People,
I have managed to now get a set of HDOS drivers to read from an IDE file in WinAPE via the SF2 emulation. I am in the process of checking these with HDOS itself.
I have also created a very simple disk editor (there is no GUI, just keyboard entries).
Attached is the ROM (+ source) with the drivers to READ/WRITE/INIT the SymbiFace-II/X-MASS device. INIT is part of the HDOS startup to get the partition geometry for the FAT32 structure definition.
Can people please try this on a REAL CPC with X-MASS/SymbiFace-II and let me know if the editor is fully function (Note: be aware that write may corrupt the media, so use it at your own peril). The read of the first sector matches the expected contents for a FAT32 partition.
The disk editor has the following commands:
'R' - Read current sector
'W' - write current sector
'S' - read sector x
'F' - forward to next sector
'B' - Back to previous sector
'P' - display next page of current sector
'E' - change contents (offset from current display data).
'I' - display Boot sector info (if it is current sector)
To start the editor
RUN "SF2DE
if people want to test this in WinAPE, remember to set it up with an IDE file associated with the SF2 interface.
if an error is displayed as a number (check code DR-SF.asm for what codes mean).
rpalmer
Quote from: rpalmer on 13:22, 20 March 16
Can people please try this on a REAL CPC with X-MASS/SymbiFace-II and let me know if the editor is fully function (Note: be aware that write may corrupt the media, so use it at your own peril). The read of the first sector matches the expected contents for a FAT32 partition.
Hi,
Here are the results with SymbiFaceII and a 4GB FAT32 partition:
(http://i68.tinypic.com/2cdak3s.jpg)
It seems valid judging the message that can be read in spanish...
for me its working correctly
if you want any particular test just tell me!
Hi Ray,
presumably there are commands such as |cd, |dir etc?
Munchausen,
The ROM posted only has the drivers to READ/WRITE/INIT the device. HDOS which has the file system contains the required commands to access the media as if it were AMSDOS formatted. The HDOS ROM is currently under going testing to make sure it works as expected.
Joseman,
The pictures seem to not correlate with an expected sector format for FAT32 (See attached images from WinAPE with a test IDE file).
Rpalmer
Joseman,
I just re-looked at your pictures and find that your device has an MBR (Master Boot Record). The sector is found in normal IDE devices formatted in Windows to assign this MBR to the first sector for multiple partitions on the same device. The pictures I have attached shows that the sector I am dealing with may be formatted with just one partition and has no need for the MBR. This would make it very difficult to know which is which under HDOS.
I will need to look further into how I can manage this situation.
rpalmer
Looks as though it works then :)
There's some detailed information here (http://www.easeus.com/resource/fat32-disk-structure.htm) on the MBR. WinAPE IDE files could in theory support multiple partitions since they are just raw sector data. The latest version of WinAPE also works with SD Cards and USB drives being selectable as well as CD/DVD/BR drives (not sure if these are actually fully functional since the current version probably only handles 512 bytes sectors, not sure).
Quote from: Executioner on 00:05, 22 March 16
....
WinAPE IDE files could in theory support multiple partitions since they are just raw sector data.
....
Yes, I created an image which is in principle a mirror of a whole harddrive ~768MB with 6 partitions (3 primary and 3 extended each 128MB), and it works fine reading it out sector wise under WinApe. :)
HAL6128,
Could you please post a few screen shots so that I can get a better picture of the image you mentioned to improve the drivers.
Rpalmer
A picture from the image? It's a simple image of a harddrive created under linux with the 'dd' command.
The first sector is a common MBR. Nothing spectacular.
[attach=2]
[attach=3]
Can someone please tell me how the SF2 emulation in WinAPE works.
I am getting nowhere trying to find where the start of the directories are in the test ide files supplied with the emulation.
I have looked at the first sector and find that is says there are &0150CF9F sectors (22073247 sectors, i.e 10 GigaBytes), yet the file is 94K. This tells me that something is seriously missing...
I calculated from the test.ide file where the directory is expected to be and find it is not there.
The specification for Boot sector data matches the original pictures I posted earlier about sector 0 of the test ide file.
The start of directory is expected to be calculated as follows:
First Sector of Directory Data = Reserved Sector Count + (Number of FATS * Number of FAT sector)
The Information I have for the formula is
Reserved Sector Count = 32
Number of FATS = 2
Number of FAT sector = &2a10 (10768)
The number i get is &5440 (21568), I read this sector and get all &FF for all bytes.
Attached is the test ide file I had played with using symbos to copy a few files and create a directory.
I also checked the first 64 sectors at the start of the test ide file and seem to locate/verify the reserved sector contents and start of the FAT.
rpalmer
Attached is the updated dsk file and ROM which i tested the test ide file with.
just run sf2de
Quote from: rpalmer on 14:59, 25 March 16
Attached is the updated dsk file and ROM which i tested the test ide file with.
just run sf2de
Do you need any test on the real deal?
joseman,
Could you please check via real hardware and manual calculations using your configuration.
rpalmer
I found my problem in reading sector data.
I had the values in REG HL back to front and was reading the wrong sector, but write still causes issues (i completes, but a read back hangs...funny)
rpalmer
People,
I have now got the drivers to read successfully sectors from the test.ide file in WinAPE via the developmental HDOS/Drivers ROMS. These are attached (see Zip file contents).
A test program "tc-sf2.bas" is included in the attached dsk file and a sanp shot of its output is attached as well.
The test of the SF2 for its catalog works with out obvious issues, but further testing to read a file will add confidence to HDOS support of the SF2 interface.
Once testing of the file read/write side is successful, i feel that HDOS will then be able to offer to the CPC users a means to access Hard discs in BASIC via the SF2.
I would also like to see if X-MASS works with these items as I understand that the X-MASS has the same I/O ports.
rpalmer
Hi
Hi,
Here is the result running tc-sf2.bas on real CPC 6128 with XMASS FAT 16.
HDos driver and HDos are in pos 5/4 in XMEM.
[attach=2]
Last line loops indefinitely.
dirtybb,
Could you also please run the "sf2de.bas" and get snapshots after it starts along with the partition information.
Your setup maybe starting with an MBR rather than the Boot sector and this may cause the initialisation to end up with the wrong values for the test file output you show.
rpalmer
sure,
[attach=2]
Info:
[attach=2]
dirtybb,
I can see from the picture that the total sectors = &367E0 (i.e 223200). The sectors per block (AKA cluster) is 4, so the number of clusters is 55800.
I have checks my code and find that the driver did not properly determine FAT16/FAT32 type. I will correct this and get back to you.
rpalmer
dirtybb,
I have what i think is a corrected driver and HDOS to support FAT16 on SF2.
Give this one a go and see if it works.
rpalmer
Hi
It does the same.
Something i did not mention yesterday, i ve to run 2 times the test program.
First, it gives this :
[attach=2]
Then i type run and :
[attach=3]
dirtybb,
The number in the geometry details appear somewhat unusual, however it might be that my understanding of differences between FAT16 and FAT32 may be the cause.
You can try the attached rom.
I should note that it is difficult to debug without actual hardware or an IDE file with your config (hopefully for not too long).
Rpalmer
Quote from: rpalmer on 13:28, 28 March 16
dirtybb,
The number in the geometry details appear somewhat unusual, however it might be that my understanding of differences between FAT16 and FAT32 may be the cause.
You can try the attached rom.
I should note that it is difficult to debug without actual hardware or an IDE file with your config (hopefully for not too long).
Rpalmer
Same results.
Before the tests, i freshly formated the XMASS with Prodatron's Tool under Symbos, and then created a few directories/copied a few files.
On its side, impDOS can access the catalogue/browse the directories.
If you need something else, just tell me
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379) if you didn't live so far away I'd send you my symbiface for testing!
Late to the party...
Here are my photos, the results are negative...
I'm using the last hdos rom, tc-sf2.bas hangs after path:
This time is a 16GB fat32 partition made with ms-dos, first fdisk /mbr, to erase de mbr, next make 1 primary partition and format it.
(http://i67.tinypic.com/344d090.jpg)
Joseman,
I am in the process of correcting the ROMs, I have managed to get the driver to not timeout with the device not being ready (see attached). It was accomplished using the IDE recalibrate command to wake up the device and return the drive back to track 0.
rpalmer
People,
I have now managed to successfully get FAT32 to work with the test IDE file in WinAPE. I am more confident that it should now work with FAT16.
Attached are the ROMs for people to please test again. Hopefully, this time they should work more reliability.
The 'test-basic.zip' is the test file I worked with in winape.
rpalmer
Hi, same result, but line doesn't loop for ever now, it ends at 16 times.
After it writes :
Items: 16
Space: 16K
I've check my XMASS, on root i've 4 directories and 14 files.
Dirtybb,
HDOS does not handle long filenames, so maybe that is why you see less items
rpalmer
I have been testing the write function of the drivers and find that when it writes my routines do not report any issues, yet when I read the sector back i get a timeout waiting for the device ready bit to be set.
I cant understand why this is so. Has anyone had the same issue within WinAPE accessing the symbiface functions?
Note: I can do successive reads with no issues.
The reason I ask of this is that for FAT16 testing i need a test IDE file formatted to FAT16 and intend to create a program to create just a thing. So without write capability I am stuck.
rpalmer
I'm sorry if this question sounds stupid, but I'm not quite sure what is the difference in approach of Syx & Ast's DOSes they're making for X-mass vs this one.
Can someone explain how all of these relate to each other? Is the key difference about this being made for SF2 and the other two I mentioned for X-mass? But conceptually all are still an Amsdos compatible driver/DOS for FAT16 and/or FAT32?
Singaja,
HDOS has been developed so that other devices can be used without the need for the file O/S needing to know anything above the device itself. All the developer needs to do is create a driver.
This means HDOS can work both with or without AMSDOS meaning a default CPC464 can access a high capacity media without AMSDOS being present.
Also HDOS has supportability for more than just SF2/X-MASS. It was designed to support:
a. CPC-IDE8255 (standard IDE via 8255 PPI chip, my own design although Yarek's may work).
b. RAM disc (512K/2Mb/4Mb or any other RAM based version - driver dependent)
c. Floppy (AMSDOS is needed for this, since the low-level routines are in AMSDOS itself)
d. Compact Flash
e. SymbiFace-II
g. X-MASS
h. SD (similar to that seen on spectrum, see drivers source).
i. Albireo SD (future).
I would also point out the I understand that Syx/Ast DOSes are hardwired for the device and that they do not seem to work via standard OS calls (i could be wrong).
rpalmer
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379) how much does HDOS cost?
munchausen, nothing its free now as I have been posting the ROMs here
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379)
Will i be able to Format the DOM Under HDOS and then take the DOM from an X-Mass and connect it to a PC and be able to load it with files.
and then reconnect it to the CPC and use it without any errors under HDOS ??
Thanks Ray
Audronic,
I had to remove the FAT initialization part of HDOS to create more room for FAT16/32 routines. The initialization can still happen, but there is no FAT compatibility.
The drivers ROM once fully tested for READ and WRITE (without issues) can be used via an external program to initialize the X-MASS/SF2 media (note: People do not FORMAT an IDE device, you only initialize it by writing a Partition/file system structure).
If people do wish to FORMAT an IDE device they must have vendor knowledge of the structure of the media, this why people don't FORMAT as they do not have this vendor knowledge.
rpalmer
PS. I now live in Perth (work moved me there 9 months ago) for those who would like to know we i live now. Previously it was in Melbourne.
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379)
Ok Thanks.
Ray
People,
Good news about the HDOS drviers for SymbiFace-II (may X-MASS too).
I can now read and write to individual sectors. This means people can use the 'sf2de.bas' program to initialize the media in a manual sense or modify the program to do all of the work.
My next stage would be do just that to create/initialize media according to desired parameters.
Attached is the drivers rom and source for SF-II interface drviers
rpalmer
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379)
I have just tried the roms in a 6128 withe X-mem and X-mass , Placed the roms in 09 and 10.
HDOS-20160402.ROM in rom slot 09
HDOS-DR-20160409.ROM in rom slot 10
I am unable to Init the hard drive (128 Meg Dom).
The 6128 and the DOM Works with BDos
Am i missing something ?
Thanks
Ray (Williamstown Victoria)
Audronic,
What actually happened?
I have removed FAT initialisation from HDOS to make room for additional FAT16 support (yet to be checked at the moment).
rpalmer
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379)
Basically it just crashed, changed screen mode, garbage on screen ??
I will reinstall the Roms and try to get some photos.
Ray
Quote from: Audronic on 04:52, 09 April 16
@rpalmer (http://www.cpcwiki.eu/forum/index.php?action=profile;u=379)
Basically it just crashed, changed screen mode, garbage on screen ??
I will reinstall the Roms and try to get some photos.
Ray
It just crashes , sometimes just hangs, other times changes screen mode.
Do i have the roms in the correct slots ??
Ray
I have HDOS in ROM 5 and driver in 6.
I have found the source as seen on the wiki forum does not properly conform to the ATA specification.
See Page 260 for Host PIO Data In or Page 264 for Host PIO Data out. It states BSY=0 AND DRQ = 1 before data transfer while the source only looked at BSY = 0 for 'Ready'. My code initially had included DRQ checks which hung in WinAPE, but removal seemed to fix the issue.
As mentioned before without actual hardware it is slow to get the software to work.
rpalmer
Quote from: rpalmer on 05:30, 09 April 16
As mentioned before without actual hardware it is slow to get the software to work.
I don't think any emulator currently emulates the ATA spec exactly. So always work with real hardware where possible and the emulators can be fixed :)
I have started on an IDE test, but it has been put on hold while I fix CRTC emulation.
Exactly. The main problem for SyX with ACMEDOS was to write it without the real hardware.
All think working fine under WinAPE, but always failled when I have tried to test it from my side.
(so, I have sent a board to him that took 4 months to reach his home... And seriously start to work)
Yes. I also agree. I finalized a Routine which initialized a IDE in Winape but doesn't work ob real Hardware. The last Winape version has enhanced IDE/ATA compability, but still not fully and for the CPC not necessary.
Quote from: HAL 6128 on 14:24, 09 April 16
Yes. I also agree. I finalized a Routine which initialized a IDE in Winape but doesn't work ob real Hardware. The last Winape version has enhanced IDE/ATA compability, but still not fully and for the CPC not necessary.
Ja, programmier mal lieber mit Originalhardware, am Emu werden einige Bit nach wie vor nicht emuliert, also kann man da schnell in Probleme laufen.
With the differences in the emulation to the hardware, i will continue with WinAPE, however I have been assisted by audronic and it seems X-MASS does not play ball either.
I think I will need to develop one version for emulation of SF-II and one for the real hardware. The real hardware version is possibly simpler to code as the routine I used to issue IDE command is simple enough to change to follow the standard.
Attach are the current drivers and updated status error code to further identify which wait routine timed out.
The routine in question is "SF2_SendCommand" which i feel can be understood by most to correlate it the the standard.
rpalmer
Quote from: rpalmer on 01:30, 10 April 16
With the differences in the emulation to the hardware, i will continue with WinAPE, however I have been assisted by audronic and it seems X-MASS does not play ball either.
The WinAPE emulation was developed to work with the SymbOS code originally. If there's something not emulated correctly that needs to be, let me know.
Executioner,
The basic issue i have found is that the if the symbiface-II emulation matches the real hardware then the real Hardware is also faulty (as by comments stating that the real hardware does not match WinAPE developed version of software).
The protocol for PIO mode is shown on Page 260 and 264 of the previously attached ATA specification. it basically should be
Wait for current device to not be busy (BSY=0)
Select new device (master/slave)
Wait for selected device to not be busy and ready (BSY=0/RDY=1)*
Issue command parameters
Issue command
wait for 'Data Request' (BSY=0/DRQ=1)
send/receive data
check if communication error
if error then exit with error
exit with no error
* is where the interface issue appears to be (sometimes BSY = 1)
Assuming only 1 sector is requested to be processed.
The other thing i found is the command (&EC) to get device geometry appears to not work and this makes the all software reliant on the media being preformatted to have the geometry within key parts of the first sector (e.g MBR or Boot Sector).
The other thing i noticed in the wiki S/W is that writing to the device is doing 'INC B' followed by 'OUTI', yet to read its 'INI' followed by 'INC B'. Why is this so, since for write the 'INC B' no longer point to the 'Data Register' for I/O when 'OUTI" is executed.
rpalmer
"The basic issue i have found is that the if the symbiface-II emulation matches the real hardware then the real Hardware is also faulty (as by comments stating that the real hardware does not match WinAPE developed version of software)."
I should have said
The basic issue i have found is that the if the symbiface-II emulation matches the real hardware then the real Hardware is also faulty (as by comments stating that the real hardware does not match WinAPE developed version of software according to ATA standards).
rpalmer
I wouldn't say faulty, more it follows a pragmatic and CPC oriented approach. Therefore a lot of ATA stuff is not necessary. Real hardware with its IDE Controller instaed yes but without necessity / usability.
What I have read in the forum is that Prodatron programed the basic drivers for SymbOS on real hardware (a Symbiface), the rest (most stuff) was written with WinApe Assembler.
Quote from: rpalmer on 13:37, 10 April 16
The other thing i noticed in the wiki S/W is that writing to the device is doing 'INC B' followed by 'OUTI', yet to read its 'INI' followed by 'INC B'. Why is this so, since for write the 'INC B' no longer point to the 'Data Register' for I/O when 'OUTI" is executed.
rpalmer
Hi,
the reason for this behavior is the microcode of the "OUTI" and "INI" instruction inside the Z80. In both cases register B is originally allotted to serve as counter, not as address register.
In the case of "OUTI" the Z80 first decrements B and puts the contents of 16Bit-register BC on the address bus afterwards. Therefore you have to increment B before "OUTI".
In the case of "INI" the Z80 first puts the contents of 16Bit-register BC on the address bus and decrements B afterwards. That's all.
netmercer,
The OUTI microcode actually outputs data to the port first the decrements both B followed by incrementing HL (see Z80 doco attached).
rpalmer
netmercer,
I just found doco to support your statement, I was WRONG I assume what i posted was the correct understanding of the OUTI instruction according to symbolic description.
rpalmer
We already discussed that a while ago :
Z80 Documentation Errors (http://www.cpcwiki.eu/forum/programming/z80-documentation-errors/msg51350/#msg51350)
gerald,
Yes I agree it was put to rest and yes I AGREE I GOT IT WRONG and have learn my ways on this issue.
But as for the SF2 issues between WinAPE and the real deal, it is difficult to articulate the problem(s) with out the real deal to compare against. This is why I would like as many people as possible to try the drivers and test software to see if I can make a version work in WinAPE and on the real kit.
rpalmer
Quote from: rpalmer on 13:37, 10 April 16
Wait for current device to not be busy (BSY=0)
Select new device (master/slave)
Wait for selected device to not be busy and ready (BSY=0/RDY=1)*
Issue command parameters
Issue command
What does "Select new device (master/slave)" entail? Sending which commands? I can't see any obvious code leaving BSY=1
executioner,
see attatched driver for SF2 (routine: SF2_SendCommand).
rpalmer