OSDev.org

Hi,

I've been waiting for it to happen, and it seems it has - the upcoming "sevice pack 1" for Windows Vista includes full support for UEFI.

Microsoft is/was the last major OS without support for UEFI, so now I'm guessing it might not take too long (6 to 12 months?) before new "white box" computers come with UEFI. The old PC BIOS might be entirely obsolete in 5 years time.

Now might be a good time to have a minute of silence to mark the passing of the PC BIOS, and for each of us to consider how easily our OSs could support UEFI....


Cheers,

Brendan

Brendan wrote: Microsoft is/was the last major OS without support for UEFI

Funny, since M$ was one of the biggest contributors to the UEFI standard. Too bad the committee didn't make a prettier standard.

Hi,

bewing wrote:Funny, since M$ was one of the biggest contributors to the UEFI standard. Too bad the committee didn't make a prettier standard.

Pretty (clean and sane) standards are not in Microsoft's best interests - Microsoft has the resources to support a complex mess, while their competition may not...

EFI was originally an Intel specification, intended for Itanium. At the time of EFI's creation it would have been in Intel's interests to make EFI clean and sane, as Intel needed to convince OS developers to support Itanium. AFAIK Microsoft's involvement (and many other companies) started when Intel turned EFI over to a consortium (who renamed it "UEFI" and AFAIK have only really made minor changes for maintenance since - it's still mostly Intel's original specification).

Of course I'm not saying that UEFI/EFI isn't complex, but the complexity seems justified to avoid compatability and portability problems (without ugly hacks).


Cheers,

Brendan

You seem happy/excited about UEFI... Why? aren't you a little concerned Brendan?

http://en.wikipedia.org/wiki/Extensible ... l_property

DRM BIOS?

*cries*

Hi,

Brynet-Inc wrote:You seem happy/excited about UEFI... Why?

I have mixed feelings...

Part of me is glad that the crappy PC BIOS is dying and part of me likes new challenges. On the other hand, part of me wishes the entire world would stop while I implemented my OS (I'm still trying to catch up to the "long mode" changes, and haven't even started with VMX/SVM).

In general I'm not too worried, as I'm confident that the latest version of my OS can handle EFI once someone writes an appropriate boot loader - all the OSs code (except the current boot loaders) should work fine.

Brynet-Inc wrote:aren't you a little concerned Brendan?

http://en.wikipedia.org/wiki/Extensible ... l_property

Paraphrased:

"According to <someone with a biased opinion and something to gain>, one of the stated goals of EFI <which I've been unable to find stated anywhere> is to “protect hardware vendors’ intellectual propertyâ€

It wasn't that paragraph that I was referring to, while it does concern me that vendors could add restrictions that effect us, hobby system level developers.

This was the part that concerns me:

Its specification is not publicly viewable and EFI site suggests that it is a supporter of the Trust Computing Group.

I'm not against closed source, I'm a proud supporter of BSD/ISC style licences... but I'd personally hate dealing with a firmware that only offers me a subset of functionality if I'm not running a Microsoft certified product. (ACPI implementors already seem to be doing this...)

I'd be worried that with a firmware like this the US (Stupid law makers that they are.) would mandate all computers run it, that is, along with its trusted computing and DRM modules.

It would just be so easy. There would be record companies, movies companies, and every other fanatic courting Congress to make laws like this. And, because such a firmware would already be in common use, it would create the perfect opportunity.

I can see this happen. Be it 'for the children', or for 'national security'. It's all bullsh*t.

I'm quite happy with my BIOS, thank you very much.

The UEFI specs can be downloaded here
http://www.uefi.org/specs/

Hi,

Brynet-Inc wrote:It wasn't that paragraph that I was referring to, while it does concern me that vendors could add restrictions that effect us, hobby system level developers.

Sorry - just every time I mention UEFI/EFI people see Ronald's misplaced quote in Wikipedia and end up talking about trusted computing and DRM.

Brynet-Inc wrote:This was the part that concerns me:

Its specification is not publicly viewable and EFI site suggests that it is a supporter of the Trust Computing Group.

It is publicly viewable.

To redistribute the specification, or to implement UEFI you need to do "the UEFI Adopters' Agreement" , which is a 3 page licence agreement that includes patent waivers and limits warranty/liability. However, IMHO an OS boot loader uses UEFI but doesn't implement UEFI - the licence agreement may be unnecessary for OS developers (who aren't writing firmware). Note: I am not a lawyer.

Trusted computing is a tool that is meant to improve security. Like everything, it can be used in "good" ways or it can be used in "bad" ways.

For example, it's possible for someone to use their clothing to strangle, choke or restrain other people, but it doesn't make sense to force everyone to be nudists because of that - it makes more sense to make laws against strangling, choking or restraining other people.

In the same way, it doesn't make sense to be "anti-trusted computing". Instead, let people use trusted computing in "good" ways (e.g. to improve security for end-users) and try to prevent people from using trusted computing in "bad" ways.

Brynet-Inc wrote:I'm not against closed source, I'm a proud supporter of BSD/ISC style licences... but I'd personally hate dealing with a firmware that only offers me a subset of functionality if I'm not running a Microsoft certified product. (ACPI implementors already seem to be doing this...)

I also don't like firmware that only offers a subset of functionality if you're not running a Microsoft certified product, but EFI is no better or worse than PC BIOS (nothing prevents a PC BIOS from offering a subset of functionality if your not running a Microsoft certified product), and (unlike ACPI) nothing in the UEFI standard cares what the OS is.


Cheers,

Brendan

Brendan wrote: In the same way, it doesn't make sense to be "anti-trusted computing". Instead, let people use trusted computing in "good" ways (e.g. to improve security for end-users) and try to prevent people from using trusted computing in "bad" ways.cares what the OS is.

The problem is that the people who often implement and provide 'trusted-computing' are in the pockets of those who would like to use it for 'bad' purposes.

Everything is run by money.

Alboin wrote:

Brendan wrote: In the same way, it doesn't make sense to be "anti-trusted computing". Instead, let people use trusted computing in "good" ways (e.g. to improve security for end-users) and try to prevent people from using trusted computing in "bad" ways.cares what the OS is.

The problem is that the people who often implement and provide 'trusted-computing' are in the pockets of those who would like to use it for 'bad' purposes.

Everything is run by money.

From my perspective, that isn't a problem. For example, if Microsoft use trusted computing to completely screw everyone, then there'll be more people willing to switch to my OS...

Of course if Microsoft use trusted computing to verify that their boot code, HAL and kernel haven't been modified (e.g. by a virus) to protect end-users from malicious code, then that'd be a "bad" thing from my perspective.


Cheers,

Brendan

Brendan wrote:According to <someone with a biased opinion and something to gain>,

Good points, in the past I was already annoyed by this blatently non-NPOV section. I added some comments on the talk page (also based on your points), see whether a discussion ensues.


JAL