alt.os.development Google Group

Re: programming a PCI device in DOS

do_not_h...@nohavenot.cmm (Rod Pemberton) — Fri, 12 Dec 2009 07:43:26 UT

In addition to the PCI documents, I found the following links. But, none of
them are really tutorials.
Dark Fiber's PM PCI BIOS example code in C:
[link]
Chris Giese's pci.c, pcibio16.zip, pcibio32.zip (from a mirror):
[link]

programming a PCI device in DOS

makar...@yahoo.com (anotherbrick) — Thu, 12 Dec 2009 22:35:02 UT

I have searched web for documents
about how to program a PCI device under DOS
but found only this
[link]
but it is not very clear
can you help about more online documents ?
where to look ?

Re: alt.os.development FAQ

james.harri...@googlemail.com (James Harris) — Thu, 12 Dec 2009 14:46:13 UT

To be clear, the FAQ is on a wiki. If anyone wants to run with Rod's
suggestion and make updates to the FAQ he or she can. Queries as to
the best update can be discussed in this newsgroup if desired.
I should say that anonymous updates record the updater's IP address.
Updates from a registered user do not. If anyone wants a userid please

Re: Function BB of Int 1Ah

t...@is.invalid (David) — Mon, 11 Nov 2009 01:44:25 UT

part of tcg / bitlocker

Martin S. posted how to create DOS .com with gas and ld.

do_not_h...@nohavenot.cmm (Rod Pemberton) — Wed, 12 Dec 2009 12:29:42 UT

On comp.os.msdos.djgpp,
This might be useful to guys on Linux, e.g., producing .com's for
FreeDOS/DOSEMU, or those who use DJGPP.
Rod Pemberton

Re: alt.os.development FAQ

do_not_h...@nohavenot.cmm (Rod Pemberton) — Wed, 12 Dec 2009 12:29:21 UT

On comp.os.msdos.djgpp:
Might be FAQ worthy. E.g., in the "What file formats are relevant?" links.
That's really two sections, isn't it? File format specifications and
producing binaries in a specific file format.
RP

alt.os.development FAQ

james.harris...@googlemail.com (James Harris - 1a) — Tue, 12 Dec 2009 22:50:13 UT

The FAQ for this newsgroup is held at
[link]
containing answers to questions on operating system development.

Re: Function BB of Int 1Ah

do_not_h...@nohavenot.cmm (Rod Pemberton) — Tue, 12 Dec 2009 02:37:17 UT

either.
another
and
try
Well, now that you know what it is, I found some stuff. :-)
"... Trusted Computing Group (TCG) Int 1Ah, sub-function BBh, the interface
that the Bitlocker Drive Encryption (BDE) feature of Microsoft Windows
Vista and later operating systems depends on."
[link]

Re: Function BB of Int 1Ah

h...@zytor.com (H. Peter Anvin) — Mon, 11 Nov 2009 17:45:04 UT

The sad part is that RBIL hasn't been updated for close to 10 years.
Although the BIOS/DOS universe hasn't moved much, it hasn't been
completely still in that time, and what is worse documentation for the
balance is generally extremely hard to find.
I guess it would take volunteers to gather information and publish an

Re: Function BB of Int 1Ah

t...@is.invalid (David) — Mon, 11 Nov 2009 01:44:25 UT

part of tcg / bitlocker

A new bochs gui project

cmk...@gmail.com (peter) — Mon, 11 Nov 2009 06:43:50 UT

[link]
Thank

Re: Help.Missing interrupt from rtl8139 network card.

jbem...@zonnet.nl (Jeroen van Bemmel) — Sun, 11 Nov 2009 11:47:31 UT

Suggest to look at the Linux driver for RTL8139 (see
[link])
Regards,
Jeroen

Re: Help.Missing interrupt from rtl8139 network card.

james.harri...@googlemail.com (James Harris) — Sat, 11 Nov 2009 11:51:32 UT

To the OP, did you get a fix? Or to anyone else who may be looking at
this thread later, it seems that to tell the 8139 which bytes your
driver has taken out of the rx ring buffer you need to advance the
8139's CAPR register - the current address of packet read - past the
packets you have processed. Are you already doing this?

Re: Function BB of Int 1Ah

do_not_h...@nohavenot.cmm (Rod Pemberton) — Fri, 11 Nov 2009 07:43:45 UT

spec
Not I.
I don't see BBh for Int 0x1A in RBIL (Ralf Brown's Interrupt List) either.
Are you sure the BBh is correct? There is one listed with 0Bh and another
that has 8Bh... A few of the AX = Bxxxh functions are related to PCI and
Plug-n-Play - which have specifications.
Rod Pemberton
PS. I added a couple of groups. Maybe someone knows. You might also try

Re: Help.Missing interrupt from rtl8139 network card.

james.harri...@googlemail.com (James Harris) — Wed, 11 Nov 2009 07:30:57 UT

Umm, on the rx overflow bit see my other post about the rx ring
buffer. It may be as simple a matter as properly acknowledging either
packet receipt or, since you are getting tx interrupts, perhaps even
packet transmission. Either ring buffer could fill up if its
interrupts are not acknowledged to the network card.

Re: Help.Missing interrupt from rtl8139 network card.

james.harri...@googlemail.com (James Harris) — Wed, 11 Nov 2009 07:02:26 UT

Just a guess but network devices usually have a ring buffer for packet
receipt. As well as receiving the packet are you clearing the packet
from the incoming ring buffer?
James

Re: Help.Missing interrupt from rtl8139 network card.

huxuelei...@gmail.com (leilei) — Wed, 11 Nov 2009 06:35:30 UT

On Nov 25, 11:18 am, "Alexei A. Frounze" <alexfrun...@gmail.com>
wrote:
It's very strange.
I only enabled the Rx Interrupt Bit in the Interrupt Mask Register.
But I have found that, the Tx Interrupt Bit, Rx Overflow Bit in the
Interrupt Status Register were set to 1 some times.
Especially, when the Rx Overflow bit in the Interrupt Status Register

Re: Help.Missing interrupt from rtl8139 network card.

alexfrun...@gmail.com (Alexei A. Frounze) — Wed, 11 Nov 2009 03:18:18 UT

Is it possible that you somehow disable interrupts from this device or
fail to acknowledge its interrupts (or other requests -- I'm not
familiar with network chips and I don't know what kind of protocols
they use) at some point? Or misconfigure it?
Alex

Re: An OS generator?

alexfrun...@gmail.com (Alexei A. Frounze) — Wed, 11 Nov 2009 03:15:27 UT

Texas Instruments offers a somewhat similar thingy, DSP/BIOS. It's a
tiny kernel, which can be configured through a GUI tool. The
configuration then translates to code and links with the app. You can
configure devices, ISRs, tasks and synchronization primitives.
Alex

Help.Missing interrupt from rtl8139 network card.

huxuelei...@gmail.com (leilei) — Wed, 11 Nov 2009 02:11:09 UT

Hi.
I am writting a small os base on x86 arch.
I have just written a rtl8139 driver for my os.the device's interrupt
vector is 11.
When I run my os, I can 'ping' it through windows for a while.
But after a few minutes, the rtl8139 network card has not produce any
interrupt.
I can assure the os is not dead.Because it can receive the interrupt

Re: An OS generator?

cr88...@hotmail.com (BGB / cr88192) — Wed, 11 Nov 2009 01:24:53 UT

yes.
absent further investigation, I think what is described is very well
possible, and I have a fairly good idea how it could work, but it is not
strictly "writing an OS" in the traditional sense.
very likely, it would work by breaking down the OS into lego-like
components, and then using the tool to figure out which blocks are needed

An OS generator?

do_not_h...@nohavenot.cmm (Rod Pemberton) — Tue, 11 Nov 2009 23:57:45 UT

It seems someone wrote an OS generator. It generates a RTOS from
information for a FSM.
"Given a set of 'tasks' written in C, WhatOS generates a finite-state
machine operating system in C, that is architecture independent and small
enough to run even on resource-limited 8-bit microcontrollers."
...

Re: Attributes of a memory request or address space request

alexfrun...@gmail.com (Alexei A. Frounze) — Mon, 11 Nov 2009 08:03:13 UT

On Nov 22, 1:36 pm, James Harris <james.harri...@googlemail.com >
wrote:
OK.
Um, OK.
Both. Physical is useful for the kernel, virtual may be useful
everywhere.
Yeah, you're right. It can only be enforced at the segment level but
as soon as the code segment overlaps with any data segment, it can be
read through non-CS. There's no read protection at the page level.

Re: Attributes of a memory request or address space request

cr88...@hotmail.com (BGB / cr88192) — Mon, 11 Nov 2009 02:52:26 UT

one possibility could be to use the PROT flag scheme from POSIX, but borrow
most of the rest from VirtualAlloc...
this is similar to what I did (for my interpreter), although it also accepts
the VirtualAlloc protection flags scheme (PAGE_*, but this depends on other
flags).
so, for example:
PROT_NOACCESS 0

Re: Attributes of a memory request or address space request

james.harri...@googlemail.com (James Harris) — Sun, 11 Nov 2009 21:36:38 UT

Good question. The first two are to allow the allocator to make better
initial placement or relocation decisions as follows.
Anticipated max size
- a hint how large the space is expected to grow or zero if
unspecified
- can be adjusted either up or down
Hard max size
- absolute size limit or zero if unspecified

Re: Attributes of a memory request or address space request

mar...@invalid.invalid (Marven Lee) — Sun, 11 Nov 2009 20:37:43 UT

James Harris wrote...
You could check what attributes are used by posix mmap().
To add to those you mentioned, you could have an alignment field for
the virtual address. I found it useful to be able to mmap() on a 64k
alignment for the slab allocator I use so as to be able to find the header/
beginning of a slab.

Re: Attributes of a memory request or address space request

alexfrun...@gmail.com (Alexei A. Frounze) — Sun, 11 Nov 2009 20:23:23 UT

On Nov 22, 11:24 am, James Harris <james.harri...@googlemail.com >
wrote:
What are the 4 above?
+ Virtual range, physical range, executable yet unreadable code.
+ God forbid, NUMA node. :)
Alex

Attributes of a memory request or address space request

james.harri...@googlemail.com (James Harris) — Sun, 11 Nov 2009 19:24:12 UT

Below is an attempt to come up with a maximal set of attributes that
might be needed in a memory request. (Really it's an address space
request as the actual memory can be allocated by a page fault routine
as needed.)
In the general case are all of these needed? Are any other attributes
needed or advisable so that an address space allocator can be most

Re: Operating Systems: Design and Implementation

cr88...@hotmail.com (BGB / cr88192) — Sun, 11 Nov 2009 16:14:42 UT

<snip>
<--
I happen to have a few commercial apps I can't live without and they
don't have a Linux port. Likewise it's often a pain to install some of
the free software on Windows. It may require something you don't know
how to get or may not work at all for reasons beyond my comprehension.
-->
I was not claiming that it was the case, I was claiming that it "should be"

Re: Discount Nike Air Max 87 Womans Sneakers,Cheap Wholesale Nike Air Max Shoes Air Max 87 shoes

qq_168...@126.com (qq) — Sun, 11 Nov 2009 05:28:40 UT

please look our website ,have more mode shoes clothing hat cap bags !
[link]
Wholesale Sell Discount:
Cheap Moncler Jacket Men's Women's Moncler T-shirts Men's Women's
[link]
Cheap Dsquared 2 shoes,Dsquared jeans t-shirt Dsquared jacket
[link]
Cheap Abercrombie & Fitch 2009 Polo A&F t-shirt bags Shorts jeans

Re: Are you formally taught or self-taught in OS devel?

james.harri...@googlemail.com (James Harris) — Sat, 11 Nov 2009 23:35:10 UT

I should say - before someone else does - that Berkeley's CS 162 may
turn out to be good at showing how to build an OS. I've only glanced
at it so far. From what I've seen it seems a bit high level but I may
turn out to be wrong.
James
P.S. The Berkeley CS 61CL is truly excellent. Among other things it

Re: Are you formally taught or self-taught in OS devel?

james.harri...@googlemail.com (James Harris) — Sat, 11 Nov 2009 23:01:51 UT

Yes, in the context of training materials I've recently found some
excellent teaching material from some of the US universities on
various subjects such as
[link]
[link]
For the first link (to Berkeley) click on the video feed links on the

Re: LibSDL as OS substructure

cr88...@hotmail.com (BGB / cr88192) — Sat, 11 Nov 2009 22:13:48 UT

oh, ok...
I may look into this...
yeah...
as is, for targetting my interpreter, I am using GCC...
I have a C library and many chunks of POSIX, but not much beyond this
(graphical stuff, ...).
so, I am mostly left uncertain as to the internal details of how I will
marshall all this stuff...
the use of "protocols" is one means I currently have available, and is

Re: Are you formally taught or self-taught in OS devel?

do_not_h...@nohavenot.cmm (Rod Pemberton) — Sat, 11 Nov 2009 21:43:00 UT

Was there something in particular you were looking for James?
If I put in (Yahoo):
+operating +system +design
So much stuff comes up, I can't begin to post it, or review it.
Adding .pdf moves up the .pdf results:
+pdf +operating +system +design
Wow, lots of good looking results there too.
HTH,
Rod Pemberton

Re: LibSDL as OS substructure

do_not_h...@nohavenot.cmm (Rod Pemberton) — Sat, 11 Nov 2009 21:31:42 UT

It's a wrapper for currently existing OSes. For a brand new OS, the LibSDL
API could be implemented as the OS' syscall API.
What it'd allow is the user to choose one or many games or OSes, written for
LibSDL, as their default computing platform. As more of these projects are
written for LibSDL, more choices of games and OSes will become available.

Re: Operating Systems: Design and Implementation

alexfrun...@gmail.com (Alexei A. Frounze) — Sat, 11 Nov 2009 21:02:31 UT

I happen to have a few commercial apps I can't live without and they
don't have a Linux port. Likewise it's often a pain to install some of
the free software on Windows. It may require something you don't know
how to get or may not work at all for reasons beyond my comprehension.
Whether it's standardization or packaging of several versions of

Re: Are you formally taught or self-taught in OS devel?

nowh...@never.at (wolfgang kern) — Sat, 11 Nov 2009 09:43:29 UT

Rod Pemberton asked:
I learned just the basic requirements at school:
1. LOGIC ;all the math. (including BIN/DEC/HEX/etc.-calculations)
2. LOGIC ;how to built gates and memory cells with electron tubes
3. LOGIC ;about complex circuitry and chip design (just theory then)
During several EE-jobs I were taught:

Re: Are you formally taught or self-taught in OS devel?

alexfrun...@gmail.com (Alexei A. Frounze) — Sat, 11 Nov 2009 09:15:34 UT

Yep. Those poor fellas that 10+ years back didn't know any English
missed out on a lot of stuff that was already available on the net.
Back then "ru-net" barely existed, let alone good programming info in
it in Russian. And we still lack many translations into Russian. I
think they still haven't translated the Foley and Vandam's book on

Re: Are you formally taught or self-taught in OS devel?

alexfrun...@gmail.com (Alexei A. Frounze) — Sat, 11 Nov 2009 09:04:56 UT

Never took a formal CS class. I don't think the ones on programming in
Pascal and on solving rather basic math problems (integrals, regular
and differential equations) count. I actually didn't even attend
lectures on Pascal, only did assignments and took the final if there
was any since I knew it already. All I know is from doing something

Re: LibSDL as OS substructure, and X11 response

cr88...@hotmail.com (BGB / cr88192) — Sat, 11 Nov 2009 06:58:23 UT

yeah...
SDL is basically just a wrapper though:
wrap GL/GLW/GLX, wrap threads and mutexes, ...
so, what it saves:
the big globs of OS-specific init code...
in my case, it would mostly be in use within the same app, but a protocol is
needed as otherwise I would need some kind of RPC, since the client and

Re: Are you formally taught or self-taught in OS devel?

do_not_h...@nohavenot.cmm (Rod Pemberton) — Sat, 11 Nov 2009 03:50:11 UT

There is all sorts of "current" stuff online. But, stuff prior to '98 or
'96 is harder to find.
1) OS development websites
2) Gaming development websites (lots of stuff on hardware, e.g., for DOS)
3) Academic papers on compilers, assemblers, memory management, register
allocation, OSes, etc.
[link]

Re: LibSDL as OS substructure, and X11 response

do_not_h...@nohavenot.cmm (Rod Pemberton) — Sat, 11 Nov 2009 03:49:33 UT

an
Linux,
anyone else find it interesting?
LibSDL seems powerful enough to me to implement an OS on top of it. A few
psuedo-OSes have been done: DOSBox and DOSEMU. Many modern games, such as
Quake 4 run on LibSDL. One of the most complicated game engines ever
created, MAME, which emulates nearly every _arcade_ video game in existence

Re: Are you formally taught or self-taught in OS devel?

cr88...@hotmail.com (BGB / cr88192) — Sat, 11 Nov 2009 03:24:19 UT

when I was doing OS dev, I was self-taught...
now I am not doing OS-dev, but am still self-taught WRT programming/...

Re: misc/RFC: design of a GUI API

cr88...@hotmail.com (BGB / cr88192) — Sat, 11 Nov 2009 01:49:52 UT

errm...
this does GUI drawing, granted...
but, I am not sure if it is how I wanted to approach GUI...
this would imply doing drawing in the interpreter (slow), and simply supply
raster graphics to the host app. this is lame...
I really don't know what you are talking about here...
basically, I am wanting a protocol, of a vaguely similar nature to the X11

Re: Are you formally taught or self-taught in OS devel?

james.harri...@googlemail.com (James Harris) — Fri, 11 Nov 2009 23:31:04 UT

Yes, self-taught. Or perhaps I should call it self-learning. I'm not
there yet. :-)
I've found there are some excellent training video lectures online for
various courses including some videos giving an overview of an OS but
none for OS design and implementation.
James

Re: Embedding assembler in a language

do_not_h...@nohavenot.cmm (Rod Pemberton) — Fri, 11 Nov 2009 22:50:22 UT

Either way.
Sure, why not? It may be slow, but it could be useful for bootstrapping an
x86 OS independent of cpu mode. To fully implement useable subset of x86
instructions, one needs:
1) math operations on one of: register, stack, or memory
2) ability to move data between register and stack
3) (optionally) ability to move data between registers

Re: misc/RFC: design of a GUI API

do_not_h...@nohavenot.cmm (Rod Pemberton) — Fri, 11 Nov 2009 22:47:30 UT

I
I've mentioned Depui GUI a few times. I'm not sure if it's developed enough
for what you intend.
Depui
[link]
His main page with a couple "typesafe" projects:
[link]
I've considered the *idea* of combining the low-level functionality of an OS

Re: Are you formally taught or self-taught in OS devel?

do_not_h...@nohavenot.cmm (Rod Pemberton) — Fri, 11 Nov 2009 22:34:28 UT

I'm self-taught in this area. You guys? Self? Schooled?
RP

misc/RFC: design of a GUI API

cr88...@hotmail.com (BGB / cr88192) — Fri, 11 Nov 2009 20:50:45 UT

well, I am idly thinking of things, and starting to consider ideas for how I
will approach this.
basically, I am considering the issue of how I will approach adding GUI
facilities to my x86-interpreter-based world.
firstly, the core issues (as I see them):
there needs to be a separate protocol and GUI API (I am currently leaning

Re: Operating Systems: Design and Implementation

cr88...@hotmail.com (BGB / cr88192) — Fri, 11 Nov 2009 18:10:03 UT

<--
No, the starting line is pretty much the same...
-->
raw DOS-like or simple Unix-like kernel...
at this point, no one will care, but they did care back in the early 90s...
<--
...but I totally agree that one cannot signle-handedly (nor even
double-handedly:) create all of what a regular user is accustomed to