Special Z80 Instructions: Difference between revisions

From Hackspire
Jump to navigation Jump to search
(Wikification of BrandonW's documentation located at http://rafb.net/p/wDBBpW40.html)
 
mNo edit summary
 
Line 2: Line 2:


==Prefixes==
==Prefixes==
It seems that there are three two-byte prefixes for every special opcode. Here are the three found in the 2.42 OS so far:
It seems that there are three two-byte prefixes for every special opcode. Here are the three found in the odd-numbered OS versions so far:
* 0EDh, 0EDh: this happens during an error in the boot code, so it might reboot to Nspire mode or something. It has data after it (possibly code).
* 0EDh, 0EDh: this happens at many points in the OS. It has data after it (specifying a specific operation).
* 0EDh, 0EEh: this is where the Flash sector erase commands were in _EraseFlash.
* 0EDh, 0EEh: this is where the Flash sector erase commands were in _EraseFlash.
* 0EDh, 0EFh: this is where the Flash write commands were in _WriteFlash.
* 0EDh, 0EFh: this is where the Flash write commands were in _WriteFlash.
Line 19: Line 19:


==0EDh, 0EDh Opcodes==
==0EDh, 0EDh Opcodes==
Each of these opcodes has a prefix of two EDh bytes and a suffix of a 10h byte.
Each of these opcodes has a prefix of two 0EDh bytes and a suffix of 10h.
A lot of these will reboot the Nspire if run from RAM. Get around this by putting a small loader in Flash (like page 0).
Some of these will reboot the Nspire if run from Flash. A lot of these will reboot the Nspire if run from RAM. Get around this by putting a small loader in Flash (like page 0).
This documentation is horribly incomplete. If you can confirm the correctness of any of this, please do.
 
* 02h: this is executed before turning off (right before BCALL _PowerOff). Don't know what it does, might not work from RAM.
* 02h: this is executed before turning off (right before BCALL _PowerOff). Don't know what it does, might not work from RAM.
* 03h: this is executed before _SetAppRestrictions...it locks the Nspire so that only the 84+SE keypad can be used, not the Nspire one. Reboots the Nspire if executed from RAM.
* 03h: this is executed before _SetAppRestrictions...it locks the Nspire so that only the 84+SE keypad can be used, not the Nspire one. Reboots the Nspire if executed from RAM.
Line 62: Line 64:


==Other Prefixes==
==Other Prefixes==
* 0EDh, 0EEh: this is where the Flash sector erase commands were in _EraseFlash.
* 0EDh, 0EEh: this replaces the Flash chip commands in _EraseFlash. If you specify a page in OS space, the sector erase will not take effect.
* 0EDh, 0EFh: this is where the Flash write commands were in _WriteFlash.
* 0EDh, 0EFh: this is where the Flash write commands were in _WriteFlash. It appears to set up emulator pointers so that the following "ldi" will write to the proper file in the Nspire file system. If you specify a page in OS space, the write will not take effect.

Latest revision as of 06:54, 9 June 2008

The Nspire's Z80 emulator interprets many invalid instructions as special opcodes.

Prefixes

It seems that there are three two-byte prefixes for every special opcode. Here are the three found in the odd-numbered OS versions so far:

  • 0EDh, 0EDh: this happens at many points in the OS. It has data after it (specifying a specific operation).
  • 0EDh, 0EEh: this is where the Flash sector erase commands were in _EraseFlash.
  • 0EDh, 0EFh: this is where the Flash write commands were in _WriteFlash.

Locations

Below are some of the locations in OS 2.42 where these opcodes are used.

00:090Eh: 0EDh,0EDh,02h,10h,18h,05h
00:0B04h: 0EDh,0EDh,05h,10h,0CDh,80h,3Bh,0CDh,0ECh,3Bh,0C3h,0F2h,3Bh
7F:4BE4h: 0EDh,0EEh (Flash sector erase from _EraseFlash)
7F:4C4Dh: 0EDh,0EFh (Flash write from _WriteFlash)
7F:6016h: 0EDh,0EDh,06h,10h
7D:73AFh: 0EDh,0EFh (Flash write duplicate?)
7D:7BDCh: 0EDh,0EDh,03h,10h,0C9h
7D:7C50h: 0EDh,0EDh,04h,10h,0C9h 

0EDh, 0EDh Opcodes

Each of these opcodes has a prefix of two 0EDh bytes and a suffix of 10h. Some of these will reboot the Nspire if run from Flash. A lot of these will reboot the Nspire if run from RAM. Get around this by putting a small loader in Flash (like page 0). This documentation is horribly incomplete. If you can confirm the correctness of any of this, please do.

  • 02h: this is executed before turning off (right before BCALL _PowerOff). Don't know what it does, might not work from RAM.
  • 03h: this is executed before _SetAppRestrictions...it locks the Nspire so that only the 84+SE keypad can be used, not the Nspire one. Reboots the Nspire if executed from RAM.
  • 04h: this is executed before _RemoveAppRestrictions...it unlocks the Nspire so that either keypad can be used, instead of just the 84+SE one. Reboots the Nspire if executed from RAM.
  • 05h: this is executed before displaying reset text on homescreen and after zeroing RAM. Might initialize USB ViewScreen as host or itself as peripheral (unlikely). Makes the LCD go gray all over. Reboots the Nspire if executed from RAM.
  • 06h: this is executed in the boot code, receives a byte through the I/O link. Returns NZ if failed. Guessing it returns in A, don't know if it blocks.
  • 08h: this is executed somewhere on page 6 (OS 2.44). Might check if the I/O link is good or something. If NZ returned, USB used. Maybe "bit 5,(iy+41h)", might mean more USB data should be read.
  • 09h: this is executed somewhere on page 74h (OS 2.44). This finishes an output of bytes in either host or peripheral mode.
  • 0Ah: this is executed somewhere on page 74h (OS 2.44). This is called before receiving bulk data. Returns C if problems.
  • 0Bh: this is executed somewhere on page 74h (OS 2.44). Returns NZ if an A cable is plugged in.
  • 0Ch: this is executed somewhere on page 7Ch (OS 2.44). Called immediately after pressing kReceive on LINK menu. If Z returned, USB attempted without 0Dh trap.
  • 0Dh: this is executed somewhere on page 74h (OS 2.44). Called immediately after pressing kReceive on LINK menu. If NZ returned, I/O used instead of USB. Equivalent: "in a,(4Ch) \ and 8"
  • 0Eh: this is executed somewhere on page 74h (OS 2.44). Gets a byte of bulk data to A from either port 0A1h or 0A2h, depending on whether calc is host or peripheral. Returns C if problems.
  • 0Fh: this is executed somewhere on page 74h (OS 2.44). You must call this after using 0Eh. It finishes outputting the byte, or something.
  • 10h: this is executed somewhere on page 74h (OS 2.44). This is called just before outputting bulk data. Use this to start it. Returns C if problems.
  • 11h: this is executed somewhere on page 74h (OS 2.44). This outputs a byte (A) to either port 0A1h or 0A2h, depending on whether calc is host or peripheral.
  • 12h: this is executed somewhere on page 7Ch (OS 2.44). Called immediately after pressing kReceive on LINK menu. This is some sort of USB init and returns NZ if failure.
  • 13h: this is executed somewhere on page 74h (OS 2.44). This is basically BCALL 5290h, full USB init.
  • 17h: this is executed somewhere on page 74h (OS 2.44). This requests incoming bulk data (the first time, I think).
  • 18h: this is executed somewhere on page 74h (OS 2.44). This is called in BCALL 5260, so maybe it's some sort of USB shutdown. Might clear any pending input/output data.
  • 19h: this is executed somewhere on page 74h (OS 2.44). Equivalent: "in a,(8Fh) \ and 5 \ cp 5" Initializes USB device if NZ returned.
  • 1Ah: this is executed somewhere on page 74h (OS 2.44). Not sure, but used in USB init. Must return NZ if USB init is okay to do. Equivalent: "ld a,(9C75h) \ bit 6,a"
  • 1Bh: this is executed somewhere on page 74h (OS 2.44). Same as 1Ah, not really sure.
  • 1Ch: this is executed somewhere on page 74h (OS 2.44). This kills the USB device. Destroys A, apparently.
  • 1Dh: this is executed somewhere on page 74h (OS 2.44). This starts an output of bytes to port 0A0h.
  • 1Eh: this is executed somewhere on page 74h (OS 2.44). This sends the byte in A out port 0A0h.
  • 1Fh: this is executed somewhere on page 74h (OS 2.44). This finishes an output of bytes to port 0A0h.
  • 20h: this is executed somewhere on page 0 (OS 2.44). I don't know, but it checks something USB-related. The OS will call the 9C1Ch callback if it returns NZ.
  • 21h: this is executed somewhere on page 76h (OS 2.44). I have no idea, but it returns something in NZ (error)? and is link or USB related. Basically "bit 0,(iy+43h)". Used in boot code (6Fh).
  • 22h: this is executed somewhere on page XXh (OS 2.44). Somehow related to 90A8h/90A9h, but I THINK that this waits a certain number of milliseconds (crystal timer replacement?). Uses BC as input.
  • 24h: this is executed somewhere on page 0 (OS 2.44). This returns whether bulk data is ready to be read. Returns NZ if so. The OS will call the 9C1Eh callback after reading it.
  • 25h: this is executed somewhere on page 74h (OS 2.44). Not sure. Might be some sort of USB error thing.
  • 26h: this is executed somewhere on page 74h (OS 2.44). This reads a byte in from port 0A1h and puts it in A.
  • 27h: this is executed somewhere on page 0 (OS 2.44). I don't know, but it checks something USB-related. The OS calls the 9C13h callback if it returns NZ.
  • 29h: this is executed somewhere on page 6 (OS 2.44). I don't know, but it checks something USB-related. If it returns NZ, the OS calls the auto-launch stuff.
  • 2Ah: this is executed somewhere on page 6 (OS 2.44). I don't know, but it does something USB-related. It MIGHT return vendor/product ID and firmware version info in HL/DE/BC.
  • 2Bh: this is executed somewhere on page 74h (OS 2.44). This finishes reads from port 0A1h. It returns something in A...might be 0FFh if there's more data to get, 0 if none.
  • 2Ch: this is executed somewhere on page 74h (OS 2.44). It looks like this requests more reads from port 0A1h.
  • 2Fh: this is executed somewhere on page 79h (OS 2.44). I don't know, but happens just before low battery message displayed in boot code, and just after on page 79h.
  • 30h: this is executed somewhere on page 0 (OS 2.44). No idea, but it's called just after the initial _getKey call on reset.

Other Prefixes

  • 0EDh, 0EEh: this replaces the Flash chip commands in _EraseFlash. If you specify a page in OS space, the sector erase will not take effect.
  • 0EDh, 0EFh: this is where the Flash write commands were in _WriteFlash. It appears to set up emulator pointers so that the following "ldi" will write to the proper file in the Nspire file system. If you specify a page in OS space, the write will not take effect.