Related
General informations:
This thread's aim is only to represent a central meeting and discussion point for BCM21553 developers and, in particular, for the open Kernel/ROM sources development for the Samsung Galaxy Pocket GT-S5300 (codenamed Cori).
Information for common users:
As already described in the previous section, if you are not a developer, please restrict your posts to the general discussion thread so that developers can maintain good communication. Every post that is not strictly respecting these rules will be reported to the forum moderators. Thanks for your understanding.
For any other BCM21553 device related question or information, please, use this thread as a reference point, instead:
[DEVS ONLY][BCM21553 series] CyanogenMod 11 for BCM21553 Development Discussion
As someone already might know, I'm streambinder, from MoltenMotherBoard team.
I have already followed some projects for the GT-S5300, but especially kept in contact
with some of the events related to the porting of ROM and Kernel sources for BCM21553 chipset based devices.
In this precise moment, the sources in my possession allow you to be able to compile
a bugfree CWM 5.0.2.8 (based on CyanogenMod 7 code) with a kernel based on the Samsung stock one.
The only - fundamental - problem was due to the fact that unless I hadn't used the prebuilt INIT binary
token in the Samsung stock firmware boot.img, the phone would not work - or, better, boot up.
This means that until the situation - regarding this issue - doesn't change, our access to the porting of custom ROM
would be barred.
Recently, I decided to give Cori another chance and rework my sources, looking at the wonderful work brought
by the BroadcomCM team on CyanogenMod 9 (in particular, thanks to @bieltv.3 and @Alberto96) and @psyke83 on CyanogenMod 11.
They've not only been able to run these two ROMs in a more or less crude way, but this developer has been able to write
the necessary strings to make the INIT binary of some of these BCM21553 devices opensource.
Strong of this informations, I readjusted some of the sources of BroadcomCM's CyangenMod 9, which includes
all the progress carried out by both the team and psyke83, in order to make them work even on Cori,
and am now next to the first test of the CWM 6.X.X.X, based on IceCreamSandwich code.
At the same time, @akhbh is working on the KitKat code based CWM.
I hope I can give more information about any progress as soon as possible.
The General Discussion thread for non-development issues is here:
*.[DISCUSSION] CyanogenMod 11 For Galaxy Pocket GT-S5300 Discussion Thread
Made a first test of CWM based on CyanogenMod 9 code.
It seems it cannot flash it as it weighs so much compared to its partition configuration value: in fact, the maximum boot partition size is set up to 5.0MB, but the compiled boot.img weighs 5.3MB.
Will have to resize its weight in order to make it fill into the partition.
@akhbh, have you had any complication in these terms, with CyanogenMod 11 sources?
@psyke83, what do you suggest to do? Do you think an increasement of boot partition would be a better idea?
streambinder said:
Made a first test of CWM based on CyanogenMod 9 code.
It seems it cannot flash it as it weighs so much compared to its partition configuration value: in fact, the maximum boot partition size is set up to 5.0MB, but the compiled boot.img weighs 5.3MB.
Will have to resize its weight in order to make it fill into the partition.
@akhbh, have you had any complication in these terms, with CyanogenMod 11 sources?
@psyke83, what do you suggest to do? Do you think an increasement of boot partition would be a better idea?
Click to expand...
Click to collapse
No, I didn't faced those complications. My boot.img weighted around 4.5 MB in lzma compression mode. In gzip, it increased to more than 5 mb.
akhbh said:
No, I didn't faced those complications. My boot.img weighted around 4.5 MB in lzma compression mode. In gzip, it increased to more than 5 mb.
Click to expand...
Click to collapse
Perfect.
Which kernel have you based your build on?
streambinder said:
Perfect.
Which kernel have you based your build on?
Click to expand...
Click to collapse
Well, I took the GeTux kernel for cori, had to change the board name though and then compile it. CWM was booting even before changing the board name but there was no reaction from the phone on trying to boot cm9/cm11
And after changing board name, a black screen on trying to boot
Another info: When tried to merge cori source into the bcm21553 common one, it did compile but gave bootloop of GT-S5300 logo on trying to boot as well as when trying to go in CWM.
Bieltv.3 recommended to use cori source instead of the bcm21553 common one so we used cori sources
akhbh said:
Well, I took the GeTux kernel for cori, had to change the board name though and then compile it. CWM was booting even before changing the board name but there was no reaction from the phone on trying to boot cm9/cm11
And after changing board name, a black screen on trying to boot
Another info: When tried to merge cori source into the bcm21553 common one, it did compile but gave bootloop of GT-S5300 logo on trying to boot as well as when trying to go in CWM.
Bieltv.3 recommended to use cori source instead of the bcm21553 common one so we used cori sources
Click to expand...
Click to collapse
I suggest to use our Kernel sources for now, too: it will automatically bypass few errors/issues/bootloops that we cannot now fight with.
The most important thing is to make INIT working from sources (hope it will be working
on my CM9 sources, too) and check that every our configuration is correctly working and
making Cori boot into recovery.
Once we'll make it perfectly working without any kind of issue, will be the right time to try
to make Cori supported with the BC21553-common kernel.
streambinder said:
I suggest to use our Kernel sources for now, too: it will automatically bypass few errors/issues/bootloops that we cannot now fight with.
The most important thing is to make INIT working from sources (hope it will be working
on my CM9 sources, too) and check that every our configuration is correctly working and
making Cori boot into recovery.
Once we'll make it perfectly working without any kind of issue, will be the right time to try
to make Cori supported with the BC21553-common kernel.
Click to expand...
Click to collapse
Okay, I will use your kernel sources and try if something is changed once I reach home. For now, I neither have this device nor a PC, as I'm in another city.
Will be keenly watching your work. Will start after reaching home around the end of September
akhbh said:
Okay, I will use your kernel sources and try if something is changed once I reach home. For now, I neither have this device nor a PC, as I'm in another city.
Will be keenly watching your work. Will start after reaching home around the end of September
Click to expand...
Click to collapse
No problem, mate.
Here you have every source in my possession:
platform_kernel_samsung_cori
platform_device_samsung_cori
Keep in contact with me, as I will need some informations by you.
Anyway I'm now making another build, keeping some not so much important binaries excluded, so that I can make the compiled boot.img fill into our little Cori's boot partition. I know it's a dirty workaround, but if it works, I'll use it untill @psyke83 will suggest me a better way to do.
streambinder said:
No problem, mate.
Here you have every source in my possession:
platform_kernel_samsung_cori | github.com
platform_device_samsung_cori
Keep in contact with me, as I will need some informations by you.
Anyway I'm now making another build, keeping some not so much important binaries excluded, so that I can make the compiled boot.img fill into our little Cori's boot partition. I know it's a dirty workaround, but if it works, I'll use it untill @psyke83 will suggest me a better way to do.
Click to expand...
Click to collapse
have u tried to build cwm v6 from cm9 source ??
cleverior.ipul said:
have u tried to build cwm v6 from cm9 source ??
Click to expand...
Click to collapse
Of course, mate. I'm working on it, right now.
It doesn't seem to boot, strange if the same INIT binary sources are working for @akhbh.
#UPDATE
In order to troubleshoot, I'll give you some info.
For his build I used these sources:
platform_kernel_samsung_cori
platform_device_samsung_cori
android_device_samsung_bcm21553-common
Applied some lines on our bcm21553-bootimg.mk, too, in order to exclude parted and mke2fs and make the compiled boot.img weigh less.
@cleverior.ipul, can you link me your kernel sources, as akhbh said he used your ones for CM11.
@akhbh, which modifies have you applied in order to compile CWM based on CM11 code? Which device tree?
#UPDATE 2
Attached my compiled boot.img.
If anyone of you would extract it (you can easily use this tool: bootimgtools - read how to use it in the README) and make a diff with the CM11 one (just extract the ramdisk of both boot.imgs and - in the terminal - use this command: diff -urN /path/to/cm9/ramdisk /path/to/cm11/ramdisk > diff.patch), would make to me a huge favour.
Let me know.
streambinder said:
Of course, mate. I'm working on it, right now.
It doesn't seem to boot, strange if the same INIT binary sources are working for @akhbh.
#UPDATE
In order to troubleshoot, I'll give you some info.
For his build I used these sources:
platform_kernel_samsung_cori
platform_device_samsung_cori
android_device_samsung_bcm21553-common
Applied some lines on our bcm21553-bootimg.mk, too, in order to exclude parted and mke2fs and make the compiled boot.img weigh less.
@cleverior.ipul, can you link me your kernel sources, as akhbh said he used your ones for CM11.
@akhbh, which modifies have you applied in order to compile CWM based on CM11 code? Which device tree?
#UPDATE 2
Attached my compiled boot.img.
If anyone of you would extract it (you can easily use this tool: bootimgtools - read how to use it in the README) and make a diff with the CM11 one (just extract the ramdisk of both boot.imgs and - in the terminal - use this command: diff -urN /path/to/cm9/ramdisk /path/to/cm11/ramdisk > diff.patch), would make to me a huge favour.
Let me know.
Click to expand...
Click to collapse
I think we didn't had significant changes. Perhaps the same as totoro. But, that resulted in the internal_sd not mounting error in cwm.
Sadly, as said before, I am away from my home city and can't provide the files to you and can't do the boot.img diffs as well
Try to ask psyke83, he might have a solution for that
akhbh said:
I think we didn't had significant changes. Perhaps the same as totoro. But, that resulted in the internal_sd not mounting error in cwm.
Sadly, as said before, I am away from my home city and can't provide the files to you and can't do the boot.img diffs as well
Try to ask psyke83, he might have a solution for that
Click to expand...
Click to collapse
Then, if you didn't make any massive change upon the sources, then I'll only try using your kernel.
Can you give me your kernel sources, mate, please?
streambinder said:
Then, if you didn't make any massive change upon the sources, then I'll only try using your kernel.
Can you give me your kernel sources, mate, please?
Click to expand...
Click to collapse
Currently, I can provide you the boot.img only. For the sources, @cleverior.ipul can provide the kernel sources coz as said earlier, his kernel is used. Well, we were working together to bring cm11 but weren't successful
akhbh said:
Currently, I can provide you the boot.img only. For the sources, @cleverior.ipul can provide the kernel sources.
Click to expand...
Click to collapse
Ok, please send it to me, will compare it with my package.
streambinder said:
Ok, please send it to me, will compare it with my package.
Click to expand...
Click to collapse
Here it is:
http://www.4shared.com/zip/1nKFbOJ2ba/ccccGetux_CM11.html
streambinder said:
Of course, mate. I'm working on it, right now.
It doesn't seem to boot, strange if the same INIT binary sources are working for @akhbh.
#UPDATE
In order to troubleshoot, I'll give you some info.
For his build I used these sources:
platform_kernel_samsung_cori
platform_device_samsung_cori
android_device_samsung_bcm21553-common
Applied some lines on our bcm21553-bootimg.mk, too, in order to exclude parted and mke2fs and make the compiled boot.img weigh less.
@cleverior.ipul, can you link me your kernel sources, as akhbh said he used your ones for CM11.
@akhbh, which modifies have you applied in order to compile CWM based on CM11 code? Which device tree?
#UPDATE 2
Attached my compiled boot.img.
If anyone of you would extract it (you can easily use this tool: bootimgtools - read how to use it in the README) and make a diff with the CM11 one (just extract the ramdisk of both boot.imgs and - in the terminal - use this command: diff -urN /path/to/cm9/ramdisk /path/to/cm11/ramdisk > diff.patch), would make to me a huge favour.
Let me know.
Click to expand...
Click to collapse
here the link source https://github.com/cleverior/android_kernel_samsung_cori
i've changed the board name. If your device can not boot after using the zImage from this source, then rename init.bcm21553.rc to init.gt-s5300.rc.
@streambinder, what is grom? As bieltv.3 said that init built grom for cori is required to fix adb over cwm recovery. If adb gets working, then possibly the black screen while booting cm11 might get fixed
akhbh said:
@streambinder, what is grom? As bieltv.3 said that init built grom for cori is required to fix adb over cwm recovery. If adb gets working, then possibly the black screen while booting cm11 might get fixed
Click to expand...
Click to collapse
Sincerely don't what are you talking about.
Anyway, have to try to understand where's the problem with the not-booting CWM.
Will try with your sources and let you know.
https://github.com/MiCode/Xiaomi_Kernel_OpenSource/tree/hydrogen-m-oss
Although the source tree name is hydrogen-m-oss, MI's staff Ivan says hydrogen and helium shares the same kernel source.
is this for real?? man I can't believe my eyes,this means real development, more ROMs...maybe I should stop dreaming and have some expert confirm,
@nijel8
@ajsmsg78
:good:
Hmm hope it's the real deal..
Sent from my MI MAX using XDA-Developers mobile app
laolang_cool said:
https://github.com/MiCode/Xiaomi_Kernel_OpenSource/tree/hydrogen-m-oss
But seems only hydrogen source is released
Click to expand...
Click to collapse
Yes its kernel source for hydrogen (android M kernel).
Now we can have better CM13 and roms based on it.
And we can compile android N also.
Confirmed with Santosh M sir (compiled android N for redmi note 3 first).
I am very much noob at compiling, lets hope developers come up with great things.:good:
Nice to hear it, if it is true...
I hope Helium will coming soon.
小米Max*全网通版*3GB/64GB*银色 - Resurrection Remix 5.7.4 build 6.10.28 by ajsmsg78
Yep it's out... Too bad I can't do anything for two more months...
Looking forward to AOSP or slim for max. too bad only hydrogen's source is out, mine is helium...
Hmmm, i will back ?
nijel8 said:
Yep it's out... Too bad I can't do anything for two more months...
Click to expand...
Click to collapse
how come?
please more roms for xiaomi max hydrogen KERNEL SOURCE IS OUT
I asked MI's staff who is also rom developer Ivan, he says hydrogen and helium shares a same kernel source.
laolang_cool said:
I asked MI's staff who is also rom developer Ivan, he says hydrogen and helium shares a same kernel source.
Click to expand...
Click to collapse
That's probably true looking at the config files... The max cpu's are set to 8.
BTW I managed to build it after fixing some compile errors but it doesn't boot... I guess .config needs tweaking but I am not a kernel expert ... Also /arch/arm64/boot/dts/qcom directory is missing and to get it to compile I had to borrow stuff from /arch/arm/boot/dts/qcom which probably is wrong... Anyway no success...
nijel8 said:
That's probably true looking at the config files... The max cpu's are set to 8.
BTW I managed to build it after fixing some compile errors but it doesn't boot... I guess .config needs tweaking but I am not a kernel expert ... Also /arch/arm64/boot/dts/qcom directory is missing and to get it to compile I had to borrow stuff from /arch/arm/boot/dts/qcom which probably is wrong... Anyway no success...
Click to expand...
Click to collapse
Ivan confirmed helium uses the same config file hydrogen_user_defconfig, and no file is missing... dts in arm64/ and arm/ linked to a same directory.
This is incredible news! Fingers crossed this will lead to some great new ROMs!
Let's hope this leads to some custom kernels
It quits with
Code:
drivers/video/msm/msm_dba/msm_dba.c:25:30: fatal error: msm_dba_internal.h: No such file or directory
for a weird reason here. The file is there, but it quits.
Edit: It works after changing #include <msm_dba_internal.h> to #include "msm_dba_internal.h" in msm_dba.c, but just like nijel8 stated it won't finish because of dts.
TEKME said:
It quits with
Code:
drivers/video/msm/msm_dba/msm_dba.c:25:30: fatal error: msm_dba_internal.h: No such file or directory
for a weird reason here. The file is there, but it quits.
Edit: It works after changing #include <msm_dba_internal.h> to #include "msm_dba_internal.h" in msm_dba.c, but just like nijel8 stated it won't finish because of dts.
Click to expand...
Click to collapse
Copy /arch/arm/boot/dts folder to /arch/arm64/boot/ and /include/dt-bindings to /arch/arm64/include keeping all already existing files(don't overwrite) and it will compile but without cpu_ctl.ko and wifi.ko modules.
nijel8 said:
Copy /arch/arm/boot/dts folder to /arch/arm64/boot/ keeping all already existing files(don't overwrite) and it will compile but without cpu_ctl.ko and wifi.ko modules.
Click to expand...
Click to collapse
From Ivan: cpu_ctl.ko is private. And wifi.ko doesn't exist in kernel/, it's in vendor/qcom/opensource/wlan/ , so MI doesn't release their module source(will not in the future I think...)
We need to get the wifi module source from qcom i.g. https://source.codeaurora.org/quic/la/platform/vendor/qcom-opensource/wlan/qcacld-2.0/, but not sure if max uses qcald-2.0 or not.
Is it possible to use Kenzo sources to fill in the gaps? I swapped my Kenzo for a Hydrogen and I'm loving this huge phone, although MIUI is still a bit clunky. The Nougat AOSP and CM ROMs for Kenzo are pretty much ready for daily use, with camera support and VoLTE being a bit buggy right now.
Both Kenzo and Hydrogen devices use msm8952 with fpc1020 fingerprint sensors, I'm not sure about the camera sensors though.
I'm working on CM14, after a lot of go through I finally zipped it.
but its not booting up.
I used following Resources
Device Tree - It needs a lot of modifications.
Kernel
Vendor Files
MSM 8996 HALs from CM14 are giving Errors, so I used HALs from OMNI
EngMod: Logcat
Bump!
No one?
WhyOrean said:
Bump!
No one?
Click to expand...
Click to collapse
Unfortunately i cant help but i bump this thread
Thanks for your effortand hope someone helps.
Try port rom 7.0 from one plus 3 with file of miui n leak
Update: Got it booting :highfive:
But it seems like its display needs to be calibrated.
It shuts down after a giving a quick glance.
This time I used Precompiled Binaries from TheMuppets
Also Hardware/qcom/display-caf and Hardware/qcom/media-caf throw errors.
So I used ones from CM13 Branch.
-> Syncing CM14.1, seems like CM14 is abandoned.
WhyOrean said:
Update: Got it booting :highfive:
But it seems like its display needs to be calibrated.
It shuts down after a giving a quick glance.
This time I used Precompiled Binaries from TheMuppets
Also Hardware/qcom/display-caf and Hardware/qcom/media-caf throw errors.
So I used ones from CM13 Branch.
-> Syncing CM14.1, seems like CM14 is abandoned.
Click to expand...
Click to collapse
good job man! :good:
Whoa, that would be great, please keep us informed!
there is no such kernel,device tree,vendor for my device.
I could build device tree via twrpbuilder tree generator,
but kernel cannot be found in opensource release center.
why am i need A605k? not F or FN or else?
that's because when i used A605FN recovery it has so many errors... and it makes me build twrp for A605K.
not for other models..
but.. i failed,
device tree built via twrpbuilder tree generator doesn't have omni.dependencies file
so, what do you guys think how can i fix those annoying things?
I know, this is not a question area, but i want to make Discussion with you guys to get a idea
Well if your trying to build a device tree and kernel source and vendor start with downloading android source in linux there are many great guides for this then download the sources for the A600F/FN or any other current model of the A600 then there will be a file in the folder for whichever source you downloaded that will allow you to pull resources from your phone like kernel recovery device vendor files ect.. use that and then import it to the same folder than try to build TWRP and test it to see if you get a working build if not then go back and try again their are many sources for learning how to compile twrp for devices that dont have it read those and try again.
Sent from my [device_name] using XDA-Developers Legacy app
tmvkrpxl0 said:
there is no such kernel,device tree,vendor for my device.
I could build device tree via twrpbuilder tree generator,
but kernel cannot be found in opensource release center.
why am i need A605k? not F or FN or else?
that's because when i used A605FN recovery it has so many errors... and it makes me build twrp for A605K.
not for other models..
but.. i failed,
device tree built via twrpbuilder tree generator doesn't have omni.dependencies file
so, what do you guys think how can i fix those annoying things?
I know, this is not a question area, but i want to make Discussion with you guys to get a idea
Click to expand...
Click to collapse
I did find the kernel and source here: http://opensource.samsung.com/recep...menu_item=mobile&classification1=mobile_phone
(page 44 for SM-A605FN - A605FNXXU1ARD7)
I would like to build it myself, but my android development knowledge is limited (i do have a solid linux knowledge though).
Is this guide obsolete (https://forum.xda-developers.com/showthread.php?t=1943625)?
If yes, anyone knows where can i find an up-to-date one?
If not, any idea where BoardConfig.mk can be found?
Where can we find a guide on how to build a twrp? I would really like to learn how to do it
If not, anyone can make a bount?
matheus_sc said:
If not, anyone can make a bount?
Click to expand...
Click to collapse
Surely they are already working on it. But it just takes some time. At the moment everything is working optimally despite fastboot. Even root is possible.
pittrich said:
Surely they are already working on it. But it just takes some time. At the moment everything is working optimally despite fastboot. Even root is possible.
Click to expand...
Click to collapse
I know i use xiaomi eu already. Only update is pain to ass
wait for it
Rumour has it some chinese dev is working on it
Any news?
matheus_sc said:
Any news?
Click to expand...
Click to collapse
Still no sign.... Orangefox also quiet...
The guy who made lots of xiaomi TWRP recoveries stopped developing.
... are kernel (or device/vendor) repos released for it?
This would help.
Do we have TWRP and/or released kernel sources from other Qualcom 888 based devices out there?
raupe said:
... are kernel (or device/vendor) repos released for it?
This would help.
Do we have TWRP and/or released kernel sources from other Qualcom 888 based devices out there?
Click to expand...
Click to collapse
This?
Xiaomi Mi 11 kernel source code is out, the device is codenamed "venus" - Gizmochina
Yesterday, Xiaomi announced Mi 11 as the world’s first smartphone powered by Qualcomm Snapdragon 888. The handset already went for pre-order soon after the launch and is set to go on sale for the first time on January 1. Ahead of its sale, the company has even released this device’s kernel...
www.gizmochina.com
I read on Xiaomi.eu somewhere that there are issues with how the partitions are made and TWRP is very difficult to make for this. They said it's gonna take either a very long time before a recovery is made, or maybe not even at all.
Danacy said:
I read on Xiaomi.eu somewhere that there are issues with how the partitions are made and TWRP is very difficult to make for this. They said it's gonna take either a very long time before a recovery is made, or maybe not even at all.
Click to expand...
Click to collapse
Yeah they changed everything over to GKI 1.0 and no one seems to be supporting it, although they found somewhere on a Chinese site a recovery that works (TWRP) if you boot it (don't flash it) and unencrypts the data. It's in the XDA forums now.
[TWRP] [venus] TWRP Recovery for Mi 11
REMOVED! Use this instead : https://forum.xda-developers.com/t/recovery-11-official-teamwin-recovery-project.4374109/
forum.xda-developers.com
mslezak said:
[TWRP] [venus] TWRP Recovery for Mi 11
REMOVED! Use this instead : https://forum.xda-developers.com/t/recovery-11-official-teamwin-recovery-project.4374109/
forum.xda-developers.com
Click to expand...
Click to collapse
*justintime* said:
Still no sign.... Orangefox also quiet...
The guy who made lots of xiaomi TWRP recoveries stopped developing.
Click to expand...
Click to collapse
Yeah Mauronofrio is out, too much work for no return. He was pretty much the last standing open-source TWRP dev out there for tons of devices. The ones from China never get source released... Which would mean that someone would have to pickup the TWRP project and re-engineer it to fit the new GKI 1.0 then 2.0 platforms that run off that ACK. That's a lot of work. I'm talking about keeping the TWRP project open-source. It would very nice if Google stepped in since they are experts on the new format, but no such luck.
mslezak said:
Yeah they changed everything over to GKI 1.0 and no one seems to be supporting it, although they found somewhere on a Chinese site a recovery that works (TWRP) if you boot it (don't flash it) and unencrypts the data. It's in the XDA forums now.
Click to expand...
Click to collapse
Interesting, that would explain also the lack of custom ROMs? But I read that the source code is released, even before the device was launched. Isn't there GKI-support to be found in there?
Danacy said:
Interesting, that would explain also the lack of custom ROMs? But I read that the source code is released, even before the device was launched. Isn't there GKI-support to be found in there?
Click to expand...
Click to collapse
Yes they released source there are just no instructions anywhere on how to build the kernel #1 and the GKI modules #2 and combine into a build.
The new Xiaomi.eu weekly includes TWRP for Venus (Mi11) as well as Star (Mi 11 Lite 5G, Mi 11 Pro, Mi 11 Ultra, MIX FOLD) (it's a bootable TWRP, but persists after installation). Seems last week's has added Alioth (Mi 10S, Redmi K40). So I would expect a TWRP for Haydn to show up soon in the weekly Xiaomi.eu releases. Seems they forgot the haydn doesn't have TWRP yet so you can't actually install it! Tried the updater app no go, it can't boot to recovery because they don't have one yet, even though on Xiaomi.eu they say it's available, can't find it anywhere.
BTW the only TWRPs I can actually find are for Venus and Star.
haydn TWRP is not working: https://androidfilehost.com/?fid=14943124697586336730 remember: fastboot boot twrp-3.5.1-haydn.img ... although I tried everything, this TWRP just doesn't work - you boot right into the system after the fastboot boot step. Tried everything I can think of, always boots to system. Guess it's still a waiting game...
mslezak said:
Yes they released source there are just no instructions anywhere on how to build the kernel #1 and the GKI modules #2 and combine into a build.
Click to expand...
Click to collapse
Just one more question since this has been bothering me; what would be needed from Xiaomi? Instructions on how to build the kernel? Have they ever done this before about something like this? Like, is it a reasonable thing to ask them to release instructions? I know some people working there, I could forward the request.
Well I've run this by many many devs.
The issue with the GKI 1.0 and ACK is that you must download the entire AOSP build environment and make sure it works with the separate ACK (Android Common Kernel) and then properly loads the GKI modules from a separate build step. Together they make 1 kernel. This ACK per Linux Version and across vendors should be AOSP and identical. Therefore, a single build should work across all Linux 5.4 based kernels.
But... most devs are just skipping AOSP and inlining the GKI modules so they can have 1 build step. Not exactly what Google intended, but way less storage, and you can use Clang or GCC or whatever you want.
I think what OEMs are posting is merely the entire combined source code, no 2 step build process. Only a single file for the QGKI config portion. Which won't build your GKI modules as it comes. A build script for this (Q)GKI portion would be extremely beneficial.
If any OEM walks through the build steps 1) how to create only the QGKI modules and 2) how to assemble them into 1 kernel with the ACK, that would be great for devs. It just hasn't happened. So we get hacked kernels for the 888 running Linux 5.4.
The Generic Kernel Image (GKI) project | Android Open Source Project
source.android.com
mslezak said:
Well I've run this by many many devs.
The issue with the GKI 1.0 and ACK is that you must download the entire AOSP build environment and make sure it works with the separate ACK (Android Common Kernel) and then properly loads the GKI modules from a separate build step. Together they make 1 kernel. This ACK per Linux Version and across vendors should be AOSP and identical. Therefore, a single build should work across all Linux 5.4 based kernels.
But... most devs are just skipping AOSP and inlining the GKI modules so they can have 1 build step. Not exactly what Google intended, but way less storage, and you can use Clang or GCC or whatever you want.
I think what OEMs are posting is merely the entire combined source code, no 2 step build process. Only a single file for the QGKI config portion. Which won't build your GKI modules as it comes. A build script for this (Q)GKI portion would be extremely beneficial.
If any OEM walks through the build steps 1) how to create only the QGKI modules and 2) how to assemble them into 1 kernel with the ACK, that would be great for devs. It just hasn't happened. So we get hacked kernels for the 888 running Linux 5.4.
The Generic Kernel Image (GKI) project | Android Open Source Project
source.android.com
Click to expand...
Click to collapse
OK, I see. Thank you for your reply. I am going to see how far I can get this. Tnx again!