3.18.63 and 4.4.79 have been released and updated on our CAF organization: https://plus.google.com/+NathanChancellor/posts/2CrBqMZYNzX
[REFERENCE] How to get an Android kernel up to date with linux-stable
- Thread starter nathanchance
- Start date
The CAF tag I used is LA.BR.1.2.9_rb1.22. I'm on Linux version 3.10.89 right now.
Last edited:
You can try using the LA.BR.1.3.6_rb1.20 branch as it is the closest to your version (3.10.84).
Great thread @The Flash!
With all the hobbyist/amateur kernel developers on here (my self included) this is arguably one of the most useful threads started on XDA
I'm upstreaming the Sony Z5Compact Copyleft kernel, which is currently 2 years behind the 3.10 branch at version 84. Iv got to 3.10.90 before my first fail to boot..... Now I just have to apply all 50 patches from v89 - v90, build and boot each one individually, to find out which commit causes the issue....
So far all of my cherry-pick conflicts (5 so far) are from Sony's devs fixing the same issues in a different way without pushing it up stream.
Thanks again!
With all the hobbyist/amateur kernel developers on here (my self included) this is arguably one of the most useful threads started on XDA
I'm upstreaming the Sony Z5Compact Copyleft kernel, which is currently 2 years behind the 3.10 branch at version 84. Iv got to 3.10.90 before my first fail to boot..... Now I just have to apply all 50 patches from v89 - v90, build and boot each one individually, to find out which commit causes the issue....
So far all of my cherry-pick conflicts (5 so far) are from Sony's devs fixing the same issues in a different way without pushing it up stream.
Thanks again!
Bisecting is exactly what that is for
Depending on the nature of the issue, I will sometimes revert the CAF/OEM fix in favor of upstreams. It's a judgement call.
@The Flash Hi, i did everything allright, upstreamed the kernel to one point, compiled, I got my kernel image, and now im stuck at repacking. I tried with 3 different mktools, and they all give the same errors. What i should do now ?
Yes, both files mkbootimg and mkbootfs are in the master folders, idk why its saying that they arent, or idk what is the problem here ...
]
EDIT: i upstreamed, flashed and phone doesnt detect simcard anymore
Yes, both files mkbootimg and mkbootfs are in the master folders, idk why its saying that they arent, or idk what is the problem here ...
]
EDIT: i upstreamed, flashed and phone doesnt detect simcard anymore
Last edited:
The latest upstream made my phone unable to connect to PC as USB storage device. The culprit might be f_mass_storage.c since they were updated, and there are more commits than the one in googlesource. Is it safe to revert all of them?
@The Flash this commit causes my leeco le pro 3 to bootloop on the splash screen
https://github.com/upstream-caf-kernels/msm-3.18/commit/e9838f9ec321bbdea48facbbe04922e78d5234c0
Hope you get freed from your speed force prison soon xD
https://github.com/upstream-caf-kernels/msm-3.18/commit/e9838f9ec321bbdea48facbbe04922e78d5234c0
Hope you get freed from your speed force prison soon xD
Go a version ahead to 3.18.38 as there is a fix for that.
Sent from my Nexus 6P using XDA Labs
I've been trying to find a guide to patch my adbd (im on Nougat 7.0, 3.18.35) to have it run as adb root with no luck so far. Upstreaming my kernel would make this harder so this would have to wait, right? Would upstreaming it break the insecured/patched adbd?
adbd is not a part of the kernel, it is in system/core, so no, upstreaming shouldn't affect that.
Getting this error when building now:
Using source from here: https://gitlab.com/Faustian_Divergence/android_kernel_xiaomi_msm8953_new
Code:
CC init/version.o
LD init/built-in.o
drivers/built-in.o: In function `mass_storage_function_cleanup':
/home/severus/Underscore/android_kernel_xiaomi_msm8953/drivers/usb/gadget/android.c:2870: undefined reference to `fsg_sysfs_update'
/home/severus/Underscore/Downloads/android_kernel_xiaomi_msm8953/drivers/usb/gadget/android.c:2870:(.text+0x363bc8): relocation truncated to fit: R_AARCH64_CALL26 against undefined symbol `fsg_sysfs_update'
drivers/built-in.o: In function `mass_storage_function_init':
/home/severus/Underscore/Downloads/android_kernel_xiaomi_msm8953/drivers/usb/gadget/android.c:2846: undefined reference to `fsg_sysfs_update'
/home/severus/Underscore/Downloads/android_kernel_xiaomi_msm8953/drivers/usb/gadget/android.c:2846:(.text+0x363dc8): relocation truncated to fit: R_AARCH64_CALL26 against undefined symbol `fsg_sysfs_update'
Makefile:955: recipe for target 'vmlinux' failed
make: *** [vmlinux] Error 1Using source from here: https://gitlab.com/Faustian_Divergence/android_kernel_xiaomi_msm8953_new
Last edited:
All I can do is recommend a git bisect as I have not much to go on with that error.
Sent from my Nexus 6P using XDA Labs
Similar threads
Top Liked Posts
-
There are no posts matching your filters.
-
162Introduction
Hello everyone! This will be a thread to assist people with getting their device's Android kernel up to date with the latest linux-stable tag from kernel.org. This process will henceforth be referred to as "upstreaming". This thread will assume that you understand how to compile a kernel as well as working with git if you are looking to do this. If you need assistance with that, please refer to one of the following threads (or search for others):
Information
Read through the linux-stable-notes README for all the relevant information!
Custom kernels that are upstreamed
ROMs with upstreamed kernels built in
Receiving help
I am more than happy to answer any questions regarding this process in this thread. You can use the repos I have merged linux-stable into as a reference for solving conflicts and seeing how the merge commit style looks. Additionally, if you feel any information is missing or can be improved upon, please let me know! I want this to be a collaborative effort.
Good luck!19I am going to start announcing when new kernel versions are available so people don't need to subscribe to the mailing list if they don't want to. So with that:
Linux 3.10.107 (announcement, tree)
Linux 3.18.59 (announcement, tree)
Linux 4.4.75 (announcement, tree)
I have also included a section for publicly available upstreamed kernels so users can choose one if they would like. They'll be broken up by device. Please suggest either your kernel or one you know is up to date, I'll be adding to some I already know of later today.15Thanks to the collaborative effort between @elginsk8r, @MSF Jarvis, @akhilnarang, and @Mazda, we have upstreamed the latest CAF branches for both 3.10 and 3.18:
https://github.com/upstream-caf-kernels
We will try to keep these up to date as best we can. 4.4 will be done in due time but not too many currently available devices are using it just yet so it's not of the highest priority. Feel free to either cherry-pick or pull these as needed. Enjoy!14
Yup, git pull is shorthand for git fetch && git merge, so it's fetching the tag then merging it. Definitely quicker if you are going fast.
To everyone, I have added section 5 and 6 under tips and tricks, going over how to easily get notified of upstream updates and known problem commits to help you in the journey upwards!14Another note I want to mention, since I see some developers missing the entire point of this and think I am attacking them/their work.
This whole thread is purely for educational purposes. I wrote this up because I saw a lot of big name developers frowning down upon and giving users a false impression of upstream Linux. This is absolutely nothing to fear. Doing an upstream bring up properly does not degrade performance, stability, or security in any way. I would love to see every developer upstream their kernels, trust me; however, I understand that for developers who support several devices, that can take a LOT of time.
Based on the latest CAF releases:
3.10.84 to 3.10.107 is around 1800 commits
3.18.31 to 3.18.61 is around 1650 commits
4.4.21 to 4.4.77 is about 2450 commits
That's a lot to test and make sure everything is working. I am perfectly okay with developers saying "no, this process isn't for me" (especially if they do try to look upstream for fixes that are critical) but only on the condition they know what they are missing out on. That's why this thread exists. Please, feel free to share this with your kernel developer/thread; I like to think I made the case extremely convincing (especially since a lot of this is backed by Greg Kroah-Hartman, who gets paid to do all of this) but don't harass them if they don't agree with it. These guys are very smart and know what is best for them (even if I or you don't agree with them).
