[media] ddbridge: fix ddb_ioctl()

There were a several problems in this function:

1) Potential integer overflow in the comparison:
	if (fio.write_len + fio.read_len > 1028) {

2) If the user gave bogus values for write_len and read_len then
   returning -EINVAL is more appropriate than returning -ENOMEM.

3) wbuf was set to the address of an array and could never be NULL
   so I removed the pointless NULL check.

4) The call to vfree(wbuf) was improper.  That array is part of a
   larger struct and isn't allocated by itself.

5) flashio() can't actually fail, but we may as well add error
   handling in case this changes later.

6) In the default case where an ioctl is not implemented then
   returning -ENOTTY is more appropriate than returning -EFAULT.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>