[9fans] old hardware, new install: adding uartisa to pcf solves ps2 keyboard issue

[Axel Belinfante <Axel.Belinfante@cs.utwente.nl>]

Summary:

at new installation from live cd on old hardware, with  pcf  kernel
the keyboard did not
work until I added a line  uartisa    to the pcf kernel config and
compiled a new kernel.


Context:

while installing plan 9 using live cd(*) on ancient hardware(**) I
noticed the following:

  1) with the 9pccd kernel on the live cd the ps2 keyboard works as
expected
    (and when the kernel is starting, at some points the numlock light -
      the only light on on the keyboard - is switched off)

  2) after installation from the live cd,
     when the 9pcf kernel prompts 'root is from' (the first prompt it
gives)
    typing on the ps2 keyboard has no effect.
    also: numlock light remains on while the kernel starts

   unplugging and replugging the keyboard at that moment
   makes it change the lights (first: all on, then all off) and I can
type again

The only significant difference seems to be in the kernel config files:
  pccd  does contain a line   uartisa   and   pcf  does not


building a new kernel with a line   uartisa   added to (a copy of)
the  pcf  file(**)
gave me a kernel where the ps2 keyboard also worked without problems,
just like the 9pccd kernel of the live cd.


Searching the archive I found some posts of 2010 with subject 'kbd
fail',
where a call to a setleds() function is mentioned - reading that made me
wonder whether my experience is related.



Axel.


*) obtained from the bell labs site, of 2011-08-06

  I also tried 9front live cd,
  but it didn't work, (error) messages I recall:
    - somehow it thinks it could only use very little memory has 8Mb (?)
    - several programs crashed using a rc boot script(?)
    - panic after a while

**)  amd-k6- 333Mhz, GA-586SGM motherboard, 64Mb memory
this hardware used to run plan 9 until a few years ago when
I found that a disk in my kenfs had crashed. I postponed fixing, a
long time.
luckily my kenfs had mirrored worm disks, one of which seems to be o.k..
also the disk that holds the cache seems ok.

**) also the string bootpcf.out was updated, to reflect name of copy