Fully libre workstation

Anybody who managed to accomplish such a thing? If so, how? Counting in microcode updates, IME, PSP, obsfucated code marked as libre and so on, is it even possible nowadays? And if possible, is it usable? The first time I got stumped was while looking for WLAN chipsets that are FSF-approved, everything on RYF and ThinkPenguin (https://ryf.fsf.org/index.php/categories/wireless-adapters) (https://www.thinkpenguin.com/catalog/networking-gear-gnu-linux) comes with drivers that are yet to be implemented in the mainline Linux kernel, and the external drivers that you can slap in via headers/dkms are non-functional from my experience. However, after spending around 3 days trying to get them to work I've found out that you can use USB tethering from your smartphone for a portable workstation, my phone (a Google Pixel 4 at the time, now a Google Pixel 6) with Graphene OS on it did the job, while it is not yet FSF-approved due to the fact that, I quote >it comes with firmware programs for installation and it appears that at least some of them are binaries without source code. It is said to be “de-Googled,” but includes a way to download and install the nonfree Google Play program. I feel like I've taken the measures to maintain that my pseudo-Ethernet isn't being tampered with by installing 100% libre software on my phone (or at least the profile where I use USB tethering) The second time I got stumped (and thus gave up) was when Leah Rowe, the maintainer of Libreboot decided to include proprietary blobs into her project, and while she did start a side-project called Canoeboot which comes with 100% libre software, and GNUboot who also does the same, my thinkpad (a T480 at the time of writing this) is not yet supported by neither of those, and this is why I'm making this thread Is anybody here actually running a 100% Libre workstation? Are there any problems that make you want to switch to a regular, non-libre workstation? If so, which?

Bump because I refuse to believe that lainchan has nobody with a fully libre setup

>>86650 If you want to keep The Man out of your machine, why enlist a man insisting themselves to be a woman? fuarrrking grifter... Anyway I do not believe there is such a thing in terms of modern hardware, outside of purpose-designed hardware such as Raptor Talos II. If you'll settle for less than modern you could try building an Apple II. It's just not a thing that can be done without serious effort or at least an honest evaluation of the economics, the spying / ecosystem lock-in pays for most of most conventional hardware so you can expect to pay 10x "the usual".

>>86650 >wireless adapters fuarrrk. I miss my Mediatek MT7601U. it was an LB-LiNK WN-151. my tip is just risk it for LB-Link WN-151 it runs out of the box with foss distros after the driver got mainlined. it's faster and more stable than most USB wifi out there, somehow either atheros/realtek is fuarrrked in windows and linux. new WN-151s use a modern realtek chip so you can guess how bad things are in the current USB wifi market and no TP-links are not an alternative they have weak power and potential backdoors. fully libre workstation is hard and you get pretty low unixbench and I'm not writing rocket and weapons systems so I have no need to (yet?). the best I could do is snag a ME disabled - certified workstation out of pure luck. also I don't plan to use libreboot as it messes with the cpu clock and lifespan, and for a laptop "workstation" lenovo's vantage 80% battery is the best thing though it could be done with certain acpi calls so I am stuck with the dilemma of that. the other only alternative is either a talos or some rare asus board is I doubt possible to obtain with my geographic location and customs.

>>86650 it is not only intel me, hdd firmware, wifi, ethernet and even bus firmware probably supports out of band management ... The amount of closed source low level firmware in the modern computer is beyond imagination ... The only way is to have a firewall that has block all by default and allow only the small traffic you for sure use (and limit your use an connectivity ..)

>>86650 >Is anybody here actually running a 100% Libre workstation? I do run Debian (which also has no proprietary blobs in the kernel by default) and disabling the proprietary firmware is trivial. I switched my wifi card to be able to not have to run proprietary firmware blobs. I also run Libreboot. The main problem is that I want to run the proprietary microcode because it fixes bugs/vulnerabilities that you can't fix otherwise and are a major security problem. So, yeah, I'm probably as libre as you can realistically get, but it's still not enough by my own judgement. The hardware not being libre/open is an issue and even if you decide to run some of those RISC-V SBCs you still require hardware that is not free like hard drive firmware. If you want a fully free computer then reusing some old laptops just doesn't cut it anymore, sure a librebooted thinkpad with a fully free distro is a lot more free than what most people have, but it's still not satisfying to have to rely on the goodwill of Lenovo, your SSD manifacturer, and all those other corpos.

Why is this place so empty? It’s got all these boards yet it’s completely dead

>>86679 >If you want to keep The Man out of your machine, why enlist a man insisting themselves to be a woman? fuarrrking grifter... What does this have to do with anything? Do they write good code? Does it work? That should be the discussion.

>>87309 Heavy-handed moderation really kills people's motivation to post.

>>86650 I've found a dealer in my own country that installs and sells libre devices though they use coreboot and debian instead of libreboot and trisquel. Also it's kind of selling really old devices(thinkpads from 2012, the most powerful device they sold was T440p, they said that's the max they could get) for 3 to 4 times the cost they're in refurbished sites. Should I fall for it?

>>87388 >sells libre devices Doesn't really sound libre if it comes with debian and coreboot Libreboot, and canoeboot by extent have been made to be accessible for the end-user, I don't know jack soykaf about hardware or electronics but the whole process went smoothly. Just make sure to not get a CH341A (the industry standard for some reason) since it can literally fry the chip (there are ways to neuter it but I assume that you want it to go smoothly so I won't elaborate on it) https://libreboot.org/docs/install/spi.html#do-not-buy-ch341a

>>87389 They say its because they don't believe in the extreme free software philosophy of RMS and believes if you want to install non free software on your device, its up to you. But they claim the core boot they use is free of binary blobs. https://mostlyharmless.io/computers/#models Its kind of hard to find these specific think pad models specified in libreboot site in my country especially since its a third world country that only recently started using computers in businesses.

>>87406 >>free software philosophy of RMS and believes if you want to install non free software on your device, its up to you. Need a citation on that, the GNU FSDG prohibits even helping out users download any sort of non-free software. https://www.gnu.org/distros/free-system-distribution-guidelines.html >What would be unacceptable is for the documentation to give people instructions for installing a nonfree program on the system [...] Debian also isn't considered an FSF complaint distribution because: >Debian also maintains a repository of nonfree software. It asserts that this software is “not part of the Debian system,” but the repository is hosted on many of the project's main servers, so people are likely to learn from Debian itself about those nonfree packages by browsing Debian's package database and wiki, and then might install them. I feel like you're just pulling soykaf out of your ass.

>>87407 >a* FSF compliant distribution

>>87407 >I feel like you're just pulling soykaf out of your ass. And I feel like you didn't bother reading what I wrote? It says 'don't believe' not 'believe'.

>>86650 >the maintainer of Libreboot decided to include proprietary blobs into her project This was done because CPU's have nonfree microcode burned into them anyway and that microcode updates patch security vulns

The solution is to stop buying proprietary hardware, or stop caring. Intel, AMD, ARM CPUs have hidden hardware and firmware that goes way beyond ME and PSP. https://www.devever.net/~hl/backstage-cast >Intel and AMD x86 platforms confer a distinct sense of “pay no attention to the man behind the curtain” in their design, essentially trying to conceal the existence of these ancillary cores. AMD CPUs, for example, not only have the PSP, but also have an ancillary core known as the System Management Unit (SMU). Before Zen it used the LM32 ISA, now it uses the Xtensa ISA. Some references seem to suggest that modern AMD CPUs actually have multiple SMUs, with different responsibilities, but it's hard to confirm this. The conclusion here is that I don't even actually know how many of these SMUs a modern AMD CPU has. Moreover, we just happen to have managed to find out about AMD's SMU; does Intel have an equivalent, more effectively hidden? Probably, though I don't know. >Intel CPUs have something called the CRBus, which is an internal memory bus for access to control registers with its own address space. You can't access this address space directly; your access to it is locked down. The CRBus is used by the Intel ME to control CPU settings Intel doesn't want you to be able to twiddle directly; I believe CPU microcode is also able to access this bus, so that MSR reads and writes (for example) can essentially be implemented as proxied, and highly restricted, access to particular registers mapped on the CRBus. Intel doesn't document the existence of the CRBus anywhere, and ordinarily you would never know of its existence; we only know of its existence due to the work of some extraordinary hackers. This is another example of “pay no attention to the man behind the curtain”. It's quite fascinating how on modern CPUs is how there is almost always a deeper level of system innards, but which is invariably hidden behind the simulacrum of pretending that the hardware is no different than a Pentium III. In the x86 case, the “curtain” has been quite strongly reinforced by vendors — and had the letters “‘hardware’ (honest)” painted on it. >>86679 >Raptor Talos II >One of the most interesting things ever to happen in the open source firmware space, the release of Raptor's Talos II/Blackbird POWER9 systems, created an interesting opportunity here. For the first time, we got to see what a modern server CPU really looks like behind the scenes. What makes the POWER9 CPUs really interesting is that there is no “curtain” in the design of POWER9. The POWER9's equivalent to Intel's CRBus, which is known as SCOM, rather than being a secret the very existence of which is omitted from mention, is instead openly accessible. >Even more interestingly, however, we have an exhaustive picture of the entire “backstage cast” of the POWER9 CPU. These are all the ancillary support cores responsible for system boot, power and thermal regulation and other functions that need to go on in the background while the system is running: >1x SBE (PPE Power ISA variant); >12x CME (PPE Power ISA variant); >4x GPE (PPE Power ISA variant); >3x IOPPE (PPE Power ISA variant); >1x PBPPE (PPE Power ISA variant); >1x OCC (PPC405); >summing to 22 ancillary backstage cores, or 46 cores in total. Thus, almost half of the cores on the CPU module are actually part of the backstage cast! The “main” cores of a POWER9 CPU are just the tip of the iceberg. >Leah Rowe Never browse personal websites of open source developers. https://web.archive.org/web/20211010235840/https://vimuser.org/littleleah.html

>>86650 https://archive.fosdem.org/2023/schedule/event/rv_selfhosting_all_the_way_down/