izaya/collapseos
1
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Remove Linker

Browse Source 
Now that the boot binary is fully cross-compiled, there's no use for
the linker anymore. Theoretically, it could still be useful, but I
can't think of a real use case.

Let's take it out of the picture. If it's ever needed again, I'll
know where to find it.
This commit is contained in:
Virgil Dupras 2020-05-24 10:22:56 -04:00
parent ed2b91411a
commit 4c1cacd8d0
12 changed files with 1 additions and 147 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
blk/001
View File

@ -2,7 +2,7 @@ MASTER INDEX
3 Usage 30 Dictionary
70 Implementation notes 100 Block editor
120 Linker 150 Extra words
150 Extra words
200 Z80 assembler 260 Cross compilation
280 Z80 boot code 350 Core words
410 PS/2 keyboard subsystem 420 Bootstrap guide

16
blk/120
View File

@ -1,16 +0,0 @@
Linker
Relink a dictionary by applying offsets to all word references
in words of the "compiled" type.
A typical usage of this unit would be to, right after a
bootstrap-from-icore-from-source operation, identify the root
word of the source part, probably "H@", and run " ' thatword
RLDICT ". Then, take the resulting relinked binary, concatenate
it to the boot binary, and write to boot media.
LIMITATIONS
This unit can't automatically detect all offsets needing
relinking. This is a list of situations that aren't handled:
(cont.)

8
blk/121
View File

@ -1,8 +0,0 @@
Cells: It's not possible to know for sure whether a cellWord
contains an address or a number. They are therefore not
automatically relinked. You have to manually relink each of
them with RLCELL. In the case of a DOES> word, PFA+2, which
is always an offset, is automatically relinked, but not
PFA+0.
Load with "122 LOAD"

1
blk/122
View File

@ -1 +0,0 @@
1 9 LOADR+

15
blk/123
View File

@ -1,15 +0,0 @@
( Skip atom, considering special atom types. )
: ASKIP ( a -- a+n )
DUP @ ( a n )
( ?br or br or NUMBER )
DUP 0x67 = OVER 0x53 = OR OVER 0x20 = OR OVER 0x24 = OR
IF DROP 4 + EXIT THEN
( regular word )
0x22 = NOT IF 2+ EXIT THEN
( it's a lit, skip to null char )
( a )
1+ ( we skip by 2, but the loop below is pre-inc... )
BEGIN 1+ DUP C@ NOT UNTIL
( skip null char )
1+
;

11
blk/124
View File

@ -1,11 +0,0 @@
( RLATOM pre-comment
Relink atom at a, applying offset o with limit ol.
Returns a, appropriately skipped.
0x24 = IF: 0x24 is an addrWord, which should be offsetted in
the same way that a regular word would. To achieve this, we
skip ASKIP and instead of skipping 4 bytes like a numberWord,
we skip only 2, which means that our number will be treated
like a regular wordref. )

16
blk/125
View File

@ -1,16 +0,0 @@
: RLATOM ( a o ol -- a+n )
ROT ( o ol a )
DUP @ ( o ol a n )
DUP 0x24 = IF
DROP 2+ ( o ol a+2 )
ROT ROT 2DROP ( a ) EXIT
THEN
ROT ( o a n ol )
< IF ( under limit, do nothing )
NIP ( a )
ELSE ( o a )
TUCK @ ( a o n )
-^ ( a n-o )
OVER ! ( a )
THEN
ASKIP ;

15
blk/126
View File

@ -1,15 +0,0 @@
( RLWORD pre-comment
Relink a word with specified offset. If it's not of the type
"compiled word", ignore. If it is, advance in word until a2
is met, and for each word that is above ol, reduce that
reference by o.
Arguments: a1: wordref a2: word end addr o: offset to apply
ol: offset limit. don't apply on refs under it.
The 0x0e and 0x2b check at the beginning is to ensure we have
either a compiledWord or a doesWord. If we don't, we do
nothing. The further 0x2b check is because if we have a
doesWord, we start 2 bytes later.
)

16
blk/127
View File

@ -1,16 +0,0 @@
: RLWORD ( ol o a1 a2 -- )
SWAP DUP C@ ( ol o a2 a1 n )
DUP 0x0e = OVER 0x2b = OR NOT IF
( unwind all args ) 2DROP 2DROP EXIT THEN
0x2b = IF 2+ THEN ( ol o a2 a1 )
1+ ( ol o a2 a1+1 )
BEGIN ( ol o a2 a1 )
2OVER SWAP ( ol o a2 a1 o ol )
RLATOM ( ol o a2 a+n )
2DUP < IF ABORT THEN ( Something is very wrong )
2DUP = ( ol o a2 a+n f )
IF ( unwind )
2DROP 2DROP EXIT
THEN
AGAIN
;

16
blk/128
View File

@ -1,16 +0,0 @@
( RLDICT pre-comment: Copy dict from target wordref, including
header, up to HERE. We're going relocate those words by
specified offset. To do this, we're copying this whole memory
area in HERE and then iterate through that copied area and call
RLWORD on each word. That results in a dict that can be
concatenated to target's prev entry in a more compact way.
This copy of data doesn't allocate anything, so H@ doesn't
move. Moreover, we reserve 4 bytes at H@ to write our target
and offset because otherwise, things get too complicated with
the PSP.
The output of this word is 3 numbers: top copied address, top
copied CURRENT, and then the beginning of the copied dict at
the end to indicate that we're finished processing.
cont. )

16
blk/129
View File

@ -1,16 +0,0 @@
( Note that the last word is always skipped because it's not
possible to reliably detect its end. If you need that last
word, define a dummy word before calling RLDICT.
We first start by copying the affected area to H@+4. This is
where the relinking will take place.
Then we iterate the new dict from the top, keeping track of
wr, the current wordref and we, wr's end offset.
Initially, we get our wr and we, withH@ and CURRENT, which we
offset by u+4. +4 before, remember, we're using 4 bytes
as variable space.
At each iteration, we becomes wr-header and wr is fetched from
PREV field. )

16
blk/130
View File

@ -1,16 +0,0 @@
: RLDICT ( target offset -- )
H@ 2+ ! H@ ! ( H@+2 == offset, H@ == target )
H@ @ WORD( DUP H@ -^ ( src u )
DUP ROT SWAP H@ 4 + ( u src u dst )
SWAP MOVE ( u )
4 + DUP CURRENT @ WORD( + ( u we )
DUP .X NL
SWAP CURRENT @ PREV + DUP .X NL ( we wr )
BEGIN ( we wr )
DUP ROT ( wr wr we )
H@ @ H@ 2+ @ ( wr wr we ol o )
2SWAP RLWORD ( wr )
DUP PREV SWAP ( wr oldwr )
WORD( SWAP ( we wr )
DUP 4 - H@ <= ( are we finished? )
UNTIL H@ 4 + .X NL ;