[NAND] CyanogenMod 7.2 from source

Search This thread
Jan 15, 2011
46
13
This is the dmesg.

For the logcat, i tried to do it via adb but it returns "Unable to open log device '/dev/log/main':No such file or directory"
So i tried to re-enable it commenting the line " rm /dev/log/main" in /system/etc/init.d/19speedy_modified but it did not solve the issue..
 

Attachments

  • newdmesg.txt
    15.8 KB · Views: 23

Elibl

Senior Member
Dec 20, 2011
686
448
i successfully compiled a build with the correct gps settings but there is somehow a problem with baseband read out for amss detection in init_gps_android_props() in gps-rpc.c on boot therefore i'm in a bootloop...

i think i'll rewrite/update whole gps code and add it directly to our device tree without amss detection stuff. i'm planning to adapt htc leo's libgps which is the same base and looks the best for me out there. i think it only needs to be changed to amss 5225. correct me if i'm wrong...

@coolrunnerii: would that be ok if i do that?

EDIT: for log: it looks like it's something with rpcrouter 15 at the first view but logcat would give more info but i can't sadly say atm how you enable it again jerry disabled it somehow...
 
Last edited:
  • Like
Reactions: Light_Sephiroth

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
This is the dmesg.

For the logcat, i tried to do it via adb but it returns "Unable to open log device '/dev/log/main':No such file or directory"
So i tried to re-enable it commenting the line " rm /dev/log/main" in /system/etc/init.d/19speedy_modified but it did not solve the issue..

Thanks for the log.

Obviously there is something wrong in the rpcrouter. At least I know where to start now.
 

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
i successfully compiled a build with the correct gps settings but there is somehow a problem with baseband read out for amss detection in init_gps_android_props() in gps-rpc.c on boot therefore i'm in a bootloop...

i think i'll rewrite/update whole gps code and add it directly to our device tree without amss detection stuff. i'm planning to adapt htc leo's libgps which is the same base and looks the best for me out there. i think it only needs to be changed to amss 5225. correct me if i'm wrong...

@coolrunnerii: would that be ok if i do that?

EDIT: for log: it looks like it's something with rpcrouter 15 at the first view but logcat would give more info but i can't sadly say atm how you enable it again jerry disabled it somehow...

I noticed that you changed to that repo. The reason I chose SP3DEV/Alex's repo was that he had commited a patch fixing the amss version detect stuff. I think the library that is actually in the roms now is a build from Alex's source.

At least his fix has to be ported to the code we choose to use.

his repo is here:
https://gitorious.org/~ast/xdandroid/asts-hardware_xdandroid_gps

Check the logs, he did some quite important stuff there.

EDIT: I got some dmesg logs myself with the lib in the november Sundream rom. I didn't see any rpcrouter errors, but still got the strange touchscreen behaviour. I will have to turn on debugging in the touchscreen driver and see why it suddenly stops working. I even got a period where swiping would work to switch between home screens, but pressing an icon didnt work.
 
Last edited:

Elibl

Senior Member
Dec 20, 2011
686
448
I noticed that you changed to that repo. The reason I chose SP3DEV/Alex's repo was that he had commited a patch fixing the amss version detect stuff. I think the library that is actually in the roms now is a build from Alex's source.

At least his fix has to be ported to the code we choose to use.

his repo is here:
https://gitorious.org/~ast/xdandroid/asts-hardware_xdandroid_gps

Check the logs, he did some quite important stuff there.

EDIT: I got some dmesg logs myself with the lib in the november Sundream rom. I didn't see any rpcrouter errors, but still got the strange touchscreen behaviour. I will have to turn on debugging in the touchscreen driver and see why it suddenly stops working. I even got a period where swiping would work to switch between home screens, but pressing an icon didnt work.

yeah alex's commits were pulled to the repo i pushed if you have a look at the earlier commits or the code... but somehow with the build from your tree baseband isn't yet available at the gps init or something like that. i wrote some deeper debug and all the stringvalues read out for baseband and stored in tok in init_gps_android_props() are zero what causes a bootloop
 

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
i successfully compiled a build with the correct gps settings but there is somehow a problem with baseband read out for amss detection in init_gps_android_props() in gps-rpc.c on boot therefore i'm in a bootloop...

i think i'll rewrite/update whole gps code and add it directly to our device tree without amss detection stuff. i'm planning to adapt htc leo's libgps which is the same base and looks the best for me out there. i think it only needs to be changed to amss 5225. correct me if i'm wrong...

@coolrunnerii: would that be ok if i do that?

EDIT: for log: it looks like it's something with rpcrouter 15 at the first view but logcat would give more info but i can't sadly say atm how you enable it again jerry disabled it somehow...

If you could do this it would be very nice of you :)

Not much seems to happen with the gps lib at the moment, so forking now would be a good idea.
 

Elibl

Senior Member
Dec 20, 2011
686
448
If you could do this it would be very nice of you :)

Not much seems to happen with the gps lib at the moment, so forking now would be a good idea.

hey how is it going?
1) i think leo gps will get a great thing could you maybe have a look too if you've got time? makefiles are driving me crazy:silly:
in the leo device tree the leo-gps is compiled to /system/lib/libgps.so. but in gingerbread the devices use the gps."bootloader_name".so gps lib if i understood right. finally gps.htcleo.so is inluded via proprietary files to system/lib/hw/ after research i think it's just the compiled already known leo-gps source. after testing i think the defines from leo's boardconfig: *gps_loc_api* need this proprietary file. all very very weird:eek:
but the gps lib is doubly this way isn't it? maybe you know a better way to implement this... i'm very unadvanced with makefiles...

on what are you currently working?

2) another thing is that i already was on the point were the module loaded completely fine at boot but know i'm just not able to find back to the makefile configuration where everything was right. it's driving me really crazy and i already spent hours... maybe my working directory is corrupted cause my virtual machine had a sudden death caused by a reboot after a windows update:rolleyes:

i'll push the defines i'm playing around with... so you can maybe have a look at

all the best

EDIT: i think i solved 2) need to compile and test. bug of shame: i forgot to define board_gps_libraries
 
Last edited:

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
hey how is it going?
1) i think leo gps will get a great thing could you maybe have a look too if you've got time? makefiles are driving me crazy:silly:
in the leo device tree the leo-gps is compiled to /system/lib/libgps.so. but in gingerbread the devices use the gps."bootloader_name".so gps lib if i understood right. finally gps.htcleo.so is inluded via proprietary files to system/lib/hw/ after research i think it's just the compiled already known leo-gps source. after testing i think the defines from leo's boardconfig: *gps_loc_api* need this proprietary file. all very very weird:eek:
but the gps lib is doubly this way isn't it? maybe you know a better way to implement this... i'm very unadvanced with makefiles...

on what are you currently working?

2) another thing is that i already was on the point were the module loaded completely fine at boot but know i'm just not able to find back to the makefile configuration where everything was right. it's driving me really crazy and i already spent hours... maybe my working directory is corrupted cause my virtual machine had a sudden death caused by a reboot after a windows update:rolleyes:

i'll push the defines i'm playing around with... so you can maybe have a look at

all the best

EDIT: i think i solved 2) need to compile and test. bug of shame: i forgot to define board_gps_libraries

I'm deligthed to hear of you progress with the GPS. Good work :) Just keep pushing all the changes you make to github.

I'm working on the sd boot issue. I couldn't make a card that worked and I am fed up with having to remove it before booting.

When you have a gps lib that works more or less we'll have to look into the touchscreen bug thet comes with the gps on.
 
  • Like
Reactions: croquis and oakes

Elibl

Senior Member
Dec 20, 2011
686
448
My guide for manipulating sdcard mbr didn't work for you? I thought it should be working with every card?!

I noticed a chmod in a *.rc file that sets permissions of sdcard path to 0000 but it's generated during compilation... i wanted to have a look at later so it may gives you an idea, maybe it's normal don't know sounds strange.

EDIT: didn't load module again:eek: but i got also no boot animation and sim request... -> deleted my whole working directory and also the corrupted git from .repo/sync/ cause i had a fetch error after repo sync -f
I hope i'll get the gps to work again soon

Gesendet von meinem U8160 mit Tapatalk
 
Last edited:

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
My guide for manipulating sdcard mbr didn't work for you? I thought it should be working with every card?!

I noticed a chmod in a *.rc file that sets permissions of sdcard path to 0000 but it's generated during compilation... i wanted to have a look at later so it may gives you an idea, maybe it's normal don't know sounds strange.

EDIT: didn't load module again:eek: but i got also no boot animation and sim request... -> deleted my whole working directory and also the corrupted git from .repo/sync/ cause i had a fetch error after repo sync -f
I hope i'll get the gps to work again soon

Gesendet von meinem U8160 mit Tapatalk

I tried the guide, but I couldn't get it to work on the only windows computer I have access to.

I have found out something that makes it boot with the SD card in. In the file init.rc in the boot.img file there is a line "exec /system/bin/sysinit". If I remove this it the phone boots with the sdcard in.

Doing this causes lots of other trouble, but it means I have at least found something close to the problem.
 

Elibl

Senior Member
Dec 20, 2011
686
448
Did you try to modify the chmod "sdcard mount path" 0000 line in init.rc?

Gesendet von meinem U8160 mit Tapatalk
 

djkarl

Senior Member
Apr 3, 2009
75
10
are flashable zips going to be updated for this? i'm X1 is gathering dust and it would be nice to bring it back in to daily use
 

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
We are working on it, but don't hold you breath ;)

I am using the X1 with the latest kernel and the sundream rom from last year, and I find it good enough for daily use. So if you want to use your X1 just flash jerrytan's icedroid zip's and the latest boot.img file from the kernel thread.

When we have some actual functionality improvements we will post zips.
 

djkarl

Senior Member
Apr 3, 2009
75
10
cheers guys i will give that ago.

you work on trying to breath fresh air in to our ageing device is appreciated
 

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
Did you try to modify the chmod "sdcard mount path" 0000 line in init.rc?

Gesendet von meinem U8160 mit Tapatalk

I tried you SD-card fix guide again and made it work with my sdcard :)

I also figured out where exactly it hangs with a card with a proper mbr:

When the wireless lan driver is initialized in the 02kovsky script in /system/etc/init.d the kernel also initializes the sdcard driver. This is because both wlan and sdcard are attached to the sdio hardware in the msm7201 chip. (sdcard is connected to channel 0 and wlan to channel 1). For some reason sdcard init hangs when the driver is able to read the partition table.

I won't do anything else with this now, and move on to other things, but if I happen upon a solution to this I will of cause publish it.
 

Elibl

Senior Member
Dec 20, 2011
686
448
I tried you SD-card fix guide again and made it work with my sdcard :)

I also figured out where exactly it hangs with a card with a proper mbr:

When the wireless lan driver is initialized in the 02kovsky script in /system/etc/init.d the kernel also initializes the sdcard driver. This is because both wlan and sdcard are attached to the sdio hardware in the msm7201 chip. (sdcard is connected to channel 0 and wlan to channel 1). For some reason sdcard init hangs when the driver is able to read the partition table.

I won't do anything else with this now, and move on to other things, but if I happen upon a solution to this I will of cause publish it.

yes we can focus on this later as it's temporarily fixed... there are more important things to do...

nice to hear that my guide finally worked for you...

i now have an independent ubuntu on my laptop alongside windows (repo sync in about 10s) :D

but the issue with leo-gps is still the same...

i read somewhere that alex wrote the baseband detection in gps-rpc to avoid modem crashes in cause of gps init while baseband isn't yet available. maybe this happens to me... applying alex's amss detection to leo-gps won't be a problem but we already know that baseband isn't yet available at gps init so this wouldn't help anything

i won't have much time the next two weeks cause of some exams at my studying place and i need to finish my laptop configuration... i'll be back afterwards...

should we first focus on detules xda-gps?????

btw some ot i'll own a new (broken) daily phone today. my huawei ideos u8160 will be replaced by a hd2/leo :laugh: how much more speed :laugh: maybe need to replace digitizer before.... hopefully i'll learn enough for the exams!!!!

all the best
 
  • Like
Reactions: oakes

CoolrunnerII

Senior Member
Jan 27, 2012
132
226
Trondheim
I figured out why the camera isn't working as it used to.

In the old roms the HTC dream propriatary libcamera.so library is used. If I cant find another solution fast I'll just add this file to the tree and we'll have to keep on using it.
 
  • Like
Reactions: taninos and kuskro

Top Liked Posts

  • There are no posts matching your filters.
  • 20
    We now have a complete working build environment for CyanogenMod7.2 for the X1/kovsky.

    Feature list:

    * Phone works (SMS, calls, HSDPA, EDGE et al)
    * Wifi works with open access points and WPA encryption. Signal meter is now working. (WEP does not work, you may get unexpected disconnects, and have to reconnect sometimes)
    * Bluetooth works, but turning bluetooth off will drain your battery very fast. (You need to reboot stop this)
    * GPS works with AGPS, but may crash your phone.
    * Camera can take pictures. (No autofocus)
    * Automatic adjustment of screen backlight
    * About 40 hours battery life.

    Issues we are working on:

    * Fix bluetooth turn off bug
    * Power usage

    Issues that will probably not be resolved:

    * FM-radio


    This is based on SP3DEV/Alex's earlier work with lots of contributions from the rest of the community.

    The android code is at: https://github.com/CoolrunnerII/
    The kernel code is maintained by Zargloub at: https://gitorious.org/~poussevin/linux-on-qualcomm-s-msm/tp-alex-linux-xperia

    If anyone feel they have the skill and interest to contribute directly to these repos please contact me and I may arrange with write permision for more people.

    Precompiled releses can be found in the second post.

    We have a google code page with a issue tracker.

    Build instructions can be found in the google code wiki
    8
    20121001:

    * First fix for wifi signal meter in wpa_supplicant. Signal level changes with distance from access point. I need feedback about absolute levels. (Should it read higher or lower?)


    cm-7-20121002


    20120920:

    * Small change to wpa_supplicant.conf that may enable WEP access points. I can't test this at the moment, so I would like some feedback on this.

    cm-7-20120920

    recovery:

    * with the same kernel as 20120919
    * with Elibl's fixes for screen roation and font size

    Mediafire: recovery.img

    20120919:

    * gps.kovsky.so with agps support (got some crashes when enabled with wifi)
    * zargloubs latest kernel with my touchscreen fix.
    * working camera

    Mediafire: cm-7-20120919


    20120902:

    * Only change from 20120902 is libcamera.so
    * Camera should work now. Using what I believe is the proprietary HTC Dream libcamera.so that has been included in all the other ROM's

    cm-7-20120902 [broken]


    20120831:

    * First release with working GPS included
    * Standard CM7.2
    * Latest kernel from zargloub (without modules, if you need them you will have to flash a boot.img afterwards)
    * RIL is working as far as I can tell
    * Camera lib seems to be broken, but I haven't tested it all that much
    * KNOWN BUG: When GPS is in use the touchscreen stops working.
    * WARNING; There has been a reported boot loop caused by booting with gps enabled. I think I have fixed this bug, but havent tested the fix. If you end up in a boot loop you will have to enter recovery and flash a zip without gps.

    cm-7-20120831 [broken]





    Initial alfa release: cm-7-20120801
    8
    I finally found out why we don't have any signal level reading for wifi. We are using wpa_supplicant_6 as all the other gingerbreads are using, only we are the only ones using the nl80211 driver in there, and that driver doesn't have any driver_cmd function. All the other ones that have the same ti wifi chipset as us are using some sort of proprietary driver. The ones having broadcom wifi are using the wext driver. driver_cmd handles signal level and power saving modes and stuff like that.

    I'll have a go at implementing the driver_cmd function. When I have the time to figure out what ioctl calls to use I may get the signal level up from the kernel space driver into wpa_supplicant and from there up to the libwifi module and all the way up to android gui. :p

    I'm trying to get hold of another X1 to open up and connect to a logic analyzer. I'll leave the power consumption issues until I do. If I can boot a phone in windows mobile and measure the voltages and current then I may have a better idea what do do to extend out battery life. Also getting the i2c bus transactions would be very helpful.
    7
    I just uploaded Build 20130310.

    This includes:

    * Battery temp readout ( Big applause for Syahmixp for the patch :D )
    * Battery driver i2c cleanup. This may fix the lost keys bug.
    * Backlight value is remembered during suspend. This will make the screen brighter faster if used outdoors.
    * tun.ko is now included in the rom and loaded at boot time.

    You can grab it from the Google code download page
    7
    I have to agree, CyanogenMod 6.2 does fit for the Kovsky. I also edit the build.prop file just to make it match the correct build number. It's a little bit of an OCD thing of mine. It would also be great if this ROM is available on ROM Manager, it would be really user friendly.

    By the way, the WiFi meter seems to be quite exact so far. And the download link in the second post leads to an older build.

    Sent from my Kovsky using xda app-developers app

    I fixed the download link now. Sorry for that :p

    I don't think I'll start work on a CM 6.2 build. If anyone want to try it feel free, I can give out write permisions for the repos so you can start you own branch.

    I think the wifi meter was a bit low, so I'll add some on it in the next release. I may be able to include power save modes also, I'm looking into it.

    The good news is that I just won an ebay auction on a X1 with a broken headset plug. When I get it I'll start some proper power supply measurement and open it up and see if I can figure out what i2c commands are sendt under windows mobile.