Just a heads up
pullmoll at t-online.de
Thu Jun 21 14:20:00 PDT 2007
After finding the "final bug", I got Salto now booting from the disk images!
What I overlooked all the time was that the CHSMERROR bit in the KSTAT
register is inverted. It is set to 1 at CLRSTAT, and is reset to 0 (not
ORed as described in the documentation), when KSTAT is written as 1.
This bug was then causing the sector task to think the word task had
detected a checksum error, and it was retrying three or four times to
load the same sector, and finally giving up.
Now it's fun to see how the headers and labels on the disks are made up as a
linked list that pulls a file's sectors into memory. It takes some time to
fill the 64K... and the bad news is: no boot seems to run after it's
It looks like I can go back to tracking down the next bug in the CPU
I suppose the disk images were verified to actually execute somewhere,
Al? Do they run in the nova-level simulation you mentioned?
More information about the Altogether-devel