vTR-DOS for Pentagon1024SL
ZXNet echo conference «hardware.zx»
Hello, boo_boo
Branded documentation from WD for the fd179x (VG93) controller.
File: wd-fd179x-ds.djvu.rar http://zx.pk.ru/attachment.php?attachmentid=4505
Hello, All
Started writing vTR-DOS for Pentagon1024SL + NEMO_IDE
In the process I came across some ambiguities
1. Where can I read about methods for executing Beta-Disk commands?
(direct programming without TR-DOS)
Type:
a) Send a sector write command to port #1F
b) Check something (such and such bit is write protection)
c) We are waiting for something (I don’t know exactly, maybe we need to wait for some bit of something)
d) We are waiting for this (DRQ data request - where #ff or #1f is read from or you can
from any)
d) Throw a byte into the data port
e) decrease the counter, if not zero, go to d)
f) ... something else
etc...
2. Interested in time delays on signals (or even better, time diagrams
signals)
That is phrase: For each DRQ request, it is necessary to write the next one in the reg.data
byte. - How many seconds (or better yet, cycles between adjacent DRQ requests)???
After how many cycles does the index mark repeat?
Hello, TmK
see attachment
File: vg93.zip http://zx.pk.ru/attachment.php?attachmentid=4504
Hello, The Exploited
The> does the glitch really creep into memory independently of trdos??
It formats the RAM disk, and for different sizes. Apparently this is
done without the involvement of TR-DOS. If you need more detailed information, please
you need to look in the description, source code, or ask questions to Dmitry Bystrov.
Hello, TmK
TmK> 1) What is the RAM disk format for Gluk, KAY and ATM???
TmK> 2) According to information about the Pentagon: you need to know the number of ROM pages
TmK> is involved in the circuit and how it is switched.
1. I can’t say anything about the RAM disk format in KAY and ATM, in Gluk-e and in
TR-DOS 6.10e RAM disk format, if I'm not mistaken, as in TR-DOS 5.12 and TR-DOS
6.05, but inverted, i.e. grows from top to bottom. In the attached archive Gluk 6.3
along with sources and descriptions (taken from the site http://alonecoder.narod.ru),
You can look there in more detail, or even better, ask Alco if
will be required.
2. Regarding ROM pages. All pages are included. Those pages where
there are basic48, basic128, TR-DOS, they switch in the standard way, but
the start occurs from the additional page (where Gluk is located). Those.
after reset-a, an additional ROM page is turned on, which is turned off
automatically when retrieving an operation code outside the ROM (i.e. if you make a CALL or
something similar to an address older than 16384), after which switching ROM pages
no different from the regular Spectrum-128.
For vTRDOS needs, you can use either the Basic128 page or the additional page. starnitsa,
but I don’t really want to exclude Gluk, because... it has a built-in commander, disk doctor
and many additional features. Using the glitch you can format the RAM disk andwork with it, so it is advisable to keep the existing RAM disk format.
File: Gluk63.rar http://zx.pk.ru/attachment.php?attachmentid=4523
Hello KingOfEvil
Kin> It formats a RAM disk, and for different sizes. Judging by
Kin> everything, this is done without the involvement of TR-DOS. If you need more
Kin> detailed information, then you need to look for it in the description, sources, or
Kin> ask questions to Dmitry Bystrov.
I doubt that he will get into all the pages. For "virtual"
formatting it is enough to clear and initialize only the zero track
- directory area and system sector. And this is a maximum of 4Kb in one single
page.
Regarding the issue of compatibility, the issue is easily resolved by replaceable page tables.
A 640Kb disk requires a table of only 40 bytes.
Accordingly, there will be one sign at the Pentagon, and another at the ATM. More precisely, she
it's already there. And at the same time it can be changed as desired. That is, with large
If you wish, you can configure the RAM disk to the glitch standard.
Hello KingOfEvil
Kin> It formats a RAM disk, and for different sizes
1) It’s just the different volume in the case of using images that didn’t give up on us.
Because The trd disk will always have 40 pages (640K), and the scl will be converted to trd
with the read only flag.
2) To be honest, I have never seen Gluck and I don’t understand what use it can have
carry for vTRdos??? Although if necessary, it’s not a problem, but the problem is that
The NEW firmware should start first, at the first start create your own
variables, define IDE parameters, etc. So another problem arises
- how to make friends with Gluck???
3) As long as you have the sources, changing the format of the disk frames is not a problem!!! So on
At this stage, I’ll think better about the accuracy of dos’a emulation because first of all he
I needed it so that on DiHalt I don’t have to worry about floppy disks, but quickly
slip images into TRDOS :-) That is. it is necessary that all sorts of im2+music loaders
worked like on a real trdos!!!
2acidrain; Naturally; 3584000 tackts/sec is just a typo
Next question:
How to protect the RAM disk from attacks by programs???
Option 1) use the corresponding bit of the additional port, limiting
memory 128k
Option 2) More advanced. When reading/writing a sector, we check the integrity
ram disk With this implementation, there are 10,000 free cycles available perwhich need to be checked. Naturally, the checksum disappears - too
long... Does anyone have any suggestions?
Hello, Maksagor
Why do you need RAM-disk compatibility with ATM, what does it give? In my opinion this is
gives nothing.
Hello, TmK
TmK> 1) That’s just the different volume in the case of using images, we don’t care
TmK> didn’t give up.
TmK> Because. trd disk will always be 40 pages (640K), and scl will be
TmK> is converted to trd with the read only flag.
TmK> 2) Honestly, I have never seen Gluck and I don’t understand what use it is
TmK> can carry vTRdos??? Although if necessary, it’s not a problem, but
TmK> the problem is that the NEW firmware should start first, when
TmK> the first start, create your own variables, define IDE parameters and
TmK> etc. So another problem arises - how is it with Glitch
TmK> make friends???
TmK>
This is not a question for me, but for AlCo. As for the benefit of the glitch - if vTRdos is in
in its entirety will be no less convenient to use than a glitch, then the need for
the glitch disappears. Then the customizer can be placed in the additional page of the ROM
and the TRD download program.
TmK> Next question:
TmK> How to protect the RAM disk from attacks by programs???
TmK> option 1) use the corresponding bit of the additional port
TmK> limiting memory to 128kTmK>
TmK> Option 2) More advanced. When reading/writing a sector, we check
TmK> integrity of the RAM disk. With this implementation there are free ones available
TmK> 10000 cycles for which it is necessary to check. Naturally control
TmK> the amount is missing - too long... Who has any suggestions?
The first option has already been implemented in Pentagon-1024SL 2.x.
As for the second, you can read the checksum not over all bytes, but
through one (2, 3, ...). Of course, the probability of error increases, but still
remains quite acceptable. Besides, no one bothers you to use both options
simultaneously, and the first one is already implemented in hardware.
Hello KingOfEvil
OPTION 1 is not very convenient in the sense that there is more than 128k of free memory.
Although for my needs
>>first of all, I needed it so that I wouldn’t have to worry about floppy disks on DiHalt
>>and quickly slip TRDOS images with demos-)
This is quite enough, but there are people who want to code in real life at 128K - it’s convenient,
if you place asm and sts in higher pages (and I think a lot more
-> magazines, games, ASMA, graph editors...) - here they either hope that everyone
OK!!! and ask for verification, or use option 2
>>As for the second, it is not possible to calculate the checksum for all
>>bytes, and after one
Write the procedure and count how many cycles it will consume!!!!!!!!!!!!!!!!!!!!!!!!!!!
(don’t forget!!!!! for example, when recording, we checked CR - it worked, we write the file, again
We count CR!!! - i.e. almost a double operation!!!)
As an option, a checksum with a step of 1024 bytes may be suitable
or change the algorithm itself.
Hello, TmK
TmK> 2acidrain; Naturally; 3584000 tackts/sec - just a typo
Yep, typo. But this is only part of the problem, just a typo, but there is also a swear word
part -
quartz in the Pentagon is EXACTLY 14 MHz, which means cycles per second
EXACTLY 3,500,000 (three and a half mullions)!
and the frame rate there is not 50Hz but a little less (because there are not 312 but 320 lines).
=))) (thanks, buddy (thank you, friend got it ;))))
Hello, TmK
Considering that most programs usually occupy memory pages from the beginning (from
addresses #c000) then you can only check the integrity at the beginning of the page.
Hello, TmK
TmK> That is. it is necessary for all sorts of im2+music loaders to work as on
TmK> real trdos!!!
They can work better and faster than on a real TRDOS. The point is that for
When working with a RAM disk, there is no need to disable interrupts at all. And between readings
Any amount of time can pass by the next bytes from the disk.
Hello Sonic
Son> They can work better and faster than on a real TRDOS
And think about how then, for example, the same Refresh will look???
There's reading video from disk...
In general, of course, ideally you would need to enter the variable Real disk speed [yes/no].
Hello, Maksagor
> I doubt that he will get into all the pages.
Did I say that he gets into all the pages? I just said what he does
this is without the help of TR-DOS.
Hello, TmK
TmK> And guess how then, for example, the same Refresh will look???
TmK> There is reading video from the disk...
TmK> In general, of course, ideally you will need to enter the variable Real disk speed
TmK> [yes/no].
What's the point? Is frame output synchronized with reading sectors from disk?
That is, the next sector is the next frame?
However, I did not mean this, but the fact that the loss can be excluded in principle
data. The delay between issuing DRQ can be left, just to eliminate the loss (not
issue the next DRQ even after a delay if the byte has not yet been
read).