I strongly suspected that it had something to do with null termination of the string. But I kept on thinking it was the job of the compiler to terminate strings with a null char.

Anyways the problem is solved by changing buffer[2] to buffer[3] & assigning last byte 0.

But while debugging, when the "sick" string "11:22:33" formed with repeatedly concatenating 2 bytes of buffer[] was passed to LCD_sendstring(), the LCD_sendstring() actually exited after passing the last '3' to lcdchar():

 void LCD_sendstring(char * message)  //good 'ol LCD_sendstring
{
    while (*message)      // Write message to LCD
    lcdchar(*message++);
}
 

and that was what made my day suck. And also doesn't strcat() append a null char to destination string? Page 313 of Cx51 Compiler User’s Guide says so!
"Description: The strcat function concatenates or appends src to dest
and terminates dest with a null character."
Same is the case with strcpy(). I initialized str_buffer with all 'A's & then after copying & appending the "sick" string, str_buffer still contained a null char. Refer screenshot [108 KB]:
http://i782.photobucket.com/albums/yy108/munish4u911ex/keil.png
So as soon as LCD_sendstring() should find a null char in str_buffer it should stop printing & exit. Isn't it?

Yes I know, but I need both data types & conversions in my much bigger code, so didnt feel like writing a dedicated third one just for this.

And I also knew that it was problem with the code only. I never blamed the compiler or 8051 ;-)