[gpsd-dev] gpsd does not recover or exit when the single device becomes unavailable

[Nuno Gonçalves]

Either when using SER or TCP, a disconnect of the device makes gpsd
hang forever.

gpsd:RAW: /dev/ttyUSB2 is known to be Trimble TSIP
gpsd:SPIN: select() {3 4 5} -> { 5 } at 1466436986.402173 (errno 0)
gpsd:UNK: polling 5
gpsd:UNK: Read 0 chars to buffer offset 0 (total 0): \x10]T\x0b
gpsd:SPIN: packet_get() fd 5 -> 0 (0)
gpsd:RAW: /dev/ttyUSB2 is known to be Trimble TSIP
gpsd:DATA: /dev/ttyUSB2 returned zero bytes
gpsd:DATA: /dev/ttyUSB2 will be repolled in 0.010000 seconds

or

gpsd:RAW: tcp://127.0.0.1:10686 is known to be Trimble TSIP
gpsd:SPIN: select() {3 4 5} -> { 5 } at 1466436959.642325 (errno 0)
gpsd:UNK: polling 5
gpsd:UNK: Read 0 chars to buffer offset 0 (total 0): \x10m\x9d
gpsd:SPIN: packet_get() fd 5 -> 0 (0)
gpsd:RAW: tcp://127.0.0.1:10686 is known to be Trimble TSIP
gpsd:DATA: tcp://127.0.0.1:10686 returned zero bytes
gpsd:DATA: tcp://127.0.0.1:10686 will be repolled in 0.010000 seconds

I don't see it is to easy to code this because it needs to be some way
to track how many devices we still have available. Anyway I think it
is important to EXIT when this happens to avoid mysterious hangs,
specially with TCP where it is more probable for it to happen.

Thanks,
Nuno