So there are two issues: a) DOS isn't supported and b) the floppy as a boot device isn't supported. That will make it a very difficult sell for mobo manufacturers. Can you imagine me buying a Linux PC with LinuxBIOS and it crashing. How am I going to repair it then?
----- Original Message ----- From: Adam Sulmicki adam@cfar.umd.edu Date: Fri, 6 Dec 2002 12:50:07 -0500 (EST) To: "Ronald G. Minnich" rminnich@lanl.gov Subject: Re: Booting from floppy
the u. md. guys will chime in here with an explanation :-)
I'm trying to use it as the opportunity to improve the FAQ :-)
From what I saw during the coverage of announcement there were quite a bit of folks unfamilar with the BIOS projects, an and this FAQ hopes to clear up the confusion.
http://www.eax.com/ADLO-FAQ.html
Big Pilot, if I may call you that, does this FAQ answers tose questions? The way I see, the part below is answered by Q5. Since I assume you are asking about LinuxBIOS. If you are really asking about linux kernel, then it is definitely wrong mailing list.
if Linux fails, how is one supposed to repair it? Or is the only option booting Linux off a floppy and repairing using that?
I'm wondering whether it is still possible to boot DOS (or a free equivalent) from a floppy if LinuxBIOS is installed on the mobo.
As for above. We had limited success with Win98 which is DOS based (we can get as far as desktop screen, but there are issues), but as the Q6 says it is still long way to go.
As for the particular susbsystem -- floppy, it has not been tested at all, and is presumed not working at the moment.
-- Adam Sulmicki http://www.eax.com The Supreme Headquarters of the 32 bit registers
So there are two issues: a) DOS isn't supported and b) the floppy as a boot device isn't supported. That will make it a very difficult sell for mobo manufacturers. Can you imagine me buying a Linux PC with LinuxBIOS and it crashing. How am I going to repair it then?
Well nobody pretends it is a finished work. As the FAQ says a milestone has been made but there's tons of work left.
On Fri, 6 Dec 2002, Big Pilot wrote:
So there are two issues: a) DOS isn't supported and b) the floppy as a boot device isn't supported. That will make it a very difficult sell for mobo manufacturers. Can you imagine me buying a Linux PC with LinuxBIOS and it crashing. How am I going to repair it then?
floppies are dead. What's wrong with a cdrom boot?
ron
"Big Pilot" bigpilot@linuxmail.org writes:
So there are two issues: a) DOS isn't supported and b) the floppy as a boot device isn't supported. That will make it a very difficult sell for mobo manufacturers. Can you imagine me buying a Linux PC with LinuxBIOS and it crashing. How am I going to repair it then?
Which hardware you can boot from depends on the bootloader you pair LinuxBIOS with. ADLO is one, that is just recently maturing into usability.
Etherboot is another.
The linux kernel is another.
9Load for Plan9 is another.
Both etherboot,Linux, and I believe 9Load have floppy support so there are cases where you can boot from a floppy.
For several of the developers the primary boot devices is the network, which makes recovering if something goes wrong quite simple, but it does assume you have a second machine.
The basic boot sequence with LinuxBIOS runs: A) LinuxBIOS sets up some hardware, and allocates the rest unique addresses. B) LinuxBIOS loads an ELF formatted bootloader from somewhere normally the same rom chip. C) The bootloader 9loader/ADLO/Etherboot/(A Linux kernel + initrd) loads and decides what to do. D) The final kernel (usually Linux) is loaded and you go about your usual computing.
The primary users so far are embedded users, and cluster users.
For use in a cluster, a primary bit of usefulness is how easy you can see what is wrong (the serial port is available before ram is initialized) and how easy it is to recover from a machine crash.
The company I work for can currently sell LinuxBIOS, and it has helped us to get some of our biggest deals. So as it is salable to some, and it is open source so fell free to pay us back, by developing the features you want.
Basically LinuxBIOS has one core design. Don't do unnecessary work. It takes some thought to see how, to make use of that, but it is not too hard to work with.
Eric
I heard an amazing data point from a company a few days ago, discussing the linuxbios issue.
They estimate they have lost $30M just this year due to their inability to support linuxbios on their boxes (no, it wasn't IBM, but it could have been). Now for the company in question, this is peanuts -- not even 1/1000 of their gross revenue. But for the salesman for the company in question, this is lost opportunity to make targets. Their salesmen are more than a little unhappy. But the companies persist in saying "there is no market" for LinuxBIOS, the same way they said "there is no market" for Linux, and the same way they said "there is no market" for Unix.
It's always amazing to me that small companies in the US, with so much more to lose, are so much more willing to take on risk than the big companies, who can waste tens of millions in the wink of an eye -- far more than it would cost them to field LinuxBIOS.
ron
What market sector? Embedded, Beowulf/cluster, ? This is really interesting stuff. Is this figure lost hardware sales?
----- Original Message ----- From: "Ronald G. Minnich" rminnich@lanl.gov To: linuxbios@clustermatic.org Sent: Friday, December 06, 2002 5:57 PM Subject: linuxbios targets ...
I heard an amazing data point from a company a few days ago, discussing the linuxbios issue.
They estimate they have lost $30M just this year due to their inability to support linuxbios on their boxes (no, it wasn't IBM, but it could have been). Now for the company in question, this is peanuts -- not even 1/1000 of their gross revenue. But for the salesman for the company in question, this is lost opportunity to make targets. Their salesmen are more than a little unhappy. But the companies persist in saying "there is no market" for LinuxBIOS, the same way they said "there is no market" for Linux, and the same way they said "there is no market" for Unix.
It's always amazing to me that small companies in the US, with so much more to lose, are so much more willing to take on risk than the big companies, who can waste tens of millions in the wink of an eye -- far more than it would cost them to field LinuxBIOS.
ron
Linuxbios mailing list Linuxbios@clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios
On Sat, 7 Dec 2002, Jeremy Jackson wrote:
What market sector? Embedded, Beowulf/cluster, ? This is really interesting stuff.
cluster only. There's lots more outside cluster.
Is this figure lost hardware sales?
Lost system sales, yes.
ron
So there are two issues: a) DOS isn't supported and b) the floppy as a boot device isn't supported. That will make it a very difficult sell for mobo manufacturers. Can you imagine me buying a Linux PC with LinuxBIOS and it crashing. How am I going to repair it then?
It's Ron's privilege to name the project as he sees fit, but when people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS. The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
The code base has turned out to be useful for much more than that, and it now supports other boot loader code, such as Etherboot. The UMD guys have just created ADLO to glue the Bochs PC BIOS together with LinuxBIOS for the hardware setup, so that various OSes (MS included) are supported. This is very new, so give it time - but I don't think we'll need to wait long, this stuff is moving fast.
On Sun, 8 Dec 2002, Peter Lister wrote:
people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS. The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
it's this way for historical reasons and it almost seems too late to change. But Garzik really called this freebios so there's no reason not to pick that up at some point.
ron
people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS. The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
it's this way for historical reasons and it almost seems too late to change. But Garzik really called this freebios so there's no reason not to pick that up at some point.
Not to mention "EtherBOOT" that boots from IDE.
Adam Sulmicki adam@cfar.umd.edu writes:
people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS. The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
it's this way for historical reasons and it almost seems too late to change. But Garzik really called this freebios so there's no reason not to pick that up at some point.
Not to mention "EtherBOOT" that boots from IDE.
That one is much less silly. The primary focus is still on booting over an ethernet network in with etherboot...
Things do tend to expand their functionality when we touch them don't they.
Another fun one is the Bochs-BIOS running on someplace besides under Bochs....
Eric
Peter Lister P.Lister@sychron.com writes:
So there are two issues: a) DOS isn't supported and b) the floppy as a boot device isn't supported. That will make it a very difficult sell for mobo manufacturers. Can you imagine me buying a Linux PC with LinuxBIOS and it crashing. How am I going to repair it then?
It's Ron's privilege to name the project as he sees fit, but when people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS.
Bogus. Linux needs a BIOS, to initialize the platform. Which we have shown the hard way. Linux simply needs not any help after it gets except on poorly designed ports.
And LinuxBIOS is a BIOS for Linux.
The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
Which is also suggested by the name LinuxBIOS. Unfortunately that does not work have as well as initially envisioned.
The code base has turned out to be useful for much more than that, and it now supports other boot loader code, such as Etherboot. The UMD guys have just created ADLO to glue the Bochs PC BIOS together with LinuxBIOS for the hardware setup, so that various OSes (MS included) are supported. This is very new, so give it time - but I don't think we'll need to wait long, this stuff is moving fast.
One small step at a time.
Eric
It's Ron's privilege to name the project as he sees fit, but when people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS.
Bogus. Linux needs a BIOS, to initialize the platform. Which we have shown the hard way. Linux simply needs not any help after it gets except on poorly designed ports.
Umm... OK - definition problem, I think. Everything needs firmware for hardware initialisation, maybe other housekeeping and booting the main OS: after that the main OS takes over. But that firmware isn't a "BIOS" - aka Basic Input Output System, which is what DOS originally need to run at all, and MS OSes need at least to boot. A BIOS, as I have always understood it (and how any halfway knowledgeable people I speak to have treated it), is precisely that firmware needed even after the main OS has taken over, but BIOS has come to mean "all firmware" in many people's minds.
Sun's have firmware - but it is not referred to as a "BIOS" (except, as mentioned, insofar as the acronym is used as shorthand for firmware). So I think it reasonable to say "Linux needs no BIOS" - most OSs don't, but the ones which have evolved on Intel didn't have much alternative. I understand that even the MS OSes are becoming more self sufficient.
The main feature of the LinuxBIOS code base is really hardware initialisation: everything else has been about extending the features of Linux (or Etherboot, Bochs). And that's as it should be.
On Tue, 10 Dec 2002, Peter Lister wrote:
Umm... OK - definition problem, I think. Everything needs firmware for hardware initialisation, maybe other housekeeping and booting the main OS: after that the main OS takes over. But that firmware isn't a "BIOS"
- aka Basic Input Output System, which is what DOS originally need to
run at all, and MS OSes need at least to boot. A BIOS, as I have always understood it (and how any halfway knowledgeable people I speak to have treated it), is precisely that firmware needed even after the main OS has taken over, but BIOS has come to mean "all firmware" in many people's minds.
Exactly correct. The idea of a BIOS goes back to CP/M, and back then an OS could not do I/O -- it asked the BIOS to do I/O. The BIOS also insulated the OS -- which only ran in a few K -- from the vagaries of all the hardware.
[[ I am pretty sure the term BIOS was invented by Gary Kildall, although there is no question that systems were built that had a BIOS-like ROM]]
So LinuxBIOS is not a BIOS in the traditional sense. But most people nowadays are not aware of this history and probably don't even care to know it. For most PC users, the BIOS is the thing that turns on the machine, sets up hardware, and starts the OS -- nothing more.
I call LinuxBIOS a BIOS because this distinction just isn't that important any more.
ron
It's Ron's privilege to name the project as he sees fit, but when people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS. The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
Personally when I first heard of LinuxBIOS I wondered what the usefulness of such a thing was. Yeah fast boots, but how is that any cheaper to buy motherboards with DoC media than to put an 8M CF card in a CF-IDE adaptor and boot from that?
Now in quantity (i.e. clusters) I think that LinuxBIOS is probably cheaper in the long run (I'm sure that Ron's done all of these analyses and so forth since it appears it's for government use) and I really am glad that Ron and the toher contributors have done what they did. It's always embarassing to write off something as being fringe and then a year or so later find yourself asking questions about it because you need to use it yourself. :-)
Regards, Andrew
Greetings,
My finding (as a commercial LinuxBIOS supplier in clusters) is that the cost difference is quite small vs booting from cf or PXE boot, etc. The big benefit in clusters is that the boot becomes MUCH more reliable.
LinuxBIOS never 'forgets' that serial console is needed, and never stops for the user to press 'any key' (and no keyboard in sight).
In other cases, I find boards that have good hardware and a good value but bugs in the BIOS make it unusable w/o LinuxBIOS.
Of course, having the source opens a number of AFAIK unexplored options for secure netbooting. G'day, sjames
On Mon, 9 Dec 2002, Andrew Kohlsmith wrote:
It's Ron's privilege to name the project as he sees fit, but when people first see "LinuxBIOS", misinterpretation does seem to be common. "LinuxBIOS" does *not* mean "a BIOS for Linux" - for one thing, Linux does needs no BIOS. The project's original intention was to put Linux in a mobo rom *instead* of a legacy PC BIOS.
Personally when I first heard of LinuxBIOS I wondered what the usefulness of such a thing was. Yeah fast boots, but how is that any cheaper to buy motherboards with DoC media than to put an 8M CF card in a CF-IDE adaptor and boot from that?
Now in quantity (i.e. clusters) I think that LinuxBIOS is probably cheaper in the long run (I'm sure that Ron's done all of these analyses and so forth since it appears it's for government use) and I really am glad that Ron and the toher contributors have done what they did. It's always embarassing to write off something as being fringe and then a year or so later find yourself asking questions about it because you need to use it yourself. :-)
Regards, Andrew _______________________________________________ Linuxbios mailing list Linuxbios@clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios
On Mon, 9 Dec 2002, Andrew Kohlsmith wrote:
Now in quantity (i.e. clusters) I think that LinuxBIOS is probably cheaper in the long run (I'm sure that Ron's done all of these analyses and so forth since it appears it's for government use) and I really am glad that Ron and the toher contributors have done what they did. It's always embarassing to write off something as being fringe and then a year or so later find yourself asking questions about it because you need to use it yourself. :-)
Actually lots of people have found uses we would not have thought of, far outside our domain of use in clusters. I have been quite surprised.
The biggest complaints I hear about the proprietary BIOSes (these are not my opinions, as a person at a USG-funded lab I am not allowed to have such opinions) - very poor code -- tons and tons of assembly that is hard to figure out - limited control -- the bios does what it is going to do, and if your hardware doesn't quite fit a PC model, you have to work around it. Example: you have something in a DIMM slot that looks like memory but is not. You're going to have to design that hardware in a special way to make sure the BIOS doesn't configure it as memory. With LinuxBIOS, that is not an issue -- you can fix this kind of thing trivially. - very high overhead costs -- initial flat fee + ongoing per-year - high unit cost -- on a $50 motherboard, you can spend a huge chunk of that on bios license. So if you get rid of the BIOS cost, you either clear more money as profit or get to be more competitive by reducing unit costs
Again, I don't endorse these claims, as I am not allowed to endorse things; but I have heard them many times from various vendors and I find them interesting.
ron
-
Adam Sulmicki -
Andrew Kohlsmith -
Big Pilot -
ebiederman＠lnxi.com -
Jeremy Jackson -
Peter Lister -
Ronald G. Minnich -
steven james