I have a Raspberry Pi 2B with Tumbleweed using an application which makes a
connection via bluetooth.
At a certain moment the reported quality of the connection, normally reported
as 80% is reported as 39%. The next time after 5 minutes the same application
tries to connect and reports a failure, the application hangs. It is invoked
with the command line:
/bin/timeout -k 30s -s SIGKILL 230s ./SBFspot <parameters>
The return code is 137, which means that the application is killed. However the
process is still visible but in a strange status.
The command "ps aux" shows it as:
sma      20441  0.0  0.0      0     0 ?        D    mei10   0:00 [SBFspot]
The command "lsof -p 20441" gives:
COMMAND   PID USER   FD      TYPE DEVICE SIZE/OFF NODE NAME
SBFspot 20441  sma  cwd   unknown                      /proc/20441/cwd
(readlink: No such file or directory)
SBFspot 20441  sma  rtd   unknown                      /proc/20441/root
(readlink: No such file or directory)
SBFspot 20441  sma  txt   unknown                      /proc/20441/exe
(readlink: No such file or directory)

I don't know what to make of this. To me it looks like the system is not
processing the kill properly. Is it a kernel issue?

Following invocations of this application hang on trying to make the bluetooth
connection, and are not killed by the timeout. They keep hanging.
Only a reboot cures the situation.

I have been using this application for several years without this problem and
was forced to use a newer version of Tumbleweed, because the used uSD broke
down. So previously the application reported the failure of making the
bluetooth connection, but ended itself normally.

In parallel I have a watchdog, each 15 minutes, which checks the bluetooth
device and reports the visible bluetooth devices, like the one the application
connects to. This does not report any malfunctioning. It also monitors the
failure messages of the application and in that case reloads the driver using
modprobe.