Dear all,
Did anyone succeed extracting kernel configuration from I9000XXJP3? Kernel version is 2.6.32.9, the vermagic is "2.6.32.9 mod_unload ARMv7"
extract-ikconfig doesn't work on it.
I succeeded extracting a zImage gzipped payload, but it seems not to contain any configuration in it (see attached).
/proc/config.gz doesn't exist, Samsung open source package (downloaded from Samsung open source site) contains only Android 2.1 Eclair or previous versions.
My target is to build tun.ko and, eventually, ext3/ext4 modules to make them working in Samsung Galaxy S I9000 with rooted I9000XXJP3.
Any idea?
Without froyo source code or a good Samsung Kernel (es. for himem capable) I think is impossible to play good with theses beta roms.
Ciao
Any news? I need tun.ko for jp3 too..
I have tried to compile the 2.6.32.9 kernel editing the .config in 2.6.9, the module tun.ko is accepted by the device, but I get a kernel panic!
redsh said:
I have tried to compile the 2.6.32.9 kernel editing the .config in 2.6.9, the module tun.ko is accepted by the device, but I get a kernel panic!
Click to expand...
Click to collapse
Did you use the stock linux kernel? Or the common from android.kernel.org?
I'm trying the same thing actually. Isn't there any default config for the processor that might work? I tried with the config from .29 but when loading the module it says wrong format.
try to build 2.6.32 with the 2.6.29 config ("yes" all missing stuff)
turn on your galaxy, adb push the tun.ko
try to load it, it will say "missing symbols" blabla
find the config options that match those symbols, enable them, recompile, try again
Great to see some people who are hacking the kernel! Keep it up!
But I'm afraid it is not going to be as easy as dropping aries_rev03_defconfig as .config in a 2.6.32 kernel tree and doing 'make oldconfig'. That's because many of Samsung's changes have not been included in the newer mainline kernel versions yet.
Samsung added quite a lot of low-level board support for their dev boards (and for the SGS, of course), did some customization and added a few drivers which you will need to forward-port to the newer kernel.
Please have a look at this thread, in which I've started a breakdown of the Samsung patches against Android Eclair's 2.6.29 kernel.
The best course of action I think is to git clone Android's kernel from AOSP, checkout the android-2.6.29 branch, apply Samsung's patches to that, then attempt to rebase your tree to a newer kernel version. (Note, you may want to start with small steps, to get a feel for what you're up against )
Note that there probably will be lots of merge conflicts which you need to resolve, and after dealing with all those, you also have to make sure that everything else that's merged still works as expected, but at least that will show you the amount of work involved. You will basically be doing all the work that Samsung is doing right now for their kernel for FroYo. It will be interesting to follow their progress on the mailing lists and on IRC.
bilboa1 said:
try to build 2.6.32 with the 2.6.29 config ("yes" all missing stuff)
turn on your galaxy, adb push the tun.ko
try to load it, it will say "missing symbols" blabla
find the config options that match those symbols, enable them, recompile, try again
Click to expand...
Click to collapse
Thats exactly what I did, but I got wrong module format... which is a fatal error. I need to invest further in the used config... maybe i did not pick up the right one properly..
miki4242 said:
Great to see some people who are hacking the kernel! Keep it up!
But I'm afraid it is not going to be as easy as dropping aries_rev03_defconfig as .config in a 2.6.32 kernel tree and doing 'make oldconfig'. That's because many of Samsung's changes have not been included in the newer mainline kernel versions yet.
Samsung added quite a lot of low-level board support for their dev boards (and for the SGS, of course), did some customization and added a few drivers which you will need to forward-port to the newer kernel.
Please have a look at this thread, in which I've started a breakdown of the Samsung patches against Android Eclair's 2.6.29 kernel.
The best course of action I think is to git clone Android's kernel from AOSP, checkout the android-2.6.29 branch, apply Samsung's patches to that, then attempt to rebase your tree to a newer kernel version. (Note, you may want to start with small steps, to get a feel for what you're up against )
Note that there probably will be lots of merge conflicts which you need to resolve, and after dealing with all those, you also have to make sure that everything else that's merged still works as expected, but at least that will show you the amount of work involved. You will basically be doing all the work that Samsung is doing right now for their kernel for FroYo. It will be interesting to follow their progress on the mailing lists and on IRC.
Click to expand...
Click to collapse
Yes that would be a lot of work. Maybe its better to just wait until they release the kernel from froyo. I read that they release their opensource stuff rather fast. But just for adding a module like ext4 I don't think you need those patches, because imho they didn't touch the fs of the kernel. We just need an adaquate kernel config and adding modules should be possible.
Phlogiston said:
Thats exactly what I did, but I got wrong module format... which is a fatal error. I need to invest further in the used config... maybe i did not pick up the right one properly..
Click to expand...
Click to collapse
Make sure you're using the same kernel version number and build number, because samsung kernels do not have the option to load incorrect module versions
Yes i've set the subversion as well but without the patches from samsung its impossible it seems. We need to wait and hope that they release the froyo kernel sources soon....
bilboa1 said:
Make sure you're using the same kernel version number and build number, because samsung kernels do not have the option to load incorrect module versions
Click to expand...
Click to collapse
Are you sure they set CONFIG_MODVERSIONS? It's off in the downloadable sources.
Just compare the output of `modinfo -F vermagic <yourmodule>` to `modinfo -F vermagic <modulewhichloads>` or to /proc/version .
did this get solved yet? I google'd for the tun.ko file for my i9000 using jp3 but nothing yet... if you have a proper one for the MoDaCo version here please attach it! ~
I think I saw someone's post with tun.ko for FroYo beta somewhere in these forums, I mean i9000 Android Dev. One of the guys here has found a way to compile kernel for jp* here. I am sure.
I actually found it attached somewhere in the forum and it was 1,5447 mb big I think it was... but it still didn't work for me so I presumed the kernel or something must have been wrong.
Hi fellows,
I was thinking, why should we stuck on 2.6.35 kernel? If we want to have a possibility to update our devices, we must have our kernels up-to-date, right?
Last night, I did a couple of scripts that can generate patches from a git tree from a start commit until an end commit. And another that can apply these patches to another tree and list the ones that wasn't applied correctly.
So, I thought if I get all patches from, let's say, 2.6.38 android kernel_common tree from 2.6.35 version commit until the HEAD and apply it to our 2.6.35 kernel, then we should get something like a 2.6.38 kernel, am I right?! I know that our kernel tree is something like a Frankenstein tree, it is a 2.6.35 with some patches from newer kernels, but these patches we can just ignore.
Last time I try to update 2.6.35 to 2.6.36 using this approach, I saw that I need to apply more than 10k patches and it will take something like a day to do here in my machine. After, the patches that wasn't applied correctly should be check. So, firstly, I want to verify if anyone see something wrong in my approach, it will save me a lot of time
Let's discuss!!!
Thanks,
Ronan
Wont porting over Kernel 3.x a be a better thing
http://git.kernel.org/?p=linux/kernel/git/kgene/linux-samsung.git;a=summary
We can diff out this n our current source(2.6.35 GB Update2), and then patch
The overcome kernel is patched with latest patches (as far as I can remember) however it wont't boot therefore it had to be spoofed with 2.6.35 to get boot. Maybe Alterbridge86 can shed more light on this.
DarkPal said:
The overcome kernel is patched with latest patches (as far as I can remember) however it wont't boot therefore it had to be spoofed with 2.6.35 to get boot. Maybe Alterbridge86 can shed more light on this.
Click to expand...
Click to collapse
He's right. My kernel is patched up to 2.6.35.13 BUT I have to spoof the kernel version as 2.6.35.7. I suspect no matter what you patch the kernel to, it will always need to be spoofed because of Samsung's proprietary crap. The RFS drivers, amongst other things, are proprietary so we use the precompiled modules from an existing initramfs. Obviously Samsung is using the 2.6.35.7 source, so those modules will not load if the source is reporting a higher version than 2.6.35.7.
I imagine you could patch the kernel up to whatever you want, however the hard part is getting the patches. I don't believe an incremental patch exists to go from 2.6.35 to 2.6.36, 37, etc.
The only thing I think you could do is get full source trees for 2.6.35 and 2.6.36 and then do a manual diff on them, creating your own patch, then try and patch that in to the samsung kernel source.
alterbridge86 said:
I imagine you could patch the kernel up to whatever you want, however the hard part is getting the patches. I don't believe an incremental patch exists to go from 2.6.35 to 2.6.36, 37, etc.
The only thing I think you could do is get full source trees for 2.6.35 and 2.6.36 and then do a manual diff on them, creating your own patch, then try and patch that in to the samsung kernel source.
Click to expand...
Click to collapse
Patches used to be availabe before kernel.org went down, i have all patches for 2.6.32.x(incremental), or maybe they are still there are i just overlooked them
Telling my point again, porting Kernel 3.0 would be better than 2.6.3x
Alterbridge86,
I wrote a script that can take all patches from a kernel tree from a specifc commit up to HEAD. It uses the log and format-patch. So, it is easy now to clone android kernel common tree, switch to, let's say, 2.3.38 branch, and take all the patches from 2.6.35 release until HEAD.
I actually got it from 2.6.36 branch (+ 11000 patches). Basically, I have the incremental patch. Now i just need some time to apply it all using another script rejecting everything that can't be patched flawlessly, built it, and verify if this "2.6.36" version works.
Now, why doesn't try 3.0? Well, i need to test the concept. The number of patches to go from 2.6.35 to 2.6.36 is more than 11000. I just can imagine how many patches exist from 2.6.35 to 3.0. I just don't have the computational power here to do this...
Anyway, I am a little out of time by now. When I have news, i will post here.
Thanks for all the advices,
Ronan
Sent from my GT-P1000L using xda premium
Bringing the thread back to life
Over days, i've tried to port 3.0, but before that, we'll have to get MTD working with 2.6.35
Tree :
https://github.com/sgt7/p1000-kernel-cm9/tree/mtd
Kernel panic
<3>[ 0.418810] samsung-onenand s5pc110-onenand: cannot get clock
<4>[ 0.418936] samsung-onenand: probe of s5pc110-onenand failed with error -2
Logs : http://pastie.org/3275285 (Thanks @ Techomancer)
Any ideas ? (havent been able to focus on this lately due to studies)
Hello, any News on porting the new Kernel?
Sent from my GT-P1000 using XDA
Hello,
In order to debug the bluetooth problem, I'm trying to compile a kernel for the Captivate Glide. I can't get it to boot properly.
@steadfasterX 's SediROM is the highest-version ROM I know of on which bluetooth works so I've got it installed (version 2.1.2) on the phone.
The steps I've taken so far:
1. I have sedikernel from https://github.com/steadfasterX/kernel_samsung_i927 at commit 088aa4109ad144c583f32da5ffba7bac200f0575
2. I copied /proc/config.gz from a running phone, which contains CONFIG_LOCALVERSION="-sediKERNEL-v1.0", ungzipped it and saved it as .config . Haven't changed anything in it yet.
3. I ran `make CROSS_COMPILE=arm-linux-gnueabihf- zImage` (also tried arm-none-eabi- (both from Debian Jessie) as well as ~/Android/Sdk/ndk-bundle/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/bin/arm-linux-androideabi- from the NDK)
4. I also made a TWRP backup of the boot partition, and used split_bootimg.pl from William Enck to split it into a kernel and initrd
5. I then ran mkbootimg with the initrd extracted in step 4 and the kernel obtained in step 3 in arch/arm/boot/zImage
6. Finally I flashed this new boot.img onto the phone with odin (Heimdall, specifically)
I see the white "SAMSUNG" logo (which is normal when booting) with a yellow-triangle with an exclamation mark in it. No further boot stage happens (no bootanim from sediROM). If I re-pack the original kernel with bootimg and flash that via odin, then after the next boot the yellow triangle changes into a blue one and sediROM boots correctly.
I'm a bit stuck at the moment, not sure what's going wrong there. Probably the kernel panics for some reason. Is there a way to read the kernel boot messages before android starts? Any help would be greatly appreciated.
Thanks!
Successful!
Thanks to a lot of help from @steadfasterX on IRC I finally managed to compile a working kernel
Here's a list of steps I had to follow in addition to https://github.com/steadfasterX/android_device_samsung_i927/blob/sediROM_CM-ICS/README.md :
1. as discussed on IRC, `repo sync` was not finding a branch. Needed to add
Code:
revision="refs/heads/ics-release"
in .repo/manifests/default.xml in the line with
Code:
CyanogenMod/android_vendor_qcom_opensource_v8
2. I removed the spurious comma in device/samsung/i927/cm.dependencies
3. In device/samsung/i927/, I changed vendorsetup.sh to point to vendorsetup.sh.nopatching rather than .patching (none of the patches could be applied)
4. lunch target was therefore the one with -eng, not -userdebug
5. I installed java 6 manually (didn't have sudo)
6. I installed schedtool
7. I adjusted the path to Java and my toolchain in build_sediROM.sh (taken from https://github.com/steadfasterX/android_buildtools )
8. I downgraded GNU make from 4.0 to 3.81
9. I installed linux-headers and created a symlink from /usr/include/asm-generic to /usr/include/asm
10. I changed the toolchain prefixes in device/samsung/i927/BoardConfig.mk to have 4.4.3 rather than 4.7
11. I built with
Code:
BUILDID=samsung/i927 LOKIFY=0 ../android_buildtools/build_sediROM.sh bootimage
12. I used the original ramdisk I had extracted from sediROM's boot image, rather than the one that was built (we discussed this on IRC too)
This got me a working kernel, yay
uukgoblin said:
Thanks to a lot of help from @steadfasterX on IRC I finally managed to compile a working kernel
Here's a list of steps I had to follow in addition to https://github.com/steadfasterX/android_device_samsung_i927/blob/sediROM_CM-ICS/README.md :
1. as discussed on IRC, `repo sync` was not finding a branch. Needed to add
Code:
revision="refs/heads/ics-release"
in .repo/manifests/default.xml in the line with
Code:
CyanogenMod/android_vendor_qcom_opensource_v8
2. I removed the spurious comma in device/samsung/i927/cm.dependencies
3. In device/samsung/i927/, I changed vendorsetup.sh to point to vendorsetup.sh.nopatching rather than .patching (none of the patches could be applied)
4. lunch target was therefore the one with -eng, not -userdebug
5. I installed java 6 manually (didn't have sudo)
6. I installed schedtool
7. I adjusted the path to Java and my toolchain in build_sediROM.sh (taken from https://github.com/steadfasterX/android_buildtools )
8. I downgraded GNU make from 4.0 to 3.81
9. I installed linux-headers and created a symlink from /usr/include/asm-generic to /usr/include/asm
10. I changed the toolchain prefixes in device/samsung/i927/BoardConfig.mk to have 4.4.3 rather than 4.7
11. I built with
Code:
BUILDID=samsung/i927 LOKIFY=0 ../android_buildtools/build_sediROM.sh bootimage
12. I used the original ramdisk I had extracted from sediROM's boot image, rather than the one that was built (we discussed this on IRC too)
This got me a working kernel, yay
Click to expand...
Click to collapse
its really long guide. You have to replace kernel modules, wifi wouldnt work without modules.
Imho best is compile kernel, modules, replace zImage + modules in this
file and flash from recovery. You dont need to compile whole android just the kernel, you dont need java and other special things, this method is much more faster.
bubor said:
its really long guide. You have to replace kernel modules, wifi wouldnt work without modules.
Imho best is compile kernel, modules, replace zImage + modules in this
file and flash from recovery. You dont need to compile whole android just the kernel, you dont need java and other special things, this method is much more faster.
Click to expand...
Click to collapse
You are right for the kernel but we discussed on IRC to build ram disk too then you need all that stuff but most important he wants to build CM too.
However I'm always a fan of making it right so building a ram disk is the correct way of creating a bootimage. But to start its ok replacing kernel and modules first.
.
Sent from my LG-H815 using XDA Labs
steadfasterX said:
Sent from my LG-H815 using XDA Labs
Click to expand...
Click to collapse
correct way is rebuild whole android, and reflash everything, wipe data/cache
You dont want to make new ramdisk without replace system. System and ramdisk bound together.
I hope you find something, and fix bluetooth or voip audio.
good luck
bubor said:
correct way is rebuild whole android, and reflash everything, wipe data/cache
You dont want to make new ramdisk without replace system. System and ramdisk bound together.
I hope you find something, and fix bluetooth or voip audio.
good luck
Click to expand...
Click to collapse
Well no and yes. I build kernels including ram disk without building system. This is not necessary as long both are compatible.
But you are right that booting fails when system and ram disk are not compatible so the correct way depends
However @uukgoblin I hope you get the fix you want :fingers-crossed:
Sent from my LG-H815 using XDA Labs
steadfasterX said:
Well no and yes. I build kernels including ram disk without building system. This is not necessary as long both are compatible.
But you are right that booting fails when system and ram disk are not compatible so the correct way depends
However @uukgoblin I hope you get the fix you want :fingers-crossed:
Sent from my LG-H815 using XDA Labs
Click to expand...
Click to collapse
would you tell me an example when kernel doesnt compatible with ramdisk but kernel does compatible with system?
bubor said:
would you tell me an example when kernel doesnt compatible with ramdisk but kernel does compatible with system?
Click to expand...
Click to collapse
No no my friend that's not what I said. :cyclops:
I meant if you have CM 9 as system you can build a boot image which includes kernel and a CM9 based ram disk without building cm9 system again. :laugh:
.
Sent from my LG-H815 using XDA Labs
Modesty - a modest custom kernel for the Samsung Tab S4
Modesty aims to provide a mildly appealing and reasonable alternative to the stock 4.4.78 kernel that comes with The Tab S4. In its pursuit of being both mildly appealing and reasonable, it will eschew features that could compromise device stability, whilst gleefully embracing low-risk, self-contained enhancements. In other words, your lowest expectation should be that this kernel will be at least as stable as the stock kernel.
Since there are currently no other custom kernel projects supporting the Tab S4, there isn't really any previous device-specific work to build on. Development of this kernel is therefore likely to be slow and steady.
"Why is this kernel called Modesty? That's crap! Why not Wolverine, Intrepid or Jupiter?"
Because it's just a operating-system kernel, not a turbo-charged supercar or a mission into outer space. Even as operating-system kernels go, this one is pretty dull. Besides, I'm a weary curmudgeon in his fifties, not a teenager.
This project has the modest aim of modestly enhancing the pleasure you derive from your Tab S4 and is therefore modestly named Modesty.
Key characteristics
Supports both the wi-fi only (T830) and wi-fi/LTE (T835) models.
Forked from Samsung's pristine kernel source code (Linux 4.4.78 for ARGH firmware at time of launch).
Regular merging of the upstream Linux kernel's linux-4.4.y branch (4.4.161 at time of public launch).
Regular merging of Samsung's updates to its modified kernel source as they are made available.
Includes @savoca's KCAL advanced colour/gamma control driver.
Includes @flar2's sound control driver to manage headphone and microphone gain.
Disables a huge amount of tracing and logging features inexplicably left enabled by Samsung in the stock release kernel. These debugging features have no place outside engineering builds.
Packed into a boot.img (boot image) taken directly from Samsung's latest stock firmware and kept as close to the original as possible. No obscure boot-time kernel configuration is stashed away here, and no changes are made to any other part of the file system at either install time or run time.
Provides a fully automated installer, with the option of interactive installation to allow manual selection of features and the ability to automatically root the device with Magisk in the post-installation phase.
Includes WireGuard VPN support (version 0.0.20180818 at the time of public launch), which will be updated as available.
Includes @Lord Boeffla's generic kernel wakelock blocker. The conservative default block-list is: qcom_rx_wakelock and NETLINK.
Utilises Westwood+ TCP congestion algorithm by default.
Includes Veno TCP congestion algorithm.
SELinux operates in enforcing mode and cannot be dynamically switched to permissive mode.
FAQ
Q. Is this kernel still actively developed?
A. No. The final ianmacd release was v1.0.0 on 21st November 2019 and no-one else has picked up maintenance of the project.
Q. Will this kernel also run on Android 9.0 (Pie) devices?
A. No. Modesty targets Android 8.1 (Oreo) and there was never an intention to update it for 9.0 (Pie).
Q. Can I overclock or underclock the CPU using this kernel?
A. No.
Q. How does interactive installation mode work?
A. If the ZIP file name contains the string _interactive or a dot-file called .modesty_interactive is present in the root of the external SD card, interactive installation mode is triggered. Please note that this mode overrides any selections implied by the archive name or the presence of dot-files on the file-system.
In interactive mode, you will be asked whether to root the device afterwards with Magisk. Selections are made using the Volume buttons. Just follow the on-screen prompts.
Q. Can I safely block wakelock X?
A. Perhaps. However, unless you know what a particular wakelock does and are certain that it is causing an actual problem on your device, I suggest you leave it alone.
Q. Why is this kernel labelled beta? Is it safe to use? And who are you, anyway? Can you be trusted?
A. My T830 has been running this kernel every day since I first rooted it, and I can therefore personally vouch for its stability on this model.
A couple of users have reported Modesty running well on the T835. Initially, it was reported that the kernel did not boot on this model, but after trying several test kernels, the user in question discovered that his machine had a non-standard firmware installation. Once this situation was remedied, Modesty booted and worked as designed.
As the person who built the kernel, I know exactly what's in it, and therefore the only risk I'm exposing myself to when I run it is that of my own incompetence. That's not true for you, however, and you should exercise due caution and at least pause for a moment to consider what you are installing, and the far-reaching powers you are about to grant this unaudited code over your device. Although I link to the source code below, you have only my word for it that this bears any resemblance to the kernel actually provided in the installer.
There are likely to be many iterations of this kernel before it sees a 1.0 release. Features may be added or removed along the way, although there is no clear roadmap at this point in time. Development will go where the needs of the users take it.
Please see the Installation section below for an important note regarding the use of this kernel in combination with stock (i.e. unmodified) Samsung firmware.
Q. Can I safely root this kernel?
A. Of course. What use would it be if you couldn't? I recommend Magisk for the task. It has a few minor issues, but as a project is very much alive, something that cannot be said about its peers. Magisk has arguably now established itself as the de facto root solution for Android devices.
It just so happens that I also produce my own builds of Magisk, which you are welcome to use. These are release builds (as opposed to debugging builds), produced from my own fork of @topjohnwu's original source, often augmented with patches. You can use anyone's builds, though.
Again, these builds work for me on various Samsung devices, but they are unofficial and you should approach them with fitting caution.
Q. Can I install Magisk at the same time as Modesty?
A. Yes. The Modesty installer allows you to automatically root your device with Magisk following installation of Modesty..
To make use of this facility, either rename the Modesty zip file to contain the string _magisk or create a file called .modesty_magisk in either the root of your external SD card or in the standard Download directory of the internal SD card. Alternatively, you can utilise interactive installation mode. See above for details.
If any of these trigger conditions is met, the installer will look in the standard internal Download directory as well as in ./Magisk (if present) on the external SD card (if present) for a suitable Magisk zip file to install. Preference is given to versioned files matching the glob Magisk-v*, in which case the latest according to lexical sort order will be used. If none is found, the installer then looks for unversioned release builds (e.g. official Canary channel release builds) called magisk-release.zip in the same locations, selecting the one with the most recent timestamp. If none is found, the installer will then try to find unversioned debug builds (e.g. official Canary channel debug builds) called magisk-debug.zip, again picking the one with the most recent timestamp. Finally, the installer falls back to looking for the most recent file called Magisk.zip or magisk.zip. If still no files have been found by this stage, chaining of Magisk is abandoned.
For example:
Code:
star2lte:/ $ ls -l /storage/0000-0000/.modesty_magisk
-rwxrwx--x 1 root sdcard_rw 0 2018-09-15 14:31 /storage/0000-0000/.modesty_magisk
star2lte:/ $ ls /storage/0000-0000/Magisk/Magisk-* | tail -n 3
/storage/0000-0000/Magisk/Magisk-v17.2-2018091001-ianmacd.zip
/storage/0000-0000/Magisk/Magisk-v17.2-2018091201-ianmacd.zip
/storage/0000-0000/Magisk/Magisk-v17.2-2018091501-ianmacd.zip
When you flash the Modesty archive in TWRP, the most recent version of Magisk that could be found will now be used to automatically root your kernel, i.e. Magisk-v17.2-2018091501-ianmacd.zip in this example.
Q. Why doesn't Modesty have its own Telegram group?
A. Because my experience of Android-themed Telegram groups is that they invariably degenerate into seething cesspits of rudeness, ignorance, superstition and — on a good day — pseudo-science. I don't wish to police such a den of iniquity. Of course, it's a free world (or so I still like to kid myself), so you are at liberty to create your own Telegram group for Modesty if you wish. Just please don't invite me to it.
Building
Building the kernel from source is beyond the scope of this document. If you want to build this kernel from scratch, for example to change its configuration, start with this handy reference tailored to building kernels for Android.
Download
See posting #2 in this thread for links to the latest and all previous versions.
Known Issues
Bluetooth HID (input) devices do not work.
Versions 0.99.11 to 0.99.22 contained a bug that caused Bluetooth HID (input) devices, such as mice, keyboards and gamepads, not to function. They could be paired with the tablet, but their input was not recognised. This bug was finally traced and fixed in 0.99.23.
Installation
Make a back-up of your existing boot partition using the custom recovery environment provided by TWRP. If your device doesn't yet have TWRP, you will need to install it first. Then, use it to flash the Modesty ZIP file. The boot image will automatically be installed in the boot partition of your device.
If your device has unmodified Samsung firmware, you will encounter problems with Bluetooth (namely delayed initialisation and forgotten pairings) after installing this or any other custom kernel. To remedy this, you will need to patch your system with modified libsecure_storage.so libraries. Some custom kernel installers actually install these without telling you, overwriting your system libraries and transparently circumventing the problem before you can run into it. This approach necessarily modifies your device's file-system, however, and that may not be what you want. At the very least, the user should be made aware what is happening to his device.
For this reason, I have instead prepared a companion Magisk module that achieves the same goal without modifying the file-system. This will allow you to run a custom kernel (not just this one, but any custom kernel) on pristine stock firmware without any Bluetooth issues. The module can be found in the official Magisk module repository, accessible from Magisk Manager on your tablet. If you install this Magisk module, you may wish to also disable the secure_storage_daemon by editing /system/etc/init/secure_storage_daemon.rc (change start to stop), as it no longer serves a purpose.
In a similar vein, you may encounter authentication errors when connecting to wireless networks after installing this or any other custom kernel. This problem is not serious and easily remedied by re-entering your passphrase for the networks you use.
Finally, if SecurityLogAgent notifies you that unauthorised actions have been detected, do not be alarmed. This is a normal consequence of having installed a custom kernel. You may wish to disable SecurityLogAgent to avoid being repeatedly notified..
Whilst the above issues are the only ones you can expect to encounter when running this kernel vs. the stock Samsung kernel, they may sound like more trouble than they're worth. In that case, you might be happier just sticking to Samsung's stock kernel. The company supplies a perfectly good kernel straight from the factory.
Configuration
You are encouraged to use either @morogoku's excellent MTweaks (a modified version of Kernel Aduitor) or @flar2's EX Kernel Manager to manage the features provided by this kernel.
Source code
Modesty's GitHub repository.
References
A useful guide to CPU governors, I/O schedulers (and more).
For more information on the some of the individual schedulers included in this kernel, you can also look under Documentation/block in the kernel source.
The WireGuard user guide, control app, home page and source code.
Credits
Thank you to everyone in the Linux kernel universe for getting us this far. Within the Android development community, I am grateful to the following people for their time-saving contributions:
@osm0sis for Android Image Kitchen, which has saved me a huge amount of work in packing and unpacking boot images.
An honorary mention must go to @Chainfire, the extent of whose benefaction to the Android community is still not fully understood or appreciated in some quarters.
Change log
v1.0.0 (final ianmacd release) (2019-11-21)
Kernel proclaimed stable. Version number incremented. No code changes since v0.99.49.
v0.99.49 (2019-11-16)
Updated to Linux 4.4.202.
v0.99.48 (2019-11-13)
Updated to Linux 4.4.201.
v0.99.47 (2019-11-11)
Updated to Linux 4.4.200.
v0.99.46 (2019-11-06)
Updated to Linux 4.4.199.
v0.99.45 (2019-10-31)
Updated to Linux 4.4.198.
v0.99.44 (2019-10-19)
Updated to Linux 4.4.197.
v0.99.43 (2019-10-08)
Updated to Linux 4.4.196.
v0.99.42 (2019-10-07)
Updated to Linux 4.4.195.
v0.99.41 (2019-09-22)
Updated to Linux 4.4.194.
v0.99.40 (2019-09-16)
Updated to Linux 4.4.193.
v0.99.39 (2019-09-11)
Updated to Linux 4.4.192.
Fixes unavailability of external SD card in Modesty 0.99.38.
v0.99.38 (2019-09-08) Release withdrawn (External SD card unavailable)
Updated to Linux 4.4.191.
v0.99.37 (2019-08-26)
Updated to Linux 4.4.190.
v0.99.36 (2019-08-12)
Updated to Linux 4.4.189.
v0.99.35 (2019-08-07)
Updated to Linux 4.4.188.
v0.99.34 (2019-08-05)
Updated to Linux 4.4.187.
v0.99.33 (2019-07-23)
Updated to Linux 4.4.186.
v0.99.32 (2019-07-12)
Updated to Linux 4.4.185.
v0.99.31 (2019-06-28)
Updated to Linux 4.4.184.
v0.99.30 (2019-06-22)
Updated to Linux 4.4.183.
v0.99.29 (2019-06-18)
Updated to Linux 4.4.182.
v0.99.28 (2019-06-12)
Updated to Linux 4.4.181.
v0.99.27 (2019-05-17)
Updated to Linux 4.4.180.
v0.99.26 (2019-04-28)
Updated to Linux 4.4.179.
v0.99.25 (2019-04-07)
Updated to Linux 4.4.178.
v0.99.24 (2019-03-26)
Updated to Linux 4.4.177.
Build only the latest revision of the DTB.
v0.99.23 (2019-03-02)
Fixed bug, introduced in v0.99.11, that caused input from Bluetooth HID devices, such as keyboards, mice and gamepads to be ignored.
v0.99.22 (2019-02-23)
Updated to Linux 4.4.176.
v0.99.21 (2019-02-20)
Updated to Linux 4.4.175.
v0.99.20 (2019-02-11)
Updated to Linux 4.4.174.
v0.99.19 (2019-02-08)
Updated to Linux 4.4.173.
v0.99.18 (2019-01-26)
Updated to Linux 4.4.172.
v0.99.17 (2019-01-17)
Updated to Linux 4.4.171.
v0.99.16 (2019-01-13)
Updated to Linux 4.4.170.
v0.99.15 (2018-12-30)
Rebased on ARK4 kernel source code and boot images.
v0.99.14 (2018-12-23)
Updated to Linux 4.4.169.
Merged four more UPSTREAM commits from android-4.4 kernel branch.
v0.99.13 (2018-12-13)
Updated to Linux 4.4.167.
Merged selected BACKPORT and UPSTREAM commits from android-4.4 kernel branch.
v0.99.12 (2018-12-05)
Updated to Linux 4.4.166.
Realtek USB Ethernet driver upgraded from v2.08.0 to v2.10.00.
v0.99.11 (2018-11-29)
Updated to Linux 4.4.165.
KCAL advanced colour/gamma control driver optimisation.
Added @flar2's sound control driver for controlling headphone and microphone gain. (Configure with MTweaks or EX Kernel Manager).
v0.99.10 (2018-11-21)
Updated to Linux 4.4.164.
Added KCAL advanced colour/gamma control driver. (Configure with MTweaks or EX Kernel Manager).
Lots of tracing and debug logging disabled, further reducing kernel size.
CONFIG_DISPLAY_USE_INFO
CONFIG_SEC_DISPLAYPORT_LOGGER
CONFIG_FB_MSM_MDSS_XLOG_DEBUG
CONFIG_SEC_FILE_LEAK_DEBUG
CONFIG_SEC_DEBUG_USER
CONFIG_SEC_DEBUG_SUMMARY
CONFIG_SCSI_UFSHCD_CMD_LOGGING
CONFIG_MSM_SMEM_LOGGING
CONFIG_PROFILING
CONFIG_DEBUG_INFO
CONFIG_SCHED_DEBUG
CONFIG_SEC_PM_DEBUG
CONFIG_CORESIGHT
Built as monolithic kernel (i.e. without CONFIG_MODULES).
Built as relocatable code (CONFIG_RELOCATABLE_KERNEL).
Assembler symbols stripped (CONFIG_STRIP_ASM_SYMS set).
Embedded kernel config (reported via /proc/config.gz) now falsely reports stock settings to allow disabling of superfluous kernel features that otherwise cause grave Android System warning on boot.
v0.99.9 (2018-11-13)
Rebased on ARJ3 kernel source code and boot images.
v0.99.8 (2018-11-10)
Updated to Linux 4.4.163.
More than 100 fixes applied from upstream AOSP android-4.4 and android 4.4-o branches.
Lots of tracing and debug logging disabled:
CONFIG_IPC_LOGGING (debug logging for IPC drivers)
CONFIG_QCOM_RTB (register tracing)
CONFIG_TRACER_PKT (for tracing IPC protocols)
CONFIG_FTRACE (kernel tracing infrastructure)
CONFIG_CPU_FREQ_SWITCH_PROFILER (CPU frequency switch profiler)
CONFIG_TRACING_EVENTS_GPIO (traces GPIO subsystem)
Fixes to allow kernel to build when above logging and tracing options are disabled.
v0.99.7 (2018-10-30)
Rebased on ARH5 kernel source code.
Reworked the v4l2 fix that restores liboemcrypto-dependent apps to working state.
v0.99.6 (2018-10-28)
v4l2 fixes to restore liboemcrypto-dependent apps to working state.
v0.99.5 (2018-10-21)
Updated to Linux 4.4.162.
v0.99.4 (2018-10-19)
Initial public release, based on Linux 4.4.161.
v0.99.3
Internal build, based on Linux 4.4.160.
v0.99.2
Internal build, based on Linux 4.4.159.
v0.99.1
Initial internal build, based on Linux 4.4.78.
It begins! Awesome to finally see a custom kernel for the Tab S4.
I want to test for you once I can get root back
ianmacd said:
Change log
v0.99.4 (2018-10-19)
Initial public release, based on Linux 4.4.161. Caution: This kernel remains completely untested on the T835.
v0.99.3
Internal build, based on Linux 4.4.160.
v0.99.2
Internal build, based on Linux 4.4.159.
v0.99.1
Initial internal build, based on Linux 4.4.78.
Click to expand...
Click to collapse
Flashed on T835 device doesnt even get past the Boot (custom device) screen is there a way to get logs without using a computer?
Sent from my Samsung SM-G950F using XDA Labs
dr460nf1r3 said:
Flashed on T835 device doesnt even get past the Boot (custom device) screen is there a way to get logs without using a computer?
Click to expand...
Click to collapse
I'll look into it this evening. I'm just about to get off a plane.
Were there any errors when installing? Was your device properly detected as a T835?
Sent from my SM-G965F using XDA Labs
ianmacd said:
I'll look into it this evening. I'm just about to get off a plane.
Were there any errors when installing? Was your device properly detected as a T835?
Click to expand...
Click to collapse
Installing just fine, correctly detected as well.
Sent from my Samsung SM-G950F using XDA Labs
dr460nf1r3 said:
Installing just fine, correctly detected as well.
Click to expand...
Click to collapse
I've gone through the ramdisk of the T835's boot image with a fine-tooth comb and can find nothing untoward. I also verified that I properly removed the dm-verity flag from the T835's device tree.
There are actually very few source code differences between the T830 and T835. Both can be built from a single tree. The only differences lie in the kernel config file and the device tree, but I am building with the default T835 configuration, and with the proper device tree for that device.
Let's try at least ruling out my installer code. Please image-flash this new boot image[/i] to your device and tell me if it boot-loops. If it does, my installer isn't the problem, because it's only used for a ZIP flash. I've already checked the installer code and can't see any bugs, so I don't think the issue lies there.
Can you also please tell me which version of the firmware your device is running? Possibly there's an issue there, too. Samsung has so far released the source to the ARGH kernel only. This seems to work fine on my ARH5 firmware, but it's uncertain whether it would still work on something based on ARI*, and I've seen that a couple of countries do now have ARI firmware available. Mind you, even if it wasn't compatible, it should still get as far as booting.
Anyway, please test that boot image and let me know your firmware version.
While I soldier on with the issues afflicting the T835 build, can anyone else verify the T830 build as working for them?
Don't be shy; I'm running it on my own device, so I'm certain that build boots.
ianmacd said:
While I soldier on with the issues afflicting the T835 build, can anyone else verify the T830 build as working for them?
Don't be shy; I'm running it on my own device, so I'm certain that build boots.
Click to expand...
Click to collapse
Sorry for the noob question if i flash this kernal on my SM-T835 will i lose DEX ? Sorry i'm not quite understanding what KERNEL does ? Thanks in advance!
N1NJATH3ORY said:
Sorry for the noob question if i flash this kernal on my SM-T835 will i lose DEX ? Sorry i'm not quite understanding what KERNEL does?
Click to expand...
Click to collapse
When the kernel is working as intended, you won't lose DeX, but it currently isn't even booting on the T835.
Only those who are able to assist in debugging the current boot failure should install the T835 build at this time.
The T830 build, on the other hand, is rock solid for me, and I encourage anyone who is capable of recovering from an unexpected bootloop to try it out.
Version 0.99.5 released.
This release updates the kernel to the latest upstream Linux.
T830 owners, install at will. T835 owners, beware: The previous release has been reported unbootable on this model, and this release is likely to be similarly afflicted. Investigations are ongoing. Until this issue is resolved, the whole project has been downgraded to alpha status.
Change log
Updated to Linux 4.4.162. Caution: This release is likely to cause a bootloop on the T835.
ianmacd said:
I've gone through the ramdisk of the T835's boot image with a fine-tooth comb and can find nothing untoward. I also verified that I properly removed the dm-verity flag from the T835's device tree.
There are actually very few source code differences between the T830 and T835. Both can be built from a single tree. The only differences lie in the kernel config file and the device tree, but I am building with the default T835 configuration, and with the proper device tree for that device.
Let's try at least ruling out my installer code. Please image-flash this new boot image[/i] to your device and tell me if it boot-loops. If it does, my installer isn't the problem, because it's only used for a ZIP flash. I've already checked the installer code and can't see any bugs, so I don't think the issue lies there.
Can you also please tell me which version of the firmware your device is running? Possibly there's an issue there, too. Samsung has so far released the source to the ARGH kernel only. This seems to work fine on my ARH5 firmware, but it's uncertain whether it would still work on something based on ARI*, and I've seen that a couple of countries do now have ARI firmware available. Mind you, even if it wasn't compatible, it should still get as far as booting.
Anyway, please test that boot image and let me know your firmware version.
Click to expand...
Click to collapse
Firmware Version ist arh5 and after flashing your img the device still constantly reboots on the start screen and doesnt even get to the boot Screen
Sent from my Samsung SM-G950F using XDA Labs
dr460nf1r3 said:
Firmware Version ist arh5 and after flashing your img the device still constantly reboots on the start screen and doesnt even get to the boot Screen
Click to expand...
Click to collapse
Thanks. That absolves the installer of any wrongdoing, at least.
Something is fundamentally wrong with the kernel for the T835.
The boot image is taken from stock firmware, and modified just enough to allow a custom kernel to boot. I very much doubt the problem lies there. A virtually identical image works for the T830.
The kernel config used is the one supplied by Samsung. The only modifications made to it are the same ones I made to the T830's.
I think my next step will be to produce a kernel built without downstreaming the 4.4.y Linux branch, so back to 4.4.78. If that works, it will indicate that an error affecting only the T835 was introduced during all of my merging of the upstream kernel.
I'll post again when I've built the kernel, which won't be for a few hours, as I'm on holiday at the moment.
Sent from my SM-G965F using XDA Labs
ianmacd said:
I think my next step will be to produce a kernel built without downstreaming the 4.4.y Linux branch, so back to 4.4.78. If that works, it will indicate that an error affecting only the T835 was introduced during all of my merging of the upstream kernel.
Click to expand...
Click to collapse
OK, @dr460nf1r3, please try this new T835 build.
This is rewound to 4.4.78, with just a few extra cherry-picked commits to enable it to build cleanly and boot without triggering dm-verity.
In other words, this kernel should be 99% identical to the one that shipped with the machine. This assumes that the source as supplied by Samsung was actually used to build the stock kernel.. They have been known to publish sources that don't match what's on the machine.
ianmacd said:
OK, @dr460nf1r3, please try this new T835 build.
This is rewound to 4.4.78, with just a few extra cherry-picked commits to enable it to build cleanly and boot without triggering dm-verity.
In other words, this kernel should be 99% identical to the one that shipped with the machine. This assumes that the source as supplied by Samsung was actually used to build the stock kernel.. They have been known to publish sources that don't match what's on the machine.
Click to expand...
Click to collapse
Thanks, flashing now will report back in a few minutes
Sent from my Samsung SM-G950F using XDA Labs
---------- Post added at 05:44 PM ---------- Previous post was at 05:17 PM ----------
ianmacd said:
OK, @dr460nf1r3, please try this new T835 build.
This is rewound to 4.4.78, with just a few extra cherry-picked commits to enable it to build cleanly and boot without triggering dm-verity.
In other words, this kernel should be 99% identical to the one that shipped with the machine. This assumes that the source as supplied by Samsung was actually used to build the stock kernel.. They have been known to publish sources that don't match what's on the machine.
Click to expand...
Click to collapse
Did not work still the same issue.. flashed via twrp to boot partition. Noob question, the boot backup i got is 64mb while your kernel hardly has 25.. whats going on here?
Sent from my Samsung SM-G950F using XDA Labs
dr460nf1r3 said:
Did not work still the same issue.. flashed via twrp to boot partition.
Click to expand...
Click to collapse
Well, that's disappointing. I was hoping that it was my screw-up, rather than Samsung's, but at this point almost everything I've done has been backed out and it still won't boot.
Noob question, the boot backup i got is 64mb while your kernel hardly has 25.. whats going on here?
Click to expand...
Click to collapse
Good question.
Your back-up is of the entire partition, including the area with no data on it, so you're getting a file the same size as the partition itself. My boot image contains just the data segment, so it's smaller.
If you pull the stock boot image from the AP file of Samsung's firmware, you'll see that it's a very similar size to mine (slightly smaller, actually):
Code:
$ unzip -p T835XXU1ARH5_T835OXM1ARH5_PHN.zip AP_T835XXU1ARH5_CL14008523_QB19263559_REV00_user_low_ship_MULTI_CERT_meta.tar.md5| tar xf - -O boot.img.\*lz4 | lz4 -dc > boot.img
$ ls -l boot.img
-rw-rw-r--. 1 ianmacd ianmacd 23593232 Oct 21 22:01 boot.img
$ file boot.img
boot.img: Android bootimg, kernel (0x8000), ramdisk (0x2000000), page size: 4096, cmdline (console=null androidboot.hardware=qcom user_debug=31 msm_rtb.filter=0x37 ehci-hcd.park=3 lpm_le)
So, what now?
I'll see which other minor changes I can back out, in an effort to arrive at a kernel built from source that is as close to stock as possible. If you're wondering Why doesn't he just build from pristine sources?, the answer is: Because Samsung's source code won't even build out of the box. Many of the kernel header files are simply not in the expected locations. Alas, this is a fairly common problem with Samsung's kernel source code releases.
I suspect the solution to this problem may actually lie in changes that have yet to be made, rather than changes made that need to be reverted. In other words, the T835 may require some kernel modifications or configuration that the T830 doesn't. Theoretically, a kernel compiled from Samsung's pristine sources should just work, but that's starting to look unlikely now.
I'm hoping that I can enable/disable a few further options in the kernel config, rebuild and produce a kernel that works for you. If, however, the problem is that the source itself is faulty, we may have to wait for a future release by Samsung to give us something that compiles into a working kernel.
But I don't intend to throw in the towel on the T835 just yet. There are still a few more things we can try.
ianmacd said:
Well, that's disappointing. I was hoping that it was my screw-up, rather than Samsung's, but at this point almost everything I've done has been backed out and it still won't boot.
Good question.
Your back-up is of the entire partition, including the area with no data on it, so you're getting a file the same size as the partition itself. My boot image contains just the data segment, so it's smaller.
If you pull the stock boot image from the AP file of Samsung's firmware, you'll see that it's a very similar size to mine (slightly smaller, actually):
So, what now?
I'll see which other minor changes I can back out, in an effort to arrive at a kernel built from source that is as close to stock as possible. If you're wondering Why doesn't he just build from pristine sources?, the answer is: Because Samsung's source code won't even build out of the box. Many of the kernel header files are simply not in the expected locations. Alas, this is a fairly common problem with Samsung's kernel source code releases.
I suspect the solution to this problem may actually lie in changes that have yet to be made, rather than changes made that need to be reverted. In other words, the T835 may require some kernel modifications or configuration that the T830 doesn't. Theoretically, a kernel compiled from Samsung's pristine sources should just work, but that's starting to look unlikely now.
I'm hoping that I can enable/disable a few further options in the kernel config, rebuild and produce a kernel that works for you. If, however, the problem is that the source itself is faulty, we may have to wait for a future release by Samsung to give us something that compiles into a working kernel.
But I don't intend to throw in the towel on the T835 just yet. There are still a few more things we can try.
Click to expand...
Click to collapse
Thanks for explaining everything for me. Id like to help you were i can but i dont have a computer by my hands right now for the next time
Sent from my gts4llte using XDA Labs
dr460nf1r3 said:
Thanks for explaining everything for me. Id like to help you were i can but i dont have a computer by my hands right now for the next time
Click to expand...
Click to collapse
Here's a new build to try. This one reverts the last few changes I originally made to the T835's kernel configuration before building. This is as close to stock as possible, whilst still being able to build.
If this doesn't boot, it pretty much means that Samsung has supplied code for the T835 that simply will not compile into a working kernel. At that point, we'll probably have to wait for updated sources. I already have a request pending with Samsung for the release of the BRI sources.
Just to be clear, the current status quo as I understand it is that the Samsung logo never starts to be written from left to right. You never get past the static screen with the device name and the word Custom. Is that correct?
ianmacd said:
Here's a new build to try. This one reverts the last few changes I originally made to the T835's kernel configuration before building. This is as close to stock as possible, whilst still being able to build.
If this doesn't boot, it pretty much means that Samsung has supplied code for the T835 that simply will not compile into a working kernel. At that point, we'll probably have to wait for updated sources. I already have a request pending with Samsung for the release of the BRI sources.
Just to be clear, the current status quo as I understand it is that the Samsung logo never starts to be written from left to right. You never get past the static screen with the device name and the word Custom. Is that correct?
Click to expand...
Click to collapse
Im sorry to tell you but this doesnt boot either in fact it doesnt even reboot the static screen. Your right about the current status quo sadly.
Sent from my Samsung SM-G950F using XDA Labs