Hi.

I just subscribed a couple of days ago and was trawling through the recent 
messages and a couple of things came to mind:

1) Any 32-bit OS cannot use int's below 0x1f for their own use.  That is why the 
BIOS is not used (apart from the fact it isn't 32-bit).  In protected mode, these 
interrupts are used or reserved by Intel to signal various exceptions such as FPU 
overflows or page faults etc.  So any BIOS that was to be used by these OS's would 
have to use int's above 0x20 or another call method like fixed addresses for entry 
to functions. (which may just be a call to the actual location) and also keep the 
interrupt controllers out of these int numbers.

2) I assume 32-bit code is used as much as possible, ie. start up in real mode and 
switch to protected mode asap. and then deal with all the rest.  It doesn't make 
much difference for boot really, but functions used by OS's should really be 
32-bit now (name one *decent* real mode 16-bit OS).

Otherwise, the project looks interested and may be of some use.
Mark Scott
[e-mail]
Department of Artificial Intelligence, Edinburgh University
God is real unless declared integer

---
OpenBIOS -- http://www.linkscape.net/openbios/
[e-mail]   Body: un/subscribe
Problems?  [e-mail]

[Lukas Ruf]

Hi folks out there

I subscribed like the one a couple of days ago. And would like to join the
discussion by this eMail:

On Fri, 27 Feb 1998 [e-mail] wrote:
> I assume 32-bit code is used as much as possible, ie. start up in real mode and
>
> switch to protected mode asap. and then deal with all the rest.  It doesn't make 
> much difference for boot really, but functions used by OS's should really be 
> 32-bit now (name one *decent* real mode 16-bit OS).

If you speak about real mode booting -- it seems to me that you think only
about booting a i386-compatible. Further it depends on wether you think to
support the PC-standard (16b RM-BIOS support code) or imlement a complete
new BIOS where no compatibilities to existing software can be guaranteed,
ie. no existing OS would run on such an incomaptible PC-BIOS reinvention.

The problem on restricting filesystem access to the B seem's quite strange
to me.  It sounds like the approach of Apple to make all software
producers produce only one thing: software only for this standard -- else
it should inlcude some hooks where simple OS-Software Calls can hang in.
-- it might be possible that I have not seen the real aim of this
filesystem discussion (I am new in this list, sorry).

By the way: Is there a location where you can retrieve all old discussion
mails ??

Regards

Lukas

---
Lukas Ruf // Neugutstr. 9 // 8002 Zurich // Switzerland // ++41-1-2813545
+++PGP 5.0 Fingerprint:4770 7EF2 6FB2 4DBE AB66  2A46 FA5D D8EF 6E84 15C2

---
OpenBIOS -- http://www.linkscape.net/openbios/
[e-mail]   Body: un/subscribe
Problems?  [e-mail]

> Any 32-bit OS cannot use int's below 0x1f for their own use.  That is why the
>
> BIOS is not used (apart from the fact it isn't 32-bit).  In protected mode, these 
> interrupts are used or reserved by Intel to signal various exceptions such as FPU 
> overflows or page faults etc.  So any BIOS that was to be used by these OS's would 
> have to use int's above 0x20 or another call method like fixed addresses for entry 
> to functions. (which may just be a call to the actual location) and also keep the 
> interrupt controllers out of these int numbers.

Intel reserved all interrupts up to 31 (decimal) for processor interrupts,
both in real and in pmode ! It's a big mistake of IBM mapping the hardware
interrupts standard on these places ! But this can simply be solved , as
all OSs do today, by reconfiguring the Interrupt Controller, a simple job.

Linux uses indeed int 80h for OS calls. But Intel has implemented in the
386 so-called call-gates, with which OS calls can easily implemented.
This is the way to code your "call to fixed addresses", as OS kernel
functions will run in kernel mode (protection level 0) and cannot be
called directly through an address by code running in user mode (PL3)
> I assume 32-bit code is used as much as possible, ie. start up in real mode and
>
> switch to protected mode asap. and then deal with all the rest.  It doesn't make 
> much difference for boot really, but functions used by OS's should really be 
> 32-bit now (name one *decent* real mode 16-bit OS).

OF COURSE. This is just on of the goals of openBIOS! Read their homepage
carefully!

########################################
#                                      # 
# Pieter Dumon                         #
#				       #
# [e-mail]               #
#                                      #
# http://studwww.rug.ac.be/~pdumon     #
########################################

---
OpenBIOS -- http://www.linkscape.net/openbios/
[e-mail]   Body: un/subscribe
Problems?  [e-mail]