1
0
mirror of https://github.com/hsoft/collapseos.git synced 2024-11-02 10:20:55 +11:00

Compare commits

...

2 Commits

Author SHA1 Message Date
Virgil Dupras
05c38424c7 recipes/rc2014: stage2 WIP 2020-04-08 09:13:46 -04:00
Virgil Dupras
4515241318 drv/acia: fix broken buffer bounds
How did this ever work?
2020-04-08 08:26:04 -04:00
3 changed files with 50 additions and 7 deletions

View File

@ -36,11 +36,15 @@ ACIA_MEM: Address in memory that can be used variables shared
; ;
: KEY : KEY
( As long as R> == W>-1, it means that buffer is empty )
BEGIN ACIAR> @ 1 + ACIAW> @ = NOT UNTIL
( inc then fetch ) ( inc then fetch )
1 ACIAR> +! ACIAR> @ 1 + DUP ACIA) @ = IF
DROP ACIA( @
THEN
( As long as R> == W>-1, it means that buffer is empty )
BEGIN DUP ACIAW> @ = NOT UNTIL
ACIAR> !
ACIAR> @ C@ ACIAR> @ C@
; ;

View File

@ -39,11 +39,11 @@ ACIA_IO
( advance W> ) ( advance W> )
HL INCss, HL INCss,
ACIAW> LD(nn)HL, ACIAW> LD(nn)HL,
DE ACIA) @ LDdd(nn), DE ACIA) LDdd(nn),
DE SUBHLss, DE SUBHLss,
JRNZ, L4 FWR ( skip ) JRNZ, L4 FWR ( skip )
( end of buffer reached ) ( end of buffer reached )
ACIA( @ LDHL(nn), ACIA( LDHL(nn),
ACIAW> LD(nn)HL, ACIAW> LD(nn)HL,
L4 FSET ( skip ) L4 FSET ( skip )
L3 FSET L2 FSET ( end ) L3 FSET L2 FSET ( end )

View File

@ -236,7 +236,46 @@ But how do we know that it really works? Because we can write in ROM!
:30 205b 2049 4e54 4552 [ INTER :30 205b 2049 4e54 4552 [ INTER
:38 5052 4554 2031 2046 PRET 1 F :38 5052 4554 2031 2046 PRET 1 F
TODO: continue We're now ready for a re-bootstrap. Here's what we're gonna do:
1. Bring `CURRENT` and `HERE` back to `0x98f`.
2. Set `CINPTR` to `icore`'s `(c<)`.
`(c<)` word is the main input of the interpreter. Right now, your `(c<)` comes
from the `readln` unit, which makes the main `INTERPRET` loop wait for your
keystrokes before interpreting your words.
But this can be changed. At the moment where we change `CINPTR`, the interpret
loop will start reading from it, so we'll lose control. That is why we must
prepare things carefully before that. We'll re-gain control at the end of the
bootstrap source, in `run.fs`, where `(c<)` is set to `readln`'s `(c<)`
`(c<)` word is the main input of the interpreter. Right now, your `(c<)` comes
from the `readln` unit, which makes the main `INTERPRET` loop wait for your
keystrokes before interpreting your words.
But this can be changed. At the moment where we change `CINPTR`, the interpret
loop will start reading from it, so we'll lose control. That is why we must
prepare things carefully before that. We'll re-gain control at the end of the
bootstrap source, in `run.fs`, where `(c<)` is set to `readln`'s `(c<)`.
At this moment, `icore`'s `(c<)` is shadowed by `readln`, but at the moment
`CURRENT` changes, it will be accessible again. However, this all has to change
in one shot, so we need to prepare a compiled word for it if we don't want to
lose access to our interpret loop in the middle of our operation.
> : KAWABUNGA!
( 60 == (c<) pointer )
0x9a0 0x60 RAM+ !
0x98f CURRENT !
0x98f HERE !
( 0c == CINPTR )
(find) (c<) DROP 0x0c RAM+ !
;
Ready? Set? KAWABUNGA!
TODO: make this work...
[rc2014]: https://rc2014.co.uk [rc2014]: https://rc2014.co.uk
[romwrite]: https://github.com/hsoft/romwrite [romwrite]: https://github.com/hsoft/romwrite