More than one million devices have already shipped with LinuxBIOS, and the growth is continuing. In his interview for the upcoming FOSDEM 2007 conference, LinuxBIOS creator Ronald Minnich talks about vendor support, the One Laptop Per Child BIOS, and his reluctance towards EFI.
LinuxBIOS Creator: EFI Is a Security Threat
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I’ve seen LinuxBIOS in news before, but if Chipset makers have gotten on board, seen LinuxBIOS and then backed off over the years, why is that?
Is there something wrong with LinuxBIOS?
Maybe they are behind other BIOS’s in terms of functionality?
I’m not picking on it, because having some things Opensourced is great, but there come a time where non-Opensource programs/technologies are better than the Opensource ones.
Most End users do not care what their BIOS is, they care that it’s going to work, with no problems.
And how is LinxBIOS in terms of Overclocking etc.
The title is misleading too, he doesn’t really talk about the security concerns of EFI, he just say it’s possible.
In this world, anything is possible.
>And how And how is LinxBIOS in terms of Overclocking etc. is LinxBIOS in terms of Overclocking etc.
If you want to play with your computer, go with Windows. Everyone with a sane mind wouldn’t do such a nonsense.
Everyone with a sane mind wouldn’t do such a nonsense.
And people with a sane one would go compile their own kernels, for example?
Sometimes I think us non-windows users are a bunch of masochists.
Comparing apples with oranges is usus nowadays I guess?
>And people with a sane one would go compile their own kernels, for example?
But this is to some extent true too, nobody in BSD would do it without need for example
But the nonsense begins, if you’re seeing a need in overclocking your hardware. Some people call this an error in reasoning.
But the nonsense begins, if you’re seeing a need in overclocking your hardware.
A few years ago, I bought a motherboard with an AMD Athlon XP 2000+ processor. But, it would not run at 2GHz by default. At the store, employees made it very clear that I would need to change the clock speed from 100 to 133 – and even the box with the CPU mentioned the number 133.
That’s because an AMD 2000+ CPU does *not* run at 2GHz. The 2000+ rating means its comparable to an Intel CPU running at 2GHz. A pretty pointless rating system actually.
Unless I’m horribly mistaken on this, in which case, correct me.
Edited 2007-02-09 00:17
I know that… It’s really 1666MHz. But without overclocking, it would be 1250.
why didn’t you buy a new motherboard with the cpu?
That’s because an AMD 2000+ CPU does *not* run at 2GHz. The 2000+ rating means its comparable to an Intel CPU running at 2GHz. A pretty pointless rating system actually.
Unless I’m horribly mistaken on this, in which case, correct me.
I’m not sure, but I think it’s not “an Intel cpu running at 2GHz” but an Intel Pentium 4 cpu running at 2GHz. My Athlon XP 2600+ @1.9 GHz is a tiny bit slower than a P4 2.6 Ghz, so it’s not that far from the truth.
Compare it to other Intel cpus, though, and the xxxx+ thing doesn’t make any sense at all. Still I can understand why AMD came up with this, because in the P4 era clockspeed actually did not mean real world speed. But people are easily fooled.
Yet AMD have to stop it some time because it would be bizarre to take the P4, one of the cpus Intel shouldn’t be proud of at all, as some kind of a golden standard for benchmarking.
That’s because an AMD 2000+ CPU does *not* run at 2GHz. The 2000+ rating means its comparable to an Intel CPU running at 2GHz. A pretty pointless rating system actually.
Pointless ?? – not if your main competitor has build up the market on pure numbers.
BTW the K7 Performance Rating scheme compares the procesor performance with the Athlon Thunderbird.
The Performance Rating was cooked up by AMD, Cyrix and Centaur and originally was called Pentium Rating. That one was a comparison with the Pentium product line. Intel didn’t liked that and took the matter to court. That is why the rating scheme today is called Performance rating and does not compare with the Intel Pentum productline.
But the Thunderbird sort of matched the PIV and thereby you get a sort-of-comparison with the PIV line.
With the arrival of the K8 productline – AMD has changed the comparison index, I have no idea of what they compare with today, but it is not the Thunderbird and it s not the Pentium product line.
>The 2000+ rating means its comparable to an Intel CPU running at 2GHz.
More precisely, to a P4 running at 2 GHz.
>A pretty pointless rating system actually.
Apparently you don’t understand marketing..
People used the clock of a computer as a measure of its performance, so Intel built CPUs with a high clock but not very efficient, AMD not willing to waste power didn’t go to the high clock rate and seemed inferior for the customer, so they provided a P4-clock-equivalent performance number, a clever idea.
“Some people call this an error in reasoning.”
Some people call it getting your money’s worth.
Some people call it getting your money’s worth.
I’d call it being desperate to eek out the last drop of performance, in the vain hopes of running your slow Windows a little faster.
Since I went the GNU/Linux route full time back in 2001, I have never overclocked a machine from that time on. Most hardware is fast enough when being managed by a well written system.
Realization: Not everybody runs Windows, but I can’t imagine the advantage of risking system damage if a non-MS system runs very well at factory defaults.
Edited 2007-02-09 16:40
Psshaw. You are not making a blanket statement…”running Windows a bit faster.” Most people who overclock do it for games not for Windows. I doubt you would even notice a performance difference in Windows per se. Sure apps will get speed up like video rendering and gaming etc etc. But then again the chips are desgined with oclockig in mind. Not everyone overclocks. I havent in my life. Because most of the time I make sure I can by the fastest processor money can buy at the time when I am buying a computer.
“Not everyone overclocks. I havent in my life. Because most of the time I make sure I can by the fastest processor money can buy at the time when I am buying a computer.”
—————————————
Some of us actually underclock so that we can run a computer that is quieter, cooler and more energy efficient – where you don’t need a huge amount of power for your apps, this is quite sensible to do. I’ll buy the fastest CPU I can afford and undervolt and underclock it down to the speed of a mid range CPU, and enjoy the silence… A BIOS that makes this easy to do is what I look for, and I am not overly concerned what that BIOS is based on…
Funny, I haven’t compled a single kernel in ten years
Why would you say that? I consider trying different OS’s, hardware and settings as playing. Your attitude doesn’t help OSS, and it doesn’t help regular users learn about OSS or different ways of doing things.
I bet you treat your users like that too, if you have any, if you don’t, thank god
I’ve seen LinuxBIOS in news before, but if Chipset makers have gotten on board, seen LinuxBIOS and then backed off over the years, why is that?
Mainly because they are just scared about potentially giving up control of their hardware, and allowing people to see how it works for reasons that are best known to them. For the smaller vendors who have let go of this bogus and paranoid IP (Intellectual Property) attitude then they’ve found that it makes no difference whatsoever. Intel is obviously the biggest culprit here.
From smaller hardware vendors’ points of view LinuxBIOS is great because it provides more of a level playing field and a ready made BIOS, with the economy of scale of lots of other people using and contributing to it (as in the Linux kernel), which they can then pick up and run with.
That approach always tends to win in the end, and is how the PC came about and how Intel got so rich. Many companies wanted to stay in a world of total proprietary and closed hardware with certified hardware from certain vendors, and Intel seems to want to take us back to that world. In the long run, no matter how big the market Intel thinks there is for things like DRMed entertainment systems (it’s not about entertainment DRM though but mainly about control), the market for more open and cheaper systems is larger and is where things naturally gravitate.
Mainly because they are just scared about potentially giving up control of their hardware, and allowing people to see how it works for reasons that are best known to them. For the smaller vendors who have let go of this bogus and paranoid IP (Intellectual Property) attitude then they’ve found that it makes no difference whatsoever. Intel is obviously the biggest culprit here.
I don’t believe this at all – for example, if Intel are worried about giving up control of their hardware they wouldn’t have datasheets on just about every chip they produce freely downloadable from their web site.
From the article, LinuxBIOS is used in (over) one million devices (embedded systems where Windows and/or backwards compatability doesn’t matter). How many desktop/server machines is it used in? My guess is probably less than 500 machines owned by LinuxBIOS developers and people experimenting with beowulf clusters.
IMHO it’s more about economics – LinuxBIOS is mostly useless, mostly buggy/incomplete (AFAIK) and rather messy for most desktop machines that end up running windows. Because of this the cost of bothering with LinuxBIOS for desktop/server manufacturers (end user documentation & tech support, reliability testing, answering emails from developers & trying to create useful responses, etc) is too high to justify considering the size of the end market (almost none).
If LinuxBIOS was actually intended as a desktop/server BIOS for Windows machines things may be completely different – it could have been used in millions of PCs and no embedded systems instead of the other way around.
I don’t believe this at all – for example, if Intel are worried about giving up control of their hardware they wouldn’t have datasheets on just about every chip they produce freely downloadable from their web site.
Intel may have datasheets to download, but getting it to work without their direct assistance is an entirely different matter. I think you’re getting the two very confused. Intel believes in apparently open hardware, but not that open. Intel also feels the need to lock up a lot of functionality within closed firmware.
If LinuxBIOS was actually intended as a desktop/server BIOS for Windows machines things may be completely different – it could have been used in millions of PCs and no embedded systems instead of the other way around.
What’s the difference?
What’s the difference?
Being designed to run any OS (like existing/standard BIOSs), rather than being restricted to a single OS (with ugly hacks to make it run some other OSs in some situations with a varying amount of success).
What kind of comment is that, and why exactly were you modded to 5 for it? I’ll agree LinuxBIOS vs. closed BIOS is about money, but that’s where it ends.
Your thoughts on LinuxBIOS being “useless and buggy” is just dumb. It’s the closed BIOS that are the hackneyed “duct tape and bubble gum” code that has existed for years and years. What you’re doing is faulting one-armed basketball player for not being able to make a decent jumpshot. Hell, the mobo manufacturers can’t even write decent setup documentation, think of what their code looks like.
If the hardware specs were easy to come by, the BIOS would exist for Windows desktops. Probably would drop the price too. That would also mean that the various (and probably intricate) ways the Windows kernel interacts with BIOS would also have to be opened up. Why would Microsoft want to do that? Not that I would think that’s a bad thing. I’d much rather see MS pull an Apple by oss’ing their kernel and rely on keeping their “eye-candy” in-house. But that’s another thread.
This argument is about IP, plain and simple. It’s about selling/upgrading more motherboards in shorter intervals, and it’s about protecting interests.
What you’re doing is faulting one-armed basketball player for not being able to make a decent jumpshot. Hell, the mobo manufacturers can’t even write decent setup documentation, think of what their code looks like.
What I’m doing is faulting mis-information that is leading normal people to think they’ll be able to use LinuxBIOS in their computer/s soon, and all the hype that doesn’t give clear and honest information about how ready LinuxBIOS is for desktop computers.
Your thoughts on LinuxBIOS being “useless and buggy” is just dumb. It’s the closed BIOS that are the hackneyed “duct tape and bubble gum” code that has existed for years and years. What you’re doing is faulting one-armed basketball player for not being able to make a decent jumpshot. Hell, the mobo manufacturers can’t even write decent setup documentation, think of what their code looks like.
Wrong again. It’s the “de-facto PC-BIOS standard” that is a mess (not any specific implementation of it), but this is what you expect with 20+ years of backward compatability. As for the quality of BIOS code, as far as I can tell most of it is good quality code (designed to comply with a crappy set of standards).
I’ve never had problems with any motherboard manufacture’s documentation, but then I’m not an average user (and neither are the PC technicians who this documentation is intended for). I’m also not sure what the documentation has to do with the BIOS’s code considering that in almost all cases the BIOS’s code is outsourced to a company like Award while the documentation isn’t. It’s like buying a motorbike with a free helmet and assuming the motorbike isn’t very good because the helmet is ugly.
If the hardware specs were easy to come by, the BIOS would exist for Windows desktops. Probably would drop the price too. That would also mean that the various (and probably intricate) ways the Windows kernel interacts with BIOS would also have to be opened up.
The way it works is that all BIOSs implement the “de-facto PC-BIOS standard”, and all OSs assume the BIOS complies with it. The intricate details of the internals of any version of Windows (or any other OS) doesn’t matter at all once the BIOS is implemented according to the relevant standards. These standards are not “closed source”, proprietory, patented or trade secrets. Anyone that is competent enough to actually write the BIOS code is competent enough to find and understand these standards.
AFAIK the LinuxBIOS project is having problems with getting information for the hardware underneath their code, not with getting information for the software above their code.
“Is there something wrong with LinuxBIOS?
Maybe they are behind other BIOS’s in terms of functionality?”
The biggest problem with LinuxBIOS starts and ends with hardware (as you should realize if you’d read the interview).
“Easy” hardware is documented; the vendor answers questions and provides sample source; there are lots of people familiar with it; the hardware is well designed, and simple to code to.
“Hard” hardware is undocumented; the vendor hides and obscures information; few if any people understand it; and it has obscure interfaces.
Even when a manufacturer has all of the information, a BIOS can be faulty. If a manufacturer is trying to obscure information, LinuxBIOS developers have a very difficult job getting things to work properly.
At this point the LinuxBIOS has reached a point of maturity that allows the developers to solve tricky problems more easily because they have a more secure foundation.
“Most End users do not care what their BIOS is, they care that it’s going to work, with no problems.”
LinuxBIOS holds the promise of doing things similar to or the same as what EFI is capable of without the same requirements of EFI. As users see what some computers with EFI or LinuxBIOS are capable of, they will care about the features.
True, most won’t have a clue that a traditional BIOS doesn’t offer the same features. So, they won’t express a desire for a particular “BIOS,” but the features will create a demand.
Is there something wrong with LinuxBIOS? Maybe they are behind other BIOS’s in terms of functionality?
Can other BIOS implementations export fine-grained power management interfaces to the OS? Can other BIOS designs suspend and resume within a second? Do other BIOS vendors allow serious overclockers to compile their own tweaked BIOS images (or download them from enthusiast sites) to get every last bit of performance? Are there any other BIOS offerings that are designed with efficient and correct operation, rather than proprietary interface abstraction, in mind?
LinuxBIOS is one of the technical reasons why OLPC will wow people. Regardless of whether you think cheap green laptops are what developing nations need to drive economic growth, OLPC is a showcase of what personal computing could be like if we said no to proprietary interfaces. It’s not about freedom in the religious sense, it’s about what the freedom allows us to do with the technology we have. The capabilities of the OLPC, especially given its price and development timeframe, will simply blow people away. And it’s because of open source and standards like LinuxBIOS.
I’m not picking on it, because having some things Opensourced is great, but there come a time where non-Opensource programs/technologies are better than the Opensource ones.
Most End users do not care what their BIOS is, they care that it’s going to work, with no problems.
I’m curious to know what you think of Asus’s uber closed-source BIOS omfgwtftechnology?
http://www.benchzone.com/page.php?al=asustek_evaluation&pg=3
“Most End users do not care what their BIOS is[…]”
May I correct your sentence?
“Most End users do not know what BIOS is”
Some of us just gotta know, granted overclocking isn’t for everybody and much like gambling you should never play with more than you can afford to lose.
While it might have to do with some or all of that… I think it also has to do with the hardware cartels reluctance to give up control…
As it is now we deal with a inferior system where everyone’s vested interest is in ensuring everything works with the systems they created.
If there was an easy way for Joe Six Pack home user to escape the claws of the de-facto standard, that allowed for faster boots, broader support, 3rd party hooks to increase functionality beyond the initial intended scope then of course they are going to balk and impede its progress…
Its the same as MP3’s / DVD / DRM… We can all find a way around it, but it is nuisance enough where most just opt to do the legit thing because the effort and learning is beyond the scope or pain threshold of circumvention…
There is not to be paranoid a conspiracy happening… More and more devices are being shipped with TPM modules.. At the moment most or optional via a setting, but now imagine industry pressure forcing a scenario where it is no longer a frivolity for corporate security but now a part of the mandatory closed specification…
The powers are walking a fine line where any major shift would accelerate a change to more open standards, but they are instead slowly getting us hooked on their new features, tying things together so by the time we realize we traded our freedom for features it is too late to go back and enjoy a life when we weren’t 100% corrupted by EULA’s, copyrights, DRM and Big Brothers right to inspect every bit, nibble and byte flowing through the pipes…
It’s all relative.. If the bios was open, we might not be limited all the way down the hierarchy of having to rely on windows drivers, or even inefficient linux ones because we still rely on essentially 1984 or prior technology of the traditional “run-of-the-mill” BIOS…
I’ve seen that guy at LACSI’06 and SC’06 – he was working on porting Plan 9 to BlueGene/L. Quite an impressive guy.
This is OpenBIOS – aka, Openfirmware; I think if people pitched a industry openstandard up against a specification controlled by Intel, chipset and motherboard vendors would think twice about the idea of getting behind EFI.
Not to be boogie-man but companies need to realise that unless EFI is open and standardised to the same degree which OpenFirmware is, they’ll find themselves in a position where they don’t want to be.
Thanks to EFI, I will not even consider purchasing a Mac, regardless of its other merits.
Based on what?
If its purely an open source thing, then fine. Personally I like a few comforts around me these days but feel free to be a pioneer if you want. As someone else pointed out, joe public doesn’t know what a BIOS is, let alone care if its open or not. Us here are all in a privileged position to care.
If you’re thinking security then why, because of some vague suggestion its not secure? No machine you can physically access is secure – you just yank the power cord and get a screwdriver out.
Because, if the computing business stops being fun, what’s the point?
DUH, I don’t know…. MONEY you f-ing retard!
Color me surprised, but doesn’t an operating system give a damn about the BIOS once it fired up all device drivers, which take care of all hardware controllin’? So why would I give a damn about a LinuxBIOS?