forked from FDOS/mouse
-
Notifications
You must be signed in to change notification settings - Fork 0
/
jwasm.txt
84 lines (60 loc) · 3.48 KB
/
jwasm.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
Notes on conversion of CTMOUSE 2.1beta4 from TASM to JWASM / MASM:
- TASM evendata pads with 0, JWASM even pads with 0fch
I do not know how to change this, I just used "db 0".
- sbb ch,ch xchg dx,si xchg cl,bh xchg si,di sbb al,al
xchg dh,bl xchg cx,si xchg cx,bx xchg dl,al xchg dl,ah
xchg dx,si xchg ah,al (but not ALL "xchg" instructions)
all swap the arguments when compiled with JWASM as
compared to compiled with TASM - I used "db ..." here.
- cmp ax, valuebelow128 uses "cmp ax,word" in TASM but
I cannot make JWASM do that - it uses "cmp ax,byte"
- adc ax,0 has the same problem as cmp ax,4 ...
- JWASM does not support testflag / setflag / maskflag
which compile into test / or / and in an "intelligent"
way. For example "testflag [bx],400h" would create a
"test [bx+1],4" in a TASM made binary...
Differences between TASM and MASM:
- MASM needs double "::" after labels to make them globally
visible, TASM has keyword "locals" to make all labels global
... however I had to use "::" at ONLY 28 places and having
the global-hints explicit makes the code more readable ...
- MASM uses "even" where TASM uses "evendata"
- MASM uses "STRUCNAME ends" where TASM uses "ends"
- the size info in "MYLABEL label byte" can be optional in TASM
- JWASM / TASM do not understand "warn" and various "%settings"
- macro options can be of type "rest" and "vararg" only in TASM
- TASM allows push and pop with multiple arguments... Dunno about
MASM but JWASM does not support this. I think NASM does...?
Other JWASM specifics I had to work around:
- JWASM seems to have pretty bad support for "record" bitfields,
I had to remove most references to them to avoid broken code
(later JWASM versions are a bit better with this)
- JWASM uses only 8 instead of 16 bytes for "foo db 8 dup (?,?)"
before version 1.8 - this bug is fixed in later JWASM versions
- JWASM does not allow structures to be named OPTION ;-) It also
can do less magic with macros compared to TASM... You can find
all sorts of related simplifications I had to use in jw-tasm.*
- in TASM you can say "offset X" instead of "offset PAIR.X"
in cases where this is not ambiguous (PAIR is a struct).
You cannot say [si].foo either, only [si+offset WHAT.foo].
You can also say "ends" instead of "STRUCTNAME ends" in
TASM and "LABNAME label byte" does not need the "byte" in
TASM if it can be inferred. Instead of ".model use16 tiny"
I had to use the JWASM -mt command line option.
- to make a COM file with JWASM, several changes were needed:
Instead of .startup I had to say: org 100h assume ds:DGROUP
and start:: (and at the end: end start). Using tlink /x /t
does not work, I have to use tlink /x and call exe2bin after
that (Watcom exe2bin) for ctmouse. However, tlink /x /t did
work for the utilities! Maybe related to the map info? The
ctmouse binary had 14e4 bytes _text, 2 _data, 5b9h const in
the version before I run exe2bin and com2exe. While exe to
com via exe sounds odd, it has use - gives smaller headers.
- you cannot say ERRIF in JWASM and I could not get the "hll"
(if_ loop_ etc) macros by Arkady nor the CODE_ or out_ ones
to work in a syntax suitable for JWASM...
- instead of PS2WHEELCODE[1] I had to say byte ptr [PS2WHEELCODE+1]
Change stats: 15 files edited, 2 files removed (file.mac, hll.mac),
787 lines removed, 1538 lines added (this often means "replaced a
line by a comment and added a new line"). Sum of lines of makefiles,
include files and source files: 9317. Unified diff: 4981 lines.