SqNOS: new article "A simple guide to writing HardwareDevices"

Hi, we just uploaded a new article to
http://people.squeakfoundation.org/article/60.html.


    SqueakNOS: A simple guide to writing HardwareDevices

SqueakNOS currently has support for PC Keyboards, PS/2 Mouse, Serial
ports and the Real Time functionality of the CMOS chip on PCs. In this
article I will use the latter as an example to explain how new
HardwareDevice could be implemented.

I will also include a list of the most wanted devices, and some open
questions I'd like to discuss with interested people. The current status
of SqueakNOS calls for collaboration on many aspects, and supporting
more hardware is one of the most important.

    we hope you enjoy it.
    gera

Hello Gerardo,

Friday, June 9, 2006, 11:44:49 PM, you wrote:

GR>     SqueakNOS: A simple guide to writing HardwareDevices

Nice!  How about a BinaryStorageDevice to mount a raw HDD partition,
so something like GemStone can run there?  No file system for you
(alas, a large identity hash is needed for this to work properly).

I liked the alarm example, and it reminded me of dealing with int70h,
which is triggered by reprogramming the RTC.  As opposed to the alarm,
which is sent until you tell it to stop, int70h will be sent only once
unless you tell it to keep going.  You can select many different
frequencies, and this does not affect the RTC's time keeping.  Maybe
it will be useful for something.

Good stuff!

--
Best regards,
 Andres                            mailto:[hidden email]

I have not had time to play with SqueakNOS, but this is really a
great project and I'm enjoying reading about it. Thanks.

Dave

On Sat, Jun 10, 2006 at 01:44:49AM -0300, Gerardo Richarte wrote:

I agree with Dave and I'm in the same situation, not time right now,
but I'm VERY interested in this project.

Hope find some free time, at least to start reading and trying.

Cheers.
gsa.

2006/6/10, David T. Lewis <[hidden email]>:
> I have not had time to play with SqueakNOS, but this is really a
> great project and I'm enjoying reading about it. Thanks.
>
> Dave

hola Andres! donde estas?
> GR>     SqueakNOS: A simple guide to writing HardwareDevices
>
> Nice!  How about a BinaryStorageDevice to mount a raw HDD partition,
> so something like GemStone can run there?  No file system for you
> (alas, a large identity hash is needed for this to work properly)
>  
    Yeah! that'll be the first step (after having low level support for
HD I/O), however, I think we will anyway have something like files on
the hard drive, at least if only to organize different memory segments
or something. However, I never ever looked at the low level details of
an Object Storage system, so I may be 100% wrong (for example, if they
work like a big memory image with garbage collection and everything).
> I liked the alarm example, and it reminded me of dealing with int70h,
> which is triggered by reprogramming the RTC.
    heh! nice, good memories from the old days coming bacl! :-)

    int 70h in DOS is the handler for IRQ 8 (IRQs for the 2nd PIC are
mapped from 70h to 78h in DOS). In the example in the article we are
using IRQ 8, which is no other than int 70h disguised as an object :-)
> As opposed to the alarm, which is sent until you tell it to stop, int70h will be sent only once
> unless you tell it to keep going.
    Weird... in our case IRQ 8 continued to be triggered until we read
register C from the CMOS/RTC. (which is no other than just an I/O port
reading). May be you were doing this just to check if the IRQ was in
fact coming from the RTC?
>   You can select many different frequencies, and this does not affect the RTC's time keeping.
    That's nice! I was not sure if this would affect the time or not. I
think I'll try to see how many IRQs per second we can handle from Squeak
without loosing any, and without killing the responsiveness. I'll report
back as soon as I do it.

    Nice to hear from you! and thanks for the ideas!
    take care

    gera

Hello Gerardo,

Sunday, June 11, 2006, 12:12:16 AM, you wrote:

GR> hola Andres! donde estas?

En mi casa, por supuesto! :)

GR> if they work like a big memory image with garbage collection and
GR> everything.

Yes, modulo details.

GR> Weird... in our case IRQ 8 continued to be triggered until we read
GR> register C from the CMOS/RTC. (which is no other than just an I/O
GR> port reading). May be you were doing this just to check if the IRQ
GR> was in fact coming from the RTC?

See attached, that's where I got the info from.  I did include int70h
functionality is some of my programs and it worked great --- with the
exception that aborting the program caused immediate crashes as the
int70h handler was not restored properly.

--
Best regards,
 Andres                            mailto:[hidden email]

Hi Richie.

I'm glad to know SqueakNOS is back. I have always been a fan of it!

>>   You can select many different frequencies, and this does not affect the RTC's time keeping.
>>    
>     That's nice! I was not sure if this would affect the time or not. I
> think I'll try to see how many IRQs per second we can handle from Squeak
> without loosing any, and without killing the responsiveness. I'll report
> back as soon as I do it.
>  
Some time ago (486 OS/2 3.0), doing low latency audio experiments in
Squeak, I was able to get about 5000 with no problems with a custom
device driver and a C plugin. I could get smooth real time dsp with a
latency of about 0.2mS. I could even control the plugin with a Morphic GUI.

The limit was the time the CPU needed to respond to the interrupt (it
was documented in the Intel manuals). I guess it must be faster these days.


Cheers,
Juan Vuletich