I was hacking on a small python script, minding my own business when emacs stopped reacting to my keyboard input. I could still switch windows and work with another applications, but emacs hang up on me.
Having spent a couple of hours stracing, grepping and awking emacs start-up syscalls, I was in the mood to debug some more.
Attaching my shell to the pid of emacs (I usually only run 1 emacs instance, hence pgrep emacs).
sudo strace -p $(pgrep emacs)
Returns the same error messages repeated.
pselect6(14, [7 13], [], NULL, {0, 19999519}, {NULL, 8}) = 0 (Timeout)
poll([{fd=7, events=POLLIN}], 1, 0) = 0 (Timeout)
write(12, "{\"id\":91,\"method\":\"get_calltip\","..., 616) = -1 EAGAIN (Resource temporarily unavailable)
pselect6(14, [7 13], [], NULL, {0, 19999519}, {NULL, 8}) = 0 (Timeout)
poll([{fd=7, events=POLLIN}], 1, 0) = 0 (Timeout)
write(12, "{\"id\":91,\"method\":\"get_calltip\","..., 616) = -1 EAGAIN (Resource temporarily unavailable)
Ad nauseum.
Using the combination of functional programming and bash magic, I nested one function call inside another to get the list of processes. pgrep -P returns the PIDs of all child processes of a given PID.
$$$$ ps -o pid,pcpu,comm -p $(pgrep -P $(pgrep emacs))
PID %CPU COMMAND
31513 97.4 python
There is only 1 child process, but it’s eating up all the CPU time and killing my battery - it drained 13% in 10 minutes that I was googling and stracing.
$$$$ kill $(pgrep -P $(pgrep emacs)
This made emacs responsive again, so I examined the emacs Messages buffer
Wrote /home/petr_tik/Coding/misc/misc/dfs.py
error in process sentinel: elpy-rpc--default-error-callback: peculiar error: "terminated"
error in process sentinel: peculiar error: "terminated"
eldoc error: (file-error Writing to process bad file descriptor *elpy-rpc [project:~/Coding/misc/misc/ python:/usr/bin/python]*)
Here is a relevant GitHub issue
Looks like it remains to be solved.