Notices by Matthew Garrett (mjg59@nondeterministic.computer), page 7

@freetar @Suiseiseki your understanding of modern microcode is simply incorrect, and whether something is software or not is not determined by the tangible form it is fixed in

@freetar @Suiseiseki what has changed is that microcode is now software targeting the CPU's underlying instruction set, and that merely happens to be fixed in ROM.

@freetar @Suiseiseki further, the description of microcode may have been true in the 8086 days, but it's not for modern CPUs. Look at how microcode updates were used to provide mitigations for Spectre vulnerabilities - that wouldn't have been possible if its functionality was so constrained.

@lxo @freetar @Suiseiseki The perspective I take is that someone choosing to design a system such that the software is sufficiently hidden you could maybe describe it as hardware is in itself a hostile act - it's a choice to take control of that system away from its owners. By choosing to treat that as acceptable we provide a perverse incentive for vendors to make devices that are harder to modify to meet the owner's needs.

@Suiseiseki "This can include, for instance, microcode inside a processor, firmware built into an I/O device, or the gate pattern of an FPGA. The software in such secondary processors does not count as product software."

Things that are in ROM are still software. The FSF says so.

@freetar @Suiseiseki "In aspects that relate to their design, those things are software; but as regards copying and modification, they may as well be hardware" - they're software, the FSF just argued that they be treated in the same way as hardware.

@Suiseiseki the CPU microcode ROM contains software, as the FSF makes clear.

@Suiseiseki It's talking about CPU microcode, which is stored in ROM.

@Suiseiseki and if it contains a ROM encoding CPU instructions, it also contains software, as the FSF makes clear in the quote I provided

@Suiseiseki "The exception applies to software delivered inside auxiliary and low-level processors and FPGAs, within which software installation is not intended after the user obtains the product"

@Suiseiseki so, by the FSF's definition, my fridge contains software

@Suiseiseki so why does the FSF call it software?

@Suiseiseki you're ignoring the bit where software in ROM is called software

@Suiseiseki So there's no need for a device that boots from CD to run free software, since it's actually hardware and therefore out of scope?

@Suiseiseki It seems like the FSF disagrees with you - RYF hardware can't include a pressed CD that include proprietary drivers (https://ryf.fsf.org/about/criteria), and it's also made clear that software in ROM is, well, software

@Suiseiseki If I give you a computer that only has a CD drive and the computer boots from that, is the computer running software?

@Suiseiseki whoops, managed to post as a top level rather than a reply, but:

The ROM is hardware, what the ROM contains is software. Otherwise you end up arguing that software on a pressed CD is hardware, which is clearly nonsense.

Them: code stored in an immutable physical form is hardware
Me: an Ubuntu live CD is hardware

@Suiseiseki Hardware executes things. Software is what is executed. Making software immutable doesn't make it not software.

@Suiseiseki Code that executes on a general purpose core is software, end of discussion