FORUMS
Remove All Ads from XDA

[WIP] Building a custom ROM for the LeapFrog Epic (CM/LOS)

750 posts
Thanks Meter: 115
 
Post Reply Email Thread
27th April 2019, 07:53 AM |#21  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
Quote:
Originally Posted by pix106

i did not use this repo.
i could build a bootable and usable cm12.1 without any patch.
thp1997 joined afterwards and added patched repos, but they were not strictly needed : they allowed WPS wifi pairing, headset icon, usb mode selection when usb is not plugged.

Hmm, maybe it's due to the blobs I have that called for those functions. I know I mentioned this earlier, but I'd probably be better off if LeapFrog somehow issued a Lollipop update with newer blobs and such -- it's 2019 and yet the latest patch they issued was only a minor update to the Kitkat ROM they're still using. :/
 
 
27th April 2019, 10:00 AM |#22  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
Looks like I made major progress lol:


I had to use this patch to make SurfaceFlinger happy, but given how others didn't need to do so on their MT8127s, it's a bit of an ugly, inelegant and atrocious method imo. But at least the results seem worth the crap. Now all that's left is to get the rest of the OS working as it should.

Maybe this is the reason why the tablet isn't able to load up any apps, other than ART doing its thing and generating .dex files of stuff in the background:

Code:
W/Zygote  (  697): Storage environment undefined; unable to provide external storage
W/Zygote  (  697): Failed to mount emulated storage: Invalid argument
E/Zygote  (  697): Cannot continue without emulated storage
F/art     (  697): art/runtime/jni_internal.cc:769] JNI FatalError called: RuntimeAbort
Latest logcat's attached which should give us an idea on what needs to be done. Still too early to celebrate but at least this is some good news.
Attached Files
File Type: zip logcat_UYT2_20.zip - [Click for QR Code] (62.3 KB, 4 views)
27th April 2019, 10:26 AM |#23  
Senior Member
Thanks Meter: 506
 
More
Quote:
Originally Posted by blakegriplingph

Looks like I made major progress lol:


I had to use this patch to make SurfaceFlinger happy, but given how others didn't need to do so on their MT8127s, it's a bit of an ugly, inelegant and atrocious method imo. But at least the results seem worth the crap. Now all that's left is to get the rest of the OS working as it should.

Maybe this is the reason why the tablet isn't able to load up any apps, other than ART doing its thing and generating .dex files of stuff in the background:



Latest logcat's attached which should give us an idea on what needs to be done. Still too early to celebrate but at least this is some good news.

Nice Maybe you'll find info in /data/tombstones, and checking which services are running (there is a service command to do so, I don't remember exactly which one it is exactly).
The Following User Says Thank You to pix106 For This Useful Post: [ View ] Gift pix106 Ad-Free
27th April 2019, 10:43 AM |#24  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
Quote:
Originally Posted by pix106

Nice Maybe you'll find info in /data/tombstones, and checking which services are running (there is a service command to do so, I don't remember exactly which one it is exactly).

'Tis beautiful man, I felt relieved when I got the graphics to show up for the first time and it turned out to be a series of silly mistakes that I should've addressed earlier. For one I dunno what got to me when I put the EGL shim code in libxlog.

I've checked out tombstones but that directory appears empty unlike before when I was still struggling with sw_sync and the framebuffer. Will take a look into it again though.
28th April 2019, 03:15 AM |#25  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
I managed to make quite some progress with the CM12.1 port:


...though while the graphics stack seems to work fine, no apps show up for some reason -- neither the first-time wizard nor SystemUI show up. Any hints on what to do with this? Also, wakelock seems a little broken especially if I leave the tablet asleep for a while, but I can't confirm this until I get everything up and running.

@Matty1993 I think you might have a clue with this considering you got later CM/LOS versions to boot on 3.4.67, like selinux and whatnot. I did quite a few corrections to the init scripts as per your advice though I might have done something wrong along the way.
Attached Files
File Type: zip logcat_UYT2_21.zip - [Click for QR Code] (54.0 KB, 4 views)
28th April 2019, 08:12 AM |#26  
Senior Member
Thanks Meter: 506
 
More
Quote:
Originally Posted by blakegriplingph

I managed to make quite some progress with the CM12.1 port:

Mind the init.mt8127.usb.rc / init.mt8127usb.rc, some references have the dot, some don't.
You'll have it fully running soon

Edit : i'm surprised you have tu put any selinux rules, it may not be running as perissive.
I don't have to set cmdline in BOOTIMG args, only in kernel args, to set ut permissive.
What does getenforce cd tells about it when running ?
The Following User Says Thank You to pix106 For This Useful Post: [ View ] Gift pix106 Ad-Free
28th April 2019, 10:17 AM |#27  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
Quote:
Originally Posted by pix106

Mind the init.mt8127.usb.rc / init.mt8127usb.rc, some references have the dot, some don't.
You'll have it fully running soon

Edit : i'm surprised you have tu put any selinux rules, it may not be running as perissive.
I don't have to set cmdline in BOOTIMG args, only in kernel args, to set ut permissive.
What does getenforce cd tells about it when running ?

I checked the SELinux status via Terminal and it does report as "permissive". Also, what do you make out of the ANR warnings? I am not honestly sure how to approach this unlike what happened before when even the graphics stack itself does not boot.
28th April 2019, 01:16 PM |#28  
Senior Member
Thanks Meter: 506
 
More
Quote:
Originally Posted by blakegriplingph

I checked the SELinux status via Terminal and it does report as "permissive". Also, what do you make out of the ANR warnings? I am not honestly sure how to approach this unlike what happened before when even the graphics stack itself does not boot.

No clue about anr, i did not have this.


Don't forget te change back init.8127.usb.rc to init.mt8127usb.rc in minit.mt8127.rc, as this is the file name and it is referenced as such in full mk

You can check which services are running, maybe there is one crashing (audio, codec, that's what i remember 2.5 years later )

In the beginning of the build process, you can see some blobs can be built from source instead, "xxxxxxxx.so overrides rule to make" or such message.
Not every one of these libs must be built from source, but i did so for a few in my test builds, you can check the cm-12.1-test branch of my device tree. Right now i'm thinking about audio libs, see what is commented as "# lessblobs - cm-12.1-test" in proprietary-files.txt
The Following User Says Thank You to pix106 For This Useful Post: [ View ] Gift pix106 Ad-Free
28th April 2019, 02:49 PM |#29  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
Quote:
Originally Posted by pix106

No clue about anr, i did not have this.


Don't forget te change back init.8127.usb.rc to init.mt8127usb.rc in minit.mt8127.rc, as this is the file name and it is referenced as such in full mk

You can check which services are running, maybe there is one crashing (audio, codec, that's what i remember 2.5 years later )

In the beginning of the build process, you can see some blobs can be built from source instead, "xxxxxxxx.so overrides rule to make" or such message.
Not every one of these libs must be built from source, but i did so for a few in my test builds, you can check the cm-12.1-test branch of my device tree. Right now i'm thinking about audio libs, see what is commented as "# lessblobs - cm-12.1-test" in proprietary-files.txt

Decided to go with a make clean + clear ccache as said cache may have messed up the output royally. I was thinking of building at least HWC and a few others from a board support package @Stricted was raving about especially as I may resort to porting a 3.10 kernel if needed. I know the stock 3.4.67 kernel is doable (as Matty1993 and a few others did manage to get Lineage on old kernels) but there's only so much you could do with a prebuilt.
28th April 2019, 02:57 PM |#30  
Senior Member
Thanks Meter: 506
 
More
In your logcat, i noticed libWVStreamControlAPI_L1.so is missing
For /dev/stpwmt permission problem, you can try to set it in ueventd.mt8127.rc
29th April 2019, 01:22 AM |#31  
blakegriplingph's Avatar
OP Senior Member
Thanks Meter: 115
 
Donate to Me
More
Quote:
Originally Posted by pix106

In your logcat, i noticed libWVStreamControlAPI_L1.so is missing
For /dev/stpwmt permission problem, you can try to set it in ueventd.mt8127.rc

Isn't that for Widevine i.e. DRM-protected video?

And idk about /dev/stpwmt; it seems as though init isn't actually doing a mknod for it on its own. I had to set that one up manually though adb shell just to get rid of those messages. It seems as though it's telling me "Get a new kernel!" or something, but idk. The only other thing I could think of is init.ssd.rc as that's what made Zygote happy for the most part. Along with this line:

Code:
E/SELinux (  124): avc:  denied  { add } for service=NvRAMAgent scontext=u:r:init:s0 tcontext=u:object_r:nvram_agent_service:s0 tclass=service_manager
Attached Files
File Type: zip logcat_UYT2_23.zip - [Click for QR Code] (63.2 KB, 2 views)
File Type: zip kmesg_UYT2.zip - [Click for QR Code] (256.9 KB, 2 views)
File Type: zip logcat_UYT2_22.zip - [Click for QR Code] (185.7 KB, 2 views)
Post Reply Subscribe to Thread

Tags
leapfrog, lineageos, mediatek, mt8127, mtk

Guest Quick Reply (no urls or BBcode)
Message:
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes