accessing the shell from PD

skrnk

It seems the commands do get executed, unless something gets printed to stdout
If anyone is having the same problem in the future; a hacky way to circumvent the issue is sending all output to /dev/null

giuliomoro

I find this sort of expected with large amounts of text being printed. Could you confirm what commands you use to trigger the issue? Also, can you try following the instructions here to upgrade to Pd 0.51 and see if the problem persists?

You should recompile your external after the update, as I am not 100% sure I preserved binary compatibility.

rph-r

Hi!
I have the same kind of issue using ggee/shell. The command used is date.
I cannot join the shell object itself I'm using here ("not allowed").

bug-shell.pd

942B

giuliomoro

@rph-r just checking first: can you not use bela_system, as suggested here https://forum.bela.io/d/4381-stream-to-icecast-with-pd/13 ?

rph-r

yes I remember about [send bela_system], but I need the return of the command date to get the current time...

giuliomoro

Do you need the current date or a monotonic time since the start of the script? The latter can be obtained with the [timer] object.

giuliomoro

You can also do this:

copy this zip archive to the board, it provides pdsend which allows to send messages to pd from outside Pd.
Then run

unzip pdutils.zip
mv pdsend pdreceive /usr/local/bin

Then create a file like this called get_time.sh in your project:

#!/bin/bash

# do magic here to get time into variable TIME
TIME=$(date +%Y.%m.%d-%T)
# send TIME to Pd's netreceive
echo $TIME | pdsend 9999 localhost udp 2>/dev/null

Then you can have a patch like the attached one (with a suitable interval in the metro):

This patch executes get_time.sh which in turn sends the current time to pd via pdsend.

rph-r

Nice workaround!
But since I need a bang on every second, should I write a shell script (or whatever) that sends directly the time on seconds to Pd?

giuliomoro

sure that's also easy enough: put something like this in a .sh file and use this procedure to run it in the background: https://learn.bela.io/using-bela/bela-techniques/running-a-program-as-a-service/

#!/bin/bash

while sleep 1; do
  # do magic here to get time into variable TIME
  TIME=$(date +%Y.%m.%d-%T)
  # send TIME to Pd's netreceive
  echo $TIME | pdsend 9999 localhost udp 2>/dev/null
done

rph-r

ok,
I finally wrote a pd object (my first one, be indulgent).
But let's say it immediately: it doesn't work very well, I get an error on each call (bang):
[n] mode switches detected on the audio thread. (n begin at 1 and is incremented on each error)
Here is the compiled object for Bela, and source code. It is based on time.h. Maybe you have an idea on how to improve. It works on linux.
https://www.dropbox.com/scl/fi/7sxt9fgy1ehd8ex4znzd5/ahora.pd_linux.zip?rlkey=jk5d8he4jh274rxwrxk2vpu9j&dl=0

The object [ahora n] works as follow:
outlets, floats, from left to right:
-year
-month
-day
-hour
-min
-sec
arg: number between 1 - 5. The number of active outlet active when inlet banged, from left to right (ex: 3 means hour,min,sec). None, 0 or >5 means all outlets actives.
inlet:
-bang: outputs time and date following creation arg.
-message year, month, day, hour, min, sec outputs corresponding element, regardless of creation arg

giuliomoro

Nice one, it probably works just fine, the issue is that time() will call into the Linux kernel and as it is called from within the audio thread it breaks the real-time guarantees of Bela, hence the "mode switch" warning.

You have several options:

start up a thread and call time() in there, have the audio thread print the result that was obtained from the other thread. If you don't want the other thread to be polling all the time, you'll need to trigger a time() call in the other thread when you receive a bang, then set a pd timer (I believe that's what it's called?) a few ms later that reads the up-to-date result of time when it's ready and sends it out of the output. This is the most generally-applicable version, which will also improve real-time friendliness of your external when running outside Bela. It may get complicated because on a multi-core processor you'll need some way of guaranteeing memory coherence so that the audio thread never sees a corrupted version of the time and has a way of knowing for sure whether it's been updated.
instead of calling time(), call __wrap_time(). This is a function provided by Xenomai's libcobalt that is safe to call from the audio thread. Before you call it you may have to declare it (depending on your compiler's options):
```
extern time_t __wrap_time(time_t *tloc);
```
I don't think you have to change your linker flags. You should be able to build it like this and at runtime it will be resolved because libcobalt is already loaded by the Bela binary.
leave the external exactly as it is. Bang it only once at the beginning of the script to get the time, ignore the mode switch message. Then use bangs to the [timer] object to measure elapsed time while the program is running. I appreciate this may have accuracy implications for long running patches (which you can obviate to by resetting the timer every so often) and if you need the year/month/day/hour/min/sec the previous option may be easier.

rph-r

giuliomoro I appreciate this may have accuracy implications for long running patches

and the running time will be... 5 years!!! ;-D (that is the time we have to guarantee the installation)

I'll check your propositions and try one... Thank you!!

rph-r

ok, I've just tried __wrap_time(). Doesn't work without extern time_t __wrap_time(time_t *tloc);. It works with it, but same errors. It seems localtime is guilty...

I will try the thread way later, but if that one works it would be nice too...

giuliomoro

Weird, it should work. Did you replace all instances of time()? Is there any other function call you are making that could be responsible?

giuliomoro

Try to remove localtime and see if the mode switches persist

rph-r

a summary:
Original code calls time() and localtime(), and generates errors
If I only call time() (with a comment on localtime()), I get the the errors as well, and obviously 0's on outputs.
If I call __wrap_time() without localtime(): NO errors!
But with __wrap_time() with localtime(), errors again.

It means both time() and localtime() are guilty. I couldn't find a way to wrap localtime the same way...

giuliomoro

rph-r But with __wrap_time() with localtime(), errors again.

Ok that's what I was suspecting . It looks like xenomai doesn't provide a wrapper foe localtime.

Try using localtime_r() instead and that may avoid the kernel call?

rph-r

ok yes, no more mode switch!
But I've just noticed time is not right: time() gives 1731277487 and __wrap_time() gives 7912 !?

giuliomoro

OK I verified that that's the case. It seems to have to do with the reference clock in use.
I verified that __wrap_clock_gettime(CLOCK_REALTIME, ...) or __wrap_clock_gettime(CLOCK_MONOTONIC, ...) give a similar result. However, there seems to be a CLOCK_HOST_REALTIME defined by Xenomai which seems to return the correct date. That define evaluates to 32, so you could do this:

extern int __wrap_clock_gettime(clockid_t clk_id, struct timespec *tp);
...

    struct timespec tp;
    __wrap_clock_gettime(32, &tp);
    secondes = tp.tv_sec;
   ...

this should give you the desired result.

BTW, there's plenty of duplicated code in your external that you may want to tidy up all in one place (namely, the calls to time() and localtime_r()). Also, it looks like when you call ahora_bang() the behaviour is to output values from all outputs starting from the one set in the creation argument. Is that expected/desired or are should you have a break at the end of each case statement?

rph-r

giuliomoro Also, it looks like when you call ahora_bang() the behaviour is to output values from all outputs starting from the one set in the creation argument. Is that expected/desired or are should you have a break at the end of each case statement?

The compiler too warned me about that, but it is the purpose: you choose to trigger the last n outputs. I confess it is a bit redundant since you can also trigger the desired outputs banging messages... But for who wants only the time, most of user a guess, it's convenient.

I'll check about duplicated code, and try CLOCK_HOST_REALTIME...