Coreboot Now Works On A ~$70 Intel Motherboard

[scratchi]

I got this board on ebay with a Pentium G860 proc for less than $100 CAD and flashed it with coreboot. Works well, even onboard video and audio...just no usb3, but all the usb2 ports work fine.

Phoronix encouraged me to do this, thanks

# dmidecode 3.1
Getting SMBIOS data from sysfs.
SMBIOS 2.7 present.
9 structures occupying 525 bytes.
Table at 0x7FEE8020.

Handle 0x0000, DMI type 0, 24 bytes
BIOS Information
       Vendor: coreboot
       Version: 4.6-2095-gdffd280b55
       Release Date: 11/13/2017
       ROM Size: 8192 kB
       Characteristics:
               PCI is supported
               PC Card (PCMCIA) is supported
               BIOS is upgradeable
               Selectable boot is supported
               ACPI is supported
               Targeted content distribution is supported
       BIOS Revision: 4.0
       Firmware Revision: 0.0

Handle 0x0001, DMI type 1, 27 bytes
System Information
       Manufacturer: GIGABYTE
       Product Name: GA-B75M-D3H

[Zan Lynx]

Our IBM server with UEFI takes over 5 minutes to boot. The OS portition (with sysvinit) takes less than 10 seconds.

So is 5 minutes the norm for UEFI while faster UEFI systems just skip initialization?

That 5 minutes is probably the ECC RAM initialization and test. A lot of servers do a full RAM test with rotating bit patterns before passing control to the operating system. I don't know for sure but I suspect they disable the CPU cache while doing it which would explain the slow speed.

[del_diablo]

will coreboot support one non-obsolete motherboard someday ?

Most likely no, unless there is a effort to select out new motherboards and dedicated reverse engineering.
I hope we get a modern AMD motherboard that supports EEC, sooner or later. Then again, its a community effort, so maybe never.

[pgeorgi]

EFI is awesome. Support for >2TB boot disks alone is a good enough incentive to get rid of the legacy BIOS

That is probably UEFI's greatest marketing hack: people think that booting from large disks is somehow reliant on UEFI. It's not.

Longer version: UEFI promoted GPT, a rather complex successor (which points out its UEFI origin) to the incidental and underpowered DOS partition table scheme that, indeed, was starting to lack free bits to represent ever increasing sector counts. Now GPT has the "protective MBR" feature, so the MBR location is typically used in some way, with enough space to keep a second level loader in there. That loader can read the GPT just fine, accessing larger disks. See GRUB2 in i386-pc mode, for example.

So yes, the UEFI Forum maintains the GPT spec, but GPT doesn't need an implementation of all the other ~3000 pages of UEFI spec to work. Question is, if they're actually a good standard maintainer - they took over ACPI and immediately restricted the usefulness of future versions of the spec through licensing: "A license to distribute, additionally reproduce, implement or otherwise use (other than to read only) the UEFI Specifications can be obtained ..." (yes, it's free - and they can cancel it anytime with 30 days notice for any or no reason), whereas before it was typical copyright legalese (don't copy, redistribute, ...) with no restriction on use, and no way to destroy your investment (whether time or money) by revoking your license.

Coreboot isn't even close to hoping to compete with BIOS, let alone EFI.

coreboot's scope is very different from BIOS or UEFI, so we don't hope to "compete" with BIOS or UEFI.

We provide BIOS or UEFI client interfaces if people need them (and the need for UEFI wasn't yet large enough to warrant the necessary large scale contributions that could turn TianoCore into something useful on real x86 hardware). We also provide other boot interfaces: FILO (a grub1 fork with various enhancements), depthcharge (ChromeOS verified boot solution), Linux as payload (so you can program your boot process with real scripting languages, and use a real network stack if you need to), and so on.

[GreatEmerald]

Our IBM server with UEFI takes over 5 minutes to boot. The OS portition (with sysvinit) takes less than 10 seconds.

So is 5 minutes the norm for UEFI while faster UEFI systems just skip initialization?

No, 5 minutes is too much as well. My current UEFI takes 10 seconds to boot.

[ahlaht]

There are some UEFI systems that voluntarily break the UEFI spec to disable things that are useless, allowing faster boots (usually such UEFI systems also have an option to turn it off, that is, allowing to make it compliant again). It's like replicating Coreboot, but from the wrong end.

Our IBM server with UEFI takes over 5 minutes to boot. The OS portition (with sysvinit) takes less than 10 seconds.

So is 5 minutes the norm for UEFI while faster UEFI systems just skip initialization?

(Also I can't really see much use for systemd on that server, heh)

[GreatEmerald]

I don't know what UEFI systems you use, but my one boots faster than my monitor.

There are some UEFI systems that voluntarily break the UEFI spec to disable things that are useless, allowing faster boots (usually such UEFI systems also have an option to turn it off, that is, allowing to make it compliant again). It's like replicating Coreboot, but from the wrong end.

[gigaplex]

you have very slow monitor

BenQ E2220HD. It's not fast, however it doesn't feel significantly slower than most monitors I've used lately.

[pal666]

I don't know what UEFI systems you use, but my one boots faster than my monitor.

you have very slow monitor

[gigaplex]

Coreboot beats BIOS and UEFI by a long shot. The technical basis for Coreboot is much better. UEFI is "everything and the kitchen sink, let's boot an entire different OS while booting because people love waiting". Coreboot meanwhile is "minimum hardware is up? Pass control, NOW!" And the latter is a much more sound idea. Why would you ever boot an OS whose only reason for existing is to boot another OS, instead of just booting the other OS? If you need multiboot, you can use GRUB as a payload directly. If not, just boot Linux directly without waiting at all. Which can result in 3-second boots, and that's from pressing the button to the desktop appearing. Impossible with UEFI.

I don't know what UEFI systems you use, but my one boots faster than my monitor.