Matthew Garrett

Brief intro into why hibernate support is disabled On Linux when secure boot is enabled! The goal of secure boot is to prevent malware from getting into your kernel, by performing cryptographic validation of the bootloader and kernel during the boot process. If malware tampers with that (and it does! Search for "bootkits") then secure boot ensures that the boot fails and you can fall back to good components instead.

Everyone who has been waiting for over a decade for Linux hibernation to work when secure boot is enabled: I'm sorry, I'm fixing it

If you're invested in there being a Linux distribution that works better with laptops then I will happily make that happen for a mere $1 million a year and you will not be associating your brand with white supremacy in the process please enquire for references and my credentials

A bit over 15 years ago, a speaker at a tech conference used a couple of slides that were inappropriately sexual given the conference's guidelines. Afterwards the speaker apologised, the organisers apologised, there were process improvements, and it shouldn't have been a big deal. And then people decided that they needed to defend the speaker. Including someone who remains a well known senior Linux kernel developer at a large company.

Man based on the recruiter mails I get it is *astonishing* how many AI companies are looking for someone to secure their bare metal servers a significant period of time after they started selling product on them

We recognize that the NSDAP is closely associated with its founder, and that raises questions about what our support means. Our contribution is made to the NSDAP, not to an individual, and is not an endorsement of any individual's personal or political views. 1Password does not endorse hateful, dehumanizing, or exclusionary views.

I cannot understand the mindset of a (I hope small) subset of the free software community that software-related freedoms can never be impinged upon, but it's fine to remove arbitrary other freedoms that have a greater impact on people's lives

Not casting shade at the people who did this, they scratched their itch, and I appreciate that a thing I wanted to exist exists! I'm just sad that knowledge wasn't gained and shared in the process and now I need to sit for a while and reexamine some of my priors

"Microscopes are good for biology, not electronics" is a take I had not imagined hearing and yet the internet has blessed me with it (along with the suggestion that I should be using a soldering gun?) Ah yes let me work on these components that are less than a mm across with my bare eyes and a tip several times larger than that

And honestly I'm one of maybe at most 3 people on the planet who is looking at this with disappointment, even though it does actually give me some pointers into doing this the way *I* think it should be done for *my* problem. And yet there's a pang in my heart that now that this has been done, there's no incentive for anyone to learn how this all actually works and document it nicely. Instead we're left with whatever the fuck https://github.com/asokero/amix-040-060-port/blob/main/docs/060-CAMPAIGN-PLAN-260805.md is.

Looking at cryptography in a place that is absolutely fine but satisfies no meaningful threat model whatsoever and deciding that the correct terminology is "emotional support crypto"

But no, it's LLM output. Which, well, fine - if someone delegated the heavy lifting to the robot, I could live with that. But sadly it's a heavily documented repo where the documents are almost unreadable, even to someone like me who's probably as close to a domain expert in Amiga UNIX as any living person at this point. I can't build a narrative around the code. I can only attempt to relate the code changes to my understanding of what was there before.

Replacement laptop obtained, microscope turning up tomorrow, fun weekend adventure of seeing if I can figure out what's wrong with the broken board (and maybe bringing it back to life) ahead of me

I would simply not complain about being personally served if my solicitors had explicitly said they were not representing me in this matter

So I was excited when I heard about a project that had got Amiga UNIX running on 68040! The source code was lost in the 90s and this wasn't a rediscovery of that, but the kernel was built with debug symbols and Ghidra could get you something fairly close to source, and also it was based off the official AT&T SVR4 code which is available in various forms, so I'd hoped it was based on all of that.

I am actually very much in favour of Omarchy because I am very much in favour of all those fuckheads going off and being in one place where they will never bother me

And, well, I can see what's going on and I can come to the conclusion that it's very far from how I'd solve these problems, and also it doesn't solve Zorro III on 68030 which is what I actually want here. It's a thing that achieves the goals of its creators (Amiga UNIX runs on 68040 and 68060 now!) but communicates very little about the story of making that happen.

I managed to get Amiga UNIX working on an old Amiga back in 2002, and I've been carrying that thing around the world since then. A few years back I managed to get X11 working on an MNT ZZ9000 card, but only at limited resolutions - Amiga UNIX never gained support for mapping Zorro III cards into addressable memory space, and you need Zorro III for 32 bit support, so without it you're limited to a framebuffer that fits in a 24-bit address space

Things that your app can do that your website can't: 1) gain access to platform cryptography and keystore primitives 2) no that's it If you're not using (1) in ways that makes my security better, fuck off

Ok, I have perhaps got an idea here? The PD controller is receiving 5V from the charger, but is not receiving 3.3V on its input line. That /should/ be coming from a TCK107AG, which seems to be getting 3.3V on its input and on its control but does *not* appear to be generating 3.3V on its output. This would probably explain the behaviour, but if so I am going to have problems because this fucker is 0.79mmx0.79mm

Ah, but there doesn't seem to be 5V coming out of the PD controller that's definitely sus

https://www.da.vidbuchanan.co.uk/blog/android-c2pa.html is both funny and inevitable. Hardware-backed key storage is intended to prevent the private key material being moved to a different system, not prevent the private key material being used in ways you don't like on the system it's stored on. You are not going to stand a good chance of preventing the device owner using that key material however they want (which is why all the fearmongering about TPMs enabling meaningful DRM has always been bullshit)

Last year, Roy and Rianne Schestowitz were found to have defamed me and were ordered to remove that defamatory content and avoid any further defamation. They chose not to follow the whole of this order, and I sought to have them held in contempt of court. Today this was upheld: https://www.brettwilson.co.uk/wp-content/uploads/2026/07/Garrett-v-Schestowitz-and-anor-No-2-2026-EWHC-1948-KB.pdf

Behaviour seems broadly consistent with the EC never booting despite appearing to be powered, and now I'm wondering what that weird white blob is at the bottom of it.

Hrm I need finer probes to get at some of these components reliably. Cheap USB microscope seems to be doing its job well, though.

I am, For Reasons, suddenly very interested in the IBM MWave DSP instruction set and whether anyone has access to any of the apparently lost documentation for that (or whether anyone conveniently already has Ghidra support for it)

Vivid dream last night where @jwz@mastodon.social's blog was actually a webcomic and the most recent entry was extremely critical of me personally and I learned about this because Scruffy from Killer Net showed up and provided his role-defining line: "He's flamed yer, called yer a wanker in all the newsgroups". Anyway. I am well adjusted.

Given the potential worst case outcome for Roy and Rianne, I am glad that they were provided legal aid to obtain competent representation. The fact that both were found to be in contempt of court anyway demonstrates that this case was not frivolous or a SLAPP. They were instructed to cease defaming me - they chose not to.

Does anyone actually have real-world examples of widescale token theft occurring through compromised middleware boxes or accidental logging of tokens?

Roy was held to have provided misleading evidence in both oral and written evidence. Rianne was held to have provided both oral and written evidence that was not correct. My understanding is that this is generally considered to not be a good thing.

I ended up rereading the GNU manifesto when writing my FOSSY talk this week and goodness all the sections about how programmers should make money are very "Some of you may die, but that's a sacrifice I'm willing to make" meme anyway this is an allegory about how LLMs and free software interact: https://www.gnu.org/gnu/manifesto.html

I genuinely have no idea how all of this resulted in Roy admitting under cross examination that he eats baby food.

Multiple grounds were found not to be in contempt due to the judge's original order (https://www.bailii.org/ew/cases/EWHC/KB/2025/3063.html) not including a date that material was required to be removed by. No decision was reached as to whether or not continuing to publish this material was a breach of the order.

Being found to be in contempt of court carries a penalty of up to two years of imprisonment. This is by far not the most likely outcome, but sentencing will take place later this year.

As mentioned, multiple grounds for being held in contempt were not upheld due to the absence of an explicit deadline in the original order. This ended up being an interesting dispute on law and I am thankful for the efforts of my solicitors and barrister. While those grounds were not decided in my favour, the judgement does not support the behaviour of the defendents - see paragraph 50 of https://www.brettwilson.co.uk/wp-content/uploads/2026/07/Garrett-v-Schestowitz-and-anor-No-2-2026-EWHC-1948-KB.pdf

Wanted: chart of chartreuse price Vs ram price

Roy and Rianne have indicated an intention to appeal. I am not a lawyer, but my understanding is that they will need to demonstrate that the judge made either an error of fact or law in his judgement. People should obviously read the judgement and come to their own conclusion as to whether or not that might be justified.

ANYWAY this whole thing is incredibly rare - contempt of court is not a common outcome, especially for defamation. Given https://techrights.org/litigation.shtml I do not have high hopes that it will result in Roy and Rianne reconsidering their behaviour, but we will see if that changes during the lead up to sentencing and their appeal. It is likely that I will be awarded costs against both, which they will not be immediately able to discharge by declaring bankruptcy because they're already bankrupt.

A key part of Roy's defense was the assertion that the maintainer of the Techrights Gemini capsule was not under his control - this was rather undermined by the conversation logged at https://techrights.org/irc-archives/irc-log-tuxmachines-151225.html#tDec%2015%2017:45:02 which clearly shows this individual acting on Roy's behalf

Rianne's defense was largely based on her not knowing that what she wrote was also published on Gemini, despite her site having a clear "Gemini edition" link. The judge found that her testimony on this topic was "not correct". I'll leave it up to the reader to decide whether someone with a degree in computer science should be expected to understand the nature of the site they publish or not.

Why is S3 such a pain I am having to capture so much state

I really wish that Mastodon would remind you that editing a post with a poll to fix a typo will remove all the votes

The cat is now, obviously, just staring out of the window as if nothing inapropriate has occurred

@llorenzin@infosec.exchange They're required to do it, but failing to do so doesn't rise to the level of contempt of court

@valpackett@social.treehouse.systems @cubeos@hachyderm.io @david_chisnall@infosec.exchange It's certainly not running on the PSP on AMD, and as far as I know the chip to cloud stuff is only in the Azure Sphere setup running on Mediatek parts - the firmware running on Windows x86 is just TPM plus a firmware update channel