Coreboot 4.12 Released - Drops Older Intel / AMD Platforms

[Guest]

Coreboot on x86 is like putting lipstick on a pig. We need a true open platform.

It's a base for Libreboot, so the project is worthwhile.

[Giovanni Fabbro]

(Full Disclosure: I work on coreboot as part of the Chrome OS firmware team, although not when the decision for coreboot was made or even with the first few generations of Chromebooks. I worked on coreboot in other companies before that. I also managed a few coreboot releases, including the just-announced 4.12.)

Is Fuschia still a thing? And would Google be looking at swapping out the firmware for that?

What's your take on TianoCore as an alternate?

[pgeorgi]

Is Fuschia still a thing? And would Google be looking at swapping out the firmware for that?

I don't work on Fuschia, so no idea? https://fuchsia.googlesource.com/fuchsia/+/master shows commits as of today though, so it's probably still being worked on?
As for swapping out the firmware, again, no idea? I don't even know what you mean by that.

What's your take on TianoCore as an alternate?

Years ago I made a comparison between a stripped-to-the-minimum coreboot and a stripped-to-the-minimum TianoCore (minimum on the file level: dropping another file would have led to a build failure on either). Both were supposed to be able to boot into a Linux kernel loaded from a virtual IDE disk on QEMU (that is, eliminating most of the complex hardware init code because there is no complex hardware init on QEMU).

Total lines of code that made up the coreboot + payload binary: 22k
Total lines of code that made up the Tianocore binary: 120k
(I didn't count the build systems or other auxiliary parts)

Does that answer your question?

[Giovanni Fabbro]

I don't work on Fuschia, so no idea? https://fuchsia.googlesource.com/fuchsia/+/master shows commits as of today though, so it's probably still being worked on?
As for swapping out the firmware, again, no idea? I don't even know what you mean by that.



Years ago I made a comparison between a stripped-to-the-minimum coreboot and a stripped-to-the-minimum TianoCore (minimum on the file level: dropping another file would have led to a build failure on either). Both were supposed to be able to boot into a Linux kernel loaded from a virtual IDE disk on QEMU (that is, eliminating most of the complex hardware init code because there is no complex hardware init on QEMU).

Total lines of code that made up the coreboot + payload binary: 22k
Total lines of code that made up the Tianocore binary: 120k
(I didn't count the build systems or other auxiliary parts)

Does that answer your question?

*shrug* not really.

[caligula]

But on other fora, the most vocal condemntations of AMD for this come from the same people who either support Intel's new-CPU-new-socket-new-chipset strategy or at best mumbled something about "well it's best to get the fully compatible chipset anyway..."

FWIW, Intel is sellling the same 14nm++++++ chip year after year. Is this fifth "generation" (= higher stock overlocking) of 14nm skylake already?

[Paradigm Shifter]

FWIW, Intel is sellling the same 14nm++++++ chip year after year. Is this fifth "generation" (= higher stock overlocking) of 14nm skylake already?

Forgive me if I'm reading this wrong, but basically this is agreeing with my subtle point about double standards?

And to be honest I've stopped keeping track of Intel's 14nm shenanigans. They've missed their 10nm roadmap targets so many times now I remain shocked that investors haven't started kicking up a fuss. But then I remember that investors are, by and large, not engineering inclined and only care about profits...

[caligula]

Forgive me if I'm reading this wrong, but basically this is agreeing with my subtle point about double standards?

Sure, I just disagreed with the first "new" in new-CPU-new-socket-new-chipset

[Paradigm Shifter]

Sure, I just disagreed with the first "new" in new-CPU-new-socket-new-chipset

Ah, I see. With "new" I meant new purchase for end user, rather than new technology, etc. But your reading also works well.

[Adarion]

...

Thanks for shedding more light on the situation.
I was guessing that more resources of the Coreboot team will go to Crombooks and we still have a lot of normal boards that would need a little polish (e.g. not having PS/2 kbd, no STR and the likes). But hopefully CPU vendors will wake up and also invest in a freedom firmware. It's just the better thing.

[NotMine999]

What I really like to show off to people is KolibriOS inside the BIOS. It's a small OS that fits on a 1.44MB floppy, so you can append the floppy image to your coreboot rom and seabios will load it from the rom. It takes like 2-3 seconds to boot (I'm using a crusty old pentium, might be faster on better proc), connects to network, has bunch of small games, text editor, webview browser and some other stuff. If it only had an SSH client, that would be really sweet. But good payload to have incase the SSD/HDD pops, you can still have a somewhat working PC with no disk...good enough to at least boot into as a sanity check to make sure nothing else on the PC is busted. I like it

Oh wun-der-ful. Yet another security vector that needs to be scrutinized very carefully.

"Oh-Gee-Wow" should never replace "Oh-Wait-What".

Just because it's buried in the BIOS doesn't mean it can't be exploited by some crafty hacker for nefarious purposes.