mjg59

16 ms·

241.
by mjg59 2y ago
The spec doesn't support it. Any USB 5.25" drive is going to either need special drivers or would have to pretend to be a generic mass storage device, which gives you a differently weird set of compatibility issues. Should be poss
242.
by mjg59 2y ago
Despite the punchline, the USB floppy spec is actually incredibly restrictive and only supports 720K, 1.25MB and 1.44MB formatted capacities. This means, ironically for a spec partially written by a friend of Raymond Chen's, that you c
243.
by mjg59 2y ago
I think your understanding is correct - for Rust to work well in Linux, the C maintainers need to maintain more discipline than has been absolutely required before (in that if they've changed those semantics then any breakage has been
244.
by mjg59 2y ago
> That is current maintainer doesn't want to be replaced with some new Rust-writing folks. Almost anyone paying Christoph is not doing so because of his involvement in the DMA subsystem - he's got a sufficiently strong involvem
245.
by mjg59 2y ago
The issue isn't just FFI - for example calling directly from Rust into C without any expression of the semantics means you have no mechanism for reasoning about ownership, and that removes many of the security benefits of writing code
246.
by mjg59 2y ago
I don't think I make this argument regularly and I wouldn't absolutely say that's the goal of the platforms themselves, but it's an effective outcome - in most cases participating in the program means accepting terms tha
247.
by mjg59 2y ago
It's probably worth mentioning that secure boot is trivially circumventable given physical access on any Intel Mac (including T2 Macs), so you want at least an M1 to feel safer here
248.
by mjg59 2y ago
X was originally an MIT development, and had no intrinsic ties to the underlying hardware (Sun implemented something called NeWS which ended up losing to X in the long run). Different vendors took the reference code and added device specifi
249.
by mjg59 2y ago
Generic VGA doesn't generally exist on EFI platforms, the firmware doesn't program the card into a state where the VGA registers are going to do anything useful. You should get a working unaccelerated framebuffer from the firmware
250.
by mjg59 2y ago
Technically not related to Solaris, since Solaris x86 existed and you wouldn't need it there, but yes, this was used on Sparc and any other CPU unable to run the card BIOS (including 64-bit x86 Linux, since virtual 8086 mode goes away
251.
by mjg59 2y ago
X does, modern X apps may not.
252.
by mjg59 2y ago
Xfree86 never did anything different to Xorg here. If you boot via CSM on a modern PC (which is really not something you should do, but, well) you should get Xorg running with the VBE backend using x86emu to execute the video BIOS, and if y
253.
by mjg59 2y ago
Microsoft is extremely chill about admin-to-kernel privesc, so that's not a meaningful barrier.
254.
by mjg59 2y ago
So don't normalise manufacturer locking. We're not going to prevent the bad thing from happening by arguing against the hardware that enables the bad thing - we're going to need to argue against the bad thing.
255.
by mjg59 2y ago
You think there's no value in your laptop being able to attest its state to your phone in order to give you confidence it hasn't been tampered with? That's something that would be entirely under your control.
256.
by mjg59 2y ago
Why do any of this rather than just have the GPU prove its identity to the streaming platform? You're adding a lot of complexity for no obvious gain.
257.
by mjg59 2y ago
If they don't have a GPU that implements this then the decrypted material would be available to the OS, which is precisely what the streaming media platforms want to avoid.
258.
by mjg59 2y ago
It could , but why? They've come up with a solution that avoids having to place any trust in the OS at all, so why introduce additional complexity and fragility?
259.
by mjg59 2y ago
On the contrary, it gives me various ways to determine that my laptop is in a trustworthy state before I type a password into it, and it makes it possible for Signal to verify that the server it's communicating with hasn't been ta
260.
by mjg59 2y ago
Joanna Rutkowska described a way to avoid this back in 2011 ( https://theinvisiblethings.blogspot.com/2011/09/anti-evil-ma... ), I extended it to be more usable in 2015 ( https://mjg59.dreamwidth.org/
261.
by mjg59 2y ago
> What do you mean "too new"? Like, you replaced your TPM? That's a thing on servers, but not laptops. I buy a GPU in 2025. I buy a new motherboard in 2026 and plug the GPU into it. How does the GPU learn about the new EK
262.
by mjg59 2y ago
There doesn't need to be a central CA, you just need to establish trust with the DRM vendor. The GPU vendors coordinate with Microsoft to make Playready work, Android devices have certs that can be validated by Google for Widevine, App
263.
by mjg59 2y ago
I'm not sure what they'd be displaying their streaming media on if they're not using one?
264.
by mjg59 2y ago
How would the GPU verify it's speaking to a real TPM? You'd need to bake the full set of legitimate EK cert CAs into it somehow (charitably let's say that's a signed blob that the driver pushes in at startup), but that&#
265.
by mjg59 2y ago
Yes, the memory controller can simply deny accesses to specific areas from the CPU while still permitting them from the GPU.
266.
by mjg59 2y ago
My fault, I meant x86 has no architectural TEE - various vendors offer their own weird things. But thanks, this is good clarification.
267.
by mjg59 2y ago
I'm extremely familiar with the capabilities of TPMs (I've worked on deploying remote attestation services at multiple companies), but here's the thing - streaming vendors don't use TPM-based remote attestation. None o
268.
by mjg59 2y ago
Not in general, Intel briefly had a program for allowing vendors to deploy apps on ME but closed it years ago. But yes, ME is involved in this for Intel iGPU.
269.
by mjg59 2y ago
The Playready docs make it clear the implementation is either in TEE or implemented in GPU hardware, and x86 has no TEE, so. You can easily find driver changelogs describing it being enabled for different hardware generations.
270.
by mjg59 2y ago
The details don't seem clear, and I don't know that there's necessarily a unique key rather than stuff being batched, but basically yeah there's a cert chain back to a "trusted" source

More ›