - LS, just all lowercase, and a space between that and "/mmc".

[allofthepeople]

i tried it both ways...

FYI, it's not "is/mmc", it's "ls /mmc" - LS, just all lowercase, and a space between that and "/mmc".

[allofthepeople]

just tried to run it from device and not card but same issue

i tried it both ways...

[NetRipper]

Would making a kernel that supports both devices increase its size significantly? Or would this more depend on what the differences are between the two..
Depends on the differences between the two. Generic parts are only kept once. I just tested it and compiling for one specific device saves 892 bytes. We'll see how it goes. I guess experience will tell. I'm new to this as well ;)

[sticks13]

Hello All,

Thanks to everyone who is contributing. I have been following this thread for some time now, checking it several times daily while I'm at work! :D

I plan on picking up a Raphael from Verizon in the next few days. I'm quickly learning the work involved to port android over to the raphael. I know everyone is working hard on this, but when might we expect a fully working version of Android on the raphael? Are we talking weeks or months here?

[NetRipper]

Hello All,

Thanks to everyone who is contributing. I have been following this thread for some time now, checking it several times daily while I'm at work! :D

I plan on picking up a Raphael from Verizon in the next few days. I'm quickly learning the work involved to port android over to the raphael. I know everyone is working hard on this, but when might we expect a fully working version of Android on the raphael? Are we talking weeks or months here?
Hi. I should warn you about the Verizon version of Raphael. If I'm not mistaken, the Verizon one is the raph500. That phone seems to have a different chipset from the other Raphaels.

The raph100, raph110, raph800 have the MSM7201A. The raph500 seems to have MSM7500A.

Although someone is already working on the raph500, I thought I should warn you anyway ;)

Pretty please correct me when I'm wrong :)

[ydoucare]

Running on my Sprint Diamond (DIAM500), with the funky colors. Appreciate all the work everyone is doing on this! Can't wait to see what is down the road.

[NetRipper]

Running on my Sprint Diamond (DIAM500), with the funky colors. Appreciate all the work everyone is doing on this! Can't wait to see what is down the road.
Sprint Diamond (CDMA) is DIAM800, not 500.

[ydoucare]

Sprint Diamond (CDMA) is DIAM800, not 500.

Then why does it clearly say DIAM500 in device info?

[Bimboy]

Then why does it clearly say DIAM500 in device info?


Go to the bootloader screen, and see what your device is there.. (Hold Vol down button while doing a soft reset, until tri-color screen appears)

[mattdh]

I second that answer. I have my Sprint Touch Diamond right here in front of me and its clearly the DIAM500.

[ydoucare]

Go to the bootloader screen, and see what your device is there.. (Hold Vol down button while doing a soft reset, until tri-color screen appears)

I'll have to try that later, but why would it differ?

[touch of jobo]

Instead of using the keyboard out GPI, is there something that can trigger just the landscape view (using the g-sensor to switch between them for diamond, perhaps)?
The g-sensor .. I wish.
but...
This kernel (http://members.home.nl/j0b0/zImage-pwrbtnswlid-jobo) maps the power button to the lid switch (not directly, like high=open, low=closed, but when it changes from low to high it toggles between open and closed / landscape and portrait) so that diamonds can rotate.
I'm not up to date with git so diamonds are mtype=1910. (I think)
Enjoy.
-Jobo

[Willis111]

Then why does it clearly say DIAM500 in device info?

My cdma telus touch diamond is labelled diam500 as well, both on the back of the phone and device info and bootloader screen.

[johnny_domino]

mines doing the cannot find /system/bin/servicemanager , disabling servicemanager thing I tried "is/mmc" like suggested it says "/bin/sh: is:/mmc: not found"

I am getting the same thing. Tried the correct spelling, didn't work. Sprint Touch Pro.

[allofthepeople]

yeah ha i dont know why i keep typing it wrong i guess im tired. regardless i hope someone can help us!

[c4software]

The g-sensor .. I wish.
but...
This kernel (http://tibook.jb.b4m.com/j0b0/zImage-pwrbtnswlid-jobo) maps the power button to the lid switch (not directly, like high=open, low=closed, but when it changes from low to high it toggles between open and closed / landscape and portrait) so that diamonds can rotate.
I'm not up to date with git so diamonds are mtype=1910. (I think)
Enjoy.
-Jobo

?? lid open and close with the Diamond? its not for the raphaels only ?

PS : Its possible to acces from linux to the Internal storage Data? mine have one partition and its formated in fat. i think its possible like say the wiki. But its seems non working on my Diamond.

[Carumir]

Hi,

sorry for asking such a dumb questions.

I managed to boot up Android on my Diamond XDA (Rebranded HTC Touch Diamond from O2).

1. I don't find this spare parts application (and going into sleep mode seems not an option?). Where do i find it to remap keys.

2. is there a general key for getting back to the Desktop of android. Whenever i open up an application i have to stay there or in a sub-menu. There seems to be no Key like "Esc" to return to upper menu, or an "Alt-Tab" Functionality, to switch to other apps?

3. My Keys don't work :-) maybe a common complaint. Only things that works is the audio keys (window pops up with volume). How can i provide help here?

Regards

[Zedith]

mines doing the cannot find /system/bin/servicemanager , disabling servicemanager thing I tried "is/mmc" like suggested it says "/bin/sh: is:/mmc: not found"

Probably made the same mistake as me, make sure the folder is named "tmp" and not "temp"

[Willis111]

... man ls (that's LS but lower case), and I think you wanted to do "ls /mmc". Caps and spacing is important. Also I think if you type "mount" it will give you a list of all mounted devices, like /dev/mmcblk0 etc. The most important thing is getting commands right, if it returns "/bin/sh:ls:*: not found" you typed something wrong

[dst6se]

Is here anyone how had compilled OPIE for TouchPro/Diamond ?
Would be wonderfull to run OPIE with VGA resolution !

[keerttttt]

Is here anyone how had compilled OPIE for TouchPro/Diamond ?
Would be wonderfull to run OPIE with VGA resolution !

Bleh, opie looks just as ugly as every other linux distro. I would'nt want that on my diamond ever!

Give me Android, complete gmail synchronisation. nice fluid animations, good looking windows. great!

[mdrobnak]

Hi,

sorry for asking such a dumb questions.

I managed to boot up Android on my Diamond XDA (Rebranded HTC Touch Diamond from O2).

1. I don't find this spare parts application (and going into sleep mode seems not an option?). Where do i find it to remap keys.

2. is there a general key for getting back to the Desktop of android. Whenever i open up an application i have to stay there or in a sub-menu. There seems to be no Key like "Esc" to return to upper menu, or an "Alt-Tab" Functionality, to switch to other apps?

3. My Keys don't work :-) maybe a common complaint. Only things that works is the audio keys (window pops up with volume). How can i provide help here?

Regards

Something I just thought of - The spare parts application may only be in the Cupcake build, But I know the screen timeout can be changed in all 1.0 builds.

-Matt

[foh1981]

Does everyone start Android by typing init.android at the BusyBox shell? Personally i got "command not found" and had to prefix it with ./

Not a bug or anything, but it's kinda strange if it's different behaviours, no?

Also, for users not familiar with Linux, you can type i<tab>.a<tab> to save a few valuable key presses on the init.android command (that's a really tiny on screen keyboard!!) :)

[foh1981]

Something I just thought of - The spare parts application may only be in the Cupcake build

Yeah, that would explain why I can't find it :P So if one would like to try the cupcake build, do I just have to download that tgz-file and change the appropriate files in the tmp folder? I downloaded the other tgz-file from the location given at http://forum.xda-developers.com/showpost.php?p=3101531&postcount=1203 and I noticed the cupcake build there too.

[touch of jobo]

?? lid open and close with the Diamond? its not for the raphaels only ?
I modified the keypad driver so that when you press the power button, the kernel tells android that you have opened or closed the keyboard. Android will believe this, and happily rotate its orientation.

[...] But I know the screen timeout can be changed in all 1.0 builds.
Yes, but not the 'End key behaviour'. Thats in Spare Parts. You have to change that to not put the device to sleep.

[...] Also, when you update your git, you will have two board files. One for raphael and one for diamond. That way you can also just enable the GPI thingy in raphael, and leave it disabled by default for any other mach's. [...] you need to have both CONFIG_MACH_HTCDIAMOND and CONFIG_MACH_HTCRAPHAEL enabled in your .config file, so that it builds both mach's.
I will update my git and only move over changes I want to keep. (I'm not sure about the investigative stuff I did in mddi, mdp tho..) I think you didn't commit the board keypad. I will produce clean diffs after I update.

Thanks,
-Jobo

Update: OK, I can confirm that changing from halibut keypad to htcraphael keypad gets rid of the safe mode.
Update: Here is a diff (http://members.home.nl/j0b0/linux-htc-msm-2.6.25-jobo20090108.diff.txt) after I moved over some modifications that I think we all want. Mostly the keypad and the gpio testing function.
The code I did in the past to investigate the raph800 colors is not in it. (I plan to move some of that over since it helps in some cases and never hurts, and it makes it easier to tinker more with it.)
Update: ...and here is the resulting kernel (http://members.home.nl/j0b0/zImage-clean-jobo).

[foh1981]

I just tried the cupcake build, and found the "Spare parts" program where all the other programs are. Also the service is working. Even made a call to my home phone.

(I'm using a DIAM100 with Telenor SE 3G.)

[allofthepeople]

some one at ppc told me this and it worked:

I finally got this working on my TP.

For sprint HTC Touch pro, this should be your default.txt

set mtype 1910
set kernel zImage
set initrd initrd.droid.cpio.gz
set ramsize 0x6000000
set ramaddr 0x10000000
set cmdline "mem=64M msmsdcc_id=3"
boot


Take note of the "msmsdcc_id=3"

This was causing my TP to not mount the memory card, and thus not find the disk images so it could not boot.



I am getting the same thing. Tried the correct spelling, didn't work. Sprint Touch Pro.

[c4software]

@ touch of jobo

And With this new kernel we can wake up the device when its go to sleep ?

Ps : Yes i know i can disable the sleep. but its for know.

[touch of jobo]

@ touch of jobo - And With this new kernel we can wake up the device when its go to sleep?
No. Thats not from lack of button control, but more like a power management issue in mddi. <-- guess
-Jobo

[c4software]

hmmm, the new kernel are not working for me. its break Android. the "home" keep crashing.

I say nothing. its my android. its seems corrupt... acore has stopped unexpectedly...

I have restore old img. and its works. I retry your kernel. Its cool. really Cool :D

Android sims a little bit more slow. maybe its really slowdown. Its same on your device ?

[allofthepeople]

First, not looking for help. Thought I would pass this along in case it helps.


I can see BusyBox shell starts and the next thing I see before the command prompt is "/bin/sh: can't access tty; control turned off"

I then init android, it boots in safe mode, and I immediately get a pop up:

"Sorry!
Application Settings (in process
com.android.settings) is not responding"

with two buttons: [Force Exit] and [Wait]

I'm getting same thing now, plus the colors are messed up on my touch pro. feels like im having a problem at every step! ugh

[GregOP]

So can someone rebuild the kernel files with the fixes for the raphael keyboard and safemode issues so that we can paste them in the /tmp folder? That'd be awesome! :)

[Bad HAL 9000]

* Delete me *

[touch of jobo]

So can someone rebuild the kernel files with the fixes for the raphael keyboard and safemode issues so that we can paste them in the /tmp folder? That'd be awesome! :)
I linked to a fresh kernel (http://members.home.nl/j0b0/zImage-clean-jobo) in the 3rd update of this post (http://forum.xda-developers.com/showthread.php?p=3138954).
-Jobo

[GregOP]

I linked to a fresh kernel (http://members.home.nl/j0b0/zImage-clean-jobo) in the 3rd update of this post (http://forum.xda-developers.com/showthread.php?p=3138954).
-Jobo

Ahh ok, I thought it would be on the first post, my bad! I used your kernel and now the keyboard is moveable and closeable and it doesn't boot in safemode, so thats good. Colors are still all 8bit looking and when you close the keyboard I have to move the screen with my thumb to the left or right to force it to redraw the area where the keyboard was. Is this normal behavior? It looked like the diamond has the colors working properly according to the wiki.

[allofthepeople]

I linked to a fresh kernel (http://members.home.nl/j0b0/zImage-clean-jobo) in the 3rd update of this post (http://forum.xda-developers.com/showthread.php?p=3138954).
-Jobo

ok so im a newbie bout this stuff... what do i do with the new kernal in reference to the currents files in the tmp folder on my phone's card?

[GregOP]

ok so im a newbie bout this stuff... what do i do with the new kernal in reference to the currents files in the tmp folder on my phone's card?

Very easy, take his newly download zImage-bla-jobo or whatever its called. Rename it to zImage. Then copy that zImage into the /tmp folder you have setup. It should overwrite the previous zImage file.

Also, for me, when I first run herat.exe it does all the loading but hangs right before the prompt, a soft reset, then a re run of the herat.exe file gets me to the command prompt line where I can type /init.android . Good Luck.