RE: Questions

[Nick Jarmany]

Title: RE: Questions

Elias (et al)

 

No, it doesn't make three devices. It means:-

 

1) Standard FLASH chip for Linux BIOS (not expensive, propriatary & physically large DoC) tailored just to boot a Linux OS from an IDE device (or, I accept, Ethernet). Last time I looked a 2M flash chip for a bloated PC BIOS was around $3.00 - pretty good in my books.

2) Compact flash based bootable Linux kernel + apps. Prices of CF devices is very competitive and performance is way ahead of DoC. Handy converters are also available that plug directly onto a standard IDE header. Last time I looked a 64MB CF device was under $35 and a 32MB device around $22 and that's 1-off.

 

This would result in:-

1) A "BIOS" that would initialise the hardware in no time flat and boot our favourite OS (Linux) from an IDE/LAN connection.

2) A boot system that is rugged, fast, compact and cheap that is also compatible with other standard IDE devices, meaning all the existing tools for these devices are available.

 

The only disadvantages with such a system are:-

 

1) Still a requirement to drastically squeeze the Linux distribution to efficiently use the CF device.

2) All flash devices (that does mean ALL) have finite write durability, so swap files and apps that like to continuously update disk files are certainly out

 

However the above disadvantages are no different to LinuxBIOS DoC only implemtations and, I believe, offers much more flexibility.

 

I am not having a dig at LinuxBIOS here, I am just saying that for me the most valuable part of the LinuxBIOS project is to effect a fast, efficient hardware checking/initialisation solution for booting a Linux kernel that, hopefully, supports a wide range of hardware platforms. Get rid of that dreadful PC BIOS! No one else is doing this and it is certainly extremely valuable work. I just feel this excellent work should not be distracted by / blurred with the actual loaded OS. This also needs attention, but for me is a separate issue.

 

Oh dear, this got rather long! Further comments/hate mail welcomed!!!

 

Best wishes all,

 

Nick Jarmany

-----Original Message-----
From: Elias Rajczyk [mailto:[EMAIL PROTECTED]]
Sent: 11 February 2002 11:54
To: Nick Jarmany; [EMAIL PROTECTED]
Subject: RE: Questions

 Doesn't that make three components to take care of : the flash chip for booting, the CF for the Linux kernel and the IDE device (say) for doing whatever you want to do. If such it sounds like increased headache. But , say, if you took the pain to develop the relevant drivers, wouldn't it be more effective to get them right into core LinuxBIOS, keeping size boundaries in mind, of course.

Elias.

-----Original Message-----
From: Nick Jarmany [mailto:[EMAIL PROTECTED]]
Sent: Monday, February 11, 2002 1:03 PM
To: [EMAIL PROTECTED]
Subject: RE: Questions

Question - why is everyone so hung up on getting everything inside the actual "BIOS" chip?

This clearly creates problems, especially when using motherboards based on "standard" PC requirements.

A simple PC BIOS Flash chip is now very cheap. Why not settle for a target of:-

1) Replacing BIOS with a LinuxBIOS in the smallest space possible, ie cheapest common flash chip.
2) Using Compact Flash (CF), or other similar technologies, placed on the IDE bus that mimics a standard disk drive.

As I understand it 1) is pretty much done. 2) May have been done but believe there is probably more support needed for trimming standard kernel + necessary driver/support modules to fit into smallest/cheapest CF devices.

Am I alone in this?

Regards,

Nick Jarmany

-----Original Message-----
From: Ian [mailto:[EMAIL PROTECTED]]
Sent: 08 February 2002 02:15
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Subject: Re: Questions

> The nice thing is that right now this is our big problem.  You have to climb
> an awfully long ways up hill to get this far.
>
> And this is only a problem on machines designed for general purpose
> uses. Embedded systems have larger ROM chips so can do more.

Yes, you guys have done an awesome job ... and is a privelege to be able
to argue about something as cool as this :-) [seriously!]

> I really want to find out what the rom situation is with AMD760 MPX
> chipset.  If I read it right this is the first non-intel system with
> LPC only support.   Which makes large roms much more practical.  But
> I still might be reading the situation wrong.
What's LPC?  A replacement for DIP32?  Actually, I was dumb-struck when
I discovered that my new ASUS board has no ISA slots :-(   All PCI ...