2012/8/10 Eli Zaretskii <eliz@gnu.org>

> From: qq510371827 <qq510371827@gmail.com>
> Date: Fri, 10 Aug 2012 17:14:14 +0800

>
> > Try invoking GDB from Emacs like this:
> >
> > M-x gud-gdb RET
> >
> > If that doesn't help, either, then I'm clueless, sorry. Perhaps
> > someone else could help.
> >
> Thanks for your patience. It finally works. However, two other new
> problem appeared after using 'gud-gdb'.
> 1).the code, gdb information and out/input datas mixed together in the same
> buffer and 'gdb-many-windows' command didn't work any more.

This is expected. gud-gdb uses the old interface with GDB, which
doesn't support gdb-many-windows.

> 2) You can refer this link:
> http://stackoverflow.com/questions/9676135/gud-gdb-emacs-24-not-working
> The same problem as mine.

No, it isn't the same. Your problem was with getting input and output
to and from the debuggee. The above URL describes a much more serious
problem, whereby Emacs doesn't show the source code of functions you
step through. I don't see any of this in your original bug report
(and it surely works for me on MS-Windows).
I'm sorry.I got it wrong.
As for the original problem: could you please post here a minimal C
program that exhibits the problem with I/O when you use "M-x gdb"? I
will then try to look into the reasons of this and the possible ways
to solve that, when I have time.

Thanks.

#include "stdio.h"

#include "ctype.h"

#include "string.h"

void Reverse(char* p,int n)

{

if(--n > 0) Reverse(p+1,n);

printf("%c",*p);

fflush(stdout);

}

int main()

{

//setbuf(stdin,NULL);

//setbuf(stdout,NULL);

// setvbuf(stdin,NULL,_IONBF,0);

// setvbuf(stdout,NULL,_IONBF,0);

//freopen("input.txt","r",stdin);

char s[255],*p,*q;

int i=0,n;

fgets(s,255,stdin);

s[strlen(s)-1] = ' ';

p = q = s; // set breakpoint at this line.

while(*p != '\0')

{

i++;

p++;

if(isspace(*p))

{

n = i;

i = 0;

Reverse(q,n);

q = p+1;

}

n = i;

Reverse(q,n);

printf("\n");

return 0;

}

Thanks in advance.