I was recently interviewed by Mike Adams on his show, we discussed the integration of Brighteon.AI (https://brighteon.ai) with our phones and laptops.

https://www.brighteon.com/3d7a3a67-2a64-4c7b-843b-1411f9088dbb

Check out this hands on demo of the model on our phone and laptop. This AI knows a TON about herbalism, natural health and survival skills, trained on 100M pages of content.

You can download the model here: https://BrightU.AI/downloads

It will run on local AI clients without an internet connection; It's free, unlimited and downloadable right now.

No deleted account found from 85 scanned users from this group 🚫👻

💬 Send /lang to change language of the bot

KernelSU v2.0.0 - Ultra Legacy

- small writeup on what I did to keep support for 3.0 ~ 5.4.

KernelSU moved from prctl to sys_reboot + ioctl for its IPC/API.
This method has to unlock sys_reboot from seccomp so that it is allowed for apps on denylist.
This can allow us to bootstrap and request a file descriptor to use via ioctl.
Basic usage: https://github.com/backslashxx/various_stuff/blob/master/ksu_fd/ksu_fd.c

Now for what I did.
Since we don't have the same seccomp-bpf subsystem as ACK 5.10+
I straight up disable seccomp for allowed apps.

Actually, I'm the one who suggested to use sys_reboot which will also keep a usable syscall for syscall-based communication since we have access to a pointer.

Thats the massive difference, now for small changes.

- /data/system/packages.list trigger
For throne_tracker trigger, we keep LSM hook on inode_rename.
This is way less overhead, way less code and way less maintenance burden.
We also have the asynchronous-ity of throne tracker since ever since, I offloaded throne tracker to a kthread.
Do you really expect me to backport fsnotify handling to 3.0?

- setuid hook
For this we also keep the LSM. same stuff.
This is way less overhead, way less code and way less maintenance burden.

- devpts
we don't hook devpts anymore. you can call pm via 'su -c pm' or su and then pm.

- sys_reboot hook
I went out of my way to meme-out other manual hook ecosystem elites.
so Shirkneko (sukisu) and simonpunk (susfs)
I convinced them to use this signature for sys_reboot hook

ksu_handle_sys_reboot(int magic1, int magic2, unsigned int cmd, void __user **arg);

This way the possibility to extend syscall-based instrumentation is infinite.
Other forks are to follow / following already.
RKSU also picked the same sys_reboot handler signature.
unsure for KSUN.

--
For 5.10+ users, while the driver on my repo will likely compile, I don't assure stability
nor I'll go out of my way to fix if theres issues. Just use upstream. This is not just due to me being lazy about it, this is also respect to upstream.

as always for LKM changes that I like, I just pass it on to kow
so for "LKM+" kind of thing, if you're willing to try bleeding edge stuff, use KowSU.

The commits are properly ordered and commented so if you want to do some integrations yourself
just cherry pick it on your own.

I guess that's it.

have fun.
- xx

Exynos fans resort to editing benchmarks to cope!

Beware of any Exynos 2600 leaks currently, as on twitter, Samsung fans (especially paid reviewers) are glazing results of a chip that doesn't exist. You'll come across the wildest justifications, including die sizes that could match an M4. All of which would be unrealistic for a mobile chip.

The funniest thing? They claim better single performance at lower frequencies than the 9500. With the same, exact, core. Oh, and better IPC too, 10% better, with the same exact core! These, to be true would again reach M4 die sizes just from the sheer cache requirements.

Follow @TechLeaksZone for future PSAs on fake leaks

all you really need is https://github.com/memstechtips/UnattendedWinstall

The changelog for ColorOS 16 is 3.5 meters long

Oct 11th OOS16 leak day 6

As it was a full merger with COS when Pete became SVP at Oppo, I'm sure the OOS16 changelog is just as long as the COS16 one, if not longer. Reading through the changelogs of CBT builds, I'm finding it very hard to come up with more topics because I can't post screenshots or screen recordings because of watermarks and I can't post pictures or videos taken because it'll give a tester's location away. So I can't show "Improves camera stability for a better photography experience" or "Fixes an issue where taking scrolling screenshots might sometimes fail."
I guess I'll make today the last day of the leaks and make it a Q&A. Keep in mind that my answer can't require a screenshot, picture, or video. I will say this: Oplus has worked on the hue difference when switching between lenses, but it's not totally fixed.

Ask away with replies to this post...

Treat Wheel 0.0.6 is here.

This version enhances ReVanced modules hiding, by bringing the concept of Treat Wheel module configuration file. By creating a tw_config file at the module's directory, with the following content, it allows Treat Wheel to recognize it is a ReVanced module, hence make it umount. If a module doesn't contain this file, it will not umount, as a limitation of how ReVanced umount system works. Because of that, users are encouraged to ask ReVanced module developers to include that file.

module_type=revanced
allow_umount=true

If you have issues with this issue, expect no fixes unless you report it. Collaborate, assist, do not leave developers in the void, like how it happened 0.0.5.

Do not enable, or disable, options you do not understand. Trust the default.

Special thanks to the translators who allowed to bring more idioms to Treat Wheel WebUI: @Katiria24, Gege, @ziomek64, @RainyXeon, @bocchi1919810, Ray. Not only them, but also the primary testers for these releases, @rrr333nnn333, @ph03nIx.

BTC: bc1qc9gpa7llp6k32au550c3jclvf8q4cam2njd6r6
Monero: 4AvbvPLD1MAXhxB6jSF1kpLBmG2VPXL1Y1DWiD7BwHipK7sokG1sRuV9poorcJyL39hUBdug6VpsFTCapMaBqvPW8dkMZAR


Is that all you care for? Hatred?
No feelings, you've been ablated.
I'll show you respect, undeserved.
Remember it's all being observed.
You are ridiculous nonetheless.

OnePlus 15 vs iQOO 15 vs Realme GT8 Pro Specsheet

Display

OnePlus 15: BOE X3, LTPO 8T, 1.5K+165Hz, 1800nits HBM, 0.12 SVM index, 2160 Hz PWM, normal polarised

iQOO 15: Samsung M14, LTPO 8T, 2K+144Hz, 2600nits HBM, 0.03 SVM index, 2160 Hz PWM, non-polarised

Realme GT8 Pro: BOE Q10+, LTPS, 2K+144Hz, 2000nits HBM, 0.07 SVM index, 1920 Hz PWM, circular polarised


Cameras

OnePlus 15: Main: Sony LYT700 (1/1.56”), Zoom: Samsung JN5 (1/2.76”), Ultrawide: Samsung JN5 (1/2.76”), Custom Oppo image engine

iQOO 15: Main: Sony IMX921 (1/1.56”), Zoom: Sony IMX882 (1/1.95”), Ultrawide: Samsung JN5 (1/2.76”), Vivo algorithm

Realme GT8 Pro: Main: Sony LYT700 (1/1.56”), Zoom: Samsung HP5 (1/1.56”), Ultrawide: Samsung JN5 (1/2.76”), Ricoh GR partnership


Battery & Charging

OnePlus 15: 7000 mAh, 100 W wired, 50 W wireless, bypass charging supported

iQOO 15: 7500 mAh, 120 W wired, 50 W wireless, bypass charging supported

Realme GT8 Pro: 7000 mAh, 100 W wired, 50 W wireless, bypass charging supported


Processor / SoC

All: Snapdragon 8 Elite Gen 5 (SM8850)


RAM & Storage

All: LPDDR5X RAM, UFS 4.1 storage


Build

OnePlus 15: IP69 rating, metal frame with ceramic coating on frame and body for better scratch resistance

iQOO 15: IP69 rating, metal frame, RGB lighting integrated into camera deco

Realme GT8 Pro: IP69 rating, metal frame, modular/customisable camera deco


Haptics & Audio

OnePlus 15: Haptics: AAC ESA1016 motor (560 mm²), Speakers: dual AAC 1115E

iQOO 15: Haptics: AAC 0916 large motor (602 mm²), Speakers: dual AAC 1511

Realme GT8 Pro: Haptics: AAC ESA0816 motor (448 mm²), Speakers: dual AAC 1115E

(Sidenote: testers suggest realme haptics feels crisper compared to OnePlus despite it's smaller size)


Biometrics & Connectivity

All: Ultrasonic fingerprint scanner, USB 3.2 Gen 1