fat, read_clusters: Don't allow memory overflow

[perlun]

This code is a bit too naïve:

chaos/servers/file_system/fat/clusters.c

Lines 75 to 100 in 97415f4

	uint32_t read_clusters(fat_info_type *fat_info, void *output, uint32_t start_cluster, uint32_t skip, uint32_t number_of_clusters)
	{
	uint32_t cluster_number = start_cluster;
	uint32_t clusters_read = 0;
	do
	{
	if (skip > 0)
	{
	skip--;
	}
	else
	{
	// log_print_formatted (&log_structure, LOG_URGENCY_DEBUG,
	// "Reading cluster number %lu", cluster_number);
	read_single_cluster(fat_info, cluster_number, (void *)
	((uint32_t) output + (clusters_read *
	fat_info->bytes_per_sector *
	fat_info->sectors_per_cluster)));
	clusters_read++;
	}
	cluster_number = get_next_cluster(cluster_number, fat_info->fat, fat_info->bits);
	} while (cluster_number != MAX_uint32_t && clusters_read < number_of_clusters);
	return cluster_number;
	}

It takes in a buffer (the output variable), but no length of the buffer. So the method has no way to prevent itself from overflowing the buffer if things go wrong...

Also, it has no way to signal to the caller that it did overflow. We would need to fix both these things. Right now, I am getting page faults because of this:

Something is messing with the cluster linking, so it reads the same cluster over and over again, leading to the buffer being overflowed.