Sometimes tty is in weird state after mpg123 exits

[Thorsten Glaser]

Sometimes, when I exit mpg123 (by pressing ^C to send SIGINT to it, which is the normal way to do so…) the prompt is in such state as pressing Cur↑ shows ^[[A, which can be fixed by then pressing ^U (invisible) and Enter, so that the cleanup that happens on every Enter fixes it. (I could also blindly type another command or use it from history. It also works, but this is annoying.)

I have absolutely no idea where that comes from, or how to debug it. I suspect mpg123 lives for a bit longer after the ^C and messes up tty state after the shell tried to restore it. Or, there is a race condition in the shell making it not restore this properly after the signal (which, after all, the shell does get).

Help welcome!

[Thorsten Glaser (mirabilos)]

Seb dixit:

>I can't reproduce it, mksh R54 with mpg123 1.23.4 work well.

I have this one:

High Performance MPEG 1.0/2.0/2.5 Audio Player for Layer 1, 2 and 3.
Version mpg123-0.59r-MirPorts-8 (1999/Jun/15). Written and copyrights by Michael Hipp.

It doesn’t happen every time, only sometimes.

>Have you tried to switch the output module with the '-o' option?

     -o s, --speaker
          Direct audio output to the speaker.

     -o h, --headphones
          Direct audio output to the headphone connector.

     -o l, --lineout
          Direct audio output to the line-out connector.

Unlikely to change anything.

>The default is likely to be pulseaudio

Nothing of that name is ever coming even close to any system I have.

No, the audio output is sunaudio of course. Or, maybe, the OSS emulation
that us BSD people have.

But, again, this is a tty issue, most likely.

bye,
//mirabilos
--
[00:02] <Vutral> gecko: benutzt du emacs ?
[00:03] <gecko> nö [00:03] <gecko> nur n normalen mac
[00:04] <Vutral> argl [00:04] <Vutral> ne den editor
 -- Vutral und gecko2 in #deutsch (NB: Editor? Betriebssystem.)

[Thorsten Glaser (mirabilos)]

This also occasionally but not reproducibly occurs when mpg123 exits by itself.

[Thorsten Glaser (mirabilos)]

I have another reproducer.

tg@caas:/etc$ sudo git show e5a6dcb8bfeef9b30f4ef844ebddd422c8f4d296 | wc
     87 247 2112
tg@caas:/etc$ sudo git show e5a6dcb8bfeef9b30f4ef844ebddd422c8f4d296
^Z

This triggers this bug… in mksh *and* in GNU bash. So this is most likely not an mksh bug.

I’d still be curious as to the reason, and consequently, whether mksh could do something to avoid this.