Disk Info is not Thread-Safe

[cetra3]

I've seen a few segmentation faults when running disk_info() on linux.

If you call disk_info() from multiple threads at once this can core dump:

#5  0x00005585db2e28cc in DFcleanup () at c/linux.c:167
#6  0x00005585db2e265f in get_disk_info () at c/linux.c:117
#7  0x00005585db2cee4a in sys_info::disk_info () at /home/cetra/.cargo/registry/src/github.com-1ecc6299db9ec823/sys-info-0.6.0/lib.rs:494

There is a global DFhashvector:

sys-info-rs/c/linux.c

Line 19 in a5d126d

static struct nlist *DFhashvector[DFHASHSIZE];

Which, if two threads are calling disk_info() the DFCleanup() call at the end of the get_disk_info() function will both try and free at the same time:

sys-info-rs/c/linux.c

Lines 159 to 172 in a5d126d

	void DFcleanup()
	{
	struct nlist *np, *next;
	int i;
	for (i=0; i<DFHASHSIZE; i++) {
	/* Non-standard for loop. Note the last clause happens at the end of the loop. */
	for (np = DFhashvector[i]; np; np=next) {
	next=np->next;
	free(np->name);
	free(np);
	}
	DFhashvector[i] = 0;
	}
	}

This results in segfaults & also double frees