797 replies to this topic

[sej7278]

tilal6991, on 13 July 2012 - 02:26 PM, said:

When compiling with v6 you only need one hack along with my patch on Gerrit. I'll edit this post in a few minutes with it.

i saw this patch of yours on gerrit (full of whitespace errors argh!)   what's the other one?

none of our blade, blade2 or skate device trees look much different (mostly cm9 plus the ALOG fixes etc.) so not sure what's my problem.

Edited by sej7278, 13 July 2012 - 04:48 PM.

[tilal6991]

That patch and add #undef __ARM_HAVE_LDREX_STREX above line 138 of bionic/libc/private/bionic_atomic_arm.h

[sej7278]

tilal6991, on 13 July 2012 - 04:52 PM, said:

That patch and add #undef __ARM_HAVE_LDREX_STREX above line 138 of bionic/libc/private/bionic_atomic_arm.h

ah that seems to be doing it, where do you find this info?

stuck at audioinwrapper.cpp now

external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp: In function 'int AudioSetVolume(int, int)':
external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp:155:61: error: invalid conversion from 'int' to 'audio_stream_type_t' [-fpermissive]
frameworks/av/include/media/AudioSystem.h:58:21: error:   initializing argument 1 of 'static android::status_t android::AudioSystem::setStreamVolume(audio_stream_type_t, float, audio_io_handle_t)' [-fpermissive]
external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp: In function 'int AudioGetVolume(int)':
external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp:165:50: error: invalid conversion from 'int' to 'audio_stream_type_t' [-fpermissive]
frameworks/av/include/media/AudioSystem.h:60:21: error:   initializing argument 1 of 'static android::status_t android::AudioSystem::getStreamVolume(audio_stream_type_t, float*, audio_io_handle_t)' [-fpermissive]
make: *** [out/target/product/blade/obj/SHARED_LIBRARIES/libSR_AudioIn_intermediates/audioinwrapper.o] Error 1

your dalvik patch with the whitespace fixed (no-op's with trailing spaces):  dalvik.patch.zip   96.63K   19 downloads

Edited by sej7278, 13 July 2012 - 05:09 PM.

[tilal6991]

sej7278, on 13 July 2012 - 05:02 PM, said:

ah that seems to be doing it, where do you find this info?

stuck at audioinwrapper.cpp now

external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp: In function 'int AudioSetVolume(int, int)':
external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp:155:61: error: invalid conversion from 'int' to 'audio_stream_type_t' [-fpermissive]
frameworks/av/include/media/AudioSystem.h:58:21: error:   initializing argument 1 of 'static android::status_t android::AudioSystem::setStreamVolume(audio_stream_type_t, float, audio_io_handle_t)' [-fpermissive]
external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp: In function 'int AudioGetVolume(int)':
external/srec/audio/AudioIn/UNIX/src/audioinwrapper.cpp:165:50: error: invalid conversion from 'int' to 'audio_stream_type_t' [-fpermissive]
frameworks/av/include/media/AudioSystem.h:60:21: error:   initializing argument 1 of 'static android::status_t android::AudioSystem::getStreamVolume(audio_stream_type_t, float*, audio_io_handle_t)' [-fpermissive]
make: *** [out/target/product/blade/obj/SHARED_LIBRARIES/libSR_AudioIn_intermediates/audioinwrapper.o] Error 1

your dalvik patch with the whitespace fixed (no-op's with trailing spaces): dalvik.patch.zip

I make it up lol .

libSRaudio needs to be disabled until our audio drivers support it.

As for the whitespace - I didn't produce the asm output. It was generated by python script which is the cause of the whitespaces.

[KonstaT]

tilal6991, on 13 July 2012 - 04:52 PM, said:

That patch and add #undef __ARM_HAVE_LDREX_STREX above line 138 of bionic/libc/private/bionic_atomic_arm.h

I ended up removing the insulting half of the ifdeffage, but this might be more elegant.

tilal6991, on 13 July 2012 - 05:24 PM, said:

I make it up lol .

libSRaudio needs to be disabled until our audio drivers support it.

Well, the compiler tells you exactly where the problem is so you don't have to make it all up from the top of your head.

There's bunch of stuff that needs to be disabled for now. libSRaudio, libaac/libFraunhoferAAC -> AACenc bits in libstagefright etc. Have you found any solutions to these?. Doesn't seem to affect too much though since audio is one the very few things I have working at the moment.

[sej7278]

KonstaT, on 13 July 2012 - 06:21 PM, said:

There's bunch of stuff that needs to be disabled for now. libSRaudio, libaac/libFraunhoferAAC -> AACenc bits in libstagefright etc. Have you found any solutions to these?. Doesn't seem to affect too much though since audio is one the very few things I have working at the moment.

lol, yeah there's a massive amount that doesn't seem to be anywhere near working - add iptables to that list. i'm giving up for now as most of the bugs (well unported code i guess) are outside our device/vendor trees.

are cyanogenmod actually starting from scratch with cm10 rather than basing it on cm9, as it seems like there would be a fair bit of work involved in porting all the cm9 fixes into the jellybean aosp code, possibly more than porting the aosp code into cm9 to make cm10.

[Hayden3430]

I feel so privileged to be a Blade user knowing that we have developers like you guys, Thanks.

[ufizo]

sej7278, on 13 July 2012 - 09:15 PM, said:

are cyanogenmod actually starting from scratch with cm10 rather than basing it on cm9, as it seems like there would be a fair bit of work involved in porting all the cm9 fixes into the jellybean aosp code, possibly more than porting the aosp code into cm9 to make cm10.

cyanogenmod said:

While Jelly Bean won't require a major rewrite of all the CM features, the source has been re-arranged and re-factored enough that a simple merge isn't possible. So we will proceed with manual merging (this was to be expected really).

The plan of attack will be to cherry-pick everything from CM9 and fix the merge conflicts and relocations. It’s tedious but the only way.

https://plus.google....sts/BMQiaHhSMte

[TheWhisp]

Hi!

How do I disable libSRaudio?
And I'm getting bunch of errors regarding export_includes and always with librpc_intermediates

make: *** No rule to make target `out/target/product/tass/obj/SHARED_LIBRARIES/librpc_intermediates/export_includes', needed by `out/target/product/tass/obj/STATIC_LIBRARIES/libloc_api-rpc_intermediates/import_includes'.  Stop.

Any ideas?

[emajar]

Hayden3430, on 14 July 2012 - 03:14 AM, said:

I feel so privileged to be a Blade user knowing that we have developers like you guys, Thanks.

+1

Lovely work going on here!

[sej7278]

TheWhisp, on 14 July 2012 - 10:41 AM, said:

Hi!

How do I disable libSRaudio?
And I'm getting bunch of errors regarding export_includes and always with librpc_intermediates

make: *** No rule to make target `out/target/product/tass/obj/SHARED_LIBRARIES/librpc_intermediates/export_includes', needed by `out/target/product/tass/obj/STATIC_LIBRARIES/libloc_api-rpc_intermediates/import_includes'.  Stop.

Any ideas?

i guess just commenting out the "include $(BUILD_SHARED_LIBRARY)" line in external/srec/audio/AudioIn/UNIX/src/Android.mk ?

i still can't get anywhere near a finished compile, even with disabling half the fscking makefiles i die after 5mins lol, my main problem seems to be AudioPolicyManager.h in the blade device tree

Edited by sej7278, 15 July 2012 - 09:30 AM.

[KonstaT]

sej7278, on 15 July 2012 - 09:30 AM, said:

i still can't get anywhere near a finished compile, even with disabling half the fscking makefiles i die after 5mins lol, my main problem seems to be AudioPolicyManager.h in the blade device tree

Most of the device specific stuff (audio, sensors, lights) is fixed in my github, only camera to go (probably won't be as easy).

Edit. libcamera compiles now too, but I run into the most unexpected issue. Makefiles and AndroidManifest.xml for the Camera.apk are missing for Jelly Bean - even from AOSP. WTF Google?

Edit2. Apparently camera is now somehow merged into gallery, still can't find it though.

[sej7278]

KonstaT, on 15 July 2012 - 03:10 PM, said:

Most of the device specific stuff (audio, sensors, lights) is fixed in my github, only camera to go (probably won't be as easy).

i've poached a few of your patches, and seem to be getting further (still stuck with stagefright/aac problems) what helped a lot was going through the main code and fixing that before worrying about device-specifics. i can't understand why they didn't just fix all the ALOG{E/V/D/I/W} stuff in one commit.

instructions so far:

apply dalvik patch

add #undef __ARM_HAVE_LDREX_STREX before #ifdef __ARM_HAVE_LDREX_STREX in bionic/libc/private/bionic_atomic_arm.h

comment out  external/aac/Android.mk, external/srec/audio/AudioIn/UNIX/src/Android.mk and hardware/msm7k/libstagefrighthw/Android.mk

delete the 6 lines between the braces in hardware/qcom/display/liboverlay/mdpWrapper.h
inline void dump(const char* const s, const msm_rotator_data_info& rot) {....}

[Felsch]

*follows guide on first page*

-downloading, all check
-almost everything else, all check.

Java version in my fedora-vm, 1.7.0...<- ... badword... now I has to get 1.6.0 which seems a pain to install... well let's try ubuntu(less of a pain to install jdk 1.6.0), and cry for all bandwidth I used.

Nice guide sej, just that small little problem ><(which I can solve ).

One thing I do have a question of: on the android site(http://source.androi...itializing.html) it states that it needs a 64-bit cpu enviroment, is that true or can it get compiled on 32-bit(as I made the fedora-vm 32bit :/)  ?

Edited by Felsch, 15 July 2012 - 07:58 PM.

[Ganster]

Not important, I normally compiled CM9 on Ubuntu x86.

[C3C0]

@sej
libstagefright: comment out lib for aac encoder in framework/av/media/libstagefright/android.mk
For msm rotator just comment out body of inline dump function in header file that's causing problem.
These are just for logging purposes.

[sej7278]

C3C0, on 16 July 2012 - 03:55 AM, said:

@sej
libstagefright: comment out lib for aac encoder in framework/av/media/libstagefright/android.mk
For msm rotator just comment out body of inline dump function in header file that's causing problem.
These are just for logging purposes.

thanks for the aac fix, seems to be compiling away now.

edit: still getting aac problems.

Edited by sej7278, 16 July 2012 - 08:29 AM.

[Ganster]

Why we can't use libstagefright(all or partially) from GB tree? As I see, problems with unimplemented functions are in codec initialization functions. May be port this functions from GB is easier than implementing missed functions to QC_OMX(it's written on C++, and it hard to reverse-engineering)? Yes, it's hard, but possible...I think.

[C3C0]

sej7278, on 16 July 2012 - 07:52 AM, said:

thanks for the aac fix, seems to be compiling away now.

edit: still getting aac problems.

Looking at my changes now and this is what I did:
1) removed libstagefright_aacenc from frameworks/av/media/libstagefright/Android.mk
2) deleted codecs/aacdec/Android.mk and codecs/aacenc/Android.mk
3) deleted external/aac

Edited by C3C0, 16 July 2012 - 11:59 AM.

[sej7278]

C3C0, on 16 July 2012 - 11:58 AM, said:

Looking at my changes now and this is what I did:
1) removed libstagefright_aacenc from frameworks/av/media/libstagefright/Android.mk
2) deleted codecs/aacdec/Android.mk and codecs/aacenc/Android.mk
3) deleted external/aac

that seems to have fixed it, had to pull in burstlams (well, teamics's) gcc 4.5 kernel fixes, not sure why i never saw that on cm9 though.

does github have an easy web-interface way of cherry-picking commits into your repo, rather than having to fork, branch, checkout and merge via the cli, there must be an easy way to do it and keep history; i think i'm over-complicating it.