Difference between revisions of "CP/M 2.2"
(→Deleting a file that doesn't exist) |
(→Console Output) |
||
(3 intermediate revisions by the same user not shown) | |||
Line 74: | Line 74: | ||
CPM2.2 doesn't use any extra memory, and since it runs in configuration 'C0' accessing extra memory is much quicker and easier than under CPM3.0. | CPM2.2 doesn't use any extra memory, and since it runs in configuration 'C0' accessing extra memory is much quicker and easier than under CPM3.0. | ||
You should ensure your code (within the TPA) is in a safe region (0000-3fff, 8000-ffff), you can then page in the memory and copy it out to your destination directly as long as the destination is also within the safe area. It is much safer and simpler to use. | You should ensure your code (within the TPA) is in a safe region (0000-3fff, 8000-ffff), you can then page in the memory and copy it out to your destination directly as long as the destination is also within the safe area. It is much safer and simpler to use. | ||
+ | |||
+ | NOTE: This can't be used on other CPM2.2 variants on the CPC. | ||
==== Console Input ==== | ==== Console Input ==== | ||
Line 81: | Line 83: | ||
==== Console Output ==== | ==== Console Output ==== | ||
− | Console output (via BDOS functions) uses TXT OUTPUT so | + | Console output (via BDOS functions) ultimately uses TXT OUTPUT so has the potential to execute firmware control codes and display CPC firmware characters. |
+ | |||
+ | However, BDOS function 2 (Console output), which has control before TXT OUTPUT, looks for some specific characters and will perform specific processing regardless of whether it is part of a control character sequence or not. | ||
+ | |||
+ | Fore example, Console Output looks for TAB (&09). | ||
+ | |||
+ | Therefore use BDOS function 6 (Direct console input/output) to avoid this processing. With this function you can use the control codes to define windows, change modes, change colours etc. | ||
==== Serial number ==== | ==== Serial number ==== | ||
Line 108: | Line 116: | ||
otherwise you're using the stack set for CCP. | otherwise you're using the stack set for CCP. | ||
− | ==== MOVCPM ==== | + | ==== MOVCPM.COM ==== |
− | MOVCPM is used to change the size of the TPA which is the area used by programs. This relocates CCP and BDOS. SYSGEN then writes this to a disc so that when this disc is booted CP/M has a different sized TPA. By reducing the TPA, memory is freed up which is not touched by CPM and which can be used by the user. | + | MOVCPM.COM is used to change the size of the TPA which is the area used by programs. This relocates CCP and BDOS. SYSGEN then writes this to a disc so that when this disc is booted CP/M has a different sized TPA. By reducing the TPA, memory is freed up which is not touched by CPM and which can be used by the user. |
CPM 2.2 is generated using MOVCPM on the CPM 2.2 disc as follows: | CPM 2.2 is generated using MOVCPM on the CPM 2.2 disc as follows: |
Latest revision as of 07:05, 15 December 2019
Operating system for the Intel 8080 and Zilog Z80 based computers by Digital Research
CP/M 2.2 was the first CP/M avaiable for the Amstrad CPC. A minimum requirement to run CP/M was a disc drive therefore CP/M 2.2 came bundled with the Amstrad DDI-1 discdrive with the BIOS in the AMSDOS ROM, making it possible to run CP/M 2.2 on the Amstrad CPC 464. Later when the Amstrad CPC 664 was released, CP/M 2.2 came together with this machine as the 664 had a built in discdrive. With the CPC6128 it was on one of the two system discs, with CPM+ on the other.
Contents
- 1 Tips
- 2 Implementations
- 3 Download
- 4 Manual
Tips
CPM 2.2 doesn't allow a program to handle BDOS errors itself like CPM+ does therefore you need to take additional steps to avoid them if you wish your program to continue:
Deleting a file that doesn't exist
First check if a file exists before deleting it:
ld c,&11 ;; search for first ld de,fcb call BDOS cp 255 jr z,doesnt_exist ld c,&13 ;; delete file ld de,fcb call BDOS doesnt_exist:
Disc change
If the disc is changed, e.g. another disc is inserted into drive A in a single disc system, then the disc will be marked as read-only. Any modification operations (e.g. delete or write) will fail and the program will be aborted with a BDOS read-only error.
To avoid this reset the disk system and re-select the current drive:
ld c,&19 ;; get current disk call BDOS push af ld c,&0d ;; reset disk system call BDOS pop af ld e,a ld c,&0e ;; select disk restoring it to read/write state call BDOS
Implementations
Vortex's implementation of CP/M 2.2
Dk'Tronics 61K TPA CPM 2.2
Amstrad's implementation of CPM 2.2
CP/M 2.2 was distributed with the DDI-1 disc interface, the CPC664, and on side 1 of the CPC6128 System_Disk discs. It provided 41K TPA.
Details of the implementation are in the DDI-1 Firmware SOFT158A.
The following attempts to give more detail than described there.
Creating a minimal system
To create a minimal CPM2.2 boot disc follow these steps:
- Boot into CPM2.2 from system discs
- Use disckit2 to format a blank vendor format disc.
- Use BOOTGEN to put the boot sector onto your new disc.
- Use SYSGEN to put CPM2.2 into the system tracks.
- Your minimal CPM2.2 boot disc is ready.
You can then add additional files using FILECOPY.
Extra memory
CPM2.2 doesn't use any extra memory, and since it runs in configuration 'C0' accessing extra memory is much quicker and easier than under CPM3.0. You should ensure your code (within the TPA) is in a safe region (0000-3fff, 8000-ffff), you can then page in the memory and copy it out to your destination directly as long as the destination is also within the safe area. It is much safer and simpler to use.
NOTE: This can't be used on other CPM2.2 variants on the CPC.
Console Input
Console input (via BDOS functions) uses firmware functions and returns firmware key codes when a key is pressed. i.e. &f0 is cursor up.
Console Output
Console output (via BDOS functions) ultimately uses TXT OUTPUT so has the potential to execute firmware control codes and display CPC firmware characters.
However, BDOS function 2 (Console output), which has control before TXT OUTPUT, looks for some specific characters and will perform specific processing regardless of whether it is part of a control character sequence or not.
Fore example, Console Output looks for TAB (&09).
Therefore use BDOS function 6 (Direct console input/output) to avoid this processing. With this function you can use the control codes to define windows, change modes, change colours etc.
Serial number
The CPM2.2 serial number appears to be stored on the disc. This is the number that is in the box on the label. Not the one with "serial number" next to it.
It's a 6 byte number stored after 'USER' (in CCP) on the disc and before BDOS in memory. D0 seems to be Amstrad's manufacturer code.
TPA
The top address of TPA can be found by reading the byte at 0007 (which is the top byte of the BDOS JP). This is the start of BDOS. CCP lives below this. If you wish to include CCP subtract &A from this.
e.g.
0005 JP &8C06
0007 is &8C. So up to &8C00 can be used.
Subtract &A to include CCP. &8C-&0A = &82. So up to &8200 can be used.
When writing a COM program you should setup the stack before execution like this:
ld hl,(&0006) ld sp,hl
otherwise you're using the stack set for CCP.
MOVCPM.COM
MOVCPM.COM is used to change the size of the TPA which is the area used by programs. This relocates CCP and BDOS. SYSGEN then writes this to a disc so that when this disc is booted CP/M has a different sized TPA. By reducing the TPA, memory is freed up which is not touched by CPM and which can be used by the user.
CPM 2.2 is generated using MOVCPM on the CPM 2.2 disc as follows:
MOVCPM 179 *
Which reports a 44KB CPM (Which is TPA+CCP+BDOS).
If you wish to reduce the TPA you need to use a smaller number. e.g.
MOVCPM 160 *
Which reports a 40KB CPM.
Once done, then use SYSGEN * to write it to a new disc.
Technical information
CP/M 2.2 uses the firmware. There are 2 BIOS jumpblocks in RAM and the "ENTER FIRMWARE" which can be patched. The normal CPM BIOS jumpblock and an extended one.
CP/M uses the firmware extensively and enters it via "ENTER FIRMWARE".
The BIOS is in the AMSDOS ROM, and this uses the jumpblocks in RAM.
The AMSDOS ROM also uses part of the RAM for XDPB storage and a sector buffer. (The BASIC facing side of AMSDOS is not active.)
The TPA is limited because the CCP, BDOS and firmware jumpblocks are in main RAM. The result is a ~33KB TPA for a CPC system which is OK.
- Amstrad CPM2.2 uses Amstrad's "System" format. This is 40 tracks, 1 side, 9 sectors per track numbered &41-&49. Each sector holds 512Bytes. There are two reserved tracks, then the directory which has 64 entries, and then the data area.
- CP/M is booted by a RSX command "|CPM" which is implemented by the AMSDOS ROM (the standard disc ROM in the DDI-1, CPC664, CPC6128 and 6128Plus).
1. The |CPM command loads track 0, side 0, sector 41 into ram at &100 and jumps to it. This is the boot sector. (The boot sector occupies &100-&2ff).
2. The boot code copies the BIOS jumpblock from the AMSDOS rom to &500. (In the AMSDOS ROM this is located at &C17F. The actual address is passed to the boot rom in BC).
3. The boot code then loads track 0, side 0, sector 42 into ram at &300 which is the "configuration" sector which defines colours, mode, and keyboard translation tables. (The configuration sector occupies &300-&4ff). The configuration sector sets the mode and displays the start up text.
4. Finally the boot code calls CPM "WBOOT".
- WBOOT loads 1 sector from track 0, side 0, sector 48 to the sector buffer. This is the first sector of platform independent part of CPM.
- It then peeks into the sector buffer and reads from offset 1,2 the address of the first JP of the CPM code. This is normally JP &9A5C.
- It adds &FCA4 to this address to get the start loading address for CPM CCP/BIOS which becomes &9700. Then it copies the sector into RAM. Therefore this address is used to relocate CCP/BIOS. This address can be indirectly defined using MOVCPM utility.
- It now loads 10 further sectors (of the platform independent CPM) to start + &200 (normally &9900). This goes up to &AD00.
- It adds &EA00 to &AD00 to get back the original start of &9700. It remembers this.
- It adds &F206 to start (&9700) to get the start of the BDOS (normally &9F06) and stores this address in the JP at &0005.
- It recalculates the BDOS jumpblock (normally at &AD00 and stores the address of the WBOOT)
- Next it copies the bios jumpblock from the ROM to this address.
- Finally it jumps into CCP and starts it (&9700).
Memory Map
CP/M runs with ram selection 0 (i.e. LD BC,&7FC0:OUT (C),C ) and assumes the upper ROM is enabled and is the one with the CPM BIOS in it (i.e. AMSDOS.)
Therefore if you patch CPM you need to:
- ensure the RAM selection state is as expected
- the upper ROM is enabled and AMSDOS ROM is selected (this can be done using firmware functions to get the current rom state and number and restore it).
- avoid touching AMSDOS/firmware areas unless you are also patching it.
- restore the alternative registers if you modify them.
The lower firmware jumpblock exists at &0000-&0040 because it is needed by the firmware jumpblock to call into the OS rom. The following changes are made for CP/M 2.2:
- &0000-&0002 => WBOOT (JP &AD03)
- &0003 => IOBYTE
- &0004 => user number/drive number
- &0005 => &0007 - BDOS (JP &9F06)
TPA is from &100-&96ff (38400 bytes, 37KB TPA). Then:
- &9700 - &acff => CPM 2.2 CCP/BDOS (This can be moved with MOVCPM.)
- &ad00 - &ad32 => CPM BIOS jumpblock. (This can be moved with MOVCPM.)
Each entry is a JP instruction. (The address of this can be calculated from CP/M if you need to patch it. Take the address at &0001/&0002 and subtract 3 to give the address of the boot JP instruction). These jumps including enter/exit firmware can be patched.
If you plan to patch it for additional hardware and need memory then you will need to use MOVCPM to reduce the TPA size.
The memory allocated by MOVCPM follows the CPM BIOS jumpblock up to &ad33 in RAM.
For CPC AMSDOS the BIOS jumpblock has the following addresses in it:
boot - C1B2 wboot - C2BE const - C2E1 conin - C2C3 conout - C2C8 list - C2D2 punch - C2D7 reader - C2DC home - CDE9 seldsk - c2f2 settrk - c524 setsec - c529 setdma - c51a read - c2f7 write - c2fc listst - c2cd sectran - c55a
- &ad33 - &ad36 => "enter firmware" jump. This doesn't move when MOVCPM is used.
- &ad37 - &ad42 => used by "enter firmware" for storage of registers etc for transitioning to firmware and restoring back.
- &ad43 - &bb00 => used by AMSDOS.
- &b900 - &be00 => main firmware jumpblock
- &be40 - &be7f => used by AMSDOS and locates the XDPB for each drive.
- &be80 - &bebf => CP/M 2.1 BIOS extended jumpblock. (functions here used by conin, conout, lstst, list, punch, reader, const and action depends on iobyte. )
- &bec0 - &bfff => stack
- &c000 - &ffff => screen
When MOVCPM is used the memory looks like this:
low firmware jumpblock, &100..<top of TPA>, CCP/BDOS, BIOS jumpblock, <free space>, &AD33 "enter firmware", firmware etc.
Patching the BIOS jumpblock
The BIOS jump block can be patched.
Note that the default boot and wboot call into the ROM and they are effectively a JP. They don't return.
- WBOOT re-reads the CCP and BDOS from the system tracks and re-initialises the jumpblocks. It doesn't re-initialise ENTER FIRMWARE.
WBOOT is triggered when a transient CP/M program is quit. (e.g. run CLOAD and then press ESC to return back to CP/M). Therefore it's problematic if patched.
- BOOT effectively performs a |CPM command from the beginning, re-loads the boot sector, re-loads the configuration sector and then re-runs WBOOT. This is also problematic to patch.
Both WBOOT/BOOT may not be patchable (to be confirmed) and if they are you need to use very specific firmware related hacks or quite involved hacks.
XSUB.COM and persistent programs in TPA
X-Sub is a persistent program which is loaded at the top of TPA. This is how it works:
- X-Sub patches the BDOS jump to it's own JP.
- The relocated X-Sub module starts with a JP and has 'xsub' immediately after it.
- X-Sub detects if it's installed by reading the BDOS JP and searching for 'xsub'. If found it is already installed, otherwise it's not. Based on this method it uses it assumes it has been installed last.
- It assumes that the address of the JP is at the start of the BDOS code. The JP at 0005, jumps to a JP at the start of the BDOS 6 bytes in (the first 5 bytes of the BDOS are the CPM2.2 serial code). It then takes the address and subtracts it's size (+ size of CCP) to find the start of the space it's allocated itself, it then copies itself to this memory and relocates the code, and then executes it's startup code.
- X-Sub's startup code installs it's own wboot JP and it's own bdos JP. It intercepts these calls and passes them on the BDOS itself.
- It then returns back to CCP to allow it to continue processing.
- X-Sub's wboot restores the dma address and jumps back to CCP. It doesn't call the wboot in the BDOS.
- X-sub's bdos JP handler filters the BDOS calls. It also remembers the dma address when used.
To have your own program like this your JP needs to be at the start of your module, you should patch BDOS jp and pass it to the one that was there before to allow it to handle the calls.
Installing persistent programs
There are various ways to install persistent programs but they are all specific to the CPC version of CP/M 2.2.
Method 1
This method installs a persistent program outside of CP/M.
- Use MOVCPM to allocate some RAM - TPA will always be reduced even when your program is not running.
- When your COM program is loaded locate the BIOS jumpblock.
- Add &33 to this to get the first byte of useable RAM - the last byte is &AD32. You can use this to check if there is enough free space for your program to be installed. There doesn't seem to be a way to determine if there is a program installed already so it looks like only 1 program at a time can be put here.
- Your program can then patch the BIOS jumpblock and re-direct to it's own functions. It is best to keep a copy of the functions you patched so that you can call them.
File:Driver.zip template for a device driver.
Method 2
This method installs a persistent program inside CP/M and uses TPA when it's installed.
- See above how XSUB installs itself. X-Sub allows the CCP to be active. - GSX will install itself, and the appropiate drivers and then runs your program. It doesn't allow the CCP to be active.
File:Cpm22 driver xsub method.zip is a template driver that install's itself like X-SUB. However, since it doesn't overwrite CCP, this means TPA is reduced more than necessary.
Links
Download
- CP/M 2.2 (DSK for Emulators)
Manual
- Entire CP/M 2.2 Manual on the Web
- Entire CP/M 2.2 Manual (in case the above link doesn't work for you)
- CPM22 p7.jpg
Page 7
- CPM22 p8.jpg
Page 8
- CPM22 p9.jpg
Page 9
- CPM22 p10.jpg
Page 10
- CPM22 p11.jpg
Page 11
- CPM22 p12.jpg
Page 12
- CPM22 p13.jpg
Page 13
- CPM22 p14.jpg
Page 14
- CPM22 p15.jpg
Page 15
- CPM22 p16.jpg
Page 16
- CPM22 p17.jpg
Page 17
- CPM22 p18.jpg
Page 18
- CPM22 p19.jpg
Page 19
- CPM22 p20.jpg
Page 20