Related
THIS IS A THREAD FOR TENDERLOIN ROM DEVELOPERS TO DISCUSS ISSUES REGARDING BUILDING, CODING, CREATING, DEBUGGING AND STABILIZING A CM10.1/JELLYBEAN 4.2.x ROM FOR THE HP TOUCHPAD. IF YOU AREN'T WORKING ON THIS, MOVE ALONG... NOTHING TO SEE HERE
Seriously, folks. This is an information exchange thread for developers. You're welcome to watch and listen, but PLEASE, please, don't ask user questions here, don't ask for ETA's, don't give huzzahs, and "I want this" stuff. Post that in the relevant ROM threads, the General or the Q&A. Thanks.
Thanks for starting this thread, shumash. Here's what I have so far:
I have built a 10.1 rom for the hp tenderloin. I am not posting the actual rom because it has a way to go. It is far from a daily driver. Instead, I am posting a link so other developers can build their own roms and maybe we can get a fully working cm10.1 rom.
My device tree is on github:
https://github.com/drmarble/android_device_hp_tenderloin/tree/10.1-ugly
On the 10.1-ugly branch is a semi working device tree for hp tenderloin cm10.1.
Many more details for building are included in the README file.
If you aren't set up to build cyanogenmod roms for the tenderloin, this won't be any use to you. It truely is developer only.
It is a work in progress. Very, very much a WIP.
Camera doesn't work.
Bluetooth doesn't work.
Thanks to the efforts of James Sullins the SDCARD mounts and sound works from the speakers.
This is heavily reliant on the work of Arne Coucheron.
His invaluable patches include:
http://review.cyanogenmod.org/#/c/34375
http://review.cyanogenmod.org/#/c/34377/
http://review.cyanogenmod.org/#/c/34374/
and his highly useful google doc:
https://docs.google.com/document/d/1cGKMXx112gGls9a6ziDTpPgObnU6ZExPy5tByqqIyI4/edit?pli=1
Without his work, we would still have no display.
Of course, everything depends upon the work of the cyanogenmod hp developers Dalingren, FatTire, jcsullins, eyeballer, dorregaray and all the rest.
All my patches are in one big, ugly patch. I included various gerrit patches that my patch is dependent on. I'll work on neatening it up so the commits are separate and readable. I just wanted to get this out there so others could work on it and we can get cm10.1 working.
I think that this is everything.
Please, lets have lots of folks working on this. We can't use gerrit because this isn't cyanogenmod. Steve Kondik feels that since we are dependent upon the legacy qcom repositories it isn't really cm. This is just a kang. We can post patches here, or link to our githubs. Or use pull requests. I also started a thread on Rootzwiki. IRC on #cyanogenmod-touchpad on Freenode is a good place to chat. Jcsullins has already contributed two essential patches that got us SDCARD mounting and sound via irc.
We can get this done together.
drmarble said:
Thanks for starting this thread, shumash. Here's what I have so far:
I have built a 10.1 rom for the hp tenderloin. I am not posting the actual rom because it has a way to go. It is far from a daily driver. Instead, I am posting a link so other developers can build their own roms and maybe we can get a fully working cm10.1 rom.
--SNIP--
We can get this done together.
Click to expand...
Click to collapse
Thanks for that informative post. What are you using for a default manifest and kernel source?
most of the information needed to build is in the README file at my github. I use the regular cm manifest and build tools. The local manifest I use is:
<?xml version="1.0" encoding="UTF-8"?>
<manifest>
<project name="drmarble/android_device_hp_tenderloin" path="device/hp/tenderloin" remote="github" revision="10.1-ugly" />
<project name="CyanogenMod/hp-kernel-tenderloin" path="kernel/hp/tenderloin" remote="github" revision="ics" />
<project name="CyanogenMod/android_hardware_atheros_wlan" path="hardware/atheros/wlan" remote="github" revision="jellybean" />
<project name="arco/android_hardware_qcom_display-legacy" path="hardware/qcom/display-legacy" remote="github" revision="cm-10.1" />
<project name="arco/android_hardware_qcom_media-legacy" path="hardware/qcom/media-legacy" remote="github" revision="cm-10.1" />
</manifest>
I just use the regular cm kernel. The addional repos work together with a couple of flags in Boardconfig.mk. Those are set in my "big_ugly_patch" which should have been broken up into several different ones. Really, you just have to set up a cm build environment, put in my local_manifest.xml, repo sync, apply the various commits listed in the README file and "brunch tenderloin".
To those who want a downloadable ROM. There isn't one yet. I don't think this is ready for anybody but home brewers. jcsullins' and others cm10 builds are much more stable and useable. This is just more fun for build-it-yourself-ers because it is newer and shinier. It's not a better android experience yet.
It may be worth mentioning, for clarity, to make sure you have included vendor/hp/tenderloin.
<project name="TheMuppets/proprietary_vendor_hp" path="vendor/hp/tenderloin" remote="github" revision="jellybean" />
I've been working on including the latest Adreno drivers but have found they are dependent on the CAF varriant, and we are stuck using legacy. Which is exactly why Steve Kondik said "I do not want to create official builds for any device that needs to use this. If your device can't support the latest driver for whatever reason, it's going to be janky and slow and probably worse than stock on 10.1."
I am trying to take some inspiration from the HTC Sensation, aka Pyramid, because that device is also msm8660 and they are utilizing the legacy display variant in 4.2.2 as well - without any complaints about video playback.
You can also use unzip-files.sh or extract-files.sh from device/hp/tenderloin to get the proprietary files. You just have to use jcsullins' latest cm10 zip file or installed rom (20130304etc.zip). These may be even more current than TheMuppets. They haven't been updated for 6 months. I know that the proprietary-files.txt file has been modified since then. I think that he updated some of the proprietary files so that sound would work. He also added some adreno libs, though we should look into the latest 4.2.2 adreno libs.
I'm trying out the Evervolv manifest, which is mostly AOSP. If it works, we may be able to get a mashup of the CM and AOSP source trees that gives us some options. So far, however, I've had to comment out the audio.primary.tenderloin package and the Settings package overlay in the tenderloin device source to get it to continue the compilation. I'll report back tomorrow.
drmarble said:
You can also use unzip-files.sh or extract-files.sh from device/hp/tenderloin to get the proprietary files. You just have to use jcsullins' latest cm10 zip file or installed rom (20130304etc.zip). These may be even more current than TheMuppets. They haven't been updated for 6 months. I know that the proprietary-files.txt file has been modified since then. I think that he updated some of the proprietary files so that sound would work. He also added some adreno libs, though we should look into the latest 4.2.2 adreno libs.
Click to expand...
Click to collapse
Are you getting any video playback with your builds using extract-flies.sh? All I ever seem to get is an infinite "loading" screen. I once used CM9 Adreno drivers in building CM10, with the same "loading" effect.
shumash said:
I'm trying out the Evervolv manifest, which is mostly AOSP. If it works, we may be able to get a mashup of the CM and AOSP source trees that gives us some options. So far, however, I've had to comment out the audio.primary.tenderloin package and the Settings package overlay in the tenderloin device source to get it to continue the compilation. I'll report back tomorrow.
Click to expand...
Click to collapse
Ok, after much thrashing, I"ve got everything compiled and waiting in the OUT folder, it stops with this:
Code:
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
out/target/product/tenderloin/obj/PACKAGING/systemimage_intermediates/unsparse_system.img: 1376/51296 files (0.0% non-contiguous), 60991/204800 blocks
Install system fs image: out/target/product/tenderloin/system.img
out/target/product/tenderloin/system.img+ maxsize=856424448 blocksize=135168 total=237901004 reserve=8650752
[email protected]:~/android/skz$
I'm guessing it has something to do with the releasetools, but I get no error. Any ideas?
If you look at my release patch I modify the release tools which changed for 10.1. I really need to separate that patch into it's separate parts. That would make it clear what you need to change. I did that months ago and really don't recall the reasons for each step, I just know that it works. The zip file format changed, as I recall. Wish I could help more. I'll get to work on cleaning up that patch and will make a cleaner branch soon.
drmarble said:
If you look at my release patch I modify the release tools which changed for 10.1. I really need to separate that patch into it's separate parts. That would make it clear what you need to change. I did that months ago and really don't recall the reasons for each step, I just know that it works. The zip file format changed, as I recall. Wish I could help more. I'll get to work on cleaning up that patch and will make a cleaner branch soon.
Click to expand...
Click to collapse
I think I've got that covered. I modified the releasetools, too, and tried both the modified Evervolv and CM versions, but what I get is that stop. I'll check and make sure, though, and I'll compile with the "-v" switch to see if there's something I missed.
I'm not sure if you are familiar with it, but you could try an application called Meld; its a diff viewer. It always comes in handy when I'm comparing code.
If you want to see what's different from the release tools in Evervolv and drmarble's repositories, you can just side by side compare them from your source locally :thumbup:
Also:
To go with the sound patch I removed one commit from the build process: # 34375 # MemoryHeapBase: ifdef for gingerbread/froyo compatibility We aren't using froyo libs apparently, thanks to jcsullins' new sound patch. This means we can remove -DBINDER_COMPAT from Boardconfig.mk.
For anyone who is not watching both threads
Sent from my SAMSUNG-SGH-T989 using xda premium
-SGA- said:
I'm not sure if you are familiar with it, but you could try an application called Meld; its a diff viewer. It always comes in handy when I'm comparing code.
If you want to see what's different from the release tools in Evervolv and drmarble's repositories, you can just side by side compare them from your source locally :thumbup:
Also:
To go with the sound patch I removed one commit from the build process: # 34375 # MemoryHeapBase: ifdef for gingerbread/froyo compatibility We aren't using froyo libs apparently, thanks to jcsullins' new sound patch. This means we can remove -DBINDER_COMPAT from Boardconfig.mk.
For anyone who is not watching both threads
Sent from my SAMSUNG-SGH-T989 using xda premium
Click to expand...
Click to collapse
Chortle, I do need to watch everything, don't I? And I do use meld. It's a great tool. What I'd like to do is to run the "tenderloin_ota_from_target_files" python script from the command line. I can just see all those compiled files sitting there waiting to be zipped up!
I've tried utilizing drivers from the HTC sensation and jcsullins CM10 03/04 build with no further success in getting video playback to function.
Shumash, any luck with getting that to compile yet? Are you trying straight Evervolv source?
Sent from my SAMSUNG-SGH-T989 using xda premium
-SGA- said:
I've tried utilizing drivers from the HTC sensation and jcsullins CM10 03/04 build with no further success in getting video playback to function.
Shumash, any luck with getting that to compile yet? Are you trying straight Evervolv source?
Sent from my SAMSUNG-SGH-T989 using xda premium
Click to expand...
Click to collapse
I got a CM10.1 to compile, but it boots to the flashy screen. I'm in process of running a clean make to ensure I didn't leave something nasty around. The Evervolv source built to the point that it needs to be zipped into the OTA and release files, but I think there's something missing in the make files that connects with the releasetools. I ran a compile with the verbose switch set, and it acted like it just reached the end of the make files. I have also successfully built a ROM from PAC-man sources, but it also boots to the flashy screen.
Try flashing this after you flash your ROM. It's just a collection of Adreno drivers which I have been experimenting with, it might help get you booting!
-SGA- said:
Try flashing this after you flash your ROM. It's just a collection of Adreno drivers which I have been experimenting with, it might help get you booting!
Click to expand...
Click to collapse
Would you check out that zip file, _SGA_? It fails to flash; says error in zip file (I don"t have any md5 or zip file checking going on).
shumash said:
Would you check out that zip file, _SGA_? It fails to flash; says error in zip file (I don"t have any md5 or zip file checking going on).
Click to expand...
Click to collapse
Just downloaded and flashed now, did not run into any errors. Hmm... I've uploaded it again under a different name :good:
-SGA- said:
Try flashing this after you flash your ROM. It's just a collection of Adreno drivers which I have been experimenting with, it might help get you booting!
Click to expand...
Click to collapse
Would you check out that zip file, _SGA_? It fails to flash; says error in zip file (I don"t have any md5 or zip file checking going on).
EDIT: I found that all but two of the Adreno egl files are missing in the ROM zip. Copied the egl files and your other files manually, and it now boots . I guess is we're going to have to copy them as prebuilt until we have good source.
shumash said:
Would you check out that zip file, _SGA_? It fails to flash; says error in zip file (I don"t have any md5 or zip file checking going on).
EDIT: I found that all but two of the Adreno egl files are missing in the ROM zip. Copied the egl files and your other files manually, and it now boots . I guess is we're going to have to copy them as prebuilt until we have good source.
Click to expand...
Click to collapse
Don't know if the attached are the same as those posted by SGA, but these are official ones from QualComm that I have been using with some sucess. They are advertised as...
"This release contains the user-mode driver binaries for Qualcomm's Adreno 3xx and 2xx GPU on Google Android 4.2 Jelly Bean MR1. It has been tested with the CAF release A8064AAAAANLGD133402.1 and M8960AAAAANLGD230306.1. Supports any Adreno 2xx and 3xx GPU on Android 4.2 Jelly Bean MR1."
How does one port ROMs to another phone? I'm looking to try and port Xylon to the Find 5.
Sent from my X909 using xda app-developers app
it's really not that difficult. just have to adjust some overlays, change some stuff in vendor and maybe grab the camera from CM if the changes haven't been made to that ROM.
if you are building from source:
Find the 4.2.2 source code of the ROM you want to port
Get the manifest.
Add the CM10.1 device/kernel tree and proprietary libs in the .repo/local_manifest/roomservice.xml
sync all the repos
(do whatever the source might need you to set up: EG paranoid android, need you to setup your device properly in vendor/pa/* files)
compile. Fix any problems(maybe missing files, etc)
you could try a non-4.2.2-based ROM to port, but you'll run into more incompatibility issues to fix
mbh87 said:
it's really not that difficult. just have to adjust some overlays, change some stuff in vendor and maybe grab the camera from CM if the changes haven't been made to that ROM.
Click to expand...
Click to collapse
So would you say just use this
http://forum.xda-developers.com/showthread.php?t=2232773
And the CM10 for our Oppo?
Sent from my X909 using xda app-developers app
newbcake said:
So would you say just use this
http://forum.xda-developers.com/showthread.php?t=2232773
And the CM10 for our Oppo?
Sent from my X909 using xda app-developers app
Click to expand...
Click to collapse
No, build from source since it's available.
mbh87 said:
No, build from source since it's available.
Click to expand...
Click to collapse
You mean Oppo's source or AOSP from Google? I presume AOSP
Sent from my X909 using xda app-developers app
assuming this is the one you want to port (it is using 4.2.2, so hopefully it'll be easier)
read their README: https://github.com/XYAOSP/buildscripts
use repo and checkout this manifest
https://github.com/XYAOSP/platform_manifest/blob/jb4.2/default.xml
use repo and sync all the repos
then see how I get the device specific files here? do the same https://git.ngvincent.com/projects/oppo-x909-cm10/wiki/Build#Get-device-specific-sources-and-build
then follow whatever instructions needed to build. if it uses CM based building (which i think they do here, then do
. build/envsetup.sh && brunch
select the find5 entry (if it;s not there, then do lunch, select find5, mka)
wait for it to fail. read error message. fix error message. repeat last 3 steps .. Google is your friend... learn to love google
---------------------------------------
you should try building CM10.1 first. there will hopefully be no errors, and you learn how to use brunch/lunch/breakfast / etc
https://git.ngvincent.com/projects/oppo-x909-cm10/wiki/Build
paperWastage said:
assuming this is the one you want to port (it is using 4.2.2, so hopefully it'll be easier)
read their README: https://github.com/XYAOSP/buildscripts
use repo and checkout this manifest
https://github.com/XYAOSP/platform_manifest/blob/jb4.2/default.xml
use repo and sync all the repos
then see how I get the device specific files here? do the same https://git.ngvincent.com/projects/oppo-x909-cm10/wiki/Build#Get-device-specific-sources-and-build
then follow whatever instructions needed to build. if it uses CM based building (which i think they do here, then do
. build/envsetup.sh && brunch
select the find5 entry (if it;s not there, then do lunch, select find5, mka)
wait for it to fail. read error message. fix error message. repeat last 3 steps .. Google is your friend... learn to love google
---------------------------------------
you should try building CM10.1 first. there will hopefully be no errors, and you learn how to use brunch/lunch/breakfast / etc
https://git.ngvincent.com/projects/oppo-x909-cm10/wiki/Build
Click to expand...
Click to collapse
Is there something wrong with the Xylon repo? I keep getting this:
Code:
Cannot fetch htc-msm8960/[some path here]
newbcake said:
Is there something wrong with the Xylon repo? I keep getting this:
Code:
Cannot fetch htc-msm8960/[some path here]
Click to expand...
Click to collapse
use google and your own intuition to help you (this is why you should build the Cm10.1 ROM first.... the knowledge you gain will help you)
most likely either github is down, or the path referencing it is invalid, or something......
but the question is, do we need htc-msm8960? most likely no, since the oppo kernel is self contained
^^^ all this comes from experience.... seriously, go build cm10.1 for the find 5 first
paperWastage said:
use google and your own intuition to help you (this is why you should build the Cm10.1 ROM first.... the knowledge you gain will help you)
most likely either github is down, or the path referencing it is invalid, or something......
but the question is, do we need htc-msm8960? most likely no, since the oppo kernel is self contained
^^^ all this comes from experience.... seriously, go build cm10.1 for the find 5 first
Click to expand...
Click to collapse
we probably dont need it, but a lot of ROMs make you download ALL their devices and corresponding files.
Well, I removed the HTC lines from manifest.xml (maybe not a good idea?) but anyway, it finished syncing after.
mbh87 said:
we probably dont need it, but a lot of ROMs make you download ALL their devices and corresponding files.
Click to expand...
Click to collapse
yeah... it's a problem with a simple fix.... if the error was something like "packages_mms_app", then it's a more serious problem.....
does removing that repo mask a larger problem? that's why I'm pushing for OP to start small first by building CM10.1, and learn the environment (instead of posting every single problem he encounters here, and asks for help)
I've only started playing with android source code in March/early april... and these are things that I've learned/picked up since then.
paperWastage said:
yeah... it's a problem with a simple fix.... if the error was something like "packages_mms_app", then it's a more serious problem.....
does removing that repo mask a larger problem? that's why I'm pushing for OP to start small first by building CM10.1, and learn the environment (instead of posting every single problem he encounters here, and asks for help)
I've only started playing with android source code in March/early april... and these are things that I've learned/picked up since then.
Click to expand...
Click to collapse
Understood. I'm syncing up the Cm10 now. Thanks!
Is is best to do this all on Linux, Mac, or Windows?
It isn't possible to do it on windows so one of the other two.
mbh87 said:
It isn't possible to do it on windows so one of the other two.
Click to expand...
Click to collapse
Cool thanks. I just "finished" my compile on my Mac 10.7.8 but there was something about gcc422 not found. I have the latest XCode installed along with Dev tools. So can anyone help cause I didn't see anything with a Google.
I presume a clean Ubuntu is preferred
Sent from my X909 using xda app-developers app
newbcake said:
Cool thanks. I just "finished" my compile on my Mac 10.7.8 but there was something about gcc422 not found. I have the latest XCode installed along with Dev tools. So can anyone help cause I didn't see anything with a Google.
I presume a clean Ubuntu is preferred
Sent from my X909 using xda app-developers app
Click to expand...
Click to collapse
it's preferable
even on ubuntu, you still have to get specific packages: need oracle version of jdk, not openjdk, need to remove some incompatible stuff
heres a guide for ubuntu 12.10
https://git.ngvincent.com/projects/oppo-x909-cm10/wiki/Android-setup
paperWastage said:
it's preferable
even on ubuntu, you still have to get specific packages: need oracle version of jdk, not openjdk, need to remove some incompatible stuff
heres a guide for ubuntu 12.10
https://git.ngvincent.com/projects/oppo-x909-cm10/wiki/Android-setup
Click to expand...
Click to collapse
I tried compiling CM10, and am even having problems with that. Google did not seem to help for this. Here is what I did:
I did the mka, then mka otapackage, and NOTHING is appearing. No zips at all.
Code:
hardware/qcom/media/mm-video/vidc/vdec/src/omx_vdec.cpp:9925:138: warning: format '%d' expects argument of type 'int', but argument 5 has type 'OMX_U32 {aka long unsigned int}' [-Wformat]
make: *** wait: No child processes. Stop.
That is the last fine before it gives me control of terminal again.
newbcake said:
I tried compiling CM10, and am even having problems with that. Google did not seem to help for this. Here is what I did:
I did the mka, then mka otapackage, and NOTHING is appearing. No zips at all.
Code:
hardware/qcom/media/mm-video/vidc/vdec/src/omx_vdec.cpp:9925:138: warning: format '%d' expects argument of type 'int', but argument 5 has type 'OMX_U32 {aka long unsigned int}' [-Wformat]
make: *** wait: No child processes. Stop.
That is the last fine before it gives me control of terminal again.
Click to expand...
Click to collapse
a few tips
typically, compilers only stop when there's an error, not warnings (unless you set flags like -Werror )
compiling is parallelized.... let's say you have a quad core, you can compile 4 projects simultaneously... when you have an error on the first make file, it stops.... but the other 3 projects will continue compiling and outputting logs until they're done (then the scheduler sees the first make file has an error, doesn't execute any more make builds and halts), so the last log file/line might not be showing you the error
For CM10.1, find5 is official now... you can just use brunch instead of breakfast/lunch & mka
paperWastage said:
a few tips
typically, compilers only stop when there's an error, not warnings (unless you set flags like -Werror )
compiling is parallelized.... let's say you have a quad core, you can compile 4 projects simultaneously... when you have an error on the first make file, it stops.... but the other 3 projects will continue compiling and outputting logs until they're done (then the scheduler sees the first make file has an error, doesn't execute any more make builds and halts), so the last log file might not be showing you the error
For CM10.1, find5 is official now... you can just use brunch instead of breakfast & mka
Click to expand...
Click to collapse
In essence, I scour the log for errors?
I just presumed as CM10 already existed on the Oppo Find 5, it would be an error-free compile.
newbcake said:
In essence, I scour the log for errors?
I just presumed as CM10 already existed on the Oppo Find 5, it would be an error-free compile.
Click to expand...
Click to collapse
you could source the log for errors
or just run the compile script again and again.... the scheduler will continue building other packages, but it'll come to a point where it can't build anything else without that broken package.... and hence, the error in the log will be obvious(since everything else is already built, no logs from them)
it should be an error-free compile, but sometimes things could happen (a small tweak in package/A gives some problems when compiling with package/B)
I would say, most likely there's some build package incompatibility in your system.
I can't get past this error
build/core/product_config.mk:227: *** _nic.PRODUCTS.[[device/samsung/i605/omni_i605.mk]]: "device/samsung/l900/full_i605.mk" does not exist. Stop.
WARNING: Trying to fetch a device that's already there
build/core/product_config.mk:227: *** _nic.PRODUCTS.[[device/samsung/i605/omni_i605.mk]]: "device/samsung/l900/full_i605.mk" does not exist. Stop.
** Don't have a product spec for: 'omni_i605'
** Do you have the right repo manifest?
No such item in brunch menu. Try 'breakfast'
real 0m0.122s
user 0m0.042s
sys 0m0.089s
apophis9283 said:
I can't get past this error
Click to expand...
Click to collapse
Do you have vendorsetup.sh in your /device/samsung/l900/ folder? If not, create it and put this in it:
Code:
add_lunch_combo omni_i605-userdebug
Do you have full_i605.mk in your device tree? Also, omni_i605.mk and i605.mk?
I had to put the i605 make file in the l900 folder. Doesnt make sense. After all that I got it to compile but no boot.
Sent from my SCH-I605 using Tapatalk 2
Also if l900 is important to build i605 then why doesnt it pull it when I brunch and pulls everything else?
apophis9283 said:
Also if l900 is important to build i605 then why doesnt it pull it when I brunch and pulls everything else?
Click to expand...
Click to collapse
Because nobody did a device bringup yet, so the compiler doesn't know that it needs some common stuff as well.
After examining the Omni repo: how can it pull i900 when the device tree does not exist in the Omni tree yet?
https://github.com/omnirom/
chasmodo said:
Because nobody did a device bringup yet, so the compiler doesn't know that it needs some common stuff as well.
After examining the Omni repo: how can it pull i900 when the device tree does not exist in the Omni tree yet?
https://github.com/omnirom/
Click to expand...
Click to collapse
Im looking at omniroms github. Android device samsung l900 is there and does have a kit kat tag. Its l900 not i900
l900 is Sprints note 2. i605 is Verizons Note 2. If having to put the i605 make file in the l900 folder is whats causing it not to boot, that means its compiling for l900 even though I want it to compile for i605. Something may be borked in build. No reason for i605 to depend on l900. Somethings just not right
Sent from my SCH-I605 using Tapatalk 2
apophis9283 said:
Its l900 not i900
Click to expand...
Click to collapse
Oh...sorry, I thought it was i900.
Anyway, you've got to create your l605 device tree and put all the l900-common references (if they exist) in it: take a look at n7100 or n7000 tree at Omni Github and you'll see what I'm talking about. There's no way for compiler to know what to do unless you tell it.
chasmodo said:
Oh...sorry, I thought it was i900.
Anyway, you've got to create your l605 device tree and put all the l900-common references in it: take a look at n7100 or n7000 tree at Omni Github and you'll see what I'm talking about. There's no way for compiler to know what to do unless you tell it.
Click to expand...
Click to collapse
My i605 folder looks identical to the way n7100 is setup. Its weird, It screams to have full_i605.mk put in the l900 device folder. It shoudlnt be doing this. Somethings borked.
All this over a typo. I found it. Compiling now without a l900 folder
In omni_i605.mk
Change
# Inherit device configuration
$(call inherit-product, device/samsung/l900/full_i605.mk)
change it to
# Inherit device configuration
$(call inherit-product, device/samsung/i605/full_i605.mk)
@apophis9283 I saw your post about No cell/data on the i605 nightly.. was that your build? SlimROM seems to have a working AOSP 4.4.2. Is there any chance that something from their project might help get OmniROM up and running on i605?
I would be willing to work on this, but have no experience or idea where to start.
Thanks.
travisn000 said:
@apophis9283 I saw your post about No cell/data on the i605 nightly.. was that your build? SlimROM seems to have a working AOSP 4.4.2. Is there any chance that something from their project might help get OmniROM up and running on i605?
I would be willing to work on this, but have no experience or idea where to start.
Thanks.
Click to expand...
Click to collapse
Data works on cm and aokp so far. I just havent looked any further into it yet. I got data on aokp by using cms device tree so i never isolated the actual problem. Omni requires an omni make file in the devices tree. So just need to pull in cms trees and make adjustments so omni will compile it
Sent from my SCH-I605 using Tapatalk 2
I quit messing with it. That roomservice file is annoying.
Sent from my SCH-I605 using Tapatalk 2
apophis9283 said:
I quit messing with it. That roomservice file is annoying.
Sent from my SCH-I605 using Tapatalk 2
Click to expand...
Click to collapse
..does that mean pulling from the other projects won't work, or just won't work without ALOT of work? I don't know what the roomservice file is.. :?
I wanted to use a local manifest. Well when I tried to compile of course there were going to be errors but omnis response is to create the roomservice file. So its hard to tell what i may need to change around. The way aokp handles it is a million times better
Sent from my SCH-I605 using Tapatalk 2
I keep getting this sync error:
Code:
Unexpected end of command stream
Fetching projects: 99% (415/419) fatal: Couldn't find remote ref refs/heads/android-4.4.2_r1
error: Cannot fetch android_frameworks_opt_mms
error: Cannot fetch TheMuppets/proprietary_vendor_htc
also, if i have a working cm official, what can i pull from it that would work with omni?
pandasa123 said:
I keep getting this sync error:
Code:
Unexpected end of command stream
Fetching projects: 99% (415/419) fatal: Couldn't find remote ref refs/heads/android-4.4.2_r1
error: Cannot fetch android_frameworks_opt_mms
error: Cannot fetch TheMuppets/proprietary_vendor_htc
also, if i have a working cm official, what can i pull from it that would work with omni?
Click to expand...
Click to collapse
The error is incomplete... I have no idea WHICH repo it's looking for that branch on since it's failing on multiple repos.
My guess would be a malformed omni.dependencies or a bad manifest in .repo/local_manifests/
I can tell you that syncing TheMuppets will result in broken builds 90% of the time.
What should I do for device props instead of muppets? And malformed omni.dependencies? That can affect it?
Sent from my One Xl using Tapatalk
pandasa123 said:
What should I do for device props instead of muppets?
Click to expand...
Click to collapse
https://github.com/DonkeyCoyote/
pandasa123 said:
And malformed omni.dependencies? That can affect it?
Click to expand...
Click to collapse
Yes, because what's in omni.dependencies gets synced as well. So if it tells repo to sync something that isn't there...
So omni is like building other ROMs? Just like building cm but change the cm parts to omni and adding twrp?
On the HTC one xl (Evita) and I really liked the multi window and thought we were overdue for an Omni build, so I wanted to try building it myself instead of waiting for one
Sent from my One Xl using Tapatalk
pandasa123 said:
So omni is like building other ROMs? Just like building cm but change the cm parts to omni and adding twrp?
On the HTC one xl (Evita) and I really liked the multi window and thought we were overdue for an Omni build, so I wanted to try building it myself instead of waiting for one
Sent from my One Xl using Tapatalk
Click to expand...
Click to collapse
In essence, yes. And I applaud you for being proactive and starting to compile for yourself.
PonsAsinorem said:
In essence, yes. And I applaud you for being proactive and starting to compile for yourself.
Click to expand...
Click to collapse
everything is syncing beautifully now...just one quick question...
for omni, should i add to the sync the CyanogenMod/android_device_htc_evita git repo with the build? i mean, with the build, the only thing that's really specified is the kernel and thats it, so there has to be more variety than just that, isn't there?
pandasa123 said:
everything is syncing beautifully now...just one quick question...
for omni, should i add to the sync the CyanogenMod/android_device_htc_evita git repo with the build? i mean, with the build, the only thing that's really specified is the kernel and thats it, so there has to be more variety than just that, isn't there?
Click to expand...
Click to collapse
Nope, not really. I'd try to sync only Omni's repos (and Google's), and leave other ROMs out of it. Assuming you're trying to build for the evita, and Omni doesn't have one for that, then you're best bet is to:
1. Fork (or create) a device tree
2. Make sure it's omnified
3. Add it to local_manifest
4. Once you have it up and running with minimal bugs, request to have Omni fork your device tree so it can be supported.
PonsAsinorem said:
Nope, not really. I'd try to sync only Omni's repos (and Google's), and leave other ROMs out of it. Assuming you're trying to build for the evita, and Omni doesn't have one for that, then you're best bet is to:
1. Fork (or create) a device tree
2. Make sure it's omnified
3. Add it to local_manifest
4. Once you have it up and running with minimal bugs, request to have Omni fork your device tree so it can be supported.
Click to expand...
Click to collapse
Omnified? This always confuses me. How do users have twrp if they have cwm installed? Will it overwrite?
And for the Evita tree, its basically me downloading cm tree locally and changing it to comply with omni? And then uploading it to my git so it can be added to local_manifest with that?
Sent from my One X using Tapatalk
pandasa123 said:
How do users have twrp if they have cwm installed?
Click to expand...
Click to collapse
They don't. It's either one or the other.
pandasa123 said:
Will it overwrite?
Click to expand...
Click to collapse
If you install it, yes. You don't have to, but it's recommended.
pandasa123 said:
And for the Evita tree, its basically me downloading cm tree locally and changing it to comply with omni? And then uploading it to my git so it can be added to local_manifest with that?
Click to expand...
Click to collapse
More or less. If you haven't seen this yet:
Porting Omni To Your Device
everything is synced and i followed the guide about the .mk files and vendorsetup.sh
at the beginning of the build, it showed these issues and i was wondering what they meant
Code:
build/core/product_config.mk:241: *** No matches for product "omni_evita". Stop.
WARNING: Trying to fetch a device that's already there
EDIT: the build died:
Code:
build/core/product_config.mk:241: *** No matches for product "omni_evita". Stop.
** Don't have a product spec for: 'omni_evita'
** Do you have the right repo manifest?
No such item in brunch menu. Try 'breakfast'
G6 Play Sources​
A huge thanks to the anonymous user who sent this post to Motorola. This issue is fixed. We can now clone all the Jeter kernel sources in one shot!!!
For Oreo:
Code:
git clone -b oreo-8.0.0-release-jeter https://github.com/MotorolaMobilityLLC/kernel-msm jeter
For Pie:
Code:
git clone -b pie-9.0.0-release-pcp29 https://github.com/MotorolaMobilityLLC/kernel-msm jeter
This will clone the kernel sources into a folder named "jeter".
Deprecated Information:
In an effort to combat Motorola's anti open source practices, I've collected all available kernel sources for the G6 Play and placed them into a single zip. Inside this zip are 43 gzips which contain all parts needed to build the kernel for our device.
Back story:
For those of you who don't know, Motorola has split the kernel sources into many parts. There is no easy way to grab them, sync them, or pull them. GitHub also lacks the ability to search for specific release tags.
Example: You can't search for "MMI-OCPS" in all of Motorola's repos, and you can't search like this inside of a single repo either.
So you have to manually open each repo one by one, and scan every page for the tag that matches the release you need. Sometimes there's nothing. Sometimes it's 5 pages down. Then you have to unpack all the kernel parts you found, and some of them need to be merged into the same folder. There is no explanation on what gets merged where. You have to try building it, get an error, then find the part you're missing in one of the zips and add it. Then repeat the process.
This is the most anti open source practice I've ever seen. I'm hoping this spurs some development for our device.
Link is now void.
Nice work putting all of these together.
kwiksi1ver said:
Nice work putting all of these together.
Click to expand...
Click to collapse
I almost have a tree going! I clobbered one together so I can rebuild twrp using Omni 9.0 sources. I'm stuck at 99% but I'm slowly chomping the bit. It seems like I can easily build a 64bit twrp if I get the last kinks out. A complete build is slowly inching closer and I can finally taste it!
Well boy howdy this is just grand news
Good news and bad news. I finally got the build to complete. However, it's not booting because for some reason, it's not building the ramdisk. I can't figure out why. No amount of searching has helped me with this. I can't find any information about it. It shouldn't even be possible as far as I can tell. It produces a recovery.img, kernel, and dt.img but no ramdisk. I'm really lost on this one.
Spaceminer said:
Good news and bad news. I finally got the build to complete. However, it's not booting because for some reason, it's not building the ramdisk. I can't figure out why. No amount of searching has helped me with this. I can't find any information about it. It shouldn't even be possible as far as I can tell. It produces a recovery.img, kernel, and dt.img but no ramdisk. I'm really lost on this one.
Click to expand...
Click to collapse
Did you have any luck figuring this out?
Sands207 said:
Did you have any luck figuring this out?
Click to expand...
Click to collapse
Not at all. I've given up on it.
Spaceminer said:
Not at all. I've given up on it.
Click to expand...
Click to collapse
Can you link me to the kernel source? I'll repo sync and try it on my dev machine
At this point I'm willing to pay someone ? 20-30 bucks to whoever gets us a decrypted twrp that ain't bugged out and works for all variants and models lol
Spaceminer said:
G6 Play Sources
In an effort to combat Motorola's anti open source practices, I've collected all available kernel sources for the G6 Play and placed them into a single zip. Inside this zip are 43 gzips which contain all parts needed to build the kernel for our device.
Back story:
For those of you who don't know, Motorola has split the kernel sources into many parts. There is no easy way to grab them, sync them, or pull them. GitHub also lacks the ability to search for specific release tags.
Example: You can't search for "MMI-OCPS" in all of Motorola's repos, and you can't search like this inside of a single repo either.
So you have to manually open each repo one by one, and scan every page for the tag that matches the release you need. Sometimes there's nothing. Sometimes it's 5 pages down. Then you have to unpack all the kernel parts you found, and some of them need to be merged into the same folder. There is no explanation on what gets merged where. You have to try building it, get an error, then find the part you're missing in one of the zips and add it. Then repeat the process.
This is the most anti open source practice I've ever seen. I'm hoping this spurs some development for our device.
Release Tag:MMI-PPP29.55-25 Pie (281mb)
Click to expand...
Click to collapse
I can't find Jeter pie source anywhere on their GitHub.
Mind throwing up a link?
Sands207 said:
Can you link me to the kernel source? I'll repo sync and try it on my dev machine
Click to expand...
Click to collapse
madbat99 said:
I can't find Jeter pie source anywhere on their GitHub.
Mind throwing up a link?
Click to expand...
Click to collapse
They don't exist in a single location on GitHub. There's literally no way to sync their sources that I'm aware of. I had to untar all the files that I have zipped together in the OP, into a rudimentary tree I found. Here's the closest thing we have.
https://github.com/SyNtheticNightmar3/android_device_motorola_jeter
Sure wish we had squid2 working on it I miss my g4 play
ninjakira said:
Sure wish we had squid2 working on it I miss my g4 play
Click to expand...
Click to collapse
I was praying for the same thing when I bought this phone.
Spaceminer said:
I was praying for the same thing when I bought this phone.
Click to expand...
Click to collapse
Right tbh that's why I even stuck with Moto is because of his work. I just popped over at the g7play forum and there doing slightly worse because the g7play is a/b partition and all they have is a dirty port of twrp for the g7
Spaceminer said:
They don't exist in a single location on GitHub. There's literally no way to sync their sources that I'm aware of. I had to untar all the files that I have zipped together in the OP, into a rudimentary tree I found. Here's the closest thing we have.
https://github.com/SyNtheticNightmar3/android_device_motorola_jeter
Click to expand...
Click to collapse
Did the g6 play start out on Nougat? I have been out of development since i got sick last year. I wish between us all we could get this mess sorted out. I'm bummed Moto doesn't have an open source website besides github where we could just download the source code and build from scratch.
Sands207 said:
Did the g6 play start out on Nougat? I have been out of development since i got sick last year. I wish between us all we could get this mess sorted out. I'm bummed Moto doesn't have an open source website besides github where we could just download the source code and build from scratch.
Click to expand...
Click to collapse
No. It came shipped with Oreo which means its treble compatible which is why you don't have to do the stuff people with phones before oreo need to do to "make" there phone treble compatible so they could flash gsi
ninjakira said:
No. It came shipped with Oreo which means its treble compatible which is why you don't have to do the stuff people with phones before oreo need to do to "make" there phone treble compatible so they could flash gsi
Click to expand...
Click to collapse
Then why is the nougat branch cloned for omni sources?
Sands207 said:
Then why is the nougat branch cloned for omni sources?
Click to expand...
Click to collapse
Hmmm I honestly couldn't even tell you why at this point ?? the development for this phone has been absolute hell from people stealing twrps to people trying to sell overpriced source and tress to making fake trees ?? like it's one giant "bruh, what the hell" moment
ninjakira said:
Hmmm I honestly couldn't even tell you why at this point ?? the development for this phone has been absolute hell from people stealing twrps to people trying to sell overpriced source and tress to making fake trees ?? like it's one giant "bruh, what the hell" moment
Click to expand...
Click to collapse
I can tell, wow
Sands207 said:
I can tell, wow
Click to expand...
Click to collapse
Yea but hey we're doing better than the G7 play guys right now so there's a win I guess ? o and hope your feeling better. Noticed you were kinda absent from all the chaos lol