Big rtl_fm (and more) improvements, testing requested

[Deron]

Hello Kyle,

I had a chance to place with it this weekend, and again last night, and 
the only additional problems I had:

- After 20-30 forced quits, the device will go completely bonkers. Ok, 
not very technical and hopefully related to the inability to stop it in 
general.

- I have no idea how to make scanning work. Is there something I am 
missing to force it to continue scanning at the next frequency? Also, 
when it finds something that breaks squelch, could it dump to shell the 
frequency using stderr?

- I'm using an R820T based device. As I understand it, the offset tuning 
is unnecessary. Looking at the code, "-E offset" should do just what I 
want. It disables the offset/90deg shift in rtl_fm, and the library 
recognizes that is unnecessary and returns -2 or some such. The (very 
minor) bug is that it reports it as an error that it is unable to set 
the offset.

So what can I do to help find/fix the quit problem?

Thanks!

Deron


Some very basic observations, hope they are useful. Some may be a result 
of my (lack of) understanding.

- Control-c will not stop it. It catches the signal, and reports "Signal 
caught, exiting!" but does not. I had to kill -9 to stop it...

- When first run, I get 30 logged errors of "xhci_hcd 0000:00:14.0: 
ERROR no room on ep ring".

- fm was much better than the old version, but -fmwb was the same. I 
narrowed it down to -o 4. That alone destroyed all improvements. By using:

     ./rtl_fm -f 95.1M -M fm -s 170k -A fast -r 32k -l 0 -E deemp - | 
aplay -r 32k -f S16_LE

It was much better. I'll play around more this weekend when I have more 
time.

Overall, very good. Night and day compared to before. Also, cpu usage 
seemed less. Top reports about 3% vs 6% from before.

Deron