C / C++ / MFC

9-Nov-06 16:23

Does anyone know how to find out the bounds of esp in a running program in a System way. At the moment I use something like the following for the Upper Bound:-

///////////////////////////////

static DWORD TOP_ESP=0;

void ChckStack(void){
int CharArr[4092];
__asm mov TOP_ESP,esp;
ChckStack();
}

int GetStackTop(){
TOP_ESP=0;
try{
ChckStack();
}
catch(...){
return TOP_ESP & 0x00004000;
}
return 0; //Never Happens Throw a bone to Bill Gates
}

////////////////////////////////////////////////////

This will give me the TOS! It relies on pushing the stack until it trashes,
and assumes that it is on Page Boundaries. Any more'System' ideas, or Any ideas to find the base of the stack?

regards

LateNightsInNewry

Stephen Hewitt9-Nov-06 16:29

Stephen Hewitt

9-Nov-06 16:29

LateNightsInNewry wrote:
in a running program in a System way

What do you mean by this?

Steve

Bram van Kampen9-Nov-06 17:20

9-Nov-06 17:20

I Meant, maybe there is an API somehere that does this in a Microsoft Approved documented way, rather than the kludge I wrote to get the allowable range of ESP. It has all to do with a custom version of longjmp, which saves away a portion of the stack before it 'jumps'. Later on, after many function calls on the cut down stack, the portion of stack is retrieved is put back.If all goes well (an there are many reasons why it should not), a bit of gerrymandering of the various registers will cause the saved process to resume where it left off. The above is ofcourse a gross simplification, we have all sorts of things to consider, starting with global variables.

In order to trap the worst of errors, it would be handy to be able to do a sanity check on the proposed value for ESP before you take the jump back. What I wrote sofar works quite well, amazingly well actually, all things considered, but, it will end up as a minor(and thankfully well hidden) part of an API for third party users. In a situation like that I feel that belts, braces, and all rescue gear available would be recommended to be in place.

Hence my request

LateNightsInNewry

Stephen Hewitt9-Nov-06 17:53

Stephen Hewitt

9-Nov-06 17:53

See here[^].

Here's some code to get the limits of the stack:

#include <windows.h>
#include <iostream>
 
int main(int argc, char* argv[])
{
    DWORD top;
    DWORD bottom;
 
    __asm
    {
       mov EAX, FS:[4]
       mov top, EAX
       mov EAX, FS:[8]
       mov bottom, EAX
    }
 
    using namespace std;
    cout << hex << showbase << "top = " << top << ", bottom = " << bottom << endl;
 
    return 0;
}

Steve

Mark Salsbery10-Nov-06 5:14

10-Nov-06 5:14

Sorry for the thread highjack, but...

Cool! Thanks for the link Smile | :)

Bram van Kampen10-Nov-06 14:34

10-Nov-06 14:34

Thanks Steve,

Your answere points in the right direction. This however appears to give the TOS Committed Page Linear Address. What I would be looking for is the TOS Reserved Page Linear Address. When trying to rebuild a saved stack, After other processes played havoc, the committed stack could well have depleted. (we were not using that portion of the stack after all). The stack adjustment we are talking about here, has been known to span several hundreds of kByte.

Hence, in this case, the stack to be restored, should not be compared to the committed pages, but the the reserved pages.

regards

LateNightsInNewry

Subclassing Etc

Bram van Kampen9-Nov-06 15:44

Re: Registry and Shortcuts

9-Nov-06 15:44

I need to send a set of WM_USER+x Messages into a Dialog, which is sitting waiting in a Modal Loop on another thread. How do I thread in an SDK WindowProcedure with MFC in such a way that I get a copy of all parameters.

regards

LateNightsInNewry

to make video

selectgood9-Nov-06 15:21

selectgood

9-Nov-06 15:21

to ues some individual jpeg pictures to make video,no matter the form of the video(avi,divx,xvid,mpeg4,mpeg2.....) Cry | :((

thank you~

Re: to make video

Christian Graus9-Nov-06 16:12

Christian Graus

9-Nov-06 16:12

The answer is not going to change. IS google broken where you live ?

Registry and Shortcuts

locoone9-Nov-06 15:17

locoone

9-Nov-06 15:17

ok i have 2 questions.
does anyone know how to check and see if a shortcut is good or points to nothing? (i know the short answer of click it lol)
and is there a list of all shortcuts in the registry somewhere?

Mark Salsbery9-Nov-06 15:39

9-Nov-06 15:39

locoone wrote:
...is there a list of all shortcuts in the registry somewhere?

I believe they are stored in special folders - check out SHGetFolderPath() API

This article may be helpful: CShortcut (Shell links to special folders)[^]

string functions in C

thathvamsi9-Nov-06 14:43

thathvamsi

9-Nov-06 14:43

Hi all,
I need help with string functions in C.

Eg.,
const char *str1 = "Apple+Orange+Pineapple";

I need to crack the list above and get them as three different strings like
str2 = Apple
str3 = Orange
str4 = Pineapple

The "+" sign separates them.
How do I do it? Can anyone help? Thanks in advance.

Christian Graus9-Nov-06 14:52

Christian Graus

9-Nov-06 14:52

strtok will tokenise a string ( split it based on a token ).

Christian Graus - Microsoft MVP - C++
Metal Musings - Rex and my new metal blog

thathvamsi9-Nov-06 15:06

thathvamsi

9-Nov-06 15:06

Thanks for your help. Is there a simple code snippet that you can send ?(if you have one). Thanks.

http://www.codeproject.com/script/comments/forums.asp?msg=1749635&forumid=1647#xx1749635xx[^]

George L. Jackson9-Nov-06 15:07

George L. Jackson

9-Nov-06 15:07

thathvamsi9-Nov-06 22:08

thathvamsi

9-Nov-06 22:08

Thanks. was very helpful.

Bram van Kampen9-Nov-06 16:55

9-Nov-06 16:55

I assume you're writing in C (not cpp).
well, copy the lot to a buffer, and scan the string. Every time you find a '+' you replace it with a 0, and store the advanced pointer in the result atrray

e.g

int Breakup(char* str,char Sep,int nMax char* Result[]){
int i,j=0;
if(str==NULL)return -1;
int len =strlen(str);
char* buf=malloc(len);
if(buf==NULL)return -2;
Result[j++]=buf;
for(i=0;i<len;i++){
if(buf[i]!=sep)continue;
buf[i]=0;
result[j++]=buf[i+1];
if(j>=nMax)return nMax;
}
return j;
}

Don't be stuck, but thee's a lot to be said for CPP!

LateNightsInNewry

David Crow10-Nov-06 3:59

David Crow

10-Nov-06 3:59

LateNightsInNewry wrote:
copy the lot to a buffer, and scan the string. Every time you find a '+' you replace it with a 0, and store the advanced pointer in the result atrray

Why not just use strtok()?

"Approved Workmen Are Not Ashamed" - 2 Timothy 2:15

"Judge not by the eye but by the heart." - Native American Proverb

Mark Salsbery10-Nov-06 5:15

10-Nov-06 5:15

DavidCrow wrote:
Why not just use strtok()?

Or maybe use strtok()

David Crow10-Nov-06 5:26

David Crow

10-Nov-06 5:26

Mark Salsbery wrote:
Or maybe use strtok()

"Approved Workmen Are Not Ashamed" - 2 Timothy 2:15

"Judge not by the eye but by the heart." - Native American Proverb

Mark Salsbery10-Nov-06 6:25

10-Nov-06 6:25

I figure if it's suggested enough times (as Christian suggested in the very first reply) then
maybe the OP will consider it. My reply should have been to the OP not to you sorry man! Smile | :)

Bram van Kampen11-Nov-06 16:05

11-Nov-06 16:05

Admittedly, Yes.
However, strtok is one of those obscure library functions on which I would spend more time debugging the peculiarities and hidden side effects, than what it would take to write it from the base up, in the knowledge of what one wants to achieve.

Once you get the picture in your head of what you want to achieve, I find raw pointer arithmetic an awfull lot clearer, and easier to debug, than the various ANSI library functions.

LateNightsInNewry

David Crow12-Nov-06 15:37

David Crow

12-Nov-06 15:37

LateNightsInNewry wrote:
However, strtok is one of those obscure library functions on which I would spend more time debugging the peculiarities and hidden side effects, than what it would take to write it from the base up...

Isn't that what looking at the CRT source code is supposed to address?

"Approved Workmen Are Not Ashamed" - 2 Timothy 2:15

"Judge not by the eye but by the heart." - Native American Proverb

Bram van Kampen14-Nov-06 14:24