lgblgblgb
commented
5 months ago I can confirm it works, at least I've tried with Xemu. However I was unaware that the first byte of the "HOME" sector must be $4C (JMP abs opcode), otherwise it returns with error what BASIC interprets as "BAD DISK ERROR", the same error message whatever the problem was (could not load the sector - real disk error -, or "only" missing $4C as the first byte). The ROM source (especially the comments) confirms this theory.
So I quickly used a hex editor to modify a D81's first sector, something like this (without using assembler, so the strange layout of the code ...):
00000000 4c 10 04 00 00 00 00 00 00 00 00 00 00 00 00 00 |L...............|
00000010 ee 20 d0 4c 10 04 00 00 00 00 00 00 00 00 00 00 |. .L............|That is basically (the sector is loaded to $0400!):
0400 JMP 0410
[...]
0410 INC D020
0413 JMP 0410The first JMP is needed to give the desired "$4C as the first byte" rule. Indeed, then BOOT SYS caused the border to flicker, so my code could run. Without $4C as the first byte, I get "BAD DISK ERROR" as I expected after learning about the "4C rule".
One interesting thing though, that the source and comments indicates something, that multiple "HOME sectors" are checked, one for "1581" one for "MS-DOS" compatibility, I can't see exactly why, or what's the purpose of this though. It would be interesting to know. I assume the "MS-DOS compatibility" means that the disk sides are handled reversed (?) it seems or something like that, but I still can't see why it's useful to have an MS-DOS formatted disk with 65xx code in the boot sector, sounds odd. That's why I am very much interested what was the reasoning of this "dual try" stuff.
BOOT SYS is under-documented and under-tested. There appears to be a need for the first byte in the boot sector to be $4c for it to succeed.
This belongs in the upcoming KERNAL API documentation, but could be added (briefly) to the BOOT SYS BASIC reference for now.