Index
> Main > flat assembler 1.65Goto page Previous 1, 2, 3, 4 Next |
| Author |
|
|
revolution 04 Jan 2006, 02:26
Today I tried 1.65.4 but still cannot get a successful compile.
Code: use32 macro lea [stuff] { common local b1,b2,b3,address address equ match reg=,[addr],stuff\{address equ addr\} match reg=,size[addr],stuff\{address equ addr\} match reg=,size =ptr addr,stuff\{address equ addr\} virtual b1=0 b2=0 lea stuff if ($-$$)=2 load b1 byte from $$ load b2 byte from $$+1 else if ($-$$)=3 load b1 byte from $$ load b2 byte from $$+1 load b3 byte from $$+2 if ~(b1=8dh & (b2 and 307o)=004o & b3=24h) & \ ;lea reg,[esp] ~(b1=8dh & (b2 and 307o)=105o & b3=00h) ;lea reg,[ebp] b1=0 end if else if ($-$$)=6 & ~address eq load b1 byte from $$ load b2 byte from $$+1 if b1=8dh & (b2 and 307o)=005o ;lea reg,[immediate] b1=100h else b1=0 end if end if end virtual if b1=08dh if ~((b2 and 7o)=((b2 and 70o) shr 3)) db 08bh,b2 or 300o else ;do nothing end if else if b1=100h db 0b8h or ((b2 and 70o) shr 3) dd address else lea stuff end if } lea eax,[eax] ;this line okay lea eax,[esp] ;<-- error: extra characters on line. |
|||
04 Jan 2006, 02:26 |
|
|
Tomasz Grysztar 04 Jan 2006, 07:57
I just keep being blind.
 Try it now. |
|||
|
04 Jan 2006, 07:57 |
|
|
revolution 04 Jan 2006, 11:13
Still having problems with 1.65.5. Going back to the LEA example:
Code: macro lea [stuff] { common local b1,b2,b3,address address equ match reg=,[addr],stuff\{address equ addr\} match reg=,size[addr],stuff\{address equ addr\} match reg=,size =ptr addr,stuff\{address equ addr\} virtual b1=0 b2=0 lea stuff if ($-$$)=2 load b1 byte from $$ load b2 byte from $$+1 else if ($-$$)=3 load b1 byte from $$ load b2 byte from $$+1 load b3 byte from $$+2 if ~(b1=8dh & (b2 and 307o)=004o & b3=24h) & \ ;lea reg,[esp] ~(b1=8dh & (b2 and 307o)=105o & b3=00h) ;lea reg,[ebp] b1=0 end if else if ($-$$)=6 & ~address eq load b1 byte from $$ load b2 byte from $$+1 if b1=8dh & (b2 and 307o)=005o ;lea reg,[immediate] b1=100h else b1=0 end if end if end virtual if b1=08dh if ~((b2 and 7o)=((b2 and 70o) shr 3)) db 08bh,b2 or 300o else ;do nothing end if else if b1=100h db 0b8h or ((b2 and 70o) shr 3) dd address else lea stuff end if } lea esi,[8] lea esi,[eax] lea esi,[eax*8] lea esi,[eax+p] lea esi,[p] lea esi,[eax*8+p] ;<-- error: missing end directive |
|||
|
04 Jan 2006, 11:13 |
|
|
Tomasz Grysztar 04 Jan 2006, 12:02
I reduced the problem to this source:
Code: if 0 dd eax*8+p end if Unfortunately I have no time now to fix it, I will look into it later. |
|||
|
04 Jan 2006, 12:02 |
|
|
Borsuc 04 Jan 2006, 12:26
Dunno if this was fixed though, didn't try latest version (I was quite busy), but here's the "invalid expression", contains no brackets:
Code: macro push reg { if reg in <ip,eip> ; do something else push reg } It gives me an error if I use push with a numeric value (which is not in ip or eip), like a label. push _end, for example, where _end is a macro local label. If I try this: Code: macro push reg { if reg eq ip pushw $ else if reg eq eip pushd $ else push reg } It works OK. dunno why it doesn't work with in directive. Hope it gets (maybe is, sorry, didn't try latest version) fixed. |
|||
|
04 Jan 2006, 12:26 |
|
|
Tomasz Grysztar 04 Jan 2006, 13:08
The bug with in operator was aleady reported by halyalvin (see previous posts), and it's fixed.
|
|||
|
04 Jan 2006, 13:08 |
|
|
Borsuc 04 Jan 2006, 13:16
sorry, didn't observe it that well in his post.. I though the bug came because he didn't use <> angle brackets in in operator. great if it's fixed
 |
|||
|
04 Jan 2006, 13:16 |
|
|
Borsuc 04 Jan 2006, 14:12
just a silly suggestion. how about some directives (!equ !define, something like that), that replace the last symbolic constant with the new one.
Thus: Code: x equ match ... { restore x define x ... } ; somewhere at the end of the macro restore x could be written as Code: x equ match ... { !define x ... } ; somewhere at the end of the macro restore x dunno, just some ideas (it could be useful, like in my posted macro with the 'false' constant, etc..). It's not that hard to read, it's pretty the same as equ or define. at least, I tried making a !define macro, and I couldn't make it work correctly like the define. when I tried !equ macro i couldn't put empty constant (i.e x !equ ) sure, you can do it manually, so it's not so important.  |
|||
|
04 Jan 2006, 14:12 |
|
|
Tomasz Grysztar 04 Jan 2006, 15:05
The variant of (posted above) "equ2" macro working also with empty values, too would look like:
Code: struc equ2 value { match ,value \{ restore . define . \} match processed,value \{ restore . define . processed \} } |
|||
|
04 Jan 2006, 15:05 |
|
|
Tomasz Grysztar 04 Jan 2006, 17:14
revolution: hunting down that bug (well, this time it was somewhere else than in the logical expression parser, actually) allowed me to also to re-optimize some of the parts of expression parser - check out the updated 1.65.5, hope it doesn't cause too much mess.
 |
|||
|
04 Jan 2006, 17:14 |
|
|
revolution 05 Jan 2006, 04:12
Today I achieved my first successful compile with 1.65.5.new. And it compiles a teeny bit faster than 1.64.
This is a good thing  |
|||
|
05 Jan 2006, 04:12 |
|
|
Tomasz Grysztar 05 Jan 2006, 07:21
I haven't yet made any reliable speed tests - perhaps Fresh timing and making some statistics would help here.
|
|||
|
05 Jan 2006, 07:21 |
|
|
decard 05 Jan 2006, 10:55
OK so I compiled Fresh. I did it a few times to get better results. I did tests on my Athlon XP 64 3000+ (2.1 GHz). FASM v1.64 - 0.7 s FASM v1.65.5 - 0.6 s So there's a only little difference. Perhaps it would be greater on slower machines, but I can't don't have any available it right now. In the evening I will try it on PIII 600. |
|||
|
05 Jan 2006, 10:55 |
|
|
RedGhost 05 Jan 2006, 15:08
yikes so many updates the last few days, hard to keep up
even though fasm is only a small bit faster, i love it that much more
i would have to say fasm currently has the best preprocessor/macroinstructions of any other language that i can think of (not only asm syntax) _________________ redghost.ca |
|||
|
05 Jan 2006, 15:08 |
|
|
vid 06 Jan 2006, 00:28
decard: (1/7)*100 percent speedup is only a little difference? Problem is that FASM is so fast already that we can't try it on existing machines :]
Wasn't here for some time, nice improvements, but please slow down with new directives... I think we already can achieve everything we need with what we have. |
|||
|
06 Jan 2006, 00:28 |
|
|
Tomasz Grysztar 06 Jan 2006, 08:13
The one I added was exactly because I was unable to achieve this particular thing - I was thinking about such directive since the time Revolution reported about problem with "x equ dqword" before the "struc POINT" definition. And The_Grey_Beast gave me the final stimulus to implement it.
|
|||
|
06 Jan 2006, 08:13 |
|
|
Madis731 06 Jan 2006, 12:36
14.28% improvement is indeed much - at least in this stadium. And when you think about these seconds (0.6s) then its miraculously fast. This is a large project and we can't even talk about seconds
 how odd is that? We're still in milliseconds
P.S. Its about 1.3seconds here with 2.6Gig Celeron |
|||
|
06 Jan 2006, 12:36 |
|
|
okasvi 12 Jan 2006, 03:14
there is 1.65.6 but seems like Tomasz forgot to mention about it here...
_________________ When We Ride On Our Enemies support reverse smileys |: |
|||
|
12 Jan 2006, 03:14 |
|
|
RedGhost 12 Jan 2006, 03:51
okasvi wrote: there is 1.65.6 but seems like Tomasz forgot to mention about it here... ah you were right it says 1.65.6 in the downloads, thanks for the update info! Quote:
its only jan 11th here, i am using the fasm of the future _________________ redghost.ca |
|||
|
12 Jan 2006, 03:51 |
|
| Goto page Previous 1, 2, 3, 4 Next < Last Thread | Next Thread > |
Forum Rules:
|