seg fault when strtok on static memory c++

gearoid_murphy · 03-14-2007, 07:58 AM

hello,
I am observing seg faults when I try to do a strtok on static memory. The code is as follows :

#include <iostream>
#include <string>
#include <string.h>
#include <vector>
using namespace std ;

char *tree = "B C H I I I";

int main()
{
char *data_ptr ;
vector <string> expr ;
char buf [64] ;
//this causes a segmentation fault
data_ptr = strtok(tree, " ");
//this works fine
strcpy(buf, tree);
data_ptr = strtok(buf, " ");

return 0 ;
}

Its not a crtical requirement to solve this but its strange. I don't see why strtoking the static memory should be a problem. I'd appreciate it if someone could shed some light on the issue?

Thanks in advance.

wjevans_7d1@yahoo.co · 03-14-2007, 12:06 PM

The problem is not static storage, but constant strings.

From the man page for strtok(3) and strtok_r(3):

Code:

       Never use these functions. If you do, note that:

              These functions modify their first argument.

              These functions cannot be used on constant strings.

              The identity of the delimiting character is lost.

              The strtok() function uses a static buffer while
              parsing, so it's not thread safe. Use strtok_r() if
              this matters to you.

The reason these functions cannot be used on constant strings (the second item in the above list) is that they modify their first argument (the first item). In many implementations of C, constant strings are stored in a read-only memory segment.

Hope this helps.

gearoid_murphy · 03-15-2007, 07:43 AM

ok, right, I had thought that constant strings were meant in the compiler context. I did not know that such memory was stored in a read only segment (or that it was even possible to force that distinction onto your own data, although I suppose there is mprotect).

Thanks for your insight.

NetCrawler · 04-29-2007, 06:36 PM

For some reason unknown to me, if you declare your static string like this:

char tree[] = "B C H I I I";

It runs through strtok() fine with no segmentation fault. I guess this definition causes the string to be placed in read/write memory.

I would have thought that both definitions were the same thing, I guess I'd be wrong...

graemef · 04-29-2007, 09:19 PM

I think that char * tree = "foo" is a pointer to read only memory, thus the contents can be read but not changed. Whereas char[] tree = "bar" is an address in read only memory, whose contents the element bar[0] bar[1] and bar[2] are not located in the read only region.

gearoid_murphy · 04-30-2007, 11:04 AM

nope, i'd say that the whole segment is marked as read only, then the individual elements will also be read only. I'd suspect that the particular compiler implementation is responsible.