[DEV] CyanogenMod 10 marvel-ALPHA

[modpunk]

Preamble

This is the development thread for HTC Wildfire S and msm7227 devices. If you have questions about the code of the ROM and want to discuss changes and stuff we have developed feel free to ask here.

I've been working on a CM9 port for the HTC Wildfire S since January 2012 and started with CM10 in August 2012. After 5 month of work this I've released the first ALPHA of CM9. Getting CM10 was esier. Neither CM10 is stable yet nor the HTC Wildfire S (Marvel) port is complete. Most of the features are working but it still needs bugfixing and polishing. We would like to improve the code and get as much as possible upstream if they accept it. There will never be an official CM10 release as ARMv6 is not supported.

I have to thank alquez who ported CM7 for the HTC Wildfire S. His work was a good starting point for CM9 and then CM10. Also thanks to Tobias Droste who was working with me on this ROM. Tobias we finally made it!
I also have to thank WinSuk and s0be working on CM9 and CM10 for the Liberty. We worked together with those guys to get CM9 and CM10 going. We will continue our collaboration in future.

Reporting Bugs

If you want to report bugs please read THIS HOWTO first. Please try to provide logs if you see errors (logcat and dmesg). If you don't know how to do that check google or ask people here to help you. Don't contact me directly.

Enabling logcat Logging
adb shell dmesg
How to use strace?

All bug reports should go to the bugtracker.

If you think you find a bug take a look at the bugtracker first. Check if it already has been reported.

Sources

You can find the source here: http://git.cryptomilk.org/


If you are a developer working on a msm7227 and want to collaborate, contact me directly.

 

[ONeill123]

When it will be released as an alpha?

I think, it must have a good version number

 

[matt5eo]

I know now the reason why there are random force closes. After I open apps they will run as background process. And if I reach the (boarder) (only 140mb ram free) the apps start to crash. The kernel thinks that it is better to force close the apps before going into an to high usage of ram.

I tried to run an auto ram manager but without luck.

Now I try the option in the Developer Settings to allow just three background processes.

Not sure why but I think its a kernel bug.

Gesendet von meinem HTC Wildfire S A510e mit Tapatalk

 

[eoghan2t7]

any idea how to fix this compile error?

target thumb C++: camera.marvel <= device/htc/marvel/libcamera/cameraHal.cpp
target StaticLib: libaudiopolicy_legacy (out/target/product/marvel/obj/STATIC_LIBRARIES/libaudiopolicy_legacy_intermediates/libaudiopolicy_legacy.a)
target thumb C++: gps.marvel <= hardware/qcom/gps/loc_api/libloc_api/loc_eng_xtra.cpp
target thumb C++: gps.marvel <= hardware/qcom/gps/loc_api/libloc_api/loc_eng_ni.cpp
device/htc/marvel/libcamera/cameraHal.cpp:40:27: error: ui/OverlayHtc.h: No such file or directory
target thumb C++: gps.marvel <= hardware/qcom/gps/loc_api/libloc_api/loc_eng_log.cpp
target thumb C++: gps.marvel <= hardware/qcom/gps/loc_api/libloc_api/loc_eng_cfg.cpp
target thumb C: gps.marvel <= hardware/qcom/gps/loc_api/libloc_api/gps.c
target thumb C: libpower <= hardware/libhardware_legacy/power/power.c
target thumb C: librpc <= hardware/msm7k/librpc/xdr.c
target thumb C: librpc <= hardware/msm7k/librpc/rpc.c
device/htc/marvel/libcamera/cameraHal.cpp: In function 'int camera_set_preview_window(camera_device*, preview_stream_ops*)':
device/htc/marvel/libcamera/cameraHal.cpp:693: error: invalid use of incomplete type 'struct android::Overlay'
device/htc/marvel/libcamera/CameraHardwareInterface.h:29: error: forward declaration of 'struct android::Overlay'
target thumb C: librpc <= hardware/msm7k/librpc/svc.c
target thumb C: librpc <= hardware/msm7k/librpc/clnt.c
make: *** [out/target/product/marvel/obj/SHARED_LIBRARIES/camera.marvel_intermediates/cameraHal.o] Error 1
make: *** Waiting for unfinished jobs....
target StaticLib: libc_nomalloc (out/target/product/marvel/obj/STATIC_LIBRARIES/libc_nomalloc_intermediates/libc_nomalloc.a)

 

[hayanno]

I'm not a dev' but I can try to hepl, I'm a student in C and when I get this error : "error: forward declaration of 'struct android::Overlay" it's that I haven't or wrong write the struct...

 

[modpunk]

any idea how to fix this compile error?

Droste just added the NoOpOverlay to our codebase.

 

[eoghan2t7]

Droste just added the NoOpOverlay to our codebase.

Lovely thanks

Sent from my Wildfire S using Tapatalk 2

 

[BigChillin]

Droste just added the NoOpOverlay to our codebase.

I know i'm not a dev and probably shouldn't even ask, but I googled noop overlay and found only one post on it in the entire interwebz. And all it said was that someone had written one and it didn't speed up server speeds. So might someone have the time to explain what it is and what it does?

 

[modpunk]

It is a UI overlay which does nothing. See the marvel device source code ...

 

[modpunk]

If you're not a developer but know how to use a shell and want to help, I will open "Tasks" in the bugtracker. These are things which need to be investigated. You can help me and work on those tasks.

https://code.google.com/p/cm9-wildfire-s/issues/list?can=2&q=label:Type-Task

I will add new ones from time to time.

 

[AllexRo]

Although a complete noob in developing, i'm willing to help as well, if instructed how to do it so, after installing the rom (from your other thread), i tried to help with the "battery current" task.
However, no matter which frequency i try, the command always returns the result 0.
I know i'm typing it correct (and works) since, if plugged, returns value 805. However, unplugged, the result is always 0.

 

[modpunk]

Thanks for the info, good to know. I thought it would show the value it uses. Looks like it is the charging value...

 

[baddboyy]

i tried to get kernel and partial wakelocks to catch what stops deepsleep. didnt used phone 22 min after unplugged from charge.
So these are statistics:

Kernel Wakelocks: (in 22 min)

21m8s : msm_serial_hs_dma
11sec : powermanagerservice
10sec : radio-interface

Partial Wakelocks :

10sec : ActivityManager-Launch
7min : audioout_1


process:
ksmd : 10 sec

alarms:
com.android.phone : wakes up 8 times
com.google.android.gsf: wakes up 4 times
android : wakes up 2 times

and others:
no data connection : 22 min
no or unknown signal : 22 min
awake : 22 min

 

[benjamingwynn]

@Modpunk

Good work, one question: how did you fix the webrtc compile errors and how did you allow both the chrome web engine and the other Javascript one to run together?

Sent from my HTC Wildfire S A510e using xda premium

 

[eoghan2t7]

@Modpunk

Good work, one question: how did you fix the webrtc compile errors and how did you allow both the chrome web engine and the other Javascript one to run together?

Sent from my HTC Wildfire S A510e using xda premium

Cm fixed the webrtc errors and the chrome and v8 engine are added to the boardconfig.mk so they both compile

Sent from my Wildfire S using Tapatalk 2

 

[modpunk]

I think the webrtc patch is still in the patches directory.

 

[modpunk]

i tried to get kernel and partial wakelocks to catch what stops deepsleep. didnt used phone 22 min after unplugged from charge.
So these are statistics:

Kernel Wakelocks: (in 22 min)

21m8s : msm_serial_hs_dma
11sec : powermanagerservice
10sec : radio-interface

Partial Wakelocks :

10sec : ActivityManager-Launch
7min : audioout_1


process:
ksmd : 10 sec

alarms:
com.android.phone : wakes up 8 times
com.google.android.gsf: wakes up 4 times
android : wakes up 2 times

and others:
no data connection : 22 min
no or unknown signal : 22 min
awake : 22 min

I've fixed some problems were rild couldn't access some kernel interfaces but it is still strange. It looks like some information from rild is wrong.

What you see in "others". Something thinks we are always without a signal and search for it, which is wrong.

 

[modpunk]

@baddboyy

If you want to help. Fully charge your phone. Turn radio on and wifi off. Shut down the phone and remove the charger.

Then turn it back on, unlock the sim and don't touch it for 30 min. Then check BetterBatteryStats and tell me what you see.

 

[modpunk]

I've fixed deep sleep. The 'msm_serial_hs_dma' kernel wakelock was the culprit. Our platform doesn't support quick switch-on/off of Bluetooth Module which can be disabled with an overlay.

 

[thelegbra]

i think that fix deserves a compile! thanks modpunk once again