fscanf only gets first 4 bytes of a hex number

Question

I'm writing a cache simulator in C and have it all pretty much done...except when I try and scan in addresses fscanf is skipping some of the digits in the hex number: it will only get 4 bytes! If I can't get the right address, the tag bits are incorrect and the simulation won't always work. The task seems pretty straight forward, but I must be missing something. Maybe something to do with fscanf format string idiosyncrasies?

The source file looks like this:

 S 00600aa0,1
I  004005b6,5
 S 7ff000398,8
 M 7ff000390,8
// and so on ...

I have tried using fgets and sscanf instead, but I get the same result.

char buffer[200];
char *pattern = " %c %x,%s\n";
int status; long address; char op; 
while ((status = fscanf(source, pattern, &op, &address, buffer)) != EOF) { 
    if (op != 'I')  { 
        fprintf(stderr,"address: %x\n",address); // DEBUG stmnt 
        simulate the cache..........................

The debug statement prints out the wrong address for lines 3. Instead of "address: 7ff000398" it writes "address: ff000398". It gets it right for line 1. Why does it only read in the first 4 bytes? 'address' is already a long and I can't find any documentation about %x behaving like this.

Answer 1

The address vairable's type is long which can store only 4 bytes and 0x7ff000398 is 8 bytes. so its store only the 4 last significant bytes and ignore the most significant. That's the reason line 1 and 2 works as expected but line 3 doesnt.

to fix it you can change the type of address to long long

Answer 2

You will need a 64 bits integer to store the result.

32 bits systems

On 32 bits system, long type may be only 32 bits long.

To be sure, you can use long long type for address.

Better, would be to use uint64_t type (defined in stdint.h on c99)

Changes

1. Change address type to long long (if needed)
2. Change %x to %lx (for long) or %llx(for long long) (two places)
3. Compile with warnings turned on.

Warnings

With warnings turned on, you should get the message:

warning: format ‘%x’ expects argument of type ‘unsigned int’, 
but argument 3 has type ‘long int’ [-Wformat=]
fprintf(stderr,"address: %x\n",address); // DEBUG stmnt
                         ~^    ~~~~~~~
                         %lx

Which gives you a good clue to solve your problem.

So, your code should look like:

#include <stdio.h>
int main(void)
{
    char buffer[10];
    char *pattern = " %c %lx,%s\n";
    int status; 
    unsigned long address; char op; 
    while ((status = scanf(pattern, &op, &address, buffer)) != EOF) { 
        fprintf(stderr,"address: %lx\n",address); 
    }
    return 0;
}

With your input, I get the result:

address: 600aa0
address: 4005b6
address: 7ff000398
address: 7ff000390

Answer 3

Since C99 you can use uintptr_t, an unsigned integer type that is capable of storing a pointer (if those addresses are pointers on the same target machine).

#include <stdio.h>
#include <stdint.h> 
#include <inttypes.h>

int main(void)
{
    char *str = "7ff000398";
    uintptr_t address;

    sscanf(str, "%" SCNxPTR, &address); // x for base 16
    printf("%" PRIxPTR "\n", address); // x for base 16
    return 0;
}