[DEV] AOSP Dev for Zte Blade

[Guest ayziaa]

Read above, the black screen is a problem for some people if not all at the moment

I know!

My Bad, I didn't pay enought attention.

[Guest furrabbit.nh]

@anyone one with the knowlege.

As for the screen issues would it be worth looking into the desire ics build, the desire uses synaptics clearpad 2000 and adreno 200 gpu, The blade also use's both of these, and in the latest desire built they have multitouch...

Edited November 24, 2011 by furrabbit.nh

[Guest t0mm13b]

@anyone one with the knowlege.

As for the screen issues would it be worth looking into the desire ics build, the desire uses synaptics clearpad 2000 and adreno 200 gpu, The blade also use's both of these, and in the latest desire built they have multitouch...

Hmmm.... where's the source of that info/linky to it, presumably its over on xda? :)

[Guest fp30]

maybe this will help for the EGL error thing?

From XDA galaxy s forum:

Originally Posted by

koaaN

I tried the honeycomb libs to, i got them all to load but then it crashes (libEGL : eglSetSwapRectangleANDROID:1460 error 3008 (EGL_BAD_DISPLAY) )

You need the hwcomposer, liboverlay, and you may need generic egl libs in system/lib/

I used the EGL libs for the system from the SDK, they seem most generic and working.

The bad display is usually that it cannot properly detect your display size. It is fixable, and a lot better than just giving up completely. I will look into it when I get home and find out how I got past the same error.

Edited November 24, 2011 by fp30

[Guest t0mm13b]

maybe this will help for the EGL error thing?

From XDA galaxy s forum:

Originally Posted by

koaaN

I tried the honeycomb libs to, i got them all to load but then it crashes (libEGL : eglSetSwapRectangleANDROID:1460 error 3008 (EGL_BAD_DISPLAY) )

You need the hwcomposer, liboverlay, and you may need generic egl libs in system/lib/

I used the EGL libs for the system from the SDK, they seem most generic and working.

The bad display is usually that it cannot properly detect your display size. It is fixable, and a lot better than just giving up completely. I will look into it when I get home and find out how I got past the same error.

The gotcha is this - the liboverlay found in the blade's device tree in the cm7 repo clashes with ICS's own version, somewhat... somewhere the compiler refused to chew up the source and spit out binaries in that arena so yeah, I guess that could contribute somewhat to the "black screen" - funny thing though is the way the screen flashes off after a period of inactivity - just like in CM7's version...

[Guest furrabbit.nh]

Hmmm.... where's the source of that info/linky to it, presumably its over on xda? :)

Hold on ill find link.

EDIT: Sorry it wasnt HTC Desire it was N1 But after a quick google it also uses adreno 200 and synaptics clearpad 2000 :) n1 and desire are pretty much same phone

Linky

Edited November 24, 2011 by furrabbit.nh

[Guest AndreiKu]

Prompt what is my mistake?

fastboot-windows.exe flash system ics4blade_system.img sending 'system' (150253 KB)... FAILED (remote: data too large)

Edited November 24, 2011 by AndreiKu

[Guest t0mm13b]

Well, just to let you know, my dropbox account has been suspended ... :lol:

got a very nice email from them to inform the bandwidth has been exceeded... so heads up, download from a mirror in the previous post by Matty-P :D

Ho hum... fair enough... :)

[Guest t0mm13b]

Prompt what is my mistake?

fastboot-windows.exe flash system ics4blade_system.img sending 'system' (150253 KB)... FAILED (remote: data too large)

There's a restriction in the fastboot in that its max is 120Mb, try the cwm update... :)

[Guest AndreiKu]

It turned out to install, but I got a black screen. If you press the button to see the effect of the old TV.

Sorry for my English, I use google translate.

[Guest fp30]

another thought:

I/blade.gralloc( 96): refresh rate = 0.14 Hz

this seems incredibly low refresh rate. is this correct?

[Guest unrandomsam]

https://dl.dropbox.com/0/view/eletp1gju7bqcyf/ics4blade_system.img - direct link still seems to be live.

[Guest KonstaT]

I've made some progress. I finally got my own build to boot all the way into Android. It was either badly patched libaudio or webrtc that caused my previous problem, but now I just removed them both from by build. It still boots into a black screen. That blade's egl driver must be a mess. All I have in my logcat is errors with Adreno200-EGL, SurfaceFlinger and gralloc. Now I'm using the EGL folder from the Hero port. Screen was flipped but was it is fixed with the patched libsurfaceflinger. It's starting to get quite hacky...

There it now just sits in ICS lockscreen saying 'Unfortunately, the process com.android.phone has stopped'. I think all the hardware keys work. Volume bar appears under that error message when keys are pressed. Also power menu appears when I long press the power key. I don't have CRT animation anymore btw.

But the actual progress part took some kernel modification. I've got touchscreen responding to touch one time - and that means just once. It's not even a pointer or anything. I always have to reboot to reproduce it. When I touch the OK button under the above error message, it turns blue. It doesn't show up in logcat for some reason(?), but I can reproduce it every time. It doesn't matter how long it's been there whether screen on or off, but it always registers the first touch.

I built a new kernel from Tom G's github with the latest changes and I applied this to /drivers/input/touchscreen/synaptics_original.c. I'm not sure which one did it, but there is an idea for someone to work on.

Edited November 24, 2011 by KonstaT

[Guest ChipZ16]

I've made some progress. I finally got my own build to boot all the way into Android. It was either badly patched libaudio or webrtc that caused my previous problem, but now I just removed them both from by build. It still boots into a black screen. That blade's egl driver must be a mess. All I have in my logcat is errors with Adreno200-EGL, SurfaceFlinger and gralloc. Now I'm using the EGL folder from the Hero port. Screen was flipped but was it is fixed with the patched libsurfaceflinger. It's starting to get quite hacky...

There it now just sits in ICS lockscreen saying 'Unfortunately, the process com.android.phone has stopped'. I think all the hardware keys work. Volume bar appears under that error message when keys are pressed. Also power menu appears when I long press the power key. I don't have CRT animation anymore btw.

But the actual progress part took some kernel modification. I've got touchscreen responding to touch one time - and that means just once. It's not even a pointer or anything. I always have to reboot to reproduce it. When I touch the OK button under the above error message, it turns blue. It doesn't show up in logcat for some reason(?), but I can reproduce it every time. It doesn't matter how long it's been there whether screen on or off, but it always registers the first touch.

I built a new kernel from Tom G's github with the latest changes and I applied this to /drivers/input/touchscreen/synaptics_original.c. I'm not sure which one did it, but there is an idea for someone to work on.

I would like to thank the Internet for these people. How boring my Blade would be without them.

Looking forward to a working ROM!

[Guest kornel88dg]

Screen not working! Is black. backlight only seen.

My screen TFT

[Guest t0mm13b]

I've made some progress. I finally got my own build to boot all the way into Android. It was either badly patched libaudio or webrtc that caused my previous problem, but now I just removed them both from by build. It still boots into a black screen. That blade's egl driver must be a mess. All I have in my logcat is errors with Adreno200-EGL, SurfaceFlinger and gralloc. Now I'm using the EGL folder from the Hero port. Screen was flipped but was it is fixed with the patched libsurfaceflinger. It's starting to get quite hacky...

There it now just sits in ICS lockscreen saying 'Unfortunately, the process com.android.phone has stopped'. I think all the hardware keys work. Volume bar appears under that error message when keys are pressed. Also power menu appears when I long press the power key. I don't have CRT animation anymore btw.

But the actual progress part took some kernel modification. I've got touchscreen responding to touch one time - and that means just once. It's not even a pointer or anything. I always have to reboot to reproduce it. When I touch the OK button under the above error message, it turns blue. It doesn't show up in logcat for some reason(?), but I can reproduce it every time. It doesn't matter how long it's been there whether screen on or off, but it always registers the first touch.

I built a new kernel from Tom G's github with the latest changes and I applied this to /drivers/input/touchscreen/synaptics_original.c. I'm not sure which one did it, but there is an idea for someone to work on.

Well done :D

You're on the road... I just sorta cba to put the netbook through another 31 hours of compile time :D :lol:

Okie...

Can the gralloc's refresh rate as pointed out by fp30 be adjusted by tweaking a external property file somewhere? egl.cfg perhaps?

I did see it mentioned about an IDC file (no idea if that would help) for the touchscreen?

By the way...

For the surface flinger,

I have this in frameworks/base/services/surfaceflinger/SurfaceFlinger.cpp

void GraphicPlane::setDisplayHardware(DisplayHardware *hw)

{

    mHw = hw;


    // initialize the display orientation transform.

    // it's a constant that should come from the display driver.

    int displayOrientation = ISurfaceComposer::eOrientationDefault;

    char property[PROPERTY_VALUE_MAX];

    if (property_get("ro.sf.hwrotation", property, NULL) > 0) {

        //displayOrientation

        switch (atoi(property)) {

        case 90:

            displayOrientation = ISurfaceComposer::eOrientation90;

            break;

#ifdef HAS_FLIPPED_SCREEN // <--!!!!

                case 180:

                        displayOrientation = ISurfaceComposer::eOrientation180;

                        break;

#endif

        case 270:

            displayOrientation = ISurfaceComposer::eOrientation270;

            break;

        }

    }

// Snip ....

And for that Android.mk in that same aforementioned path above for surfaceflinger:

ifeq ($(BOARD_HAS_FLIPPED_SCREEN),true)

        LOCAL_CFLAGS += -DHAS_FLIPPED_SCREEN

endif

That ensures the screen becomes the right way round... :)

The webrtc can still be compiled in - no problemo! xD

As for the touchscreen, there's two or three of them compiled in, so maybe you need to add the hack to the other drivers, cypress_i2c_rmi.c, synaptics_i2c_rmi.c and probably another one... the synaptics_original.c (not sure if that's actually being referenced in the make?) IIRC, there was FocalTech compiled in as well...

But if you look in the kernel's drivers/input/touchscreen/Makefile, there's references to mXT224_atmel which is perhaps also compiled in....

Edited November 24, 2011 by t0mm13b

[Guest fp30]

I've made some progress. I finally got my own build to boot all the way into Android. It was either badly patched libaudio or webrtc that caused my previous problem, but now I just removed them both from by build. It still boots into a black screen. That blade's egl driver must be a mess. All I have in my logcat is errors with Adreno200-EGL, SurfaceFlinger and gralloc. Now I'm using the EGL folder from the Hero port. Screen was flipped but was it is fixed with the patched libsurfaceflinger. It's starting to get quite hacky...

There it now just sits in ICS lockscreen saying 'Unfortunately, the process com.android.phone has stopped'. I think all the hardware keys work. Volume bar appears under that error message when keys are pressed. Also power menu appears when I long press the power key. I don't have CRT animation anymore btw.

But the actual progress part took some kernel modification. I've got touchscreen responding to touch one time - and that means just once. It's not even a pointer or anything. I always have to reboot to reproduce it. When I touch the OK button under the above error message, it turns blue. It doesn't show up in logcat for some reason(?), but I can reproduce it every time. It doesn't matter how long it's been there whether screen on or off, but it always registers the first touch.

I built a new kernel from Tom G's github with the latest changes and I applied this to /drivers/input/touchscreen/synaptics_original.c. I'm not sure which one did it, but there is an idea for someone to work on.

good skills. does ddms not give more information than logcat? maybe that can be of some help?

also i'm just throwing some ideas out now but I noticed something about alpha transparency =1 in the build.prop from t0mm13b's build. Is this necessary maybe this has something to do with the balck screen? (probably not since you got it to work using different egl files but its a thought) :rolleyes:

Edited November 24, 2011 by fp30

[Guest Matty-p]

Well, just to let you know, my dropbox account has been suspended ... :lol:

got a very nice email from them to inform the bandwidth has been exceeded... so heads up, download from a mirror in the previous post by Matty-P :D

Ho hum... fair enough... :)

Yea you get 10gb of bandwith per file in /public so thats ~68 downloads

i have taken the liberty of changing you original download link in post 2 to the mirror as at the moment when they click your old dropbox link it 404's and people therefor can't download it without digging through the thread - no offence taken if you want to take it out/change it in any way OFC

[Guest t0mm13b]

@Matty-P: cheers for that... :)

[Guest Matty-p]

@Matty-P: cheers for that... :)

@Tom your welcome :-) i hope you didn't mind it was just for user experience/clarity as the 5:30 > is generally the time where downloads and views go crazy as everyone gets home from work here in the uk but having said that a surprising amount of visitors are from different countries nowadays.

[Guest lsde]

Hi,

I've loaded the cwm update and the blade is bootlooping at :

F/libc    (  217): Fatal signal 7 (SIGBUS) at 0x00000000 (code=128)

see the full logcat:

https://gist.github.com/1391922

I'm TFT, GEN2

[Guest t0mm13b]

Hi,

I've loaded the cwm update and the blade is bootlooping at :

F/libc    (  217): Fatal signal 7 (SIGBUS) at 0x00000000 (code=128)

see the full logcat: https://gist.github.com/1391922 I'm TFT, GEN2

Weird! Okie, suggest you try wipe the cache/data in cwm, format the system partition via fastboot, and try again, unless something is missing from the cwm update... the error code 128 is intriguing as general consensus of that

#define EKEYREVOKED 128 /* Key has been revoked */

- btw this was taken from the standard linux's errno.h.. so something is amiss there that caused everything to keel over and just get hit by BUS... :) (You did re-TPT it I presume)

As a fyi, I converted the system.img into a cwm update via xda's dsi kitchen script and booted fine after flashing it with Clockwork...

Edited November 24, 2011 by t0mm13b

[Guest Sami Beck]

ICS=Winning

[Guest fp30]

ICS=Winning

:) what exactly do you mean?

[Guest lsde]

Weird! Okie, suggest you try wipe the cache/data in cwm, format the system partition via fastboot, and try again, unless something is missing from the cwm update... the error code 128 is intriguing as general consensus of that

#define EKEYREVOKED 128 /* Key has been revoked */

- btw this was taken from the standard linux's errno.h.. so something is amiss there that caused everything to keel over and just get hit by BUS... :) (You did re-TPT it I presume)

As a fyi, I converted the system.img into a cwm update via xda's dsi kitchen script and booted fine after flashing it with Clockwork...

thanks, /system format did the trick..

but i'm still one of the blackscreen-ers