PPCGeeks

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)

MassStash 03-03-2011 11:53 PM

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

Originally Posted by steveholtam (Post 2064118)
ACL, or another, I want to test the new libhtcgeneric.so file for the RIL, but I am not sure the best way to swap the new file with the old file. Could someone point me towards some instructions? I tried swapping the file using 7-zip, but that did not seem to work. Should it have?

don't worry bout it bro, thanks tho....

steveholtam 03-04-2011 03:18 AM

Re: NAND Boot Testing - 01-07: Panel power on/off fixes
 
Well, I've been messing with the files available, and here are my findings. The 1/9 NBH has 3g on my 400. The brand new 3/3 NBH build still does not have 3g. I did not mess with the libhtcgeneric.so file manually, hoping it was in the 3/3 build. I like how these builds have the long press power = recent progs list. The latest build is snappier though then the first ones. Much more so. Waking up the phone seems a bit slower, and you have to do the screen touch trick everytime. I also get the rapid green/orange blinking led.

Just to confirm, here is my system info as shown on the phone:
2.2.1
2.6.27.46 autobuild@Builds#50
FRX04

If I had data, I would run this full time! Im hoping some of this helps someone.

[ACL] 03-04-2011 04:34 AM

Re: NAND Boot Testing - 01-07: Panel power on/off fixes
 
Nate: Here is a little info to help out. I cant replicate since i have no service but you might be able to do it.

/etc is a simlink to /system/etc (and ril needs to access /etc/ppp)

originally the ril bombs out because /system is read only. Now if we change it to rw, then it bombs out due to permissions. This isnt even because of pppd yet since this part of the code is before.

I'm not sure what the exact permissions are but we need to get it set. Also we can move the ppp folder to data so we can rw without messing up system.

So thats a new experiment to try.

Lmiller1708 03-04-2011 10:31 AM

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

Originally Posted by [ACL] (Post 2064203)
Nate: Here is a little info to help out. I cant replicate since i have no service but you might be able to do it.

/etc is a simlink to /system/etc (and ril needs to access /etc/ppp)

originally the ril bombs out because /system is read only. Now if we change it to rw, then it bombs out due to permissions. This isnt even because of pppd yet since this part of the code is before.

I'm not sure what the exact permissions are but we need to get it set. Also we can move the ppp folder to data so we can rw without messing up system.

So thats a new experiment to try.

So I tried to replicate the privileges the best I could with a new system...
Still can't get it to work. And still had to redo the pap and chap files to get them writable. Once I did that my service kept dropping completely out...


My next step was going to try this approach again: ln -s /system/bin/pppd /bin/pppd

Also what do you think about just copying everything from system/etc to /etc... This is what's done with Rootfs. And this might help with the permissions. :)

[ACL] 03-04-2011 11:09 AM

Re: NAND Boot Testing - 01-07: Panel power on/off fixes
 
Ok so this is the order or execution in the ril.

1) talk to modem and make sure you are ok
2) check pap and chap files make sure you can open them
3) attempt to fire up pppd

Now you notice pppd is the last step. Funny thing is we arent even getting there, we are failing at least once on 1 and then we keep failing on 2. Only reason i know this is because of the rils i made. They are very verbose so we get to see everything

Unfotunately we cant copy because android is natively setup this way. /etc will always be a link for /system/etc. But thats where it ends. We decide about the internals. So we can move ppp to data if we want to and link it back to /etc

The other method is to recompile pppd and have it look somewhere else. Now arrg was able to bring up data once after much hacking which was manual. But it was also short lived.

At this stage i say we try anything.. i finally got a boost mobile account and will try to activate today. So we will continue this through the weekend.

steveholtam 03-04-2011 01:21 PM

Re: NAND Boot Testing - 01-07: Panel power on/off fixes
 
Just wondering why data works on the Full Package build from 1/9 from (post #1), but not the later builds? Shouldnt that help figure out where it broke by comparing the two?

Lmiller1708 03-04-2011 01:25 PM

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

Originally Posted by steveholtam (Post 2064329)
Just wondering why data works on the Full Package build from 1/9 from (post #1), but not the later builds? Shouldnt that help figure out where it broke by comparing the two?


Because It's still running off the SDcard and using the Rootfs, Jjust like Haret. :)

But yes it will help in comparing them.

natemcnutty 03-04-2011 01:48 PM

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

Originally Posted by steveholtam (Post 2064118)
ACL, or another, I want to test the new libhtcgeneric.so file for the RIL, but I am not sure the best way to swap the new file with the old file. Could someone point me towards some instructions? I tried swapping the file using 7-zip, but that did not seem to work. Should it have?

If you are using the new full NAND build, the best way is to create a tgz named androidupdate.tgz with the folder structure as it is on the NAND. For libhtcgeneric-ril.so, the folder structure inside the tgz should be /system/lib/libhtcgeneric-ril.so

Put that in your andboot folder, reboot, and use replimenu or a custom install-seq.sh to apply it. I've removed replimenu from mine and added an androidupdate.tgz installer to init if it finds it, but I'm not sure if we want to commit that or not.

natemcnutty 03-04-2011 01:54 PM

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

Originally Posted by [ACL] (Post 2064284)
Ok so this is the order or execution in the ril.

1) talk to modem and make sure you are ok
2) check pap and chap files make sure you can open them
3) attempt to fire up pppd

Now you notice pppd is the last step. Funny thing is we arent even getting there, we are failing at least once on 1 and then we keep failing on 2. Only reason i know this is because of the rils i made. They are very verbose so we get to see everything

Unfotunately we cant copy because android is natively setup this way. /etc will always be a link for /system/etc. But thats where it ends. We decide about the internals. So we can move ppp to data if we want to and link it back to /etc

The other method is to recompile pppd and have it look somewhere else. Now arrg was able to bring up data once after much hacking which was manual. But it was also short lived.

At this stage i say we try anything.. i finally got a boost mobile account and will try to activate today. So we will continue this through the weekend.

Great information. I think that if we have it working with rootfs, we should have no issue porting that over to initrd or elsewhere and using symlinks or whatever we need to do. I'm slowly testing out changes at the moment to see what breaks what. Next on my list is to try swapping out android.init and see what changes I need to make to get that one working.

[ACL] 03-04-2011 01:59 PM

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

Originally Posted by natemcnutty (Post 2064339)
Great information. I think that if we have it working with rootfs, we should have no issue porting that over to initrd or elsewhere and using symlinks or whatever we need to do. I'm slowly testing out changes at the moment to see what breaks what. Next on my list is to try swapping out android.init and see what changes I need to make to get that one working.

Our initrd automatically checks for this already. It looks for a file called "init" on the system. If not found, then it executes its own android.init. Good for cookers who want to deploy a custom init if needed. So we should include the android.init inside /system but rename it as init :-)

BoominSVX got me a boost account so ill be active tonight to continue the good fight.. also the recovery is coming along fine. Not sure if im going to git it or not.


All times are GMT -4. The time now is 06:39 PM.

Powered by vBulletin® ©2000 - 2025, Jelsoft Enterprises Ltd.
©2012 - PPCGeeks.com


Content Relevant URLs by vBSEO 3.6.0