!UUYziobKGGxpovWyAN:nixos.org

Robotnix

244 Members
Build Android (AOSP) using Nix | https://github.com/danielfullmer/robotnix75 Servers

Load older messages


SenderMessageTime
29 Oct 2024
@samueldr:matrix.orgsamueldrright, the last four were qualcomm01:06:47
@oak:universumi.fioak 🏳️‍🌈♥️I think Qualcomm announced recently they will lock the secure side of their SoCs so it's only going to be running TEE OS signed by Qualcomm keys in the future. Even for companies that make phones with those chips01:07:41
@oak:universumi.fioak 🏳️‍🌈♥️So phone makers won't be able to implement their own TEE OS on Qualcomm hardware01:08:16
@atemu12:matrix.orgatemu12Is anyone actually doing that?01:08:29
@atemu12:matrix.orgatemu12I'd assume everyone takes the vendor's software to get the device to work01:09:06
@oak:universumi.fioak 🏳️‍🌈♥️Well there are alternatives for what you can run as TEE OS using TrustZone01:09:08
@oak:universumi.fioak 🏳️‍🌈♥️Vendor's software is usually a pain01:09:50
@atemu12:matrix.orgatemu12I believe that01:10:03
@oak:universumi.fioak 🏳️‍🌈♥️But yeah at least in the past Qualcomm was very bad at mainlining their stuff to kernel01:10:16
@oak:universumi.fioak 🏳️‍🌈♥️They basically just forked the kernel at some point and hacked it to work with their SoC01:10:53
@atemu12:matrix.orgatemu12I think that's gotten better01:11:01
@atemu12:matrix.orgatemu12LOS just pulls regular upstream 4.19 into the FP4 kernel repo01:11:41
@atemu12:matrix.orgatemu12Straight from Greg01:12:00
@atemu12:matrix.orgatemu12 So it can't have be diverged that much anymore 01:12:25
@atemu12:matrix.orgatemu12I think Google's been applying some pressure in this regard the past couple of years01:12:50
@oak:universumi.fioak 🏳️‍🌈♥️Past few years I worked with the NVIDIA Jetson Orin (Tegra) stuff and that was also somewhat painful01:13:31
@oak:universumi.fioak 🏳️‍🌈♥️NVIDIA does really low quality software releases01:14:27
@atemu12:matrix.orgatemu12Yeah Nvidia is something I don't wanna deal with in my personal systems either and that's just for the GPU driver01:14:33
@rosscomputerguy:matrix.orgTristan RossEvery ARM machine is cursed in some way01:14:36
@atemu12:matrix.orgatemu12The least cursed might be Apple surprisingly01:14:59
@samueldr:matrix.orgsamueldr
In reply to @rosscomputerguy:matrix.org
Every ARM machine is cursed in some way
(the ISA doesn't matter... all machins are cursed in many ways)
01:15:12
@rosscomputerguy:matrix.orgTristan Ross
In reply to @atemu12:matrix.org
The least cursed might be Apple surprisingly
Because Asahi?
01:15:17
@rosscomputerguy:matrix.orgTristan Ross
In reply to @samueldr:matrix.org
(the ISA doesn't matter... all machins are cursed in many ways)
Ik, I'm more of meaning vendor wise lol
01:15:31
@oak:universumi.fioak 🏳️‍🌈♥️Well, Asahi seems like a lot of work :)01:15:41
@samueldr:matrix.orgsamueldr
In reply to @atemu12:matrix.org
Is anyone actually doing that?
I believe most OEMs are, at least a downstream from the BSP
01:15:45
@samueldr:matrix.orgsamueldr* I believe most big OEMs are, at least a downstream from the BSP01:15:52
@rosscomputerguy:matrix.orgTristan Ross
In reply to @oak:universumi.fi
Well, Asahi seems like a lot of work :)
But it's made good progress and it's very stable
01:15:59
@atemu12:matrix.orgatemu12Are you running it bare-metal?01:16:22
@rosscomputerguy:matrix.orgTristan RossYes01:16:30
@atemu12:matrix.orgatemu12Neat01:16:33

Show newer messages


Back to Room ListRoom Version: 6