True - but as the other HN thread (that I can't seem to find anymore ><) pointed out, if Microsoft's requirements were planned as an attack against free software, it would be an unnecessarily complicated and ineffective one: There is nothing stopping anyone from writing Linux drivers for the new RAID mode - and I guess if more models with this setup show up, eventually someone will write one.

On the other hand, Microsoft could have easily gotten a much more "robust" lockdown - by simply demanding mandatory, unchangeable Secure Boot like on the ARM tablets. But they didn't do that.

> - but as the other HN thread (that I can't seem to find anymore ><) pointed out,

Flagged as a dupe. It's here: https://news.ycombinator.com/item?id=12545966

One guy has done a nice work decompiling the BIOS and found that skipping over the hidden BIOS pages is hard-coded into the firmware. See this https://forums.lenovo.com/t5/Linux-Discussion/Installing-Ubu...

Can they not modify that BIOS code to not skip those pages?

I'm sure anyone with enough knowledge to decompile a BIOS would have obviously thought of that idea though. I'm guessing the BIOS images are signed or something?

Regardless, this smells evil to me...

>On the other hand, Microsoft could have easily gotten a much more "robust" lockdown - by simply demanding mandatory, unchangeable Secure Boot like on the ARM tablets. But they didn't do that.

Didn't you hear that Secure Boot is basically broken?

Those leaked provisioning certs can be blacklisted on new hardware

Is it meant as contra Linux though? Or just pro RAID?

RAID on a machine with a single SSD (and no provision to add another)? It'd make far more sense if the BIOS was setup so the RAID option was always disabled.

Edit: it'd be rather amusing and saddening if this whole controversy was caused merely by someone misunderstanding/misreading a perfectly sensible requirement to disable the RAID mode, and instead disabled the other options. Given the level of communication fluency among some programmers, I would not be surprised if this were the case; and the fact that the machine still boots into Windows because it has drivers might've let this slip past QA...

The choice seems to be between AHCI and sorta-NVMe, with the latter enabled in 'RAID' mode.

Sure, real/standards-conforming NVMe would be better -- and wouldn't require driver shenanigans in either Windows or Linux -- but there should be a performance advantage to 'RAID' mode in this case.

If it was meant contra Linux, it obviously failed as the Linux installer booted just fine.

Also making it impossible to install Windows 10 downloaded directly from MS.

Or the bug is that it fails to properly store the initial change?