where and how do you set the PORT character pointer?
Ed.
Glen wrote:
>
> Dear list,
>
> I have been having trouble with something very odd. Here is my global struct:
>
> struct{
> int BAUD;
> int PARITY;
> int DATA;
> int STOP;
> int FLOW;
> int FLAGS;
> char *PORT;
> }comm_settings;
>
> When I call a certain function, the char * variable in the struct gets
> killed. Observe:
>
> <snip>
>
> fprintf(stdout,"PORT = %s\n", comm_settings.PORT);
> fprintf(stdout,"FLOW = %d\n", comm_settings.FLOW);
>
> send_error = send_data(buffer);
>
> <snip>
>
> comm_settings is a global struct with PORT being a char *. The call to
> fprintf() above displays the correct value. In the call to send_data(), we
> have this:
>
> <snip>
>
> if((port_fd = open(comm_settings.PORT, comm_settings.FLAGS)) == -1){
> fprintf(stderr, "error: port %s could not be opened: ", comm_settings.PORT);
> perror("");
> exit(1);
> }
>
> <snip>
> At the beginning ot the send_data() function, the value of comm_settings.PORT
> is "/dev/ttyS1". After entry into send_data(), it gets changed to
> "*\021@v/ttyS1". This is weird, but gets better. After the call to open()
> above, it changes to "?\025@Luy?" (I can't type some of these chars, but this
> is close). Sometime before/during/after the call to fprintf() it, it changes
> to an empty string and fprintf() prints nothing. The rest of the code not
> shown is very straightforward.
>
> I have _never_ seen or heard of this kind of 'struct smashing'
> before. Anyone got any ideas?
>
> Glen
>
> P.S. Yes, this will most likely be POSIX termios compliant and LGPL'd, if
> anyone was wondering.
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 19:22:56 EDT