Intel Puts Out TianoCore For Broadwell, Bay Trail

[phoronix]

Phoronix: Intel Puts Out TianoCore For Broadwell, Bay Trail

TianoCore, the open-source UEFI implementation backed by Intel that was just talked about in Google Pushes "Project PIANO" Into Coreboot, already has support for Intel's yet-to-be-released Broadwell and Bay Trail processors...

http://www.phoronix.com/vr.php?view=MTI4ODY

[zanny]

I think the way the article is worded might conflate what this means - it is pointless to have TianoCore support for processors where the chipsets they require can't be intialized by coreboot. Hell, coreboot still can't start any of the X58 chipset boards, or anything more recent.

TianoCore requires something to initialize the hardware, and coreboot can't do that on any modern Intel chipset from the last half decade. It is also mostly Intel's fault for not providing any documentation for coreboot devs to work with, where AMD has done a (barely) better job at it - some chipsets as recent as Llano are supported. Barely.

I think the only reason coreboot is even news is because unlike the chipset black box of desktops, Google has more influence on ARM devices running Android, and can pressure chipset manufacturers into using coreboot for Chrome OS. But that won't make Intel or AMD budge and be more proactive in supporting it.

[efikkan]

I've been wanting Coreboot for a long time to avoid the crappy EFI. But if EFI is implemented in TianoCore then I don't see much use of Coreboot anymore, except for it being open of course.

[zanny]

I've been wanting Coreboot for a long time to avoid the crappy EFI. But if EFI is implemented in TianoCore then I don't see much use of Coreboot anymore, except for it being open of course.

TianoCore can't initialize hardware. It must be the payload of something that can actually set up the RAM and various system busses. Coreboot is the only major open source attempt to do that, and it has woeful support on anything from the last half decade. And coreboot itself is not necessarily a complete boot environment, you need to give it something to payload, be it TianoCore (for EFI functionality), SeaBIOS, (for classic BIOS / AHCI functionality), a bootloader, or a kernel (which wouldn't have AHCI tables, or tunables in a bios, or efi variables to interact with).

[efikkan]

TianoCore can't initialize hardware. It must be the payload of something that can actually set up the RAM and various system busses. Coreboot is the only major open source attempt to do that, and it has woeful support on anything from the last half decade. And coreboot itself is not necessarily a complete boot environment, you need to give it something to payload, be it TianoCore (for EFI functionality), SeaBIOS, (for classic BIOS / AHCI functionality), a bootloader, or a kernel (which wouldn't have AHCI tables, or tunables in a bios, or efi variables to interact with).

I see why TianoCore would need Coreboot, but I don't see why Coreboot need EFI. Are you aware of any drivers which rely on EFI?

[zanny]

I see why TianoCore would need Coreboot, but I don't see why Coreboot need EFI. Are you aware of any drivers which rely on EFI?

Nothing depends on EFI, but having coreboot EFI support means you can payload any OS that uses it. Otherwise you need software specifically targeted by coreboots payload process. Same thing with legacy bios - coreboot can boot directly to grub, but if you don't want to use grub having seabios on top means you can boot any classic bios OS.

[pgeorgi]

I think the way the article is worded might conflate what this means - it is pointless to have TianoCore support for processors where the chipsets they require can't be intialized by coreboot. Hell, coreboot still can't start any of the X58 chipset boards, or anything more recent.

In fact, coreboot supports Sandy Bridge and Ivy Bridge, both newer than X58.

I think the only reason coreboot is even news is because unlike the chipset black box of desktops, Google has more influence on ARM devices running Android, and can pressure chipset manufacturers into using coreboot for Chrome OS. But that won't make Intel or AMD budge and be more proactive in supporting it.

AMD and its partners are actively contributing to the tree. As for Intel, see http://www.techonline.com/electrical...-Architecture- they acknowledge it, support it somewhat (with binary components) but won't provide any source. Documentation was "ok" for some corporate developers (which is how the support was done up to now), but these days Intel doesn't publish (and maybe even write?) relevant documentation anymore.

[pgeorgi]

TianoCore [...] already has support for Intel's yet-to-be-released Broadwell and Bay Trail processors...

No, it hasn't. The code in question is some generic glue code, and what Intel released is the wrapper code for the components that Intel (as usual) won't publish (I'll be happy to be proven wrong).

The situation with UEFI is weird: Intel writes their driver code against an ancient EFI version ("EDK1117" - 1117 means november, 17th. They mean 2006).

Then they provide compatibility code (as part of Tiano) so this code can be used with current EDK2 (which contains secure boot support). This compatibility code, a frozen copy thereof, is what they published here. All current UEFI vendors use current EDK2, as you can discern from their press statements.

I assume they do this so some of their partners can use the ancient code based they forked from (without the new compatibility package) - eg. Apple, which still uses some EFI 1.x version - with current drivers, but don't know for sure.

[zanny]

In fact, coreboot supports Sandy Bridge and Ivy Bridge, both newer than X58.

Got any sources? I've only ever referred to coreboot support listings, but I've never seen anything past LGA 775 chipsets.

[pgeorgi]

Got any sources?

http://review.coreboot.org/gitweb?p=...868bf6b95fdd3e

I've only ever referred to coreboot support listings, but I've never seen anything past LGA 775 chipsets.

The wiki pages are manually maintained. As wikis go, they're outdated.