[RESPONDED] Coreboot on the Framework Laptop

Why would framework not try and crowdfund it ? I’m not sure I can deliver Coreboot on my own with no support from Framework nor any other devs help, as well as AMD support too most likely.

Right now, I’m just trying to compile what is available for the Framework 13th AMD and see if I can run it using QEMU.

1 Like

That’s perfect for developping I guess thanks !

Yes I’m referring to him, I don’t know if I’ll be of much help but m willing to check if I can help move this forward.

Thanks for the contribution you might do once this gets somewhere concrete

I’m happy to pitch in too

Right now, I’m just trying to compile what is available for the Framework 13th AMD and see if I can run it using QEMU.

Just a note, coreboot (and firmware in general) is hardware specific, so a coreboot ROM built for the Framework 13 AMD will not run in QEMU.

6 Likes

Azalea port update - Framework 13 AMD 7040 series

Hi All, I’m really happy to have all the interest here, so I thought I’d write an update. Sorry it got a bit long.

For anyone who is interested in testing, I’ll try to get a binary ready for you over the next week or so. I’m working with my team at AMD to get some binaries needed for the initial release pushed to the amd_blobs repo. This might take a bit of time as they test interactions between different versions of the binaries.

The first version of the Framework port will be using the FSP binary implementation that was used with previous AMD chromebooks and not openSIL. I don’t know yet when the Phoenix openSIL might be released publicly, but it will be a Proof-of-Concept only, similar to the released Genoa package. Once that’s released publicly, the plan is that the Framework implementation will be updated to test with either FSP or openSIL.

Before sharing the binary with anyone, we need to be sure that anyone testing with it has a hardware solution in place to restore their existing firmware if things go badly for them. I don’t want to brick anyone’s board if their system doesn’t boot with my image. My previous recommendation still stands, and shouldn’t be too difficult for anyone with decent soldering skills.

This requires that you have a SPI ROM programmer that will program 1.8V. I’d
recommend desoldering the SPI ROM chip from the board and soldering down
a socket so you can swap roms back and forth quickly. I think you’d want to buy
additional SPI ROM chips and set up a way to flash the chips off the board, then
drop them into the socket.

Known issues:

As I said before, suspend won’t be implemented, at least not initially, and maybe not ever. This includes both S3 and S0i3. The OS’s hibernate functionality is probably your best bet.

I haven’t tested USB4, so I suspect it will be broken and that those ports will only work as USB3. We should be able to get that working eventually, but I don’t want to promise anything.

Boot time will also initially be a problem. coreboot doesn’t currently update the APCB data with the information about the SoDIMMs installed, which means that the system will completely re-initialize memory on every boot instead of using the saved APOB data. This wasn’t a problem on chromebooks because they all used soldered-down memory.

Those issues aside, most of the basic functionality should be working. I do appreciate the offers to help test - that’ll help find any of those harder to nail down issues. Not dealing with suspend also makes this significantly easier. During testing, the hardest issues are always those relating to power management. coreboot uses a very small set of SMI handlers, which was historically another place for hard to solve issues.

Finally, for anyone not familiar with coreboot, there aren’t really a lot of setup options supported compared to a typical BIOS or UEFI firmware. Various people at coreboot are currently working to expand this and have a full-featured setup engine, but that isn’t currently available. Changes to functionality mostly get done at build time. If there are options that are important to people, I’d be happy to help coordinate work between people to make those options available.

I do want to note that I was supplied a Framework 13 device by AMD at Framework’s request, but as I said before AMD is not otherwise sponsoring this work. I work in the group that does the core firmware, which currently has the coreboot expertise to do this work. The rest of my team is helping as well, just not with doing the actual port.

If it’s only funding that’s needed (ie. not needing a special non-fuse-blown unit from FW?), perhaps consider crowd-funding this effort? I for one would be interested in pitching in to make this happen regardless of official plans.

If someone wanted to raise some money for additional features beyond what I have planned to be done by one of the amazing coreboot consulting firms, that’d be great. Please do keep in mind that the azalea port likely won’t ever be more than a proof-of-concept simply because of the state of the AMD FSP and openSIL codebases for the Phoenix chip. I don’t begrudge Framework at all for not paying to get a coreboot port done - that’s expensive and they’re still a small company. I appreciate what they’re already doing in pushing to make things open.

Martin

36 Likes

This is very exciting. I’m currently working on a different coreboot-related project, but after that’s dialled in, I might be willing to treat my AMD FW13 as a guinea pig for testing. I run Linux on it exclusively, and while suspend is nice to have, I bet I could live without it. I’ll keep my ears open.

2 Likes

Hello @Mario_Limonciello I haven’t been able to receive a response from Martin by mail, I don’t know if you have any way to communicate with him, I just want to know if I can help him any way move forward the Coreboot port. As well as discuss with him the current blockers and risks of trying to use it on a real laptop (maybe not usable or breaks everything (USB-C/Power Management, etc.).

I’ve sent Martin a mail a week ago, if he’s busy at the moment, IMO it’s okay I can wait more time for a response.

EDIT : I just saw @Martin_L_Roth’s update post, sorry for the ping

I thought it was possible to try and build it like the Coreboot starter/starting tutorial where you need to get Coreboot and the Seabios payload running in QEMU. But you’re most likely right, thanks for the information, I clearly wasn’t aware.

The starting tutorial is actually mostly intended to ensure your build environment works properly, and the QEMU board is an easy target to test this for as you don’t need to flash any hardware (but as mentioned, this QEMU ROM will only run on QEMU and a Framework ROM will only run on a Framework). The general flow of building coreboot for a particular board (like the Framework laptop) will be similar though, but you will need to change the configs appropriately.

2 Likes

Thanks for the details.

I’ve done flashing hardware and dumping firmware through SPI flash before, so this part is okay for me, as well as taking precaution around backing up the known good/and official BIOS before doing any hardware mod.

I’ve got also the proper hardware for flashing at 1.8V.

azalea port likely won’t ever be more than a proof-of-concept simply because of the state of the AMD FSP and openSIL codebases for the Phoenix chip

Will openSIL be better in a few years ? Won’t this allow this PoC to be more polished in the future once openSIL is way more ready for consumer laptops ? Or maybe become usable daily ?

1 Like

And there’s a mention on Phoronix, which links to the thread. This is very cool! Hopefully something Framework can build on for the future, perhaps. Or at least serve as a model Coreboot for AMD platform in general.

7 Likes

i wish Fw will release an official coreboot

3 Likes

Me too, me too. Intel Coreboot aswell (and I dont mean the chromebook edition)

To whomever tagged me, everything @Mario_Limonciello is correct.

4 Likes

Just for your information: Congratulations!

Arguably most exciting with the new platform support in Coreboot 24.05 is supporting the Framework 13 AMD laptop. This is the early support for the Framework 13 AMD and still does require binary blobs but at least this is progress being made and better than the default proprietary BIOS. It’s an experimental Coreboot port being worked on by various AMD firmware engineers and other stakeholders as an unofficial project.

20 Likes

Wow, awesome news!

2 Likes

Saw it too … and it is really nice. Waiting for the FW16 Coreboot :slight_smile:

3 Likes

Absolutely fantastic news!! Am I reading correctly that existing Framework 13 laptops available for sale today could have Coreboot installed on them? If not, what steps remain.

Many, many thanks to all the folks who put effort into this.