Since Nextbit has stubbornly decided that the Robin will not officially support Robin, I must turn to custom roms to properly make use of the hardware in this phone. To that end, I have been having difficulty locating a rom with the necessary GPU libs to support Vulkan, and I am not ready to experiment with patch zips. Can anyone provide a list of roms they know include it?
https://insider.razerzone.com/index.php?threads/vulkan-support.23169/
Hi, I was wondering if there is any way to make the Vulkan API as deafault instead OpenGL. Or maybe a way to change it when android boots up.. A command line to do it would be awesome since I could use Tasker to automate the process. Thanks in advance and sorry about my terrible english.
ROM: HAVOC OS 2.1 (Pie)
Unfortunately even Skia is not enabled by default or supported completely atm on Oreo or Pie so Vulkan is still far away/ wip & likely to be the default only in the next upcoming Android version.
Hi everyone,
I use a Wireguard always-on VPN on my OnePlus 7T.
I'm rooted with magisk.
I would like to get better performance and less battery drain with my VPN, so I heard it could run using a kernel module, instead of a user-land code.
When I go in Wireguard app, I get the error "No modules are available for your device".
I would like to know what is necessary to get such module ? Who is in charge of making it ? (Wireguard developers ? OnePlus developers ? Magisk developers ?) And how can we help to get this module done ?
Thank you.
Florent
Yeah you'd need to bake it into the kernel. Find a kernel for a snapdragon 855 device that has wireguard support and ask them. I'm sure someone has made a guide.
Why do we need a *new* kernel ? We can't build kernel modules on Android ?
OnePlus has published 7T kernel sources : https://github.com/OnePlusOSS/android_kernel_oneplus_sm8150/tree/oneplus/SM8150_Q_10.0
Use Blu spark kernel it says wireguard support
This is the stock kernel that ships with glassrom (or will ship with it)
5g variants are not yet supported
It is based off kirisakura kernel with additional hardening from my side.
You get this:
All the features from kirisakura kernel
Removed qualcomm's rmnet drivers
COMPAT_VDSO is disabled to enable full vDSO ASLR
KSPP patches have been applied
Clang control flow integrity (https://source.android.com/devices/tech/debug/cfi)
Backward edged control flow integrity:
Strong protections enforced by shadowcallstack (https://source.android.com/devices/tech/debug/shadow-call-stack)
Weak protections enforced by adding stack canaries to everything and ensuring ASLR is of a decent enough quality
Compiled with -O3 and Polly for maximum performance
Wireguard driver has been removed
AVB depends on the ROM. Flashing it on glassrom/oxygenos will definitely cause it to boot with enforcing AVB. On other ROMs this shouldn't happen
Selinux forced enforcing patch from Samsung
Yama is enabled and set to SCOPE_NO_ATTACH
Uses sdfat driver to provide vfat and exfat drivers
Todo:
Port Linux-hardened patch
fix fingerprint on oos
Notes:
Flashing it on oxygenos will break dt2w
Flashing the kernel regardless of ROM or device combination will break twrp ramdisk boot. The only way to boot twrp is using fastboot boot, installing it to the ramdisk will always lead to a kernel panic. This is not a bug and will not be fixed
Download: see release post https://forum.xda-developers.com/showpost.php?p=81105101&postcount=8
Source:
https://github.com/GlassROM-devices/android_kernel_oneplus_sm8150
Donations:
Most of the hard work was done by @Freak07 so check out his thread and buy him a coffee
anupritaisno1 said:
This is the stock kernel that ships with glassrom (or will ship with it)
5g variants are not yet supported
It is based off kirisakura kernel with additional hardening from my side.
You get this:
All the features from kirisakura kernel
Removed qualcomm's rmnet drivers
COMPAT_VDSO is disabled to enable full vDSO ASLR
KSPP patches have been applied
Clang control flow integrity (https://source.android.com/devices/tech/debug/cfi)
Backward edged control flow integrity:
Strong protections enforced by shadowcallstack (https://source.android.com/devices/tech/debug/shadow-call-stack)
Weak protections enforced by adding stack canaries to everything and ensuring ASLR is of a decent enough quality
Compiled with -O3 and Polly for maximum performance
Wireguard driver has been removed
AVB depends on the ROM. Flashing it on glassrom/oxygenos will definitely cause it to boot with enforcing AVB. On other ROMs this shouldn't happen
Selinux forced enforcing patch from Samsung
Yama is enabled (does nothing significant for now)
Todo:
Set Yama to level 3 (breaks magisk)
Port Linux-hardened patch
Notes:
Flashing it on oxygenos will break dt2w
Flashing the kernel regardless of ROM or device combination will break twrp ramdisk boot. The only way to boot twrp is using fastboot boot, installing it to the ramdisk will always lead to a kernel panic. This is not a bug and will not be fixed
Download:
https://mirror.apexcdn.net/files/glassrom/unsigned.zip
Source:
https://github.com/GlassROM-devices/android_kernel_oneplus_sm8150
Click to expand...
Click to collapse
Fingerprint is broken on oos
Kaz205 said:
Fingerprint is broken on oos
Click to expand...
Click to collapse
Yeah sorry about that. I'll make a version for oos soon
I did test it for a short while on oos but did not test it enough
Merged in the latest kernel from kirisakura git and also merged in 4.14.156
It boots fine but I don't have a good internet connection to be able to upload it
Will do so soon
anupritaisno1 said:
Merged in the latest kernel from kirisakura git and also merged in 4.14.156
It boots fine but I don't have a good internet connection to be able to upload it
Will do so soon
Click to expand...
Click to collapse
Thanks! Does this one work with OOS?
MrGimpGrumble said:
Thanks! Does this one work with OOS?
Click to expand...
Click to collapse
I eventually plan to stop supporting oos
OOS is proprietary for one and such a system is almost never secure. And if you don't believe me just look at their past vulnerability announcements. Almost all oxygenos vulnerabilities come from the fact that oneplus finds loopholes around Google's CTS. Who knows what other holes they've opened up that Google forgot to add checks for
Further, oos has many "memory optimisation" drivers that directly try to access ram and break most of the security features I'm implementing. Most custom ROMs do not have these and the drivers can be safely disabled
I will also add that this kernel is almost functionally identical with kirisakura kernel. Yes I might merge upstream slightly faster but other than that there is no difference that you would notice. The only difference is that I'm enabling all the security features that must be enabled - especially CFI and shadowcallstack which come standard on any Google pixel device
As for wireguard I just think running a VPN in kernel space is a very bad idea. Not to mention I have confirmed that on Android the tunnel leaks ipv6 traffic if you're not careful and no, disabling ipv6 is not the solution. The userspace go implementation is much safer and I mean it. The userspace implementation almost never leaks ipv6 traffic. Not to mention Go is a much safer language than C
okay new update is in the attachments
changes: linux 4.14.156
upstreamed to oos open beta 6 (doesn't mean fixed fingerprint yet)
upstreamed wifi driver and audio driver to latest caf tag (LA.UM.8.1.r1-12200-sm8150.0)
yama is now at level 3
all upstream changes from kirisakura. except for wake gestures as lineagehw seems to already have those
oos users should disable smart boost from settings
okay new build is here
changelog:
linux 4.14.157
upstreamed sdfat driver
fixed a weird kernel panic that happened on anything other than oxygenos when the device was fast charging from a very low battery
anupritaisno1 said:
okay new build is here
changelog:
linux 4.14.157
upstreamed sdfat driver
fixed a weird kernel panic that happened on anything other than oxygenos when the device was fast charging from a very low battery
Click to expand...
Click to collapse
work on pa?
ryshd296 said:
work on pa?
Click to expand...
Click to collapse
Please test it and let me know
It should boot on any ROM that can enforce selinux
anupritaisno1 said:
Please test it and let me know
It should boot on any ROM that can enforce selinux
Click to expand...
Click to collapse
This sent me into an immediate Qualcomm crash dump upon booting on both stock OOS and Omni for OnePlus 7t global variant.
Previous releases as well, not just the newer release.
scott.hart.bti said:
This sent me into an immediate Qualcomm crash dump upon booting on both stock OOS and Omni for OnePlus 7t global variant.
Previous releases as well, not just the newer release.
Click to expand...
Click to collapse
Please duplicate the crashdump message exactly
Especially send the "PC at" line and the error message if present
If the error message is blank please mention that it is
If you get a PC at __cfi_check_fail message please mention this
@scott.hart.bti still waiting for your response
Please send the crash log if possible
Do i need ma
gisk companion for this?
psychemisha said:
Do i need ma
gisk companion for this?
Click to expand...
Click to collapse
No you don't
However somewhere around ob4 maintaining compatibility with oxygenos became next to impossible without breaking custom ROMs
I think most users are still on OOS. If not I can just release builds for custom ROMs
Costum plz
is the development stopped?
I've been trying to update the kernel version on the HD 10 Suez 2017. I am using the sources from this link https://github.com/chaosmaster/android_kernel_amazon_suez/tree/lineage-16.0 on the master Branch I cannot get the the device to boot when building a kernel. However if I use the Android 16 branch, I can get the device to boot but I have an irregularity that I cannot seem to figure out. The screen will go black. It doesn't turn off it just goes black that's if you stop touching the screen or you stop moving the device. aside from that I have not been able to see if anything else is affected but I can get the kernel versions updated at least two 3.18.22 but I haven't continue further because of this the weirdness. I have not tried any other sources as of the moment but was wondering if anybody is run into this problem.
Would this help at all? It’s the googlesource for MTK kernel 3.18
Have you tried building lineage 17 or 18 yet?
https://android.googlesource.com/kernel/mediatek/
https://www.xda-developers.com/alldocube-x-10-5-inch-android-8-1-oreo/
Mt8176 running Oreo 8.1. Supposedly project treble compatable too. Could this mean 8.1 for the HD10 and/or treble support too?
Fix MediaTek no display bug
MediaTek has a family of SoCs that are used in (among other devices) several Chromebooks, such as the MT8173 used in the Lenovo N23 Chromebook and Acer Chromebook R13, plus the MT8183 in the Lenovo IdeaPad Duet Chromebook. We are working hard to have better upstream support for these devices, especially Enric Balletbo, who has been involved in kernel and Chromebook upstream support for a long time. The displays on MT8173 based devices were not functional when using the mainline kernel, this has been rectified with this kernel release.
The cause of this defect was a conflict between the clock and DRM drivers. He found that two different drivers - the MT8173 clocks controller driver, and the MT8173 DRM driver were using the same compatible string mediatek,mt8173-mmsys. This is a problem since in such cases only the first matched driver will be probed. The other one, in this case the DRM driver, never gets probed and so the display was not made available for use. In a collaborative effort between Enric Balletbo and the MediaTek SoC maintainer Matthias Brugger, a top level mmsys driver (Multi-Media subsystem) was introduced to register the clock and DRM drivers. See 1, 2.
Linux 5.8 probably fixed your display bug. What is the chance of moving this tablet up to android 10/11?
source:
https://www.collabora.com/news-and-blog/news-and-events/kernel-5.8.html