EFI and Linux: the future is here, and it's awful - Matthew Garrett

[rgzdev]

Secure Boot: How to use GPL'd code without contributing anything back.

[BernofMiracles]

No machines can host a 100MB UEFI BIOS on earth today. Not even servers. Flash part physics in play here. Besides, not everyone is using the same Tiano code. Even in Tiano there are EDK and EDK II. Let alone all the BIOS vendors. Besides why are the Linux guys worry about the actual BIOS code in the first place, whether 1 file or 7600 files? Linux consume APIs provided by BIOS, Linux does not produce BIOS code.

[zzmgck]

And how long until Intel and AMD processors refuse to execute firmware unless that firmware is signed? That's the next logical step for this "secure boot" stuff. After all, we can't have people hacking or replacing their BIOS to thwart secure boot.

[thomasggpm]

@zzmgck As of now, AMD is not only releasing it's hardware datasheets to the public, it also actively supports Coreboot development and with it _all_ current platforms.

So as far I am concerned, AMD right now are clearly the good guys and for the forseeable future there will be enough AMD hardware we can safely buy.

[zzmgck]

@thomasggpm AMD datasheets available to the public? I'll believe it when I see it. Where can I get the "AM3 Processor Functional Data Sheet", AMD document number 40778?

[aliancemd]

Microsoft has a very good history of trying to stop Linux progressing. He says that they have their point, but I think that there is a pretty big probability that they use secure boot mostly to make sure they dominate the market(cause from stats Linux and Mac grow faster than Windows now). I don't say that's the reason but it can be...

[BernofMiracles]

Tiano is Intel's reference implementation of UEFI. UEFI Spec does not require the use of Tiano implementation. Besides, do not tribute implementation bugs as spec being "awful".

No one system implements everything in the 1500-page UEFI Specification or use all 7601 files of Tiano code. The modularity and flexibility allow the pick and choose of the needed code, that is crucial to the faster boot time enabled by UEFI. Nobody is putting on a system a 100MB UEFI BIOS. UEFI can be in the 100K's.

[GreatEmerald1]

@BernofMiracles I'm pretty sure the vendors don't really care. If they did care, we would have pretty well-written BIOS all around already, and there would be few bugs in Tiano itself. So we will be lucky if the vendors don't actually introduce new bugs to the system!

[kazetsukaimiko]

@BernofMiracles Except in reality, almost everyone is using that very, 100MB reference implementation. He said it was this bad because it just is this bad.

[BernofMiracles]

test

[veritanuda]

Hmm ok so GG started blocking links so time to disguise. HyperTexTransPort colon slash slash coreboot org

[veritanuda]

Isn't this precisely what Coreboot  is set to address? Methinks everyone should get behind this project and make sure the OEM are compatible.

[AlexFromHowest]

Incredibly interesting. Thanks!

[Garegin]

this guy is the hope for the entire linux mac community. how do I donate to him so that I can get grub to work on a mac properly?