Slashdot Mirror
UEFI Formed to Replace BIOS
anonymous cow-herd writes "Businesswire reports that several leading technology companies including Intel, AMD, Microsoft, IBM, Dell and HP and others have formed the Unified EFI Forum. The non-profit corporation will assume responsibility for the development and promotion of the EFI specification, a pre-boot interface originally developed by Intel that is intended to replace the aging PC BIOS."
Read more about EFI here.
They are a bios firm.
http://www.insydesw.com.tw/
It doesnt adapt well to serial consoles..
It has limitations on which parts of the disk it can boot from..
It's not scriptable..
It can't be configured in any ways other than what the "setup" program makes available to you..
OpenFirmware as used by SUN is much nicer, you can run diagnostics, write scripts, and get some low level information about the hardware attached... You can control the whole system from a serial console easily, and even install the OS from there..
You can also explicitely boot from any partition on your disk, instead of requiring a bootloader in the MBR to do the selection for you.
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
Yea right...
:-) ).
Linux has been booting on EFI Itanium boxes since the beginning, even before there was a 64-bit Windows (outside MSFT labs, that is
EFI is certainly not pretty, but it's still a great improvement.
The source code for Intel's implementation of EFI can be found at http://www.tianocore.org/
Also, this standard should finally allow seemless integration of new hardware onto the linux desktop. The main hurdle for desktop linux has always been lack of seemless driver integration.
And there's a link on the main page of the Intel EFI page.
In soviet russia, You ask not what country do for you, but what you do for country!
Oh wait...
What's wrong with the PC BIOS anyway? ... On a more sinister note, there's no mention in TFA of DRM and the idea of "trusted" computing.
According to the Overview page, Microsoft's listed as the only OS maker. First, why isn't Apple among the lineup? Novell? Red Hat Linux? Perhaps it's because they're not part of the real circle of friends...
Enter Microsoft's Trusted Computer Platform. According to the TCPA FAQ, the companies belonging to the alliance are: "Microsoft, Intel, IBM, HP and AMD". And let's take a look here...yep, they're all there. But what are they really planning?
According to the specifications page, nothing's listed as far as features that are to be included (" The UEFI specification is in development"). But currently, since there is no mention as to the true intent of this new technology, and right now the BIOS isn't broken, why reinvent the wheel? Load times are now less than three seconds, which is a tremendous step from BIOS beginnings. New equipment continues to be supported through new BIOS updates. So what do these companies need that the current BIOS can't give them?
Enter DRM. According to Microsoft's Patent on their DRM-supported OS, Microsoft has a few issues with the current BIOS...This AEGIS model requires a tamper-resistant BIOS that has hard-wired into it the signature of the following stage. This scheme has the very considerable advantage that it works well with current microprocessors and the current PC architecture, but has three drawbacks.
1) First, the set of trusted operating systems or trusted publishers must be wired into the BIOS.
2) Second, if the content is valuable enough (for instance, e-cash or Hollywood videos), users will find a way of replacing the BIOS with one that permits an insecure boot.
3) Third, when obtaining data from a network server, the client has no way of proving to the remote server that it is indeed running a trusted system.
So, Microsoft admits that there are flaws that prevent them from using the BIOS in their Trusted Computing platform. But create a new way of booting a computer, protect the technical details from public view, and put the power of the DMCA behind it, and you have a nice foundation into the DRM frontier.
OpenFirmware is SUNs brainchild, IBM (and Apple) adopted it in the powerpc development process
I'd write it in C, with some assembly hooks for the immediate boot process. The BIOS for modern PCs has to do a lot more than it used to: It has to initialize the hardware (which was the original intent, and is fine in assembly), but even that is getting much more complicated now, with networking, wireless, and video. Imagine having to edit BIOS options on a cluster of 300 PCs, it'd take you weeks using Award/AMI BIOS that don't have anything like serial or network console controllability. The big big use I see for something that isn't assembly is the booting process. That's the Achilles' Heel of the BIOS. Ever tried to boot off of a PCI-based SCSI adapter/drive? You probably can do that, but if you have two SCSI cards, you're screwed. The BIOS doesn't know or care how to tell the difference. USB boot is sketchy at best, and even CD-boot varies from manufacturer-to-manufacturer. The Award-BIOS source code is full of patches upon patches to support different quirky hardware that would be much better implemented in C. Things like large hard drive support would almost be trivial in C, but in assembly, you've go to change large swaths of code for larger bit-amounts for drive size. The newest size is 48-bit, which is a large number of TB, but drives keep getting bigger. The code itself is a gigantic mess of thousands of files with a loose grip on reality. There are circular dependencies from hell and code that just shouldn't exist.
-Jesse
Nothing says "unprofessional job" like wrinkles in your duct tape.
One thing I want to see in a BIOS replacement is for the new firmware to go into protected mode directly as one of the first things it does after setting up whatever it has to set up in real mode.
Then OS's that run on the new firmware standard would come in with a pre-defined protected mode setup ready to go and not have to mess around with switching into protected mode (OS's like windows and linux will need to be ported anyway)
I am not a systems programmer (I have programmed assembly but only as a userland programmer) so I dont know if doing this is actually possible or not.
Something else I want to see is a complete end to all limitations on what storage devices you can boot from and where on those devices you can boot from. (for example, any limitations on not being able to boot from partitions starting later than on the disk which I seem to remember used to be a problem)
You could even add a complete bootloader into the BIOS that would be able to read the boot sector from any hard disk partition, floppy disk (although in the ideal world, the floppy would disappear from the PC just like it has from the mac), optical media, USB storage device or whatever and boot that directly without the need for programs like GRUB and LILO and others to let you pick what to boot with.
By removing all the other legacy crap no-one really uses anymore (e.g. serial and paralel ports) you could create a new PC system without any legacy stuff. Done right, the only things that should care about the changes are operating systems like linux and windows plus device drivers for certain kinds of hardware.
It seems to be a fear of control. When you have complete co-operation between every single layer of a machine, the ability of those in co-operation to dictate terms increases dramatically. If the new CPUs and Mobos only work with EFI, and EFI only lets you boot into DRMed material, and they refuse to license their DRM methods for reasonable amounts, then they can functionally decide what can be done with the computers they create and sell.
I totally agree. You can write an absolutely tiny Forth interpreter and define the rest in Forth words.
I don't think we need to get too fancy and, it could even support multiple machine architectures, since once the interpreter is loaded, you're running in Forth.
MSBPodcast.com The opinions expressed here are my own. If you don't like 'em... Think up your own stuff.
we wrote the virtual machine in C - still small, and it's less maintenance that way
www.openbios.org
Apple is using EFI/UEFI for their new Intel Inside Macs, but they haven't joined the group yet.
With my PC, if I press a key, I get to a menu where I can choose from a number of boot methods, including network boot. My laptop's BIOS also has a special feature that allows me to boot straight off a CD/DVD by pressing a different key. With the right boot disk it's a breeze to, say, save files from affected partitions to USB/CD/DVD media, (re)install a bootloader or fix the partition table, without having to touch another computer.
So no, that's not really the key functional difference. It's the internals and their flexibility that counts far more. For starters, the PC BIOS requires some of the hardware to initialise to what's essentially a legacy mode, retained for compatibility with the IBM PC circa 1983 and resulting in some rather useless weight in modern operating systems.
It's just the processor being switched out. ,the new macs will still be macs, just with a different proc. So, why do they need to embrace PC technology for their non-pc hardware?
Macs arent becoming pc's, they're just using X86, they're still macs. It's not like you're gonna pop one open and bam, looks like your standard pc now, it's gonna be the same hardware, just running a different cpu arch.
x86 != pc
a PC could have a sparc cpu for example. x86 is just commonplace.
So,
Just to make it clear, Open Firmware implementations exist for Intel hardware:/ open-fw.pdf
http://www.firmworks.com/open_firmware/literature
Why change?
This page does a fairly good job of describing the limitations of the current legacy PC BIOS, and the ways EFI aims to improve the situation
/
http://www.kernelthread.com/publications/firmware
Open Firmware is a specification, not an implementation. In case you want an open-source implementation of Open Firmware, check out OpenBIOS. OpenBIOS can also be used as a payload for LinuxBIOS.
Intel has been implementing EFI in all it's high-end servers already. Itanium2 and Blade servers have it built in, and it's making it's way to the desktop. Personaly, I like the current ver. of EFI, but if DRM find's it's way in there, to hell with it.