[BRLTTY] Nos braille ouput through Orca since 6.2

[Didier Spaier]

Issue solved (somehow).

In our main init script, after having started brltty, the was this line:
rm -rf /var/run/*

Just moving it up (before starting brltty) did the trick.
It looks a little weird though that the issue occurred even with:
-W /run/brltty
in the command line or:
writable-directory=/run/brltty
in brltty.conf

An aside I am not convinced that the benefit of this least privilege thing
is worth the additional complication. Maybe that's related to the fact that
I never could read in full "man capabilities" before falling asleep :-)

Oh and the brltty process started early (by root) still can't be killed with
a SIGTERM, but that's a minor inconvenience, I think. Anyway I have 
added -9
to the kill command of our daemon manager in the stop function.

Have a good week-end everybody.
Didier

Le 06/03/2021 à 18:56, Didier Spaier a écrit :
> Le 05/03/2021 à 02:23, Dave Mielke a écrit :
>  > [quoted lines by Didier Spaier on 2021/03/05 at 01:03 +0100]
>  >
>  >>> Thanks to Tony, here goes: https://slint.fr/brltty/brltty.log
>  >>
>  >> Do you need more information to investigate?
>  >
>  > As I think I said, at the time, that log showed brltty terminating 
> properly
>  > when receiving SIGTERM. What I need is a log that sows brltty failing to
>  > terminate on SIGTERM.
> 
> No, I  don't remember you said that. I have uploaded a file twice in the
> same location (with the same name, hence maybe the misunderstanding),
> did you look there after the second time? If I remember correctly Tony
> killed it with a SGKILL not a SIGTERM. Tonny, do you confirm?
> 
> Anyway Philippe (in CC) reproduced the issue, but this time brltty not
> starting by a script, but only by udev.
> 
> In the use case alsobrltty works in the console but not through Orca.
> 
> Additionally, from a tty:
> 1. Killing brltty with SIGTERM leads to no input on the device, but the
> brltty process remains.
> 2. Starting brltty again starts a second process, but the device still
> gets no input
> 3. Then 'killal -9 brltty' works and starting brltty the device gets
> the expected input.
> 
> I can still request a debug log in this context, and mount / read-write
> from start just editing a line in grub.cfg, but then how should I modify
> the udev rules to include the debug options? Should I modify udev.rules
> or udev-wrapper and how?
> 
> Best regards
> Didier