mirror of
https://github.com/hsoft/collapseos.git
synced 2024-11-27 12:58:09 +11:00
385 lines
10 KiB
Forth
385 lines
10 KiB
Forth
( Configuration words: RAMSTART, RS_ADDR )
|
|
H@ 256 /MOD 2 PC! 2 PC!
|
|
|
|
( STABLE ABI
|
|
Those jumps below are supposed to stay at these offsets,
|
|
always. If they change bootstrap binaries have to be
|
|
adjusted because they rely on them. Those entries are
|
|
referenced directly by their offset in Forth code with a
|
|
comment indicating what that number refers to.
|
|
)
|
|
|
|
H@ ORG !
|
|
|
|
0 JPnn, ( 00, main )
|
|
0 JPnn, ( 03, find )
|
|
NOP, NOP, ( 06, unused )
|
|
NOP, NOP, ( 08, LATEST )
|
|
NOP, ( 0a, unused )
|
|
0 JPnn, ( 0b, cellWord )
|
|
0 JPnn, ( 0e, compiledWord )
|
|
0 JPnn, ( 11, pushRS )
|
|
0 JPnn, ( 14, popRS )
|
|
JP(IY), NOP, ( 17, nativeWord )
|
|
0 JPnn, ( 1a, next )
|
|
0 JPnn, ( 1d, chkPS )
|
|
NOP, NOP, ( 20, numberWord )
|
|
NOP, NOP, ( 22, litWord )
|
|
NOP, NOP, ( 24, unused )
|
|
NOP, NOP, ( 26, unused )
|
|
0 JPnn, ( 28, flagsToBC )
|
|
0 JPnn, ( 2b, doesWord )
|
|
NOP, NOP, ( 2e, unused )
|
|
NOP, NOP, ( 30, unused )
|
|
NOP, NOP, ( 32, unused )
|
|
NOP, NOP, ( 34, unused )
|
|
NOP, NOP, ( 36, unused )
|
|
NOP, NOP, ( 38, unused )
|
|
NOP, NOP, ( 3a, unused )
|
|
|
|
( BOOT DICT
|
|
There are only 5 words in the boot dict, but these words'
|
|
offset need to be stable, so they're part of the "stable
|
|
ABI"
|
|
)
|
|
'E' A, 'X' A, 'I' A, 'T' A,
|
|
0 A,, ( prev )
|
|
4 A,
|
|
L1 BSET ( EXIT )
|
|
0x17 A,, ( nativeWord )
|
|
0x14 CALLnn, ( popRS )
|
|
RAMSTART 0x06 + LD(nn)HL, ( RAMSTART+0x06 == IP )
|
|
JPNEXT,
|
|
|
|
NOP, NOP, NOP, ( unused )
|
|
|
|
'(' A, 'b' A, 'r' A, ')' A,
|
|
PC L1 @ - A,, ( prev )
|
|
4 A,
|
|
L1 BSET ( BR )
|
|
0x17 A,, ( nativeWord )
|
|
L2 BSET ( used in CBR )
|
|
RAMSTART 0x06 + LDHL(nn), ( RAMSTART+0x06 == IP )
|
|
E (HL) LDrr,
|
|
HL INCss,
|
|
D (HL) LDrr,
|
|
HL DECss,
|
|
DE ADDHLss,
|
|
RAMSTART 0x06 + LD(nn)HL, ( RAMSTART+0x06 == IP )
|
|
JPNEXT,
|
|
|
|
'(' A, '?' A, 'b' A, 'r' A, ')' A,
|
|
PC L1 @ - A,, ( prev )
|
|
5 A,
|
|
L1 BSET ( CBR )
|
|
0x17 A,, ( nativeWord )
|
|
HL POPqq,
|
|
chkPS,
|
|
A H LDrr,
|
|
L ORr,
|
|
JRZ, L2 BWR ( BR + 2. False, branch )
|
|
( True, skip next 2 bytes and don't branch )
|
|
RAMSTART 0x06 + LDHL(nn), ( RAMSTART+0x06 == IP )
|
|
HL INCss,
|
|
HL INCss,
|
|
RAMSTART 0x06 + LD(nn)HL, ( RAMSTART+0x06 == IP )
|
|
JPNEXT,
|
|
|
|
'E' A, 'X' A, 'E' A, 'C' A, 'U' A, 'T' A, 'E' A,
|
|
PC L1 @ - A,, ( prev )
|
|
7 A,
|
|
L2 BSET ( used frequently below )
|
|
0x17 A,, ( nativeWord )
|
|
IY POPqq, ( is a wordref )
|
|
chkPS,
|
|
L 0 IY+ LDrIXY,
|
|
H 1 IY+ LDrIXY,
|
|
( HL points to code pointer )
|
|
IY INCss,
|
|
IY INCss,
|
|
( IY points to PFA )
|
|
JP(HL),
|
|
|
|
( END OF STABLE ABI )
|
|
|
|
( Name of BOOT word )
|
|
L1 BSET
|
|
'B' A, 'O' A, 'O' A, 'T' A, 0 A,
|
|
|
|
PC ORG @ 1 + ! ( main )
|
|
( STACK OVERFLOW PROTECTION:
|
|
To avoid having to check for stack underflow after each pop
|
|
operation (which can end up being prohibitive in terms of
|
|
costs), we give ourselves a nice 6 bytes buffer. 6 bytes
|
|
because we seldom have words requiring more than 3 items
|
|
from the stack. Then, at each "exit" call we check for
|
|
stack underflow.
|
|
)
|
|
SP 0xfffa LDddnn,
|
|
RAMSTART SP LD(nn)dd, ( RAM+00 == INITIAL_SP )
|
|
IX RS_ADDR LDddnn,
|
|
( LATEST is a label to the latest entry of the dict. It is
|
|
written at offset 0x08 by the process or person building
|
|
Forth. )
|
|
0x08 LDHL(nn),
|
|
RAMSTART 0x02 + LD(nn)HL, ( RAM+02 == CURRENT )
|
|
RAMSTART 0x04 + LD(nn)HL, ( RAM+04 == HERE )
|
|
HL L1 @ LDddnn,
|
|
0x03 CALLnn, ( 03 == find )
|
|
DE PUSHqq,
|
|
L2 @ 2 + JPnn,
|
|
|
|
PC ORG @ 4 + ! ( find )
|
|
( Find the entry corresponding to word where (HL) points to
|
|
and sets DE to point to that entry. Z if found, NZ if not.
|
|
)
|
|
|
|
BC PUSHqq,
|
|
HL PUSHqq,
|
|
( First, figure out string len )
|
|
BC 0 LDddnn,
|
|
A XORr,
|
|
CPIR,
|
|
( C has our length, negative, -1 )
|
|
A C LDrr,
|
|
NEG,
|
|
A DECr,
|
|
( special case. zero len? we never find anything. )
|
|
JRZ, L1 FWR ( fail )
|
|
|
|
C A LDrr, ( C holds our length )
|
|
( Let's do something weird: We'll hold HL by the *tail*.
|
|
Because of our dict structure and because we know our
|
|
lengths, it's easier to compare starting from the end.
|
|
Currently, after CPIR, HL points to char after null. Let's
|
|
adjust. Because the compare loop pre-decrements, instead
|
|
of DECing HL twice, we DEC it once. )
|
|
HL DECss,
|
|
DE RAMSTART 0x02 + LDdd(nn), ( RAM+02 == CURRENT )
|
|
L3 BSET ( inner )
|
|
( DE is a wordref, first step, do our len correspond? )
|
|
HL PUSHqq, ( --> lvl 1 )
|
|
DE PUSHqq, ( --> lvl 2 )
|
|
DE DECss,
|
|
LDA(DE),
|
|
0x7f ANDn, ( remove IMMEDIATE flag )
|
|
C CPr,
|
|
JRNZ, L4 FWR ( loopend )
|
|
( match, let's compare the string then )
|
|
DE DECss, ( Skip prev field. One less because we )
|
|
DE DECss, ( pre-decrement )
|
|
B C LDrr, ( loop C times )
|
|
L5 BSET ( loop )
|
|
( pre-decrement for easier Z matching )
|
|
DE DECss,
|
|
HL DECss,
|
|
LDA(DE),
|
|
(HL) CPr,
|
|
JRNZ, L6 FWR ( loopend )
|
|
DJNZ, L5 BWR ( loop )
|
|
L4 FSET L6 FSET ( loopend )
|
|
( At this point, Z is set if we have a match. In all cases,
|
|
we want to pop HL and DE )
|
|
DE POPqq, ( <-- lvl 2 )
|
|
HL POPqq, ( <-- lvl 1 )
|
|
JRZ, L4 FWR ( end, match? we're done! )
|
|
( no match, go to prev and continue )
|
|
HL PUSHqq, ( --> lvl 1 )
|
|
DE DECss,
|
|
DE DECss,
|
|
DE DECss, ( prev field )
|
|
DE PUSHqq, ( --> lvl 2 )
|
|
EXDEHL,
|
|
E (HL) LDrr,
|
|
HL INCss,
|
|
D (HL) LDrr,
|
|
( DE conains prev offset )
|
|
HL POPqq, ( <-- lvl 2 )
|
|
( HL is prev field's addr. Is offset zero? )
|
|
A D LDrr,
|
|
E ORr,
|
|
JRZ, L6 FWR ( noprev )
|
|
( get absolute addr from offset )
|
|
( carry cleared from "or e" )
|
|
DE SBCHLss,
|
|
EXDEHL, ( result in DE )
|
|
L6 FSET ( noprev )
|
|
HL POPqq, ( <-- lvl 1 )
|
|
JRNZ, L3 BWR ( inner, try to match again )
|
|
( Z set? end of dict, unset Z )
|
|
L1 FSET ( fail )
|
|
A XORr,
|
|
A INCr,
|
|
L4 FSET ( end )
|
|
HL POPqq,
|
|
BC POPqq,
|
|
RET,
|
|
|
|
PC ORG @ 0x29 + ! ( flagsToBC )
|
|
BC 0 LDddnn,
|
|
CZ RETcc, ( equal )
|
|
BC INCss,
|
|
CM RETcc, ( > )
|
|
( < )
|
|
BC DECss,
|
|
BC DECss,
|
|
RET,
|
|
|
|
PC ORG @ 0x12 + ! ( pushRS )
|
|
IX INCss,
|
|
IX INCss,
|
|
0 IX+ L LDIXYr,
|
|
1 IX+ H LDIXYr,
|
|
RET,
|
|
|
|
PC ORG @ 0x15 + ! ( popRS )
|
|
L 0 IX+ LDrIXY,
|
|
H 1 IX+ LDrIXY,
|
|
IX DECss,
|
|
IX DECss,
|
|
RET,
|
|
|
|
'(' A, 'u' A, 'f' A, 'l' A, 'w' A, ')' A, 0 A,
|
|
L1 BSET ( abortUnderflow )
|
|
HL PC 7 - LDddnn,
|
|
0x03 CALLnn, ( find )
|
|
DE PUSHqq,
|
|
L2 @ 2 + JPnn, ( EXECUTE, skip nativeWord )
|
|
|
|
|
|
PC ORG @ 0x1e + ! ( chkPS )
|
|
HL PUSHqq,
|
|
RAMSTART LDHL(nn), ( RAM+00 == INITIAL_SP )
|
|
( We have the return address for this very call on the stack
|
|
and protected registers. Let's compensate )
|
|
HL DECss,
|
|
HL DECss,
|
|
HL DECss,
|
|
HL DECss,
|
|
A ORr, ( clear carry )
|
|
SP SBCHLss,
|
|
HL POPqq,
|
|
CNC RETcc, ( INITIAL_SP >= SP? good )
|
|
JR, L1 BWR ( abortUnderflow )
|
|
|
|
L3 BSET ( chkRS )
|
|
IX PUSHqq, HL POPqq,
|
|
DE RS_ADDR LDddnn,
|
|
A ORr, ( clear carry )
|
|
DE SBCHLss,
|
|
CNC RETcc, ( IX >= RS_ADDR? good )
|
|
JR, L1 BWR ( abortUnderflow )
|
|
|
|
|
|
PC ORG @ 0x1b + ! ( next )
|
|
( This routine is jumped to at the end of every word. In it,
|
|
we jump to current IP, but we also take care of increasing
|
|
it by 2 before jumping. )
|
|
( Before we continue: are stacks within bounds? )
|
|
0x1d CALLnn, ( chkPS )
|
|
L3 @ CALLnn, ( chkRS )
|
|
DE RAMSTART 0x06 + LDdd(nn), ( RAMSTART+0x06 == IP )
|
|
H D LDrr,
|
|
L E LDrr,
|
|
DE INCss,
|
|
DE INCss,
|
|
RAMSTART 0x06 + DE LD(nn)dd, ( RAMSTART+0x06 == IP )
|
|
( HL is an atom list pointer. We need to go into it to
|
|
have a wordref )
|
|
E (HL) LDrr,
|
|
HL INCss,
|
|
D (HL) LDrr,
|
|
DE PUSHqq,
|
|
L2 @ 2 + JPnn, ( EXECUTE, skip nativeWord )
|
|
|
|
( WORD ROUTINES )
|
|
|
|
PC ORG @ 0x0f + ! ( compiledWord )
|
|
( Execute a list of atoms, which always end with EXIT.
|
|
IY points to that list. What do we do:
|
|
1. Push current IP to RS
|
|
2. Set new IP to the second atom of the list
|
|
3. Execute the first atom of the list. )
|
|
RAMSTART 0x06 + LDHL(nn), ( RAMSTART+0x06 == IP )
|
|
0x11 CALLnn, ( 11 == pushRS )
|
|
IY PUSHqq, HL POPqq,
|
|
HL INCss,
|
|
HL INCss,
|
|
RAMSTART 0x06 + LD(nn)HL, ( RAMSTART+0x06 == IP )
|
|
( IY still is our atom reference )
|
|
L 0 IY+ LDrIXY,
|
|
H 1 IY+ LDrIXY,
|
|
HL PUSHqq, ( arg for EXECUTE )
|
|
L2 @ 2 + JPnn, ( EXECUTE, skip nativeWord )
|
|
|
|
PC ORG @ 0x0c + ! ( cellWord )
|
|
( Pushes the PFA directly )
|
|
IY PUSHqq,
|
|
JPNEXT,
|
|
|
|
PC ORG @ 0x2c + ! ( doesWord )
|
|
( The word was spawned from a definition word that has a
|
|
DOES>. PFA+2 (right after the actual cell) is a link to the
|
|
slot right after that DOES>. Therefore, what we need to do
|
|
push the cell addr like a regular cell, then follow the
|
|
linkfrom the PFA, and then continue as a regular
|
|
compiledWord.
|
|
)
|
|
IY PUSHqq, ( like a regular cell )
|
|
L 2 IY+ LDrIXY,
|
|
H 3 IY+ LDrIXY,
|
|
HL PUSHqq, IY POPqq,
|
|
0x0e JPnn, ( 0e == compiledWord )
|
|
|
|
|
|
PC ORG @ 0x20 + ! ( numberWord )
|
|
( This is not a word, but a number literal. This works a bit
|
|
differently than others: PF means nothing and the actual
|
|
number is placed next to the numberWord reference in the
|
|
compiled word list. What we need to do to fetch that number
|
|
is to play with the IP.
|
|
)
|
|
RAMSTART 0x06 + LDHL(nn), ( RAMSTART+0x06 == IP )
|
|
E (HL) LDrr,
|
|
HL INCss,
|
|
D (HL) LDrr,
|
|
HL INCss,
|
|
RAMSTART 0x06 + LD(nn)HL, ( RAMSTART+0x06 == IP )
|
|
DE PUSHqq,
|
|
JPNEXT,
|
|
|
|
PC ORG @ 0x22 + ! ( litWord )
|
|
( Similarly to numberWord, this is not a real word, but a
|
|
string literal. Instead of being followed by a 2 bytes
|
|
number, it's followed by a null-terminated string. When
|
|
called, puts the string's address on PS )
|
|
RAMSTART 0x06 + LDHL(nn), ( RAMSTART+0x06 == IP )
|
|
HL PUSHqq,
|
|
( skip to null char )
|
|
A XORr, ( look for null )
|
|
B A LDrr,
|
|
C A LDrr,
|
|
CPIR,
|
|
( CPIR advances HL regardless of comparison, so goes one
|
|
char after NULL. This is good, because that's what we
|
|
want... )
|
|
RAMSTART 0x06 + LD(nn)HL, ( RAMSTART+0x06 == IP )
|
|
JPNEXT,
|
|
|
|
( filler )
|
|
NOP, NOP, NOP, NOP, NOP, NOP,
|
|
|
|
( DICT HOOK )
|
|
( This dummy dictionary entry serves two purposes:
|
|
1. Allow binary grafting. Because each binary dict always
|
|
end with a dummy entry, we always have a predictable
|
|
prev offset for the grafter's first entry.
|
|
2. Tell icore's "_c" routine where the boot binary ends.
|
|
See comment there.
|
|
)
|
|
'_' A, 'b' A, 'e' A, 'n' A, 'd' A,
|
|
PC L2 @ - A,, ( prev )
|
|
5 A,
|
|
|
|
H@ 256 /MOD 2 PC! 2 PC!
|