ibmpc: scribbles on odder machines

diff --git a/Kernel/platform-ibmpc/README.nonclones b/Kernel/platform-ibmpc/README.nonclones
new file mode 100644 (file)
index 0000000..3add041
--- /dev/null
+++ b/Kernel/platform-ibmpc/README.nonclones
@@ -0,0 +1,68 @@
+Not all early MSDOS machines are IBM PC
+
+It would be nice to cover some of this with BIOS mode but the BIOS is not
+sufficient for our keyboard and serial I/O drivers, and not ideal for video.
+In addition our hadcoded assumption of 640K lowmem maximum is wrong for
+these systems many of which could actually go higher. We may be able to get
+some useful info from stuff like FOSSIL drivers for the serial. Keyboard may
+be a challenge
+
+Many of these systems had their own firmware (eg Apricot) or disk loaded
+BIOS which may be a challenge!
+
+
+Dec Rainbow:
+       Keyboard is an 8251A 4800 8N1
+       Dual MPSC serial
+       Non PC Video
+       62K of RAM shared with a Z80
+       2K Z80 only (low 2K not shared)
+       Rest of memory (> 640K allowed) is 8088 only
+       Timing via Vblank 
+       Z80 can be interrupted (im0) by the 8088 (write to F7)
+       Z80 - drivers an RX50
+       RAM to EC000 (or so)
+       NVM at ED000 (256 x 4bit)
+       Video at EE000, attributes at EF000
+       I/O map in table 4/4
+       Z80A has I/O mappings to poke other CPU and FDC
+       Video has h/w scroll under line/reverse/bold/blink and double h/w
+       Lines of 80 or 132 chars followed by FF line attributes and next ptr
+Zenith Z100/Heathkit H100
+       BIOS source available in docs
+       Custom keyboard and serial drivers needed
+Apricot/Sirius
+       Very non standard indeed, BIOS is loaded with OS, Xi and later
+       have a custom different BIOS layer !
+MBC-550        
+       BIOS not in ROM
+       http://www.seasip.info/VintagePC/sanyo.html
+       no HD option
+
+How to handle keyboards via BIOS - sick thought department.
+
+The keyboard BIOS isn't designed to be called from an IRQ nor are the
+keyboard BIOS routines re-entrant (even on an original IBM PC). However
+if we *only* call the BIOS keyboard routine with the keyboard IRQ masked
+at the end of the keyboard IRQ then no keyboard re-entrancy occurs and we
+can just run the official BIOS vector then loop emptying out the buffer into
+an OS buffer, then iret.
+
+Similar approach for timer hooking
+
+Doesn't work for serial as usually not IRQ driven, but for those that are
+might
+
+
+Do we need a config area
+
+console    drivertouse
+kbd       driver
+ttyS0..3   driver for each / port / irq
+printer    driver
+disk0..3   driver to use
+fd0..3    driver to use
+emm       driver/params
+
+(joystick etc ... net ...)
+