00:14 <Jookia> apritzel: bunnie made that slide of fear vs known threats, and people are really scared about the extremely hard to implement attacks like hardware backdoors, chip doping, etc- but ignoring things like buggy boot ROMs

00:46 <apritzel> yeah, I agree. To top this off: in contrast to Intel CPUs, to the best of my knowledge no one has actually demonstrated a successful Spectre/Meltdown attack on a vulnerable(!) Arm based CPU

05:20 <KREYREN_oftc> > <apritzel> and I find it amazing that software people are always scared about CPU errata, when there are so many more bugs and vulnerabilities, with a much higher hit rate, lurking in software < I can fix software, but i can't fix CPU chipsets

05:22 <KREYREN_oftc> > norton> KREYREN_oftc: https://0x0.st/Hde5.txt < Thanks! will be reading through it! ^-^

05:22 <KREYREN_oftc> That looks really good I don't see anything that would be indicative of an issue which is great! ^-^ Once I figure out the kernel for Teres I will figure out a way to maintain these boards too bcs thats really cool

05:23 <KREYREN_oftc> also more than likely enables better development of teres-2 as i can use OLIMEX's SOM standard with the A20 on it

08:34 <KREYREN_oftc> fraolt, do you have any issue/lore that is relevant to the anx6345 being broken on ~6.7.2+ ? i would like to document it

08:34 <KREYREN_oftc> unless those mentioned are already sufficient https://linux-sunxi.org/Olimex_Teres-A64#Broken_ANX6345_on_Linux_6.5.2B_.28non-LTS_version.29

08:47 <fraolt> My current understandig is, that the bug in 6.5 and 6.6 are two different bugs. diego71 tracked down the specific commit in 6.5 that broke the panel (see his post on the sunxi mailing list). The reason here is probably that pre-6.5 the SoC was able to closely fulfill the panels request for a specific rate (because tcon was using pll-video0-2x as it's parent). 6.5 only allowed pll-mipi as the parent and pll-mipi was probably not able to give

08:47 <fraolt> the rate as closely as requested by the panel.

08:48 <fraolt> In 6.6 I added support for pll-mipi to set a closer rate by a) overshooting and b) setting the parent rate.

08:49 <fraolt> It seems that allowing to overshoot fixes the issues on teres-I but the fact that it can set the parent rate, breaks it.

08:51 <KREYREN_oftc> 6.7.4 reported broken by gentoo without the provided hotfix patch btw

08:51 <fraolt> That's the only explanation I have for this weird behaviour. I'm currently at a loss, why setting the parent rate breaks it. As per your suggestion, I asked in the olimex forum if somebody can spare their teres-I. The post is under moderation.

08:52 <KREYREN_oftc> fraolt, olimex post noted i try to ask some users who want to gift the device for development

08:54 <KREYREN_oftc> fraolt, you said in the post that you are located in germany, are you by any chance close to any hackerspace?

08:55 <fraolt> Yes, but I have no relationship with any.

08:56 <KREYREN_oftc> i will likely be self-manufacturing few teres-1 boards soon for the teres-2 development, they should be identical in terms of components, but the wiring will be different (basically turning that design into a System On a Module) that i intent to distribute to hackerspaces

08:56 <KREYREN_oftc> fraolt, can you DM me which one? i try to prioritize it

08:57 <maz> apritzel: errr... yes, we have demonstrated it a while ago. full Meltdown was shown on partner silicon, including across VMs.

08:58 <maz> and it is hardware you have access to (it's called mystery-roach for a reason).

09:00 <fraolt> KREYREN_oftc: I'm on matrix and don't know how to cross-DM. :-D Can you drop me a mail? https://lore.kernel.org/lkml/20230807-pll-mipi_set_rate_parent-v6-0-f173239a4b59@oltmanns.dev/

09:00 <KREYREN_oftc> @5m5z3q888q5prxkg:chat.lightnovel-dungeon.de

09:53 <diego71> fraolt: if you need more test, you can email me

09:58 <fraolt> diego71: Thanks, will do. Let's see how far we can get with remote debugging. :-)

14:48 <norton> Kreyren have you got my A20 Micro vulnerability report?

15:22 <KREYREN_oftc> norton, yes i did thanks! <3

15:23 <KREYREN_oftc> it's a major help to get the software side of OLIMEX's products better

17:59 <norton> meta-k,j,j,k

18:50 <machinehum> KREYREN_oftc: Just hackerspaces in Germany?