compare_disks.c

#include <libgen.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#define CHUNK_SIZE (512*1024)
#define START_GB ((off_t)2928) // ~13 % of 22 TB

int main(int argc, char* argv[])
{
	if (argc != 3) {
		printf("%s: <disk1> <disk2>\n", basename(argv[0]));
		return EXIT_FAILURE;
	}

	FILE* f1 = fopen(argv[1], "r");
	if (f1 == NULL) {
		perror(argv[1]);
		return EXIT_FAILURE;
	}

	FILE* f2 = fopen(argv[2], "r");
	if (f2 == NULL) {
		perror(argv[2]);
		fclose(f1);
		return EXIT_FAILURE;
	}

	char* buf1 = malloc(CHUNK_SIZE);
	if (buf1 == NULL) {
		perror("buf1");
		fclose(f2);
		fclose(f1);
		return EXIT_FAILURE;
	}

	char* buf2 = malloc(CHUNK_SIZE);
	if (buf2 == NULL) {
		perror("buf2");
		free(buf1);
		fclose(f2);
		fclose(f1);
		return EXIT_FAILURE;
	}

	const char* zero = calloc(1, CHUNK_SIZE);
	if (zero == NULL) {
		perror("zero");
		free(buf2);
		free(buf1);
		fclose(f2);
		fclose(f1);
		return EXIT_FAILURE;
	}
	
	off_t offset = START_GB * 1024 * 1024 * 1024;
	fseeko(f1, offset, SEEK_SET);
	fseeko(f2, offset, SEEK_SET);

	while (fread(buf1, CHUNK_SIZE, 1, f1) == 1 &&
               fread(buf2, CHUNK_SIZE, 1, f2) == 1)
	{
		if (memcmp(buf1, zero, CHUNK_SIZE) != 0 && memcmp(buf1, buf2, CHUNK_SIZE) == 0) {
			offset = ftello(f1) - CHUNK_SIZE;
			printf("Found identical data at offset 0x%lX\n", offset);
		}
	}

	free(buf2);
	free(buf1);
	fclose(f2);
	fclose(f1);

	return EXIT_SUCCESS;
}