• Introducing XDA Computing: Discussion zones for Hardware, Software, and more!    Check it out!

Development [Kernel][30.09.2021][Android 11] Kirisakura 1.2.0 for Asus ROG Phone 5 aka "Anakin"

Search This thread
nothing to do with CAF.

the lag issue was because of not building a QGKI kernel and therefore a config for msm_performance didn´t get enabled.
The crash issue was caused by a commit of a mmap backport series. I compared the code around that to the zenfone 8 and my kernel there. No crash there, but the rog5 does not like that one commit for whatever reason.

Thanks for reminding me about wireguard! :) I forgot to run the script to patch the kernel repo to include it before release.

I´m about to release an update so I´ll push my updated source.

Sorry for the crashes and lags 😇:rolleyes:

I was trying to give you a pass, but I can take it back if you didn't want it lol. Pretty sure I know which config you are talking about. I had wondered why it wasn't enabled, so that explains it.

The ROG Phone 5 is super finicky. It doesn't help that they released two (or three, depending where you're from) models and then immediately pushed out a second series with two more. All of them have slightly different hardware that Asus is trying to keep on a single code base.

Hm. I´m not the biggest gamer. Is there a game where I can easily check that?

I just tried these two videos with my google pixel buds:


It seems to be what I would expect of BT headphones there.

has a bunch of tactile (effect) UI components that make it easy to pick out the lag.
 
  • Like
Reactions: mikalovtch

Freak07

Recognized Developer / Recognized Contributor
Jan 2, 2011
5,068
16,083
I was trying to give you a pass, but I can take it back if you didn't want it lol. Pretty sure I know which config you are talking about. I had wondered why it wasn't enabled, so that explains it.

The ROG Phone 5 is super finicky. It doesn't help that they released two (or three, depending where you're from) models and then immediately pushed out a second series with two more. All of them have slightly different hardware that Asus is trying to keep on a single code base.



has a bunch of tactile (effect) UI components that make it easy to pick out the lag.

I pushed my source :) I took the pass don´t worry. :)

CONFIG_MSM_PERFORMANCE_QGKI needs to be explicitly enabled.
For some reason when extracting the defconfig from boot.img or from /proc/config.gz and using that to build the kernel it doesn´t get enabled.
It seems there´s a lot for me to be learned about CAFs QGKI build "system". There might be more surprises.

Yes, the ROGs are super finicky I completely agree.


I´ll check the game and bluetooth on my end. thanks :)
 
Last edited:

Freak07

Recognized Developer / Recognized Contributor
Jan 2, 2011
5,068
16,083
Update to 1.1.0

Hey guys and girls,


I hope everyone is having a good day or starting into one. Here´s the next update.
I forgot to include wireguard in the release. That´s now added.

Since we´re not building a QGKI kernel, but rather a standalone kernel an important config that feeds information to the powerhal daemon didn´t get fully built, since the QGKI dependency wasn´t met. This resulted in lags in games like NieR.

Also fix certain games like 7DS (com.netmarble.nanagb) crashing frequently on startup.

thanks to @twistedumbrella for reporting the bugs and noticing wireguard missing.


More changes in the Changelog below:


Changelog:
- add wireguard
- enable qgki part of msm_performance to fix lags in certain games
- fix some games ( for example 7DS (com.netmarble.nanagb)) frequently crashing during startup




Download:



Instructions can be found in the OP!
But look at this post too :)

Have fun, enjoy the kernel and your phone.

Donations are not mandatory but very welcome.
If you like my work and want to buy me a coffee/green tea to keep development going: http://paypal.me/freak07
 
I pushed my source :) I took the pass don´t worry. :)

CONFIG_MSM_PERFORMANCE_QGKI needs to be explicitly enabled.
For some reason when extracting the defconfig from boot.img or from /proc/config.gz and using that to build the kernel it doesn´t get enabled.
It seems there´s a lot for me to be learned about CAFs QGKI build "system". There might be more surprises.

Yes, the ROGs are super finicky I completely agree.


I´ll check the game and bluetooth on my end. thanks :)

I believe their build system is completely integrated. Their chain of configs put completely together is nothing like what you pull from their kernel and even that can't compile as is, which is a likely indication they have some of the configuration in the system side of things and build it as a single process (rather than kernel binary style). That is the reason I opted for export ASUS_BUILD_PROJECT="ZS673KS" instead of the usual process of changing around some ifdefs that I saw elsewhere. I don't trust that they didn't throw a random curve somewhere I'd miss it.

Yours come up as "-qgki-perf" using extract-ikconfig because of the spoofing commit.
 

Freak07

Recognized Developer / Recognized Contributor
Jan 2, 2011
5,068
16,083
I believe their build system is completely integrated. Their chain of configs put completely together is nothing like what you pull from their kernel and even that can't compile as is, which is a likely indication they have some of the configuration in the system side of things and build it as a single process (rather than kernel binary style). That is the reason I opted for export ASUS_BUILD_PROJECT="ZS673KS" instead of the usual process of changing around some ifdefs that I saw elsewhere. I don't trust that they didn't throw a random curve somewhere I'd miss it.

Yours come up as "-qgki-perf" using extract-ikconfig because of the spoofing commit.

the spoofing commit will show the defconfig that I extracted from the stock kernel, that´s correct.

Interestingly on zenfone 8, CONFIG_MSM_PERFORMANCE_QGKI was enabled without passing it explicitly in the defconfig I use to build the kernel with. It seems on the zf8 repo having CONFIG_QGKI and CONFIG_MSM_PERFORMANCE was enough to enable it. Checked .config in the out folder there.
For some reason not the case on the ROG 5 source.
I pass the same flag in my script.

Either way it´s fixed now :)
 
the spoofing commit will show the defconfig that I extracted from the stock kernel, that´s correct.

Interestingly on zenfone 8, CONFIG_MSM_PERFORMANCE_QGKI was enabled without passing it explicitly in the defconfig I use to build the kernel with. It seems on the zf8 repo having CONFIG_QGKI and CONFIG_MSM_PERFORMANCE was enough to enable it. Checked .config in the out folder there.
For some reason not the case on the ROG 5 source.
I pass the same flag in my script.

Either way it´s fixed now :)

Two display and two sound drivers with conflicting definitions... all I have to say...
 

Anubarak16

Senior Member
Oct 24, 2014
137
42
ASUS ROG Phone 5
  • Like
Reactions: twistedumbrella
Last edited:
  • Like
Reactions: Freak07

Freak07

Recognized Developer / Recognized Contributor
Jan 2, 2011
5,068
16,083
In normal media there is actually not that much of a delay but every game where you swing a sword that makes a sound will face a hard delay. For example https://play.google.com/store/apps/details?id=com.direlight.grimvalor

@Freak07


Not sure that it eliminates the issue, but the delay is noticeably reduced

thanks for the heads up @twistedumbrella, appreciate it. I tried the commit, but it doesn´t make a difference for me.
I resorted to using the game titan quest. Pressing the pause icon, and then pressing the "continue" button in the interface, but doing a hold press instead of a tap-press, produces a sound without the interface vanishing.
If I use 3,5mm headphones the sound is basically synced with me touching the screen, while on my pixel buds over bluetooth there´s, as to what I can perceive and tell apart, exactly the same delay with or without the commit.

Now I think it´s interesting that there´s not the same kind of delay at all over the youtube test I posted earlier.
Two guesses, maybe it´s either, both combined or none.
There might be sound processing in the background tailored to games and only being active there. Might increase latency in that case.
Second guess, additionally to the "usual" bt delay it´s more perceivable on the ROG 5 when running games due to the very low touch latency.

I´ll compare it with my other phones, how they behave in the same game with the pixel buds over bluetooth to get a general impression how it looks on other devices.
But it would certainly be nice if it could at least be reduced in some way. Let´s hope we find a way :)
 
If I use 3,5mm headphones the sound is basically synced with me touching the screen, while on my pixel buds over bluetooth there´s, as to what I can perceive and tell apart, exactly the same delay with or without the commit.

The more I thought about it, my last test was also on full stock. Any improvement I had could have been that. That is still hope that some is possible, at least.
 

Freak07

Recognized Developer / Recognized Contributor
Jan 2, 2011
5,068
16,083
The more I thought about it, my last test was also on full stock. Any improvement I had could have been that. That is still hope that some is possible, at least.
Yes. I tried your kernel, my kernel and compared them. I think @Anubarak16 might also be interested in that.
Compared to stock, both yours and my kernel is slightly improved when I use the Google pixel buds. At least to my perception it's noticable, maybe I'm biased though. I think others can try too. (Pixel buds use AAC, SBC)

I also found time now try with the Sony WH-1000XM3. Same scenario with the game I mentioned. They support SBC, AAC, aptX, aptX HD and LDAC.
Compared to the pixel buds there's barely any latency. There's still a bit latency compared to the 3.5mm jack, but not a real "delay". I guess if one plays "competitive" games it's still way too much delay, but it's not as disruptive to the experience when swinging a sword in an RPG compared to when I use the pixel buds.
The difference between the pixel buds and the Sony makes it very difficult for me to say there's a difference between stock kernel and custom kernels there, as the delay is way less on the Sony headphones.

There's also the possibility that it's related to the QCOM BT stack and we can't do anything in the kernel.

@twistedumbrella | @Freak07

Asking because I couldn't locate where it's mentioned;
Which models can this kernel work with ?

* ROG 5
* ROG 5 Pro
* ROG 5 Ultimate
* ROG 5s
* ROG 5s Pro

The kernel "should" work fine on all non 5s phones. However I can only test the regular ROG 5.
There's the possibilty it might break the touch zones on the back cover and rear display on the pro/ultimate. Without a device I can't confirm.

The 5s uses a more recent CAF snapshot and might (very possibly) require a separate kernel. I guess to confirm that somebody needs to try.
 
  • Like
Reactions: Anubarak16
Yes. I tried your kernel, my kernel and compared them. I think @Anubarak16 might also be interested in that.
Compared to stock, both yours and my kernel is slightly improved when I use the Google pixel buds. At least to my perception it's noticable, maybe I'm biased though. I think others can try too. (Pixel buds use AAC, SBC)

I also found time now try with the Sony WH-1000XM3. Same scenario with the game I mentioned. They support SBC, AAC, aptX, aptX HD and LDAC.
Compared to the pixel buds there's barely any latency. There's still a bit latency compared to the 3.5mm jack, but not a real "delay". I guess if one plays "competitive" games it's still way too much delay, but it's not as disruptive to the experience when swinging a sword in an RPG compared to when I use the pixel buds.
The difference between the pixel buds and the Sony makes it very difficult for me to say there's a difference between stock kernel and custom kernels there, as the delay is way less on the Sony headphones.

There's also the possibility that it's related to the QCOM BT stack and we can't do anything in the kernel.

That's also a good point. My first test was with a Wonderboom speaker and the second was with Airpods Pro (last year's model). The Airpods would have much better hardware than a speaker that is now about four years old. At 2AM, firing up the Wonderboom wasn't ideal.

The kernel "should" work fine on all non 5s phones. However I can only test the regular ROG 5.
There's the possibilty it might break the touch zones on the back cover and rear display on the pro/ultimate. Without a device I can't confirm.

The 5s uses a more recent CAF snapshot and might (very possibly) require a separate kernel. I guess to confirm that somebody needs to try.

I would think the only thing they didn't recycle is the processor. As long as the current base can also support the 888+, it should run. That's not to say they didn't go and change something under the hood.

@JazonX can you extract the ikconfig from the stock S kernel? That would be a quick and easy way to verify the hardware.
 
Last edited:

JazonX

Senior Member
Dec 16, 2009
1,833
727
Xiaomi Mi 11 Ultra
The kernel "should" work fine on all non 5s phones. However I can only test the regular ROG 5.
There's the possibilty it might break the touch zones on the back cover and rear display on the pro/ultimate. Without a device I can't confirm.

Just installed on ROG 5 Ultimate Global - A Version (ZS673KS) : Boots and works fine.
Installed using TWRP.

Not sure where the mapping or enabling the option for the back cover touch zones exist. Never used it, and I doubt I ever will since it's a hassle with the size of the phone, and I have to remove the cover to do so. Not a dealbreaker at all.

Was and is an active user of your work on the ROG II.

Thank you so much for your hard work and support. :)

ROG5.png
 
  • Like
Reactions: Freak07

Freak07

Recognized Developer / Recognized Contributor
Jan 2, 2011
5,068
16,083
Would think the only thing they didn't recycle is the processor. As long as the current base can also support the 888+, it should run. That's not to say they didn't go and change something under the hood.

@JazonX can you extract the ikconfig from the stock S kernel? That would be a quick and easy way to verify the hardware.
Yes, I'm unsure about the newer base (especially changes in the kernel device tree and other kernel side changed that might depend on updated firmware).

Maybe somebody will try and report back that owns a rog5s device.
In case you've never used 'less' before (since I don't typically use it)

Code:
less -R "ROG Phone 5 to S.txt"

That will display the color-coded diff from the text and 'q' will quit the view
Nice thanks, I appreciate it! I'll look at it when I'm back home at my PC. :)
Just installed on ROG 5 Ultimate Global - A Version (ZS673KS) : Boots and works fine.
Installed using TWRP.

Not sure where the mapping or enabling the option for the back cover touch zones exist. Never used it, and I doubt I ever will since it's a hassle with the size of the phone, and I have to remove the cover to do so. Not a dealbreaker at all.

Was and is an active user of your work on the ROG II.

Thank you so much for your hard work and support. :)

View attachment 5411159
Thanks for the kind words :) I really appreciate it. Glad to read posts like this from time to time!
Enjoy the ride. :cowboy:
 

MotoE2014

Senior Member
Jul 17, 2015
290
34
ASUS ROG Phone 5
ASUS ZenFone 8
Sorry for the inconvenience, I have a Rogphone 5 16/256 with Bootloader unlocked only, I have read the instructions but I have not understood some things well: I have the stock firmware with Bootloader unlocked, I just need to install the twrp, root and kernel and enough? Do you need to do some wipes? Thank you so much to those who will help me.
 

Top Liked Posts

  • There are no posts matching your filters.
  • 2
    Update to 1.2.0

    Hey guys and girls,

    I hope everyone is having a good day or starting into one. Here´s the next update.
    The kernel is updated for the latest firmware:
    18.0840.2109.176. Make sure to update to the latest firmware before flashing the kernel to have no compatibility issues.
    Some device specific changes, which should fix a touchscreen issue was in the source code.
    All other security related updates on the kernel side, were already included by merging CAF-Upstream previously.
    Caf-Upstream was also merged, which includes several improvements from QCOM to different subsystems like the scheduler, gpu driver etc. Details on github as usual.

    The link to the WW firmware can be found here:


    More changes in the Changelog below:


    Changelog:
    - Import ASUS_I005_1-18.0840.2109.176 source code
    - Linux-Stable Upstream to 5.4.150
    - Merge CAF (QCOM) upstream up to tag LA.UM.9.14.r1-18300-LAHAINA.0
    - Scheduler improvements from QCOM and Google
    - Enable NAP power saving state for GPU
    - updates from kernel common for several subsystems
    - for details please check github


    Important: If flashing the kernel via TWRP, flash the DLKM magisk module found in the download folder that matches the kernel version.

    Download:



    Instructions can be found in the OP!
    But look at this post too :)

    Have fun, enjoy the kernel and your phone.

    Donations are not mandatory but very welcome.
    If you like my work and want to buy me a coffee/green tea to keep development going: http://paypal.me/freak07
    1
    updated post #3 with some information on how to use the battery saver/cpufreq limiting feature
    1
    Thanks @Freak07. Really appreciate your work. I can also confirm that flashing the new ROM update and the kernel update by TWRP works fine. Just tick the option to reflash TWRP after flashing ROM. Then reboot recovery then flash Kernel -> Magisk alpha -> kernel again then reboot system
    1
    Thanks @Freak07. Really appreciate your work. I can also confirm that flashing the new ROM update and the kernel update by TWRP works fine. Just tick the option to reflash TWRP after flashing ROM. Then reboot recovery then flash Kernel -> Magisk alpha -> kernel again then reboot system
    It works, but the magisk helper module is still broken I suppose if flashing from twrp. Need to update the flashable module for that and adjust the instructions.

    Thanks for the reminder :)
  • 6
    Kirisakura-Kernel for the ROG Phone 5

    Hello everyone,

    To keep it short: Here is Kirisakura - Kernel for the Asus ROG Phone 5 aka Anakin. The ROG Phone line-up internal codenames are traditionally derived from Yedi-Characters of the Star Wars universe. Three years ago the 1st gen ROG Phone bore the generic title "Yedi", the ROG Phone 2 was codenamed "Yoda" and the Rog Phone 3 bore the title "Obiwan".
    I would appreciate if everybody that flashes the kernel, reads at least once through this opening post and the following ones.

    Kirisakura - Kernel is designed to bring a handful of beneficial features to the device, while ensuring excellent performance and smoothness to get you safely through the day!
    If you expect a custom kernel to magically improve your devices battery life manifold and this is your only priority then this might not be the right place for you.
    However the kernel is built to keep battery life close to the same as you would experience the on stock kernel while being tuned to perform better in various UI related benchmarks. The same benches are used by google to verify and ensure UI smoothness on the Pixel devices. The scheduler is tuned to yield lower jitter and reduce jank frames.
    There´s also a cpu-battery saver mode that cuts back the CPU max- freqs, but without the disadvantages like delayed notifications from built into the system user-space power-saving modes.
    If that got your curious, I welcome you to continue reading if you´re still interested!


    This project aims to keep most of the subsystems updated, way ahead of the stock kernel, thereby improving security and performance, while keeping stability as the foucs during testing! This includes Linux-Stable, CAF-Upstream and kernel/common.


    A very positive development. Google pushed for CFI and SCS to be enabled by default in the android-common kernel. In this years generation ( 5.4 linux kernel base/ SD 888 ) CFI and SCS are enabled by default. It was a lot of work to get these features working nicely on the kernels I did develop on for the past few year. So it´s nice to see this is now there by default in everybody can profit from it.
    I put the explanations and links in hide tags here.

    Kernel Control Flow Integrity (Kernel-CFI), which is achieved by linking the kernel with LLD and Link Time Optimization (LTO), more precisely ThinLTO, are quite unique security features.
    The recently released Kirisakura-Kernel for the OnePlus 8 Pro was featured on the XDA-Portal due to the inclusion of CFI. The featured article explains CFI in great details, is easy to understand and definitely worth to read.
    If possible the latest Kirisakura-Kernel releases for various devices aim to include CFI in enforcing mode so users can benefit from this security features.

    Another security feature is Shadow Call Stack (SCS).
    SCS is another security patchset that is aimed at preventing attacks via return oriented programming (ROP).
    ROP is a technique where the attacker gains control of the kernel stack to overwrite function return addresses and redirect execution to carefully selected parts of existing kernel code.
    If you´re interested here are a few good links:
    Google Security Blog explaining SCS.
    LLVM doc about SCS
    Android DOCs documenting SCS

    If reading about linux or caf/qcom upstream in the paragraph above got you curious, have a read about Linux-Stable and why it is important here. The stable-process is not the same for every subsystem, but the general idea, rule of thumb and benefits are applicable for other subsystems as well.
    Read more about the Release-Model following this link.
    This conversation on the LKML mailing list also shows the power of Linux-Stable and why it´s very beneficial to not just say it´s not worth and a potentiol source for bugs:
    Starting with the Initial Message , the following reply and a good summarizing statement.


    Quick explanation of CAF-Upstream. CAF is short for Codeaurora-Forums. This is the place where development from Qualcomm for their SoCs happens. This includes the Snapdragon 888 that´s built into the different SKUs of the ROG Phone 5.
    The kernel-bases available on the Codeaurora-Forums, are basically the foundations OEMs use, to build their own kernel additions on top for devices featuring Qualcomm SoCs.
    The Qualcomm developers push regular updates to the bases of the different SoCs, a bit similar to how Linux-Upstream works. OEMs usually stop updating the initial base after the phone is released.
    This kernel focuses amongst the other features to provide the latest CAF-Updates merged in a regular manner, to provide updates, improvements and enhancements for SoC specific drivers and subsystems! This might take a while, because a lot of device specific testing is done on my end to ensure stability.

    The kernel includes a lot of improvements and contributions from other developers as well. Without this kernel would not exist.
    A big part of improvements originate from @arter97´s, @kdrag0n´s and @Sultanxda´s work. Many others contributed in some way or another to this kernel.
    A big thanks to all of them at this place!

    Now lets continue with a list of features in the next paragraph!


    Main Features:
    - Based on latest kernel sources from Asus for Android 11 and intended to be used on the latest Asus Stock-Firmware (currently 18.0840.2109.176)
    - compiled with Clang 13.0.1 and built with -O3 speed optimizations
    - Linux-Stable-Upstream included to 5.4.150
    - upstreamed CAF base for the Snapdragon 888 (qcoms internal codename "lahaina")
    - include important fixes/improvements from kernel/common
    - tune scheduler for improved jitter and less jank frames
    - reduce power consumption during deep sleep
    - remove sleep debugging from Asus for release build
    - disable various debugging mechanisms for drivers which are not needed on a perf build (for example touchscreen, fingerprint reader, display, binder)
    - disable various debugging configs, that are not needed in a perf kernel build according to google.
    - disable selinux auditing (we don´t have to adress selinux denials running a stock rom)
    - display tweaks to reduce power consumption on statically rendered images
    - disable unnecessary CPU errata for our specific SoC
    - disable per cgroup pressure tracking to reduce overhead
    - free up RAM by removing reserved mem regions used only for debugging
    - Flashing the kernel will keep root!
    - Flashable via EXKM, FKM or TWRP (if available) on a rooted system!
    - Anykernel zip is based on the Anykernel3 release from @osm0sis


    EAS related features:
    - Implement dynamic WALT-Window size depending on current FPS, collecting information about CPU-Load to provide more precise information for the scheduler
    - EAS implementation patched up to latest CAF state
    - updates from kernel/common
    - scheduler updates from sd888/ 5.4 kernel
    - scheduler update from sony xperia 1 III kernel aiming to improve WALT

    CPU related features:
    - Power saving workingqueues enabled by default (toggleable in EXKM)
    - Change various drivers (WLAN, MM, audio, charger, power, thermal, glink etc) to use power efficient workingqueues. This compliments EAS in general
    - improvements to cpufreq/times/stats
    - vdso32 patches as found on Google Pixel devices (brief documentation here) (CAF 5.4 default, keep for documentation)
    - remove event debug logging
    - disable broken irq detection, disable debugmon
    - improve suspend/resume from audio driver (caf-upstream for audio-kernel)
    - cpuidle: wake only cpus for which qos has changed (this allows cores to keep idle states better)
    - disable debugging during sleep states as per qcom recommendation
    - remove RTB logging to reduce overhead
    - disable trace_printk as it shouldn´t be used on production/perf kernel builds
    - disable edac and coresight (following google on the pixel kernels)
    - disable jtagv8 debugging as recommended by qcom for perf builds to avoid potential power, performance and memory penalty
    - disable s2idle in favor of deep idle state -> improves deep sleep drain (google initially used 2idle for improved suspend/resume latency on pixel 3 and pixel 4, but disabled it on pixel 5. qcom followed suit for sd 888, but apparently didn´t see google reverting it on pixel 5. Other OEMs such as Xiaomi or OnePlus already disabled s2idle on their SD888 devices in favor of deep- read more about s2idle vs deep here)
    - raise LTO-inlining -> more info

    GPU related features:
    - GPU driver updated to latest CAF-State
    - gpu/kgsl improvements from pixel 4/pixel 5 kernel
    - add low_prio_worker thread for gpu driver (Improves performance when tearing down processes (eg, via LMK) and not blocking currently running UI from submitting work to the GPU
    - remove POPP functionality from GPU driver to avoid constant fight about downscaling with msm-adreno-tz
    - increase gpu worker thread priority
    - don´t wait for fenced GMU writes

    File System related features:
    - upstream BFQ-MQ to latest state in linux-mainline and default to 128KB readahead
    - improve memory allocations in kernfs
    - ufs improvements from pixel 4/ pixel 5 kernel
    - improve GC as to not interfere while user interaction is happening and prevent possible jitter or latency spikes

    Audio related features:
    - provide option to override audio profile based on impedance measurement (run 16ohm headphones in 64ohm mode, dangerous use with caution!)
    ----> more information/explanation how to control the feature linked here

    Display related features:
    - Backlight Dimmer (reduce minimal brightness below stock limit)

    Network related features:
    - Wireguard Support (Details)
    - advanced TCP algorithms enabled
    - use bbr as default TCP congestion algorithm (fasted algo according to this excellent research from @kdrag0n found here )
    - improved networking drivers from caf-upstream (especially wifi)
    - remove unnecessary wakelock from WLAN driver to reduce thousands of very short wakeups during a 7 hour deep sleep period
    - improved bluetooth performance (¹, ²)
    - Enable support for TTL spoofing

    Memory related features:
    - Speculative Page Faults left enabled (1. 2) (CAF 5.4 default, keep for documentation)
    - Fix various memory leaks in different subsystems
    - improved mm with patches from mainline
    - patches from kernel/common
    - improve memory allocations in various subsystems, video encode/decode, kernfs, cgroup
    - Backport to enable CONFIG_HAVE_MOVE_PMD (see here) , thanks @arter97
    - Backport to enable CONFIG_HAVE_MOVE_PUD (commit here) , thanks @arter97
    - free more RAM by removing unused reserved memory regions
    - mm: Micro-optimize PID map reads for arm64 (improves game performance, especially unity based games such as genshin impact)

    Security related features:
    - Control-Flow-Integration (CFI) and Link-Time-Optimization (LTO) (CAF 5.4 default, keep for documentation)
    - Use ThinLTO which was first used in the Pixel 4 XL Android R-Kernel instead of full LTO (CAF 5.4 default, keep for documentation)
    - Shadow Call Stack (SCS) Security-Feature ( 5.4 default, keep for documentation)
    - enable init_on_alloc for even more security, more information can be found in the commit message (CAF 5.4 default, keep for documentation)


    Misc Features:

    Wakelock Blocker:
    - advanced wakelock blocker with the ability to block any wakelocks (dangerous, use with caution)
    - please read [URL="https://arstechnica.com/gadgets/2018/08/p-is-for-power-how-google-tests-tracks-and-improves-android-battery-life/"]this for further info
    - blocking a kernel wakelock should only be done in case of firmware incompatabilities (WiFi network at work (can´t be changed) causes deep sleep to not work on the phone.) That´s the only use case I see for this feature. The kernel does not features this to improve battery life!

    Other features:
    - clean up asus drivers and fix a few warnings in kernel log
    - disable logging and debugging to make the kernel more lightweight in various places (includes binder, mm, vma, gpu, ipa etc)
    - fix several coding issues detected by newer Clang-Toolchains
    - enlarge dmesg size to ease debugging


    Flashing Guide, Download and Changelog

    Requirements:
    - unlocked Bootloader (this will void warranty and kill future OTA updates on all asus phones at the moment)
    - USB-Debugging in developer options enabled
    - latest adb and fastboot binaries
    - working adb and fastboot environment
    - magisk root
    - a backup of stock boot.img, vendor_boot.img and dtbo.img so you can go back to stock conditions (simplest way is TWRP or to extract from stock firmware)

    How to flash the Kernel:
    1. Download the latest kernel.zip and make sure you have properly updated to the latest matching Asus Stock-Firmware (check the feature list for the current firmware the source is based on). When the ROG Phone 5 receives an OTA update it takes Asus a while to release the source code for the new OS and me a while to build a new kernel. If you don´t want to face any issues, wait until I either release an updated kernel or give green light because there were no kernel changes.
    Always make sure to run the firmware the kernel is built for!
    If you feel adventurous and try in advance, make sure you have a backup ready!

    2. Flash the kernel.zip via latest TWRP (if available), EXKM or FKM app and do a full reboot.
    2a. Important: If flashing the kernel via TWRP, flash the DLKM magisk module found in the download folder that matches the kernel version.

    3. Reboot and profit.


    DOWNLOAD:
    Download is located always in this folder:

    Important: Read after Download
    Please take a look at the second post after flashing the kernel!

    Changelog:
    Android 11

    1.0.0 Initial Release
    1.1.0 https://forum.xda-developers.com/t/...-rog-phone-5-aka-anakin.4333363/post-85637825
    1.2.0 https://forum.xda-developers.com/t/...-rog-phone-5-aka-anakin.4333363/post-85709075

    Donations:
    Donations are not mandatory but very welcome if you want to support development or just buy me a coffee/tea :)
    If you like my work: http://paypal.me/freak07

    Credits:
    Asus for the development device, giving me the opportunity to create this project!
    @osm0sis for all his work, including the ak3 installer!
    @tbalden for being the best HTC, Pixel, OnePlus and now Asus wingman!
    @LeeDroid and @mwilky for their awesome roms and work I used on multiple devices!
    @Captain_Throwback for all the mentoring and guidance!
    @Eliminater74 for bringing me into the game and the Inspiration
    @nathanchance for his upstream guidance and assistance
    @RenderBroken for helping me out
    @flar2 for all his work
    @joshuous for all the help he provided to me in the past!
    @arter97 for giving me advice
    @kdrag0n for his help and advices!
    @topjohnwu for magisk!


    Source Code: https://github.com/freak07/Kirisakura_ANAKIN_ROG5
    4
    Update to 1.1.0

    Hey guys and girls,


    I hope everyone is having a good day or starting into one. Here´s the next update.
    I forgot to include wireguard in the release. That´s now added.

    Since we´re not building a QGKI kernel, but rather a standalone kernel an important config that feeds information to the powerhal daemon didn´t get fully built, since the QGKI dependency wasn´t met. This resulted in lags in games like NieR.

    Also fix certain games like 7DS (com.netmarble.nanagb) crashing frequently on startup.

    thanks to @twistedumbrella for reporting the bugs and noticing wireguard missing.


    More changes in the Changelog below:


    Changelog:
    - add wireguard
    - enable qgki part of msm_performance to fix lags in certain games
    - fix some games ( for example 7DS (com.netmarble.nanagb)) frequently crashing during startup




    Download:



    Instructions can be found in the OP!
    But look at this post too :)

    Have fun, enjoy the kernel and your phone.

    Donations are not mandatory but very welcome.
    If you like my work and want to buy me a coffee/green tea to keep development going: http://paypal.me/freak07
    3
    F.A.Q:

    Question:
    Is root preserved when flashing this kernel?
    Answer: Yes, the AnyKernel.zip will detect root and keep it.

    Question: Safetynet does not pass on my phone since I unlocked the phone, why is this so?
    Answer: Google introduced hardware backed attestation recently. Unfortunately the old kernel tricks to still get safetynet passing won´t work. Instead you will have to rely on some magisk modules. Short guide in post #4 below.

    Question: How do I return back to stock or another kernel.
    Answer: Extract boot.img, vendor_boot.img as well as dtbo.img from matching firmware you are on (you can do so by using this tool or similar ones found on XDA and when googling around) and flash them via fastboot. If you want to keep root flash back the magisk_patched_boot.img. Or dirty flash your rom and re-root.

    Question: Why is having Magisk installed mandatory for this kernel?
    Answer: The kernel uses a ramdisk overlay to apply some settings after boot. If you are not rooted these settings will not get applied and you miss some of the optimizations.

    Question: How to report bugs properly?
    Answer: Have a look at post #3 in the linked thread. The linked guide is a pretty good starting point.
    Before reporting any bug make sure you´re running on a stock configuration. That´s means you´re not using any mods, tweaks in kernel managers or other root tweaks , magisk modules, scripts or other modifications that alter various functions like sound mods, data traffic, sleeping behaviour, scheduler, magical battery tweaks etc.
    Try to describe the issue as detailed as possible! Give your exact setup, like rom, magisk version, kernel version.
    Is the issue reproducible? Does it happen frequently?
    Provide logs, otherwise debugging is a lot harder. If you can already reproduce the issue and provide logs it greatly limits the amount of time I have to spent until I figure out how to reproduce it.

    If the device force reboots/randomly reboots (that means you see the bootloader unlocked screen!) provide a .zip of /asdf.
    Easiest way is immediately after booting up, with a root explorer navigate to /asdf. Copy the contents to your internal storage, zip it up and send it to me.

    If the issue happens while the device is running provide a dmesg plus a logcat that you take while/shortly after the issue happens. This will log what´s running in the current session. Try to be as detailed as possible what´s happening when the issue appears. It will also help me in reading the log!
    3
    Github Repository containing buildable standalone stock kernel

    Building the ROG Phone kernels standalone, was always tricky and the ROG Phone 5 is no different.

    I was asked for help in getting the stock source buildable a while back.
    Now I got hands on a Rog Phone 5 so I created a branch on my github, which is essentially a barebones stock kernel, just configured to be built standalone.
    Toolchains can be obtained from googles repos. GCC you need to reset the commit history.
    Buildscript is called build_kirisakura.sh.
    Please check my AK3 installer as to not miss the kernel modules that should be included.



    Have fun and good luck on any project started.
    3
    I was trying to give you a pass, but I can take it back if you didn't want it lol. Pretty sure I know which config you are talking about. I had wondered why it wasn't enabled, so that explains it.

    The ROG Phone 5 is super finicky. It doesn't help that they released two (or three, depending where you're from) models and then immediately pushed out a second series with two more. All of them have slightly different hardware that Asus is trying to keep on a single code base.



    has a bunch of tactile (effect) UI components that make it easy to pick out the lag.

    I pushed my source :) I took the pass don´t worry. :)

    CONFIG_MSM_PERFORMANCE_QGKI needs to be explicitly enabled.
    For some reason when extracting the defconfig from boot.img or from /proc/config.gz and using that to build the kernel it doesn´t get enabled.
    It seems there´s a lot for me to be learned about CAFs QGKI build "system". There might be more surprises.

    Yes, the ROGs are super finicky I completely agree.


    I´ll check the game and bluetooth on my end. thanks :)