Index
> Main > Stack balanced multicat string |
| Author |
|
|
hopcode 22 Jan 2009, 06:11
See please,also the updated version in my following post requiring no macro
Hallo All, starting from this interesting thread,i have got this idea for a multicat string routine to concatenate multiple strings passed on the stack.It could be improved a lot. I think that, starting from this here,a further development could replace completely the wsprintf API function. Here the code: Code: ;;ü------------------------------------------ö ;;| catstr code for macro @catstr | ;;| by hopcode[mrk] | ;;#------------------------------------------ä catstr: ; ESI = source; EDI = dest; xor ebx,ebx pop edx pop ecx pop esi @@: inc ebx lodsb test al,al jnz @b mov edi,esi dec ebx .@1: pop esi dec edi @@: inc ebx lodsb stosb test al,al jnz @b dec ebx dec ecx jnz .@1 jmp edx ; ;...the macro to preserve ebx/esi/edi ; macro @catstr destbuf,[args] { common local numparams numparams = 0 push ebx push edi push esi reverse push args numparams = numparams +1 common push destbuf push numparams call catstr mov eax,destbuf mov ecx,ebx pop esi pop edi pop ebx } ;------usage ;------ ;.data szDestbuffer db 100 dup (0) szDir db "mydir",0 szSubdir db "mysubdir",0 szSlash db "\",0 szFile db "myfile",0 ;.code @catstr szDestbuffer,szDir,szSlash,szSubdir,szSlash,szFile ; ; return EAX=pointer to destbuffer ; return ECX=num chars copied in destbuffer ;---------------------------------------------------- Enjoy, hopcode Last edited by hopcode on 23 Jan 2009, 16:50; edited 2 times in total |
|||
22 Jan 2009, 06:11 |
|
|
baldr 22 Jan 2009, 13:06
bitRAKE,
Inner loop copies NUL terminator for concatenated ASCIIZ inside destination.  |
|||
|
22 Jan 2009, 13:06 |
|
|
bitRAKE 22 Jan 2009, 16:08
Should prolly add a DEC EDI after the last POP ESI - that'll fix the string length as well (was off by one, counted NUL).
|
|||
|
22 Jan 2009, 16:08 |
|
|
hopcode 23 Jan 2009, 16:36
Here my working solution using the bitRake's "PUSH 0" hint (thanks).
This solution (anyway not my final) preserves API regs EBX/EDI/ESI but doesnt require a wrapping macro.Also,it could be cut and copied in a DLL without extra coding. Added a little (not exaustive) control on the parameters before PUSH 0 Max num stackable strings 0FFh Follows code and usage. BTW:/to Anyone/Do you know weather this "double stacking" technik as a real used name, and/or old and well known related examples ? Code: ;;ü----------------------------------------------ö ;;| catstr code to concatenate multi strings | ;;| by hopcode[mrk] version 0.1 | ;;| Datum 23/01/2009 | ;;#----------------------------------------------ä catstr: push ebx push edi push esi call .2 pop esi pop edi pop ebx pop edx add esp,eax jmp edx .2: add esp,0CH+4+4 ; point to params \ xor ecx,ecx ; \ = push ebx/edi/esi + \ pop edi ; \this proc + callling proc xor eax,eax push edi inc ah dec ecx repne scasb pop ecx neg ecx jmp .1 .0: lodsb stosb test al,al jnz .0 .1: inc ah pop esi dec edi test esi,esi jnz .0 shr eax,6 add ecx,edi sub esp,eax ;mov eax,[esp] ;dont touch!!: pointer to destBuffer sub esp,0CH+4+4 ret ;--------usage: ;-------------------------------------------- szDestbuffer db 100 dup (0) szDir db "mydir",0 szSubdir db "mysubdir",0 szSlash db "\",0 szFile db "myfile",ß ;.code stdcall catstr, szDestbuffer,szDir,szSlash,szSubdir,szSlash,szFile,0 ; ; return ECX=num chars copied in destbuffer ;---------------------------------------------------- Enjoy, hopcode |
|||
|
23 Jan 2009, 16:36 |
|
|
edfed 24 Jan 2009, 02:35
using a liked list would be good too.
Code: reserve many bytes to create the new string byte from (n)string(i) byte = nul? { (n) = last? {byte = nul} byte to dest(j) } byte to dest(j) j++ i++ etc etc... |
|||
|
24 Jan 2009, 02:35 |
|
|
bitRAKE 24 Jan 2009, 05:16
hopcode, it should be okay in Windows, but common convention specifies memory less than ESP as volatile (undefined) - should be able to "push x1 x2 ... xN/add esp,4*N" between any two instructions without effecting stack usage. Assembly coders know otherwise though.
In Windows ESP can be used generally if a stack isn't needed by a thread.
Additionally, on Windows the string address cannot be less than $1000. So, pushing zero first is only one option - could push the argument count, or stack correction value - because the caller knows. Code: pop esi cmp esi,$1000 jnc .0 add esp,esi retn  |
|||
|
24 Jan 2009, 05:16 |
|
|
revolution 24 Jan 2009, 05:23
When using functions with a variable number of parameters it is common practice for the caller to restore the stack: ccall. I think it complicates things to have the function restore a variable number of parameters.
bitRAKE wrote:
|
|||
|
24 Jan 2009, 05:23 |
|
|
bitRAKE 24 Jan 2009, 05:28
Oh, yeah: jmp edx, then.
|
|||
|
24 Jan 2009, 05:28 |
|
|
hopcode 24 Jan 2009, 06:56
/to edfed/ Hallo edfed,i like your idea. Thanks
/to bitRake/ As i have time i will try exaustively the proc in stacks+threads.Till now in threads, it seems all ok. . (Only to the reader's attention...it is important...) . Do you know weather this "double stacking" technik as a real used name, and/or old and well known related examples ? . Do you like the matrjoska ? Here... Prinzip der Matrjoschkas
|
||||||||||
|
24 Jan 2009, 06:56 |
|
|||||||||
|
baldr 24 Jan 2009, 21:26
revolution wrote: When using functions with a variable number of parameters it is common practice for the caller to restore the stack: ccall… _________________ "Don't belong. Never join. Think for yourself. Peace." – Victor Stone. |
|||
|
24 Jan 2009, 21:26 |
|
|
revolution 25 Jan 2009, 04:36
baldr wrote:
Perhaps we should ban all macros because they are too much like going "high-level"? |
|||
|
25 Jan 2009, 04:36 |
|
< Last Thread | Next Thread > |
Forum Rules:
|