The Latest Open-Source AMD Firmware / Coreboot Happenings In Early 2021

[pal666]

warranty

has nothing to do with it. you can reflash your motherboard with something without any vendor opensource work. reflashing non-vendor-provided image should void warranty

[michaelb1]

I should also point out how complex and how much effort has been in these 'closed source' implementations. especially the high end motherboards for overclockers and enthousiasts. Their GUI's and option granularity has been amazing. switching to coreboot is a bad move, worst case they'll lose alot of functionality.

The only reason why a coreboot doesn't have a "Fancy GUI" with bells and jingles, is because its' pragmatic users didn't feel the need to create one, + for a lot of features like IOMMU it makes sense to enable but not to disable, and even with a wide range of my use cases I rarely touched a lot of BIOS settings more than once. But, if the need for an advanced coreboot GUI arises, of course it will be created. I.e. https://kingsgambit.dasharo.com/ workstation with a coreboot-based firmware has a custom GUI

[f0rmat]

This could be easily resolved by some "firmware restore" mechanism: many motherboards already have a "Dual BIOS" feature. Or, maybe just put a USB CH341A programmer + DIP8 remover + a couple of spare BIOS chips to a motherboard's package, together with a good instruction: the total price of these dirt cheap components is less than $5.

I remember getting a DIP8 remover and ordering a custom BIOS chip so that a Tyan server motherboard I had on a home server would boot dual core Opterons. When it first came out, the motherboard BIOS would not support dual core Opterons, only single core. (IIRC it was Opterons, but maybe it was an Athlon MPs - it was over 10 years ago.) It was rather cheap - less than 10 USD if I remember correctly.

[michaelb1]

So now i come back to a large OEM, and someone puts out a new EFI based on Coreboot and someone on Reddit or TikTok says it makes your gaming 200% better. A new sideload industry begins to blossom and that OEM is now suddenly confronted with a large warranty expense because once again thousands of noobs thought they were going to get 2 more fps in Call of Duty and end up with a non-bootable system and end up at the Geek Squad desk looking to dump their mistake.

This could be easily resolved by some "firmware restore" mechanism: many motherboards already have a "Dual BIOS" feature. Or, maybe just put a USB CH341A programmer + DIP8 remover + a couple of spare BIOS chips to a motherboard's package, together with a good instruction: the total price of these dirt cheap components is less than $5.

[michaelb1]

does this coreboot wants to be something universal compatible with all motherboards (current or future ones)

Yes, the maximal compatibility is desired, although in practice the coreboot devs could encounter various hardships like a different set of the non-AMD controllers (i.e. Super_I/O ) which require a specific code to work with them. However, if the AMD-related source code components would've been open source, that would've made the task of adding a support for a new motherboard - much easier.

Some of us are forced to dual boot (that paycheck thing), so does anybody know if this will work with Windows? Unfortunately, my financial circumstances cannot really justify me buying a separate computer just for Windows and have one just for linux (I wish I could).

Looking at the motherboards currently supported by a coreboot opensource BIOS, their ability to boot Windows depends on i.e. how good is the ACPI implementation: if ACPI for a specific board is too quirky, it could be that Linux still can boot on it while Windows is having bluescreens and other problems. So the ability to boot Windows varies between the different coreboot boards. However, with the support of a high quality coreboot consulting like https://3mdeb.com/ with their https://dasharo.com/ service, it could be guaranteed that a coreboot-supported board will be able to boot Windows as well.

[michaelb1]

As a Lenovo product manager told me, "what value does Coreboot provide?...Lenovo already owns or has rights to any EFI or BIOS structure it desires and can modify it to any customer requirement"

But, for that "can modify" to turn into "did modify", the customer requirement should be critical enough (i.e. a really major firmware bug) and reported during the first half of that product's official lifecycle. Otherwise: Lenovo, like any other laptop company, won't bother meeting the customer requirement: why spend time/money for that when a new 100500th laptop model is knocking the door? However, this common short-sighted approach results in a poor customer satisfaction and the product not reaching its' full potential (i.e. there are a lot of laptops with broken IOMMU).

The benefit of using Coreboot for end users is well understood, but outside Google, what would drive them (or any other large vendor) to embrace it?

If the end users will be able to access the source code of BIOS and fix its' problems by themselves, that will lead to a higher customer satisfaction with the less money/time resources spent by a vendor himself.

The benefit of using Coreboot for end users is well understood, but outside Google, what would drive them (or any other large vendor) to embrace it?

That's because there are rarely detailed changelogs for the BIOS updates / significant changes improving the user-experience; so the people rarely see a reason to upgrade. However, if the sources would've been open, there'd be a higher chance of major changes happening and giving more reasons to upgrade.

[intelfx]

Linksys pushed out their GPL changes for the Broadcom based WRT and it blossomed into an entire side load industry. Sales very much went up as people wanted to take advantage of the opportunity. But their warranty expenses tripled as noobs consistently bricked their routers and returned them to Circuit City/Best Buy/CompUSA and took advantage of generous return policies.

I remember walking through a warehouse at a now defunct retailer and there were 2 large groups of shelves containing hundreds and hundreds of white boxes. I asked him what all that was? He said those are Linksys WRT's that were bricked and returned, subsequently reflashed and reboxed and being sold as refurbished. Key item on the box was "30 day warranty" so Linksys didn't have to repeat the cycle.

So now i come back to a large OEM, and someone puts out a new EFI based on Coreboot and someone on Reddit or TikTok says it makes your gaming 200% better. A new sideload industry begins to blossom and that OEM is now suddenly confronted with a large warranty expense because once again thousands of noobs thought they were going to get 2 more fps in Call of Duty and end up with a non-bootable system and end up at the Geek Squad desk looking to dump their mistake.

Again I get the benefit of Coreboot being available to anyone, but until an OEM finally does what Linksys did years ago, they came out with a WRT that was expressly designed and sold to be sideloaded. Perhaps an OEM will feel adventurous and create a SKU that is "Coreboot Only".

Sounds like a problem with over-generous return policies and/or not enough scrutiny.

[ezekrb5]

As a Lenovo product manager told me, "what value does Coreboot provide?...Lenovo already owns or has rights to any EFI or BIOS structure it desires and can modify it to any customer requirement"

The benefit of using Coreboot for end users is well understood, but outside Google, what would drive them (or any other large vendor) to embrace it?

There is no money in it for them.

It has no warranty improvement benefits.

It doesn't extend/reduce the life of the hardware in any fashion (unless misconfigured)

A majority of customers use EFI/BIOS in consumer perhaps once maybe twice in the devices' lifetime. Traditional consumers rarely patch.

In the business space, perhaps a little more than that, at setup, for a repair or a patch event.

In other words, unless someone demands it (like Google) there is no biz case.

So again why would a major OEM spend all the money to go Coreboot other than to satisfy a small libre community?

The same granularity on the system handling can be achieved with coreboot as well, (for example on System's76 utility).

However the main reason to want coreboot is the security and privacy implication, the amount of vulnerabilities found in closed source firmware is HUGE. Also it forbids the user the ability to stop the intel management engine/ AMD Secure Procesor; which is an important lyability, as it's propietary code that has full access to anything happening in cpu, cache, ram, storage and network.

Just because most people out there doesn't care about this security issue, it doesn't mean they are right. The amount of users supporting the propietary and proven malicious code in past ocassions (incluiding lenovo's superfish and LGs television android firmware), is what drives the sales, it doesn't mean they are right.

[f0rmat]

It's generally possible to boot Windows with coreboot, both in PCBIOS and UEFI modes. This requires some care with the configuration that has to be done for each individual board and not all (as in: only a few) coreboot developers verify their work with Windows.

So yes, possible, but some boards may require some development effort.

Thanks for the insight.

[pgeorgi]

does anybody know if this will work with Windows?

It's generally possible to boot Windows with coreboot, both in PCBIOS and UEFI modes. This requires some care with the configuration that has to be done for each individual board and not all (as in: only a few) coreboot developers verify their work with Windows.

So yes, possible, but some boards may require some development effort.