- Ctrl-u now clears the current line (I really needed this
bad... :)
- The Noname version of MILO is now able to read the impure area.
This means that the memory size will be determined automatically.
Also, the CPU cycle and timer interrupt frequency in the hwrpb are
now initialized properly. Doing a "cat /proc/cpuinfo" should
give you correct values in those fields. Unfortunately, I haven't
found the impure area on the Cabriolet yet.
Also, if you were having problems with the keyboard on a Noname booted
via MILO: search no further! I think what happened is that MILO
didn't eat up the key-release code of the last keystroke. This
confused the initialization code in the Linux keyboard-driver and
resulted in a message "Failed to change to scanmode 2" and a stuck
keyboard. My solution was to change the initialization code to eat up
the extraneous scancode if one is present. Seems to work well. If
you need this patch, let me know (otherwise, it should show up in one
of the next kernel patches).
Unfortunately, I'm still having problems with interrupts on the Noname
when using MILO. In particular, I often get "scsi0: INTFLY with no
command pending" or "ide0: spurious interrupt" messages. It may be
that something is flakey with the hw that I have, but the interesting
thing is that the same kernel works just fine when booting it via SRM.
Maybe there are some differences in the PALcode? I especially wonder
what the MILO PALcode does w.r.t. machinecheck handling.
--david