Google Sponsors the LinuxBIOS project

Rockgod noted that "The LinuxBIOS project aims to take down the last barrier in Open Source systems by providing a free firmware (BIOS) implementation. LinuxBIOS celebrates its Sixth anniversary this year, and has an installed base of over 1 million LinuxBIOS systems. With the One Laptop Per Child (OLPC) project, that number is expected to exceed 10 million users in 2007. LinuxBIOS supports 65 mainboards from 31 vendors in v1 and another 56 mainboards from 27 vendors in v2"

Comments

It's not the last barrier

[ClosedSource]

to Open Source systems since the microprocessor and other PC hardware is not open.

Re:It's not the last barrier

[McNihil]

So the sparc processor doesn't count?

http://www.sun.com/processors/opensparc/

Enlightened yet?

Re:It's not the last barrier

[squiggleslash]

They opened up SPARC decades ago, and while they didn't start with a GPL'd reference implementation, they have released a number since.

This is just the way Sun works. It has nothing to do with them deciding whether the architecture is a viable player or not.

Re:It's not the last barrier

[T-Ranger]

The SPARC processor has (nearly) always been "open" for the 1990+/-5 definition of "open". Its design is managed by SPARC International, which besides Sun includes TI, Cypress Semiconductor, and Fujitsu.

But anyway...

The processor of a system is. Being "open" to change doesn't really get you anything. If you have enough money to do a production run of a modern CPU, then the costs of buying into SPARC International, or the reference design of MIPS, or an IBM POWER, etc, etc, is nothing. Getting a custom chip is more then just sitting around and thinking. BIOS is software. Writing a better BIOS is a matter of sitting around and thinking. (Well you see what I'm getting at here..) Getting a custom chip produced is not feasible for anyone but the most rich geeks, beyond the time sitting down with Verilog. Getting a custom BIOS, beyond sitting down in an IDE, is trivial.

Re:It's not the last barrier

[smittyoneeach]

What would be the point of a GPL hardware implementation to the individual user?
Who would have the skill (to say nothing of the fab) to make a change to the hardware, and then distribute it?

Re:It's not the last barrier

[DragonWriter]

"Hardware" doesn't have "source code", so long as the software is open source, that would be removing all the barriers to "open source" systems.

It wouldn't be the last barrier to "open" systems, which would require some analogus open regime for hardware, but it wouldn't be "open source".

why would a major manufacturer of motheboards

[BrentRJones]

Why would a major manufacturer of motheboards want to stay away from Linux for BIOS?
What do Award and Phoenix have better than Linux?

Re:why would a major manufacturer of motheboards

[runderwo]

LinuxBIOS is not compatible with legacy DOS-based PC operating systems and the GPL does not allow for proprietary extensions.

Re:why would a major manufacturer of motheboards

[Lumpy]

Why would a major manufacturer of motheboards want to stay away from Linux for BIOS?

What do Award and Phoenix have better than Linux?

two simple reasons.

1 - they do what they are told by the OS and content industries. "Trusted computing" is a buzzword they spent lots of money on.
2 - A linux Bios will not have the ability to lock the user on DRM or Os choices. something they desperately want at Microsoft. D oyuo think a company will make a motherboard that microsoft will refuse to support their os on? how about one that will never run windows VistaXP2 with "Protect you from you" technology? because the Bios does not refuse to boot an OS without a Microsoft certificate?

Try and buy yourself an ATX Alpha processor motherboard or Power PC motherboard. They exist but are insane priced because nobody buys them but uber geeks and research/science people....

Do you want your next Linux computer to cost you 3 times as much because your Motherboard costs $1800.00?

Re:why would a major manufacturer of motheboards

[Laur]

LinuxBIOS is not compatible with legacy DOS-based PC operating systems

According to the ADLO page there is no reason that LinuxBIOS couldn't boot up DOS-based OSes with a bit of tweaking, it just seems that no one has bothered yet.

Re:why would a major manufacturer of motheboards

[kfg]

Try and buy yourself an ATX Alpha processor motherboard or Power PC motherboard. They exist but are insane priced because nobody buys them but uber geeks and research/science people....

Similarly 12/24/48VDC ATX power supplies are also available off the shelf; at about three hudred bucks a pop. But if you're the sort of uber geek research/science person who really needs one, well, that's what you pay.

However, if, eventually, enough people get sick enough of the MS locked down systems there just might be a market for an outside player to hit one out of the park with an open system. That's why MS and its familiars are trying hard to get laws passed to forbid it.

KFG

when I read "guts-like-to-be-free-too"

[jimstapleton]

I immediately pictured a guy walking around with a gaping hole in his torso, with all of his internal organs dangling about, dragging along, behind him, etc.

An I the only one?

Extensible Firmware Interface

[smitty97]

uh.. EFI & TianoCore ?

Re:Extensible Firmware Interface

[Laur]

Don't forget a link to the OpenBios project, which aims to create a FLOSS implementation of the Open Firmware standard.

Re:Extensible Firmware Interface

[DeathPenguin]

uhhh: https://www.tianocore.org/nonav/servlets/LegalNoti ces?type=TermsOfService

"You acknowledge and agree that You will not, directly or indirectly: (i) reverse engineer, decompile, disassemble or otherwise attempt to discover the underlying source code or underlying ideas or algorithms of the Software; (ii) modify, translate, or create derivative works based on the Software; (iii) rent, lease, distribute, sell, resell or assign, or otherwise transfer rights to the Software; or (iv) remove any proprietary notices in the Software."

This is a compulsory licensing agreement which developers must agree to before even hopping on board with Tianocore.

Any place to buy the MOBO with lb pre-installed?

[WindBourne]

If a company is selling mobos with these on it, now is the time to speak up. It strikes as this will be free advertisement. If not, this might be the time to start selling.

Eternal game of catch-up?

[Kadin2048]

Aren't the BIOS/firmware revisions specific to various motherboard models?

I always assumed that they were made by Award and Phoenix in conjunction with the mobo/chipset manufacturers, because the BIOS was specific to a particular configuration of parts, and wouldn't be interchangeable.

So if you did write an "open source BIOS," how would you keep it up to date with the multitude of different chipsets and motherboards? Wouldn't each one require its own modified version? Seems like, unless the major motherboard manufacturers commit to using LinuxBIOS, that they'll forever be playing catch-up, trying to modify and QA their revisions against new pieces of hardware. Which I guess isn't a bad thing, but it seems like it'll never be mainstream that way.

Re:Eternal game of catch-up?

[sgtrock]

If a motherboard vendor chooses to use the FOSS BIOS, then I would assume that the vendor would recognize that it was in their interest to make sure that any changes to the hardware elements on their motherboard lines would necessarily require that they demand that their parts vendors work with the FOSS BIOS project to make sure that the low level drivers are working correctly.

I suppose it's possible that such a motherboard vendor might want to donate engineering time and samples to the project as well. They would have to weigh the cost of that effort against a host of other costs; licensing costs to use Award or Phoenix, the size of the expected market for the combined product, etc. Show them that the FOSS BIOS will work for MS Vista and they'll have a real incentive to push for it. Tell them that the market will be limited to just Linux and *BSD and they'll probably lose interest really fast.

Re:Eternal game of catch-up?

[drinkypoo]

This isn't just a random open source BIOS though. It's based on the linux kernel, and all the hardware support that entails. Or well, as much as you can cram into your kernel image. This kernel then bootstraps to another kernel (or through ADLO, apparently can run WinXP or Win2k's NTLDR.) The only bootloaders it in fact supports are NTLDR and LILO, apparently.

Re:Eternal game of catch-up?

[Kadin2048]

Only a very small minority of people, even among open-source-software users, are capable of interpreting source code and modifying it.

Also, even a person that did know code, would need to be very comfortable (I would think) to write or change something that could potentially brick their mobo.

So whether the code is modifiable or not, really isn't relevant to all but a few users, at least in the direct sense. There are indirect benefits of having code available (see my other post in this thread concerning the indirect benefits of OSS to non-coders), but most people are going to look at the piece of software as a unit, and ask whether it works, and if it doesn't, they're going to move on to something else. The benefits of OSS are rarely so great as to make hiring a skilled programmer to modify it for you worthwhile.

It sounds from TFA that they have a nice automated QA system set up, where new revisions get tested against actual hardware automatically, but they're going to have to sustain an awfully high level of effort, in order to keep creating and testing new software revisions to cope with all the new boards that get released to the market every month.

I'm not panning the project; I really hope they succeed. It just seems like yet another project that probably won't have direct support from the hardware manufacturers, and as a result will always be one step behind mainstream usability. Perhaps that's okay -- maybe "mainstream usability" is overrated. But it's something worth considering.

Re:Eternal game of catch-up?

[kfg]

So whether the code is modifiable or not, really isn't relevant to all but a few users, at least in the direct sense.

Who said anything about users?

It just seems like yet another project that probably won't have direct support from the hardware manufacturers

Yes. That is what; and only what, is required.

. . .maybe "mainstream usability" is overrated.

It is. It can be a positive hinderence if you're not mainstream.

KFG

Re:Eternal game of catch-up?

[davidsyes]

But, then microsoft would just starve the BIOS and MOBO makers of marketing dollars. Sure, make it compatible/supportive of Vista, but them mshaft will just make the OS kernel check the BIOS maker ID-- IDs assigned by mshaft. As much as I'd love to see the hardware more Linux-friendly, I have no doubt it'll still be some time off.

Heck, I have TWO EZ-Cam webcams I bought back in 2001 or 2002. The designers sometimes allow their contract manufacturers to change up components for almost-the-same depending upon the prices in the commodities or components market.

It happened when I worked for a maker of multiplexers/demultiplexers. During burn-in, whole racks of equipment would just go to shit because the contract manufacturer sometimes sourced crappy (but supposedly- or almost-identical) components to put on the MOBOs. My director was a smart guy and traced the problems to that. Dell does the same thing. I had one Dell tech on the phone and seduced/challenged him into admitting that Dell from time to time changes board components even in the same make/model of machine without telling the consumers/purchasers of the equipment. Even when I was in the company IT department before moving to Customer Support & Manufacturing, my IT manager saw I was having problems with one or 2 boxes. He told me those were a pair Dell had quietly made changes to and were to be replacing them at Dell's cost. So, TWICE in the same company, in under 2 years, I got burned by secret changes.

What does it mean? Anyone buying boatloads of hardware for some large lab or emulation or scientific project might have some wayward machines having spurious, untraceable problems. All because a component on the MOBO was switched. I HAD to get the Dell guy on the speaker phone during a tech support call, and it was timely. I worked for a friend who told me I didn't know what the hell I was talking about. Having been insulted and demeaned in front of others by him (some non-Dell MOBO machines were just reFUSING to be Ghosted (Norton) and he blamed it on me and incompetence even tho I told him of my experiences with Dell and other products, and that the problem HAD to be in the MOBO he had just bought, a comment to which he took great exception), I was on the phone with Dell and got my over-the speaker answer. Swapping the MOBO on the NON-DELL box solved the problem. I never got an apology from that day.

As for my EZ-Cam cameras, one has a shitty Conexant chip. No distro of Mandrake since 2001, no Mandriva, no Ubuntu, or Suse, or any other distro I tried could get that damn camera to shoot or show images. They DID SEE the camera, but could not activate it. Meanwhile, another camera of the same markings, DID work. Years ago I opened both and took pictures and found what I suspected: different components, yet same model/type, etc.

So, it won't be that hard for mshaft to software discriminate against Linux-friendly BIOS code unless... well, unless, I suppose, that code is cross-licensed with Novell/Suse and mshaft, too.

Hopefully, Linux devs will become so nimble that the BIOS makers might be bypassed altogether. But, the hardware makers will still likely solder or epoxy on some firmware-laden chip that will refuse to work with Linux BIOS footprints.

Hitting two sacred cows at once...

[Otter]

With the One Laptop Per Child (OLPC) project [CC], that number is expected to exceed 10 million users in 2007.

1) Given that yesterday's news was that OLPC managed to produce a whole 10 computers, and that we're now halfway through November 2006 -- yeah, I can't see how they could possibly fail to hit 10 million in 2007!

2) Has Googlefawning now hit the point where it's no longer necessary for Google or the Slashdot story to explain exactly what it is that "Google sponsors" means? (Apparently they paid for a build system. Take that, Gates Foundation!)

Re:Hitting two sacred cows at once...

[Ant P.]

10 computers? That's odd, I could've sworn they made a few more than that.

Re:LinuxBIOS has a problem

[mengu]

Google's cache of the page:

http://72.14.209.104/search?q=cache:VJTK5OuL8OIJ:l inuxbios.org/index.php/Supported_Motherboards+Moth erboards+supported+in+LinuxBIOS&hl=en&gl=us&ct=cln k&cd=1

BIOS + DRM = lockin ..

[rs232]

http://news.zdnet.co.uk/software/0,1000000121,3911 6902,00.htm
http://slashdot.org/hardware/03/09/04/1427237.shtm l?tid=137&tid=185

Info on Linux BIOS in actual usage?

[tji]

I have seen this mentioned every so often here, and I am interested in trying it out. But, the stuff I read blurs the line between what I think of as BIOS functions and the actual OS. So, I am not sure if it's worth trying out or not.

Does anyone have pointers to good information, or experience themselves? The kind of questions I have are:

- Do I still have the configuration capabilities that you expect in a Phoenix/Award BIOS? En/Dis-able integrated devices, Fan Control, ACPI en/dis-able, etc.

- The articles all say that LinuxBIOS boots a linux kernel very quickly. Is this into a limited BIOS setup environment, or is this the actual kernel for the Operating System that you're running? If it's the latter, don't kernel upgrades become more difficult/dangerous? (Are there any docs which go through the system bootstrap process step by step?)

- Is AMD64 (in 64 bit mode) supported?

- Beyond the Linux hobbyist incentive to try out new things, are there any other major advantages to using LinuxBIOS on my home Linux server (which is a supported board)? Do I lose anything my current Award BIOS offers?

Re:Info on Linux BIOS in actual usage?

[Shewmaker]

You can edit a plain text file in user space and write it to CMOS with at utility like lxbios or cmos_util. The options I've seen are: boot sequence related, ECC memory related, power on after failure, debug level, cpu throttling, and NMI related. I didn't see anything about the enabling and disabling of devices or fan control, but I'm sure it depends on how much effort the developers have put into a particular chipset/motherboard.

LinuxBIOS supports several different types of payloads: Linux, Open Firmware, Etherboot, etc. If you are using a Linux kernel payload, then you probably don't want to be upgrading it often. In that case, you can set up the first kernel to kexec a second kernel (before kexec, there was a patch called the two kernel monte).

AMD64's 64-bit mode is definitely supported.

It's not trivial (yet) to boot a version of MS Windows with LinuxBIOS, but using Linux as a BIOS can give all sorts of benefits. One very interesting capability for people running beowulf clusters is that you can boot over any network device that Linux supports (e.g. Myrinet or Infiniband). That may not mean anything to a regular home user, but the point is that you have a whole lot more flexibility in what you can do. Even if you don't want to make it boot your home system over your wireless LAN, it does increase your freedom and it prevents people from nibbling away at the freedom you already have.

I would say freedom from future DRM really is the biggest incentive for trying out LinuxBIOS at home. You can avoid Intel's EFI standard (they're pushing for it to be on all desktops and servers), which will enable companies to inflict DRM on you. Linus has made some very good points about why EFI is not good. One way to look at EFI is that it is basically an OS, and not a very good one.

There are several white papers and tutorials that do a good job of explaining how LinuxBIOS works. Look at the LinuxBIOS documentation section.

Re:Why bother?

[eno2001]

Actually, I think of it in almost the opposite way... Why isn't the base OS kernel sitting on an eeprom on the MoBo talking directly to hardware and thereby completely obviating BIOS? I remember back on my Atari ST, they had a 512k ROM that had GEM/GEMDOS in it. If I didn't pop a floppy into the drive, the system still booted to a desktop using the ROM image. If I did pop a floppy into the drive, then the OS loaded off the floppy. The main point being that all interfacing to basic hardware on the ST was handled by the OS itself. There was no BIOS. I remember being very confused by this when I moved to the PC. I turned it on expecting to get a Windows 3.1 desktop that I could then use to format the HD and install the OS with, or as a second best option, a DOS prompt that would allow me to format/partition/install the OS. I remember when I turned it on and all I got was a "NO ROM BASIC" error, I called the store and complained that they'd sold me a damaged system and the OS seemed to be missing from the motherboard. I was incredulous when they told me that I actually needed to boot an OS from a floppy. That seemed so backwards based on where I was coming from.

Fast forward 12 years and here I am a big *nix head and everything... I can chuckle about my misconceptions to an extent. But I STILL to this day believe that the OS kernel should talk to and provide low level support for the MOBO components. Yeah, it would make the OS more complex, but I think it would also allow great flexibility and longer life to a system since many of it's functions would be in software with fewer limitations imposed by BIOS.

Re:Any place to buy the MOBO with lb pre-installed

[Laur]

Try starting with the Products page at LinuxBIOS.

Harddisk encryption

[Ignatius]

A customizable BIOS with the Kernel in flash would be the proper place to setup user authentication, software harddisk encryption, firewall rules and VPNs. If supported by the kernel (AFAIK OpenBSD has such a feature; don't know about Linux), you could switch the OS into a secure mode after boot up and initialization where it is no longer possible to change certain settings before you even access the harddrive.

It's basically as close as you can get to "tamper-proof" by a software-only approach and for notebooks, it would provide some reasonable theft protection, esp. if combined with a "this notebook is the property of ....." banner on startup and some epoxy over the bios flash rom ...

Re:Why bother?

[elgaard]

> Why would I care about the BIOS?

For the same reason you care about other programs being open. E.g.

- Fixing bug. Eg hibernate problems.

- Checking for bugs and backdoord.

- Improving it to your needs. E.g., I would like to be able to boot from USB-disks or a CFlash card in a PCI-adaptor.
Or I could remove unnecessary stuff and put in a shell. Or an SSH server i the BIOS.

- Performance. My BIOS is slow. It does a lot of unnecessary things.

- Consistency. Next time I get a new computer, it would be nice to have the same bios. A company might prefer to use the same BIOS on all computers.

Re:Why bother?

[Hast]

Operating systems tend to ignore things which the BIOS tells them because they are not reliable. It's a lot easier and more robust to have the OS detect disks and memory than the BIOS.

So it takes the BIOS quite a lot of time to do something which isn't used anyways.

Re:Why bother?

[arcade]

I'll give you an answer, from my point of view.

Have you ever used Solaris? On Sun Hardware? It's great to able to send the OS a 'break', and get an OK> prompt, where you can configure low-level stuff. It's one of the things that makes me love Sun+Solaris way more than I love linux.

Even better would be if we could have a standard co-processor-thingie listening on the serial port, like on SUN, where "lights-out-management" could be done. I really like that feature of Sun hardware too :p

HP's ILO and Dell's DRAC really doesn't cut the muster compared to Sun's LOM/ALOM/whatever-they-call-it-today :)

Of course the last two paragraphs can't be addressed by linuxbios, but the first one can. It's one of those things I really, really miss in Linux.

OSS has benefits to non-coders.

[Kadin2048]

End users aren't going to modify their processors and have them fabricated, but then again, "end users" for the most part, aren't going to open up the source code to their applications and make any sort of nontrivial adjustments to them, and recompile them.

Writing code and recompiling a piece of software is almost as much a black art to most people, as designing a microprocessor and fabricating a chip is.

Source code is meaningless gibberish to most users, regardless of whether that source code describes hardware or software. Code written in VHDL is just a slightly more arcane strain of gibberish than C, but still meaningless.

Most people (who have even the foggiest idea of open source) benefit from it indirectly: by having higher-quality products to begin with, and having them available from more vendors, and having a guarantee that if a vendor tanks, that their product stands a better chance of being supported by somebody else (because another company or organization can take it over). This would also be true with hardware. An open and well-documented chip design would be available, were it popular, from a variety of vendors, and even if one vendor went out of business, the design would survive. These benefits exist even to people who cannot understand code, and exist for both hardware and software.

opencores.org

[RAMMS EIN]

http://www.opencores.org/

Re:Why bother?

[J.R. Random]

If you actually read the LinuxBIOS web site, you will see that a prime motivation was to allow remote updating of the BIOS on Linux clusters. It beats attaching a keyboard to each of 256 motherboards and updating them one by one.

Manufacturers of embedded systems are likely to be interested in a BIOS that is free and fast.

It is not so clear what the benefits are for Joe and Mary desktop user. I'm sure most Linux users will continue to use the BIOS that comes with their board.

OLPC isn't using LinuxBIOS anymore

[burns210]

As of a few weeks ago, the OLPC project isn't using LinuxBIOS anymore, they have moved to OpenFirmware from Sun, which was recently open sourced. Sorry to burst the bubble.

Re:OLPC isn't using LinuxBIOS anymore

[LinuxOnEveryDesktop]

As of a few weeks ago, the OLPC project isn't using LinuxBIOS anymore, they have moved to OpenFirmware from Sun, which was recently open sourced. Sorry to burst the bubble.

That's incorrect. The OLPC project is still using LinuxBIOS to boot the machine, but they are using OpenFirmware as a payload. I have seen both OpenFirmware and a linux kernel payload demonstrated on the OLPC revA developer boards. They chose OpenFirmware because it can fit in a smaller ROM chip.

If you don't understand what that means, just read up on things at http://linuxbios.org/