PPCGeeks - NAND Testing - 05-25 Update: New LK, Recovery.img, Kernel Updates through Recovery

PPCGeeks (http://forum.ppcgeeks.com/index.php)

- Android On TP2 Development (http://forum.ppcgeeks.com/forumdisplay.php?f=319)

- - NAND Testing - 05-25 Update: New LK, Recovery.img, Kernel Updates through Recovery (http://forum.ppcgeeks.com/showthread.php?t=134598)

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

No data with the last update. I'm working on installing the Android SDK now for full adb.

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Quote:

Originally Posted by steveholtam (Post 2065231)

No data with the last update. I'm working on installing the Android SDK now for full adb.

the latest build just went up so if u didnt just try it then its already the old one. they pump out new builds avery few hours these days its amazing how much proress there is, but either way go pull it from here

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Still no data using the 213238 nbh and update files. I've got the adb shell logcat stuff all figured out as well. Just took all day! Here is the latest logcat -b radio log attached.

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Nate/Lmiller : weird thing just happened.. i got some file corruption on /cache.

cyangen has two dalvik-caches and one had a bunch of nonsense and i would get oob errors on the dmesg log. Im thinking since cache is at the end of nand, you think we have the same issue as hd2 where they cant write to the last 24mb of nand ? So far this happened once and i guess im glad it was just the cache. Had it been the data it would be difficult to troubleshoot.

/data/dalvik-cache <= Used by market apps (stored in /data/app)
/cache/dalvik-cache <= Used by System apps (stored in /system/app).

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Quote:

Originally Posted by steveholtam (Post 2065253)

Still no data using the 213238 nbh and update files. I've got the adb shell logcat stuff all figured out as well. Just took all day! Here is the latest logcat -b radio log attached.

damn

fopen /etc/ppp/pap-secrets failed = 2

Means the files are not found. Have you figured out how to adb yet ? this might be a 1 line fix .. adb into the device and do the following

ls /system/etc

ls /data/ppp

Print out that you see. If /data/ppp has files, if it does then check if there is a ppp link inside /system/etc. If no link is found then remount system as rw using the command below.

mount -o rw,remount -t yaffs2 /system

Once thats done you can make the llink manually

ln -s /data/ppp /system/etc/ppp

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Quote:

Originally Posted by [ACL] (Post 2065218)

wow i didnt realize git didnt have my cache directory. Thats some buggy shit. I made a directory and addeded but git does not have it. So when you check the gz. its there .. odd. Thanks for the fixes bro.. I guess its smarter to have the script create it. So these are good commits.

I have a BIG todo for you.. :-)

when we install new modules we just put em in data/modules. Thing is they should really be in data/modules/<uname of kernel>. Then link it back to data modules. This is so we can have different versions of modules for more than one kernel. This is a must since if you upgrade a kernel and wipe your old modules you wont be able to go back incase its a bad kernel. Also we cant use mount bind like we do now. Sometimes it locks the partition busy so we need to implement this and use links for the final product. You can do it :-)

hmm.. maybe we need lmiller to adb and verify. Can you ssh into your phone once wifi is enabled? should give u full shell.

Edit: thanks to this im going to re-push out a new autobuild. shortly. Cache is important for cyanogen. they use it for alot.. so good call on that partition.

Sounds great. I also added the installer script to git so we can keep that consistent. I'll look into the uname stuff tomorrow. I almost always prefer ln -s or cp -a over mount --bind. :)

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Quote:

Originally Posted by natemcnutty (Post 2065266)

Sounds great. I also added the installer script to git so we can keep that consistent. I'll look into the uname stuff tomorrow. I almost always prefer ln -s or cp -a over mount --bind. :)

thanks dood.. also if you can test with the cache directory that would be awesome. I havent been able to replicate the issue of corruption yet.. (edit: just recreated it.. looks like cache seems to get corrupt).

I do know we cant format the recovery parititon table. Something about it being protected .. :-\

anwyays off to bed.. thanks for the commit..

21.352783] msm_nand_read_oob 1ff31000 800 10 failed -74, corrected 0
to
[ 62.875152] msm_nand_read_oob 1ff3f800 800 10 failed -74, corrected 0

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Quote:

Originally Posted by [ACL] (Post 2065257)

I doubt we have the same problem as the HD2 because they can't even format that section, but I guess we could test it pretty easily by taking some space away from the data partition and see if the issue goes away. Could it be running out of space on the cache? What if we upped it to 64 MB instead?

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

So you guys still can't write to the cache partition?
If so try changing this line on the MAKEDEV's

mknod -m 0666 /dev/mtd/mtd4 c 90 6
to
mknod -m 0666 /dev/mtd/mtd4 c 90 8

I stated this before but never looked at the push, but I was also uncertain about it... Sorry about that. :)

But maybe the cache getting corrupt is something different... Kinda like the losing of some widgets on the Android home screen?

As for not being able to write to the recovery partition are you still using the same starting place as before. 0x4000000?

I won't be able to test for a while today for I forgot my phone at home. :(

Re: NAND Boot Testing - 01-07: Panel power on/off fixes

Quote:

Originally Posted by natemcnutty (Post 2065214)

That looks awesome ACL. Any chance you can push all of your changes to bootenv (feel free to overwrite my commits)? I pulled down your changes and a few things were missing. I was initially not getting the cache partition to format and mount because it was not created yet. Once I did that, it got me further to where Android started initializing, but then it keeps rebooting.

Your files from the autobuild seem to boot up just fine, so I'm thinking there is something not committed in initrd or something like that. Also, no data for me either, but not real way to show anything without ADB. I'm sure it still has to do with pppd, but I'm just not quite sure where to begin on that.

Edit: Also, your androidinstall does not have a /system/bin/su

Nate if you keep getting your Android install stuck in a bootloop it means you are most likely having some permission issues.
Specifically this file: /system/etc/dbus.conf
Here is what the Vogue changes them too...
chown 1002:1002 /system/etc/dbus.conf
chmod 0444 /system/etc/dbus.conf

@ACL,
How are you creating your andriodinstall.tgz?
I'm thinking we are having issues with permissions coming right from the tgz.

Here is mine... Probably not right but it works once I run the fixperm script :)

Code:

echo "Making androidupdate.tar"

cd androidupdate

chown -R 0:0 *

tar -cvzf ../androidupdate.tar *

chown -R lance:lance *

cd ..

echo "Moving androidupdate.tar"

cp androidupdate.tar ~/Desktop/Android/1.Complete_Build/ANDROID/



echo "Making androidinstall.tar"

cd androidinstall

chown -R 0:0 *

tar -cvzf ../androidinstall.tar *

chown -R lance:lance *

cd ..

echo "Moving androidinstall.tar"

cp androidinstall.tar ~/Desktop/Android/1.Complete_Build/ANDROID/



cp install-seq.sh ~/Desktop/Android/1.Complete_Build/ANDROID/



echo "Completed..."