GNU Linux-libre 6.3-gnu Cleans Up New ath12k WiFi Driver, Other New Additions

[phoronix]

Phoronix: GNU Linux-libre 6.3-gnu Cleans Up New ath12k WiFi Driver, Other New Additions

Following this afternoon's release of the Linux 6.3 kernel, GNU Linux-libre 6.3-gnu has already been released as this downstream kernel flavor that removes support for loading binary-only modules and trying to de-blob other driver code that otherwise depends upon "non-free" code...

https://www.phoronix.com/news/GNU-Linux-Libre-6.3

[Franco Castillo]

And how do you make ath12k work without firmware?

[Barley9432]

And how do you make ath12k work without firmware?

That's the neat thing, you don't.

[dylanmtaylor]

I myself strongly prefer open source, but I don't get the level of almost religious need for fully open source software to the point where you're not willing to use non-free firmware, and therefore can't get a fully functional device. It's not ideal, but most people are far more pragmatic than this.

[mirmirmir]

I myself strongly prefer open source, but I don't get the level of almost religious need for fully open source software to the point where you're not willing to use non-free firmware, and therefore can't get a fully functional device. It's not ideal, but most people are far more pragmatic than this.

Someone need to be on the extreme. Like remember we were afraid saying things on telephone because someone might tap it? Today we are fine with knowingly being spied. And people's resentment to this project shows us that we are fucked.

[andyprough]

I myself strongly prefer open source, but I don't get the level of almost religious need for fully open source software to the point where you're not willing to use non-free firmware, and therefore can't get a fully functional device. It's not ideal, but most people are far more pragmatic than this.

My devices are fully functional with Linux-libre. I guess you just have to decide what kernel you want to use and plan your hardware accordingly. Kind of like with all other kernels. Amazing, eh?

[phoron]

And how do you make ath12k work without firmware?

Two options:

• If you're rich: You buy Qualcomm and release that firmware as free software. Or if previous contractual history prevents you,you tell staff to make sure that's the plan for next devices.
• if you're poor: You don't buy hardware from Qualcomm until someone takes the first option. Eventually you don't buy any hardware, save money, and don't have to charge devices, update systems and risk intrusions.

Thanks a lot to linux-libre for their work. I always felt it should be the other way round. Upstream should be free as linux-libre and downstrean could add proprietary stuff if wanted. But since linux don't want it like this, linux-libre is needed.

[Developer12]

Scheduled reminder that linux libre have modified their kernel to openly lie to their users about whether they're vulnerable to various CPU defects and security issues, only because the default kernel message suggests installing a microcode patch to fix it.

Again reminded of https://lists.gnu.org/archive/html/info-gnu/2018-04/msg00002.html, in which a GNU project removes a warning message informing users that their CPU microcode leaves them vulnerable to CPU microarchitectural attacks

https://twitter.com/mjg59/status/1377402589386432512

https://twitter.com/mjg59/status/1377403213368815617
https://news.ycombinator.com/item?id=29674846

And, of course, it's not like there weren't megabytes of nonfree microcode in there when you bought your CPU:

https://twitter.com/mjg59/status/1129249130004377600

https://twitter.com/mjg59/status/1129124275464441856

https://twitter.com/mjg59/status/1129124367542018049
Refusing to patch it doesn't make you any more free in any way, and removing the ability to do so from your users is stripping them of their freedom.

[idrisz19]

I myself strongly prefer open source, but I don't get the level of almost religious need for fully open source software to the point where you're not willing to use non-free firmware, and therefore can't get a fully functional device. It's not ideal, but most people are far more pragmatic than this.

First you must understand why we use the term "free software" instead of "open source". As a matter of principle, don't you agree that one must've surrender or compromise one's own freedoms? You wouldn't sign away your freedom of speech when purchasing a microphone, so why would you sign away your computing freedom when procuring software? If a populace were to tolerate the normalization of signing away freedoms merely to use a product, would that populace still be considered just?

Disclosure: I have Guix (Linux-libre) on one hard-drive, and EndeavourOS (vanilla Linux and Linux-zen) on the other, and I use EndeavourOS more often than Guix. Even so, I think that we should admire the ambition of those seeking to build 100% free software ecosystems for themselves.

[phoron]

Scheduled reminder that linux libre have modified their kernel to openly lie to their users about whether they're vulnerable to various CPU defects and security issues, only because the default kernel message suggests installing a microcode patch to fix it.

I think I should leave you alone, but here I go for once...

Pedantic reminder that silence is no lie.
Maybe I would have acted otherwise, maybe I would have informed the user their hardware had a reported vulnerability and they should consider using something else.
Or maybe the message should say "Intel sold you a vulnerable CPU, go sue them!". But I might not dare say so much because I could be sued by Intel.
But I think the blame is on Intel for not opening up more their designs.
Anyone who is not Intel can't tell how many vulnerabilities a patch introduces.

And why should the kernel tell me about Intel's vulnerabilities? Should maybe my browser tell me ? or my boot loader ? or my mail user agent?

The reason linux tells of the vulnerability (it's not its responsibility anyway, but it makes some sense) is because linux is not only a kernel but it is also a firmware (in this case microcode) upgrade tool. Since it has the functionality of patching the microcode, it may make sense to warn that using that functionality may be desired in some cases (so the user can do whatever it needs or wants to do to obtain new patches). And Intel, as a linux contributor may find this more convenient than replacing all those sold CPUs with new fixed hardware. The linux community allows them to do as they do and keep their trade secrets, but that does not bind anyone downstream.

Once you remove the functionality of patching microcode, it makes much less sense to keep the warning to patch microcode. Maybe the reminder should be given by any tool the user wants to use to patch microcode (none, or maybe UEFI, or maybe a dedicated OS in a boot media they only use for that purpose, or something else).

And, of course, it's not like there weren't megabytes of nonfree microcode in there when you bought your CPU:

That's again not linux-libre's fault.

Refusing to patch it doesn't make you any more free in any way, and removing the ability to do so from your users is stripping them of their freedom.

That's like saying if I write a game without an embedded word processor I'm removing from my players the freedom to write novels.
Linux-libre does nothing to stop anyone from installing linux or any other kernel, so it doesn't restrict their freedom. It simply tries to do what it is advertised to do,
not what a hardware vendor would like it to do.

Maybe you just dislike the choices linux-libre has made, because you consider security as a necessity for freedom and you do believe non-free, opaque microcode patches improve security. If that's so maybe you should use linux instead of linux-libre, as you likely do. But that's no reason to criticize linux-libre to offer a different compromise.

Now, sorry if I'm maybe just overreacting, you didn't make such a big fuzz, just exposed your views like anyone else, most likely with intent to inform or help others.
I just tried to point out it's nothing illogical, just a different policy. As you seem to acknowledge this has already been discussed, so I'll leave it at this and simply hope we can agree to disagree.