[KERNEL] [9] A2N Kernel For Galaxy S8/S8+/Note8 [EXYNOS] [STABLE + BUG-FREE] [HIGH-END] [V. 2022.05.13]

[Ryujin96]

Do you mean all apps menu ? So not just the apps places on the homescreen.

Yea, only apps menu (app drawer), not on home screen. Maybe a force close?

 

[ace2nutzer]

OK, never faced this on S8. I believe there is something wrong with your phone, ROM or user apps. Later i'll show you how to find out what is using so much RAM. Killed by lmk.

 

[Ryujin96]

OK, never faced this on S8. I believe there is something wrong with your phone, ROM or user apps. Later i'll show you how to find out what is using so much RAM. Killed by lmk.

Hmm. Yea its weird it happens to me this issues.

 

[ace2nutzer]

Latest build still ok ? - For me zero issues so far with zRam 90 %.
Also no glitches after coming from deep sleep.

How to check RAM usage:

with ADB or terminal emulator app, run:

top -s 6 -n 1

and see "RES". Value is in MB.

 

[Ryujin96]

Still ok with the latest test build, so far i know no problems do i need the su command before running the other command in terminal emu?

 

[ace2nutzer]

Not needed. You can also remove "-n 1" and it'll run for ever, so you have a "RAM monitor". To stop it, press "Q" or "Strg + Q".

 

[Ryujin96]

Not needed. You can also remove "-n 1" and it'll run for ever, so you have a "RAM monitor". To stop it, press "Q" or "Strg + Q".

Dont know which they are from

 

[ace2nutzer]

Ohh you need first "su" cmd. Here with adb works without root.
Then use landscape mode, otherwise you won't see which package it's on the corresponding line.

 

[Ryujin96]

Ohh you need first "su" cmd. Here with adb works without root.
Then use landscape mode, otherwise you won't see which package it's on the corresponding line.

I dont see anything suspicious here. Seems normal for me

 

[ace2nutzer]

Yeah, nothing wrong.

 

[Ryujin96]

Yeah, nothing wrong.

Got a random reboot with screen off 2 min after i checked social media.

 

[Ryujin96]

Just out curiosity, which error does it give everytime in kmsg?

 

[ace2nutzer]

Just out curiosity, which error does it give everytime in kmsg?

It's not everytime the same error, but similar. But it think it's just one bug.

Spoiler: kernel oops message

<0>Unable to handle kernel NULL pointer dereference at virtual address 00000000
<2>sec_debug_set_extra_info_fault = KERN / 0x0
<1>pgd = ffffff800a42c000
<1>[00000000] *pgd=00000008fdffe003, *pud=00000008fdffe003, *pmd=0000000000000000
<0>Internal error: Oops: 96000005 [#1] PREEMPT SMP
<4>Modules linked in: cifs md4 dns_resolver ntfs fscache usb_f_ptp usb_f_mtp_samsung
<0>exynos-snapshot: exynos_ss_get_reason 0x0 (CPU:7)
<0>exynos-snapshot: core register saved(CPU:7)
<0>FEMERR0SR: 000000007f030000, FEMERR1SR: 0000000000000000
<0>LSMERR0SR: 0000000000000000, LSMERR1SR: 0000000000000000
<0>TBWMERR0SR: 0000000000000000, TBWMERR1SR: 0000000000000000
<0>L2MERR0SR: 0000000000000000, L2MERR1SR: 0000000000000000
<0>exynos-snapshot: context saved(CPU:7)
<6>exynos-snapshot: item - log_kevents is disabled
<6>TIF_FOREIGN_FPSTATE: 0, FP/SIMD depth 0, cpu: 0
<4>CPU: 7 MPIDR: 80000003 PID: 937 Comm: kswapd0 Tainted: G        W       4.4.111-a2n-g5bbb3595 #13
<4>Hardware name: Samsung DREAMLTE EUR rev06 board based on EXYNOS8895 (DT)
<4>task: ffffffc87706b800 task.stack: ffffffc87768c000
<4>PC is at obj_malloc.isra.11+0x34/0x198
<4>LR is at zs_compact+0x24c/0x7a0

<0>Process kswapd0 (pid: 937, stack limit = 0xffffffc87768c028)

<0>[<ffffff80081d66e8>] obj_malloc.isra.11+0x34/0x198
<0>[<ffffff80081d7880>] zs_compact+0x24c/0x7a0
<0>[<ffffff80081d7dec>] zs_shrinker_scan+0x18/0x2c
<0>[<ffffff8008199da0>] shrink_slab.part.7+0x220/0x2d8
<0>[<ffffff800819cd9c>] shrink_zone+0x2ac/0x2d4
<0>[<ffffff800819dc10>] kswapd+0x4d4/0x7b0
<0>[<ffffff80081097f4>] kthread+0x10c/0x110
<0>[<ffffff80080c9ba0>] ret_from_fork+0x10/0x30

I'm not very skilled at Linux memory management.
It's very low-level stuff.
I do suspect it has something to do with the movable pages or virtual memory.
There is a little mess ...
Samsung partially added patches from newer kernels for movable pages for the stock driver zSwap, but not for zRam. Now we want to use zRam, because it's faster. Another problem is, there is no proper documentation about added patches on top (commits). And this makes it very hard to track bugs. Also it's hard to upstream this kernel. Because most of the patches would just fail, due to some patches already applied and so on. Same problem for device specif patches, no commits at all. Samsung did that mess on Galaxy Ace 2 and i was hoping they would do a more cleaner job on Galaxy S8, but the same mess here.

If kernel source were clean like official Android kernel source, and just added device specif patches on top, even without proper commits messages, it would be possible to filter out device specific patches, and then simply adding that patch on top of latest Android kernel. Then we never had stability issues. But well, we will fix it anyway.

Uploaded new test kernel .. disabled

CONFIG_ZRAM_MIGRATION_SUPPORT
and
CONFIG_MALI_DMA_FENCE

Bsically CONFIG_MALI_DMA_FENCE is disabled in stock kernel, don't know why Samsung didn't enabled it, since it compiles fine without any error.
And CONFIG_ZRAM_MIGRATION_SUPPORT was originally enabled for zSwap, but adapted from me to work with zRam too. So disabled, maybe i did some mistake here.

Both are optimizations for better performance, maybe one of them is a culprit.

And there are further 2 optimizations that i suspect that it could make also problems .. but we will disable them in next test build if still unstable. After that. i'm out of ideas.

Those 2:

CONFIG_RBIN
CONFIG_ION_RBIN_HEAP

 

[rtyuakatsuki]

Report dev exprience using test kernel 30 november i had some lag that stuck for a while come smooth for while too when open facebook whatsapp and other social media apps for other apps game and other almost same problem but no more then 1 second in social media apps it took 3 or 5 sometime
To day i will test new test kernel

 

[Ryujin96]

Alright, its kinda ****ty indeed, but yea its fixable, only takes some time with testing and fixing and repeat but yea installed the new test kernel

 

[Ryujin96]

@ace2nutzer maybe its a idea to talk with a other kernel dev if the problems persists, dont wanna be a **** or something but just a advice and maybe its useful did the same thing for my case with my linux servers and game panel

 

[ace2nutzer]

First lets test those 2 test kernels.

 

[Ryujin96]

First lets test those 2 test kernels.

Yea first that, but in mean while i dont have any random reboot yet.

 

[Ryujin96]

Day 1, stable. No random reboots besides a shutdown becus i run out of accu and some lagging for a few secs so now and then

 

[Ryujin96]

Random reboot while checking apps.

kernel 12-02

Zram 90%