Fedora Developers Discussing Possibility Of Dropping Legacy BIOS Support

[Giovanni Fabbro]

Not necessarily. If I remember correctly, SeaBIOS - default payload of opensource coreboot BIOS - is 32-bit.
It's also written mostly in C. And there's nothing UEFI can do that BIOS can't do at least theoretically.
If you can't find some UEFI's (anti-?) feature in a coreboot+SeaBIOS, maybe it wasn't important enough to be added.
GPT and >2.2TB support - everything is possible in the opensource BIOS, if there's enough dedication

What is the first line here?: https://en.wikipedia.org/wiki/SeaBIOS

[michaelb1]

What is the first line here?

1) Wikipedia isn't always correct
2) " open-source implementation of a 16-bit x86 BIOS" - implementation of original could surpass this original
3) 32-bit PCI BIOS calls support
And the latest seabios git does run the post code and bootup interrupts (int 0x19/0x18) in full 32 bit mode

[Giovanni Fabbro]

Thanks for the detailed response. Oddly I have a mega core Xeon Lenovo that calls it UEFI but it's a legacy boot menu and appears to be supporting EFI 1.2.

I already threw out a collection of Acer's, Dell's and others that had EFI v1 and were just impossible to setup properly beyond Windows.

There is no EFI 1.2. There was only 1.0 and 1.1. The version jumped to 2.0 and became UEFI. 2.1 is where Secure Boot was added, hence any OEM system (possibly not low-end Chinese tablets and the like because those are for the 3rd-world markets and don't need certification) that shipped with Windows 8 or newer had at least UEFI 2.1 because Microsoft mandated that all systems include UEFI with Secure Boot. I believe they've bumped the OEM requirements up to a newer version for more recent Windows releases (2.3.something IIRC).

Most native EFI systems will have a CSM for legacy BIOS support although specialty systems may not have switches present to enable it ("weird" targeted computers like Microsoft Surface's maybe).

Apple has a weird implementation too although newer Mac's generally don't have a problem with it. On those, you can install natively with UEFI and not use Boot Camp. Boot Camp is just a macOS utility to shrink your Mac partition and create one for Windows, and allows you to create a driver disk. There's nothing fancy about it and it isn't involved in the boot process. If you buy a current Mac, you can wipe it and install Windows without involving Boot Camp, but getting Windows drivers for Mac hardware components has previously been a headache without having Boot Camp make that driver disk. I don't know if that's changed. Apple Software Update for Windows does provide driver updates, but I don't know if it will get core drivers if you don't have them loaded, and AFAIK Apple doesn't submit drivers to Windows Update. This is going to be moot once they switch over to ARM-based chips though. Don't expect Microsoft to release Windows-on-ARM for public consumption because it's tied up in their special hardware OEM and embedded channels.

[AmericanLocomotive]

Dropping the "legacy BIOS" support is a horrible idea:
Not just there are a lot of "legacy BIOS" PCs, especially in a corporate world where the upgrades are slower than in the domestic environments.

The corporate world typically runs on a 4-6 year upgrade cycle. Almost every PC in a corporate environment will be running UEFI by now.

There are also a lot of really modern PCs running a coreboot firmware with a SeaBIOS payload - which is a modern "legacy BIOS" written in C.

I really think you overestimate the "a lot" of modern PCs running coreboot. Just look at coreboot's compatibility list - it's absolutely minuscule. Mainly just some Thinkpads, a couple of Macbooks, a handful of 2011-Era SandyBridge (or older) Intel boards and some old AMD APU boards. Linux has ~2% of the desktop marketshare, and of that Fedora is only about 2% of the Linux distribution marketshare. Then of those Fedora users, only a very small percentage will be bothering with things like coreboot. I really don't think that indirectly dropping support for SeaBIOS would matter very much at all to Fedora's marketshare.

I ultimately agree that dropping Legacy BIOS support wouldn't be a great idea at this point, but not for the reasons you posted. Corporate users upgraded their systems a long time ago, and really almost no one users coreboot in the grand scheme of things.

[Giovanni Fabbro]

1) Wikipedia isn't always correct
2) " open-source implementation of a 16-bit x86 BIOS" - implementation of original could surpass this original
3) 32-bit PCI BIOS calls support
And the latest seabios git does run the post code and bootup interrupts (int 0x19/0x18) in full 32 bit mode

A BIOS starts a processor in 16-bit mode - that includes SeaBIOS. UEFI is typically made x64 for native support of 64-bit operating systems, which is generally how TianoCore is built, although low-memory (1-2GB RAM) Clover Trail, Bay Trail, and Cherry Trail tablets have had 32-bit UEFI to meet Windows requirements, and booted a 32-bit OS image. Windows requires the UEFI bit depth to match the OS version, and UEFI boot is restricted to GPT while BIOS boot mode is restricted to MBR. Linux doesn't have either restriction. That stuff you mention is what most BIOS's do. (It also supports 48-bit LBA, but that doesn't mean that SeaBIOS runs in "48-bit mode"). Windows 9x ran in 32-bit protected mode, but still launched from a 16-bit BIOS.

[Giovanni Fabbro]

There are also a lot of really modern PCs running a coreboot firmware with a SeaBIOS payload

Absolutely untrue.

The only considerable number of systems that use it are Chrome OS computers whose marketshare is pretty much non-existant outside of the US education market.

Also FYI: Chrome OS itself doesn't use SeaBIOS to boot.

[michaelb1]

The only considerable number of systems that use it are Chrome OS computers whose marketshare is pretty much non-existant outside of the US education market.

If you'd look at coreboot board-status reports (one of the few ways to estimate a coreboot's internal marketshare), >90% of them report SeaBIOS.
This truly shows the uselessness of UEFI: when the people really have a choice - like in a coreboot world - the majority of them choose SeaBIOS.
People didn't got UEFI consciously - it got shoved down their throats together with a new PC,
just like no-one in sane mind chose a Intel ME or AMD PSP hardware backdoors consciously.
Many people don't even know that their PC is running a UEFI instead of BIOS, and don't give a crap what UEFI is

[Giovanni Fabbro]

If you'd look at coreboot board-status reports (one of the few ways to estimate a coreboot's internal marketshare), >90% of them report SeaBIOS.
This truly shows the uselessness of UEFI: when the people really have a choice - like in a coreboot world - the majority of them choose SeaBIOS.
People didn't got UEFI consciously - it got shoved down their throats together with a new PC,
just like no-one in sane mind chose a Intel ME or AMD PSP hardware backdoors consciously.
Many people don't even know that their PC is running a UEFI instead of BIOS, and don't give a crap what UEFI is

Nobody chooses SeaBIOS. Google mandates that on Chrome OS devices for legacy boot modes which are locked away to the end user without a lot of modification. SeaBIOS isn't even used by Chrome OS - it's locked away for developer-use only.

How much is 90% of zero, again?

[michaelb1]

Google mandates that on Chrome OS devices

Nobody cares about Chrome OS and their users - most of them aren't choosing a SeaBIOS vs Tianocore consciously (getting these google netbooks for dumb browsing of the Internet etc and many don't even know what is BIOS or UEFI), aren't the active members of coreboot community and I see only a few board_status reports from their machines.

Meanwhile, the majority of those who are upgrading their Thinkpads etc with a coreboot, are consciously choosing SeaBIOS as their coreboot payload. And last time I looked through the board status reports, >90% of them contain SeaBIOS at their config. If you don't believe me, you're welcome to check for yourself - https://review.coreboot.org/admin/repos/board-status Maybe that's because UEFI is a SystemD of a BIOS world - a horrible bloatware by design and by code, even a Tianocore. So the people aren't using UEFI when they're able to make a personal conscious choice, they're choosing a SeaBIOS

[AmericanLocomotive]

If you'd look at coreboot board-status reports (one of the few ways to estimate a coreboot's internal marketshare), >90% of them report SeaBIOS.
This truly shows the uselessness of UEFI: when the people really have a choice - like in a coreboot world - the majority of them choose SeaBIOS.

You're neglecting the fact that SeaBIOS is the coreboot default payload. It's also the most mature, least buggy and most easy to compile payload available for coreboot right now. Many people experience issues getting TianoCore to work correctly or even compile.