[KERNEL][9] A2N Kernel [10h SOT][Easy Guide Step by Step][2021.02.06]

predict9 · Jan 26, 2021 at 6:40 PM

Lets gooo nice update bro keep up

ace2nutzer · Jan 26, 2021 at 6:40 PM

rostifaner said:
@ace2nutzer
On previous version I have got random soft reboot when calling.
Here is last kmsg.
Love your work, keep it up)
UPD
Also on newest release got random reboot

As for the soft reboot: you need to provide dmesg next time.
And for random reboot on latest kernel: what did you the last 5 minutes ?
Did you change zRam size ? - or did you enable / disable wifi ?
Also was it shortly after a fresh reboot ?

rostifaner · Jan 26, 2021 at 7:09 PM

ace2nutzer said:
And for random reboot on latest kernel: what did you the last 5 minutes ?
Did you change zRam size ? - or did you enable / disable wifi ?
Also was it shortly after a fresh reboot ?

Always use 50% zram and with new version I set only 2.4 ghz band. Change settings yesterday.
Last 5 minutes: wifi on, bt connected, swiping Instagram feeds then press power button and after ~10 sec it goes to reboot. It was ~15 minutes after fresh reboot.
Upd
For now I haven't seen new cases

ace2nutzer · Jan 27, 2021

rostifaner said:
Always use 50% zram and with new version I set only 2.4 ghz band. Change settings yesterday.
Last 5 minutes: wifi on, bt connected, swiping Instagram feeds then press power button and after ~10 sec it goes to reboot. It was ~15 minutes after fresh reboot.
Upd
For now I haven't seen new cases

So the only changes is zram & wifi band ? - So anything else is just kernel/app defaults (also no init.d scripts ?

I've checked the last_kmsg2 ...

This caused the reboot:

osmaniyeliapo · Jan 27, 2021 at 4:50 AM

Can I use it on Android 10? Will it work?

recoman · Jan 27, 2021 at 5:13 AM

Report kernel a2n 20210126 :
Devices s8 plus
Rom hades v12, magisk v21,4, twrp 3310 in general a2n kernel
Kernel overhearting use gg chrome, fb, youtube battery drain.
Use gaming super overhearting ....
Kernel best smoothness. Thx mate dev

ace2nutzer · Jan 27, 2021 at 8:47 AM

recoman said:
View attachment 5200145 View attachment 5200147Report kernel a2n 20210126 :
Devices s8 plus
Rom hades v12, magisk v21,4, twrp 3310 in general a2n kernel
Kernel overhearting use gg chrome, fb, youtube battery drain.
Use gaming super overhearting ....
Kernel best smoothness. Thx mate dev

The stats and temps on your screenshots looks quite normal.
As for battery draining: check your real battery capacity in power menu from a2n kernel demo app.
Btw: make sure to enable MagiskHide in Magisk Manager app and see if it's cooler. My phone is pretty cool when using YouTube for hours in fullscreen mode.

Devil90 · Jan 27, 2021

I noticed that at boot when the samsung logo appears a few times it moves position up and then returns to its place.
In addition, always in the boot if I press the home button it vibrates is normal?
it also occurs to you @ace2nutzer?
Currently stock rom.

ace2nutzer · Jan 27, 2021 at 2:55 PM

Yes sometimes, so far i didn't investigate since it's cosmetics.
But in current build there is no little CPU OC ... later i'll upload new build.

rostifaner · Jan 27, 2021

ace2nutzer said:
So the only changes is zram & wifi band ? - So anything else is just kernel/app defaults (also no init.d scripts ?

also set cpu dvfs = 2002, gpu dvfs = 455, torch brightness = 10. Other changes by default without any scripts.

ace2nutzer said:
Try to uninstall HD VideoBox app and see if it's stable, let me know then.

Done!
Upd
Got another reboot but before deleting that app and seems it cause by other reason

pranil60 · Jan 27, 2021 at 6:35 PM

I overclocked my gpu using init.d, it shows maximum frequency as 572mhz in some hardware checker app( I tried two apps ) but in both of them it never goes over 546mhz when they monitor real time changes in frequency, during playing games.
Things I changed:
1 user_max_clock and user_min_clock
2 gpu_max_clock and gpu_min_clock

Things I observed
1 gpu_volt has default voltage number that required for 546

I'm requesting you to give me last hint.

And I found some bugs in 26-1-2020 kernel:
1 logo changing position and coming back to original position
2 moderate heating while playing games such as pubg upto 50 degrees(it actually feels)
3 Sometimes freezes screen, unable to touch and even buttons won't work until I go for force power off
and everything seems right, I will report if I found some more..

moreformusic · Jan 27, 2021 at 7:42 PM

So I made some testing this month and I concluded that Android 9 (Pie) is quite unoptimised despite kernel performance. I've tested fluidity and stability at some games in stock Pie ROM (with and without A2N kernel), and on some other custom ROM's based on Pie, and then on Hades ROM based on Q (Android 10) and I've seen that despide some errors, in hard fps games it was best, no lag, stable fps, no heating (despite lower CPU and GPU performance), etc.. So my question is how to make this stock rom as optimised as Q? I tried debloating as good as possible, but I think it's not just by that, I don't have very much development skills but that's my 5 cents

Devil90 · Jan 27, 2021 at 7:57 PM

I also point out a slight delay in unlocking with fingerprint (App A2n control on default), I did the test by returning to stock kernel and it does not occur.

Luiz2004 · Jan 28, 2021

parabéns pelo bom trabalho. e se você conseguir rodar o gpu em 839 MHz no máximo dele sem o travamento do smartphone ficaria muito bom.

Translation: Congratulations for the good work. and if you can run the gpu at 839 MHz at most without locking the smartphone it would be very good.

ace2nutzer · Jan 28, 2021

rostifaner said:
Got another reboot but before deleting that app and seems it cause by other reason

Yeah, it looks like there is a hardware / software compatibility issue.
This is what i get from your log:

Code:

cat last_kmsg.txt | grep "mmc_host mmc0"
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030043, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:-84, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000

if i do the same grep on my S8 last_kmsg, then i get this:

Code:

dreamlte:/ # cat proc/last_kmsg | grep "mmc_host mmc0"                                                                                                
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 300000Hz, actual 297619HZ div = 42)
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 300000Hz, actual 297619HZ div = 42)
<6>mmc_host mmc0: card inserted

According to my experience and how know, your board has a different PLL chip and divider 42 is not supported. This is extremely strange, and shows us that the S8 was shipped with different mainboards (to make my work harder, thanks @samsung)

Honestly i really hate the S8 from the first day of kernel development, exactly because of the fact that if something works on my phone it doesn't automatically mean that it'll work for every S8 / S8+ Exynos.
I can remember that i slightly increased the clock of the bus speed that probably requires a higher divider for the MMC flash (in order to avoid OC of the MMC). So now to find out if this is the reason, try with stock kernel:

index - powered by h5ai v0.28.1 (https://larsjung.de/h5ai/)

xda.mister-freeze.eu

i can also upload a test kernel with stock bus speed ..

Or even better you can check from time to time if you got sometimes proper output like mine .. by using ADB ... you even don't need to wait for a random reboot .. just reboot 2 times in a row and then here we go with ADB ...

Type this:

Code:

adb shell
su
cat /proc/last_kmsg | grep "mmc_host mmc0"

If you see everytime errors, then better flash stock kernel for now until i provide a test kernel later.
Should work wiithout wipe data !

pranil60 said:
I overclocked my gpu using init.d, it shows maximum frequency as 572mhz in some hardware checker app( I tried two apps ) but in both of them it never goes over 546mhz when they monitor real time changes in frequency, during playing games.
Things I changed:
1 user_max_clock and user_min_clock
2 gpu_max_clock and gpu_min_clock

Only change user_max_clock.
gpu_max_clock changes automatically according to the DVFS, and that's the reason why i have added user_max_clock in order to don't mess it up with DVFS and user max clock.
u can post the script that you created in /system/etc/init.d .. maybe there is something wrong.

Things I observed
1 gpu_volt has default voltage number that required for 546

You don't need to care about this interface, it's quite inaccurate and doesn't show the real actual gpu voltage, but it show basically the last used voltage just before the gpu went to powersave mode of 0 MHz.

If you want to check what is your Max supported GPU Freq, then you should check sysfs interface

Code:

gpu_asv_table

The highest Freq that has voltage assigned, is supported.

1 logo changing position and coming back to original position

Yeah i know, it's like a already said only cosmetics, i'll check when i've more time.

2 moderate heating while playing games such as pubg upto 50 degrees(it actually feels)

Yeah that's quite normal, also DVFS is coming from the ROM, so if the ROM decides too late to tell the kernel it should throttle, then we have this little heating up issue.
Maybe better to shut up with the ROM and to code a low-level DVFS in kernel with tunable when it should start to throttle, since it doesn't work properly.
Anyway, here on stock ROM i don't face really hard heating up even if run 3 times in a row 3DMark "Sling shot extreme" with Vulkan .. (more heat

)
Also honestly i don't like it too how it works, when throttling should start then both at the same time CPU/GPU will throttle, instead to throttle only CPU or GPU as needed .... this is somehow not very efficient in terms of "best performance"
So it's definitely possible to make it better.
Which ROM are you using ?

Sometimes freezes screen, unable to touch and even buttons won't work until I go for force power off

ohh that sounds painful and for sure i would never upload such a bad kernel, it's basically not usable then. Such issue mostly happens if this kernel is installed on non stock ROM from Samsung or wrong user config for kernel or last bad one would be if there is filesystem corruptions caused by hardware issue itselfs or caused by too much random reboots ... (every random reboot basically has the potential to corrupt the filesystem because it's a dirty reboot without to sync dirty data from RAM that should be usually written to the MMC.

If you Guys really want 100 % stable phone, then every time a random reboot happened or let me say when it starts to freak out with app force closes, freezes and random reboots .. .then the only fix is to wipe data of /data /sdcard and if inserted as well for /sd-card.
/system (the ROM files) will not corrupt because it's mounted "ro" (read-only)

I know it's painful but i don't know more user friendly ways.
Those are hard facts that not every developer will tell you.

and everything seems right, I will report if I found some more..

You are welcome.

moreformusic said:
So I made some testing this month and I concluded that Android 9 (Pie) is quite unoptimised despite kernel performance.

Right, stock software was never fully optimized. And so yeah the kernel can't add missing optimizations from the ROM, only can optimize over stock kernel.

I've tested fluidity and stability at some games in stock Pie ROM (with and without A2N kernel), and on some other custom ROM's based on Pie, and then on Hades ROM based on Q (Android 10) and I've seen that despide some errors, in hard fps games it was best, no lag, stable fps, no heating (despite lower CPU and GPU performance), etc.. So my question is how to make this stock rom as optimised as Q? I tried debloating as good as possible, but I think it's not just by that, I don't have very much development skills but that's my 5 cents

This is a good question, and above i've already give the answer.
Of course performance is in relation with the heat. Looks like Android 10 DVFS is more aggressive, but results in lower performance. Some users prefer a low temp, and some don't care about temp as long as it gives maximum possible speed and is of course stable. So what we should do ?

Devil90 said:
I also point out a slight delay in unlocking with fingerprint (App A2n control on default), I did the test by returning to stock kernel and it does not occur.

Hmm tested right now and i can't notice a delay, sorry

Luiz2004 said:
Congratulations for the good work. and if you can run the gpu at 839 MHz at most without locking the smartphone it would be very good.

Thanks, but it depends on the hardware, see above description.

ace2nutzer · Jan 28, 2021 at 4:16 AM

@rostifaner
flashed stock kernel and it's the same for me:

Code:

<6>[    1.503567]  [0:    kworker/0:1: 1222] mmc_host mmc0: card inserted
<6>[    1.528284]  [5:  kworker/u16:0:    6] mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>[    1.588560]  [6:  kworker/u16:0:    6] mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 300000Hz, actual 297619HZ div = 42)

So according to this, the only significant change would be int_freq ... set from 400 @ 667 MHz.
i've uploaded a test kernel for S8 with int_freq = 400 MHz.
But honestly i still doubt that this is the culprit, i do more believe that it's caused by Filesystem error(s) or CPU/RAM failures.

Test Kernel attached here.

pranil60 · Jan 28, 2021 at 6:19 AM

I'm using stock rom and I was not familiar with init.d script before 2 days ago. But somehow I learned my own from scratch and reached upto here.

1 is my current config having the same conditions as I mentioned in previous post
2 was my experimental script where I was actually able to reach 572mhz even after removing gpu_volt command it hitted 572mhz.. but temperature was such that I served two cups of tea to my guests with extra sugar. So I decided to turn back to normal setting.
And I don't know how to change DVFS, it would be great if you tell me it will save my experiment time.

And some more bugs that I observed and like to let you know:
1 when both fast wireless and fast wired charging is connected mobile chooses wireless over wired charging which is slower than wired. BTW I would just disconnect cable if wireless charging or simply put my device little eccentric over wireless charger pad, no problem for me actually

And most IMPORTANT thing, THANK YOU very much. Even without above modifications kernel works so good. My primary intention to apply this kernel was to improve PUBG performance, but want more stability so wanted to overclock gpu further more, so...
Again Thank You!

rostifaner · Jan 28, 2021

@ace2nutzer
Flashed your test kernel then reboot 2 times and I get this output

C#:

dreamlte:/ # cat /proc/last_kmsg | grep "mmc_host mmc0"
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: card inserted
<6>mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030043, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000

But its strange for me also, Im using your kernel more than 1 year and got strange behaviour with random reboots after 06112020 version, as I remember int_freq has changed much earlier(maybe im wrong). So the last stable version for me its 27092020 and according to your git commit maybe its ram issue due to defconfigs: disable PGTABLE_MAPPING(Just my guesses).
UPD
Flashed your stock kernel and have same output

C#:

dreamlte:/ # cat /proc/last_kmsg | grep "mmc_host mmc0"
<6>[    1.082106]  [0:    kworker/0:1: 1299] mmc_host mmc0: card inserted
<6>[    1.112109]  [7:  kworker/u16:2:   53] mmc_host mmc0: Bus speed (slot 0) = 25000000Hz (slot req 400000Hz, actual 390625HZ div = 32)
<6>[    1.329442]  [5:  kworker/u16:2:   53] mmc_host mmc0: Bus speed (slot 0) = 200000000Hz (slot req 200000000Hz, actual 200000000HZ div = 0)
<6>[    1.329979]  [4:  kworker/u16:2:   53] mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030043, EN_SHIFT = 0x00000000
<6>[    1.330075]  [4:  kworker/u16:2:   53] mmc_host mmc0: Tuning error: cmd.error:-5, data.error:-115 CLKSEL = 0x03030003, EN_SHIFT = 0x00000000
<6>[    1.330225]  [4:  kworker/u16:2:   53] mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030044, EN_SHIFT = 0x00000000
<6>[    1.330352]  [5:  kworker/u16:2:   53] mmc_host mmc0: Tuning error: cmd.error:0, data.error:-84 CLKSEL = 0x03030004, EN_SHIFT = 0x00000000

UPD2
Also maybe its rom issues? soon Im going to install latest( G950FXXSBDUA3 ) version and see what happens with your last kernel

predict9 · Jan 28, 2021 at 1:07 PM

Idk what yall saying but i have 0 problems with this kernel only after the update i get random reboot idk why but i dont mind cuz its reboots after a day or something so its all good for me

predict9 · Jan 28, 2021 at 1:09 PM

I want 1 thing tho if u got time to make can put something like a custom profile thanks ss keep uppp the good workj

[KERNEL][9] A2N Kernel [10h SOT][Easy Guide Step by Step][2021.02.06]

Senior Member

Recognized Developer

Senior Member

Recognized Developer

Member

Senior Member

Recognized Developer

Member

Recognized Developer

Senior Member

Attachments

Member

Senior Member

Member

New member

Recognized Developer

Recognized Developer

Attachments

Member

Attachments

Senior Member

Senior Member

Senior Member