Expression pedal freeze Pipedal midi input.

[faucon2001]

I have made a pedal board for Pipedal with an Arduino Uno and mocoLufa : 8 switches and 1 pot (expression pedal)
Pipedal runs on a RPi5 with bookworm, and the sound card is a Steinberg UR22-mk2

Every thing works perfectly until I heavily use the expression pedal controlling a Wha wha emulation.
If I use continuously the expression pedal after more or less 1 min the midi input will stop working, Pipedal won't freeze and continue to respond to the web UI, but midi communication is lost.

I have checked with a midi monitor that the Arduino midi messages were correct and not flooding. Messages are continuous without lag and they are sent only if a change of position occurs (no ghost value flooding).
I did the same with the Arduino hooked to the RPi, and dumped with amidi the midi input of the RPi : everything OK and no freeze.
After a midi freeze, stoping and restarting Pipedal service reset the midi input.

I believe that I am not the only one to use an expression pedal, so what could I have done wrong ?

[rerdavies]

Various thoughts....

• Are you getting audio xruns?

• Are the messages debounced? Is it possible that the device is sending huge bursts of alternating CC values when the controller is sitting right on the threshold of one value and the next? I don't think there's actually any code to recover from MIDI overruns; it just relies on the fact that it has generous buffers to work with. So this could be a problem. When I wrote it, I had old-fashioned MIDI cables in mind, which are SLOW. There are various 64k buffers involved, which might not be adequate if MIDI messages are being sent over faster transports.

• there's a lot of high-risk code around midi program changes. Is there anything else going on at or around the time that midi stops. Has a MIDI program change occurred at at all prior to the issue occurring?

• Could you take a look in system logs to see if any errors are being reported: Reboot, replicate the problem, and then run the following command at a shell prompt on the Pi 5.

    journalctl -b 0 | grep pipedal 

• You may also want to look more generally in system logs to see whether ALSA ('asound' in logfiles, I think), or pipewire are reporting issues

• I would be interested to know what happens if you disconnect and reconnect the MIDI device without restarting Pipedal. PiPedal is aware of midi devices going offline and coming online, and should open a new midi connection if the device has been selected as an input. If it recovers, that may help me narrow down where the issue might be happening. If Pipedal recovers, that that definitely narrows down the places in which problems could occur.

In the meantime, I'll look at the code to see whether I can see any obvious problems.

[faucon2001]

1 - No, there is no audio xruns, no drop nor crackle.

2 - I am using the library from Notes and Volts : https://www.notesandvolts.com/2016/03/arduino-midi-controller-potentiometers.html
There is no debounce parameter for pots, but I believe that t's set by default in the library. Examination of midi output dump while moving the expression pedal, shows a smooth progression and continuous values, without freeze nor drop. 150 CC messages per second were sent by the expression pedal over 1 min period.
https://ibb.co/xqHqL2gH

With Pipedal running, same test makes the midi freeze after 30-40 sec.

3 - I use PC at Arduino's boot to reset Pipedal to preset 1., and to navigate between presets. When using the expression pedal, there is no PC nor any other CC sent.

4 - 5 - logs right after boot and after a midi freeze
Pipedal after boot : https://pastebin.com/FNuhaVmd
Pipedal after midi freeze : https://pastebin.com/keQTUYR2
Alsa after boot : https://pastebin.com/Edjf8edY
Alsa after midi freeze : https://pastebin.com/XA7BNdb0

Everything seems normal, the only error message I have come form Alsa, but I don't know what it means :

(udev-worker)[403]: controlC0: Process '/usr/sbin/alsactl -E HOME=/run/alsa -E XDG_RUNTIME_DIR=/run/alsa/runtime restore 0' failed with exit code 99.
Nov 08 14:39:28 pi systemd[1]: alsa-state.service - Manage Sound Card State (restore and store) was skipped because of an unmet condition check (ConditionPathExists=/etc/alsa/state-daemon.conf).

6 - If I disconnect and reconnect the Arduino, it doesn't reactivate midi. The interface ( MocoLufa) is recognized but messages are not received. The only way to reactivate it, is to stop and restart Pipedal.

[onirob]

Hi, I think I may have reproduced a very similar issue.

In my case I first noticed it while testing a wah pedal plugin I coded, but after further testing it seems not to be specific to that plugin as it appears to happen with expression pedals in general.

The symptoms are very similar to what was described here: pipedal itself keeps running normally, audio continues, the web ui still responds, and alsa still appears to work if I test it from the command line (I run headless raspberry os lite as recommended), but pipedal's midi input stops reacting. Only solutions is restarting it, but it is shortlived since a couple of minutes of heavy expression pedal use can trigger the issue again.

So I tried looking at pipedal's code, and my understanding is that a midi event counter in the alsa driver code is not reset, so it can be flooded by continuous midi ccs. I made a small modification there, rebuilt pipedal and with that change, I tested it for about one hour while heavily using two expression pedals, and I could not reproduce the freeze anymore. CPU usage, latency, and xruns remained normal in my setup.

This is only tested on my hardware (RPi5, UMC204HD, FCB1010), so I would not claim it is a definitive fix yet, but I can open a pr with the patch if that would be useful, or provide more details/testing first if preferred.

[rerdavies]

Yes please! I have had trouble replicating it with the MIDI equipment I have at my disposal. So a fix would be marvelous. A Pull Request or diff. Either works (Pull request is ever so slightly better, as it automatically adds you as a contributor to the project). If making a pull request, please add your preferred copyright notice underneath mine at the top of the file, for each file that you modify. I need your changes to be MIT licensed (easier than doing a copyright assignment). If you add your name, I can (i think) reasonably assume that your contributions are appropriately licensed for use in Pipedal. The preferred form is without a year, as the year is no longer required in copyright notices. No need for comments in code on what changes were made, unless they particularly need explanation. Thanks, Robin Davies.

…

On Tue, Jun 16, 2026 at 12:19 PM Roberto Figliè ***@***.***> wrote: Hi, I think I may have reproduced a very similar issue. In my case I first noticed it while testing a wah pedal plugin I coded, but after further testing it seems not to be specific to that plugin as it appears to happen with expression pedals in general. The symptoms are very similar to what was described here: pipedal itself keeps running normally, audio continues, the web ui still responds, and alsa still appears to work if I test it from the command line (I run headless raspberry os lite as recommended), but pipedal's midi input stops reacting. Only solutions is restarting it, but it is shortlived since a couple of minutes of heavy expression pedal use can trigger the issue again. So I tried looking at pipedal's code, and my understanding is that a midi event counter in the alsa driver code is not reset, so it can be flooded by continuous midi ccs. I made a small modification there, rebuilt pipedal and with that change, I tested it for about one hour while heavily using two expression pedals, and I could not reproduce the freeze anymore. CPU usage, latency, and xruns remained normal in my setup. This is only tested on my hardware (RPi5, UMC204HD, FCB1010), so I would not claim it is a definitive fix yet, but I can open a pr with the patch if that would be useful, or provide more details/testing first if preferred. — Reply to this email directly, view it on GitHub <#420?email_source=notifications&email_token=ACXK2DAMNLYGGUGJ6PVVEOL5AFXPLA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGI3DCMJUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-17326114>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACXK2DHHCNDRHMNZHN5TH6D5AFXPLAVCNFSNUABHKJSXA33TNF2G64TZHMZTSNRTHE4DGNJTHNCGS43DOVZXG2LPNY5TSMJRGM4TEM5BOYBA> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/ACXK2DHBF4MBEVX6KGS4XET5AFXPLA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGI3DCMJUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSVGM33PORSXEX3JN5ZQ> and Android <https://github.com/notifications/mobile/android/ACXK2DD5HVM366KWTL5WECT5AFXPLA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZTGI3DCMJUUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>. Download it today! You are receiving this because you commented.Message ID: ***@***.***>

[onirob]

Yes, I’ll do it as soon as I can. I’ll prepare a pull request and add the copyright notice as requested. The change is very small, so it should be easy to review.

Also, I’d like to take this opportunity to thank you for this wonderful project!

Roberto Figliè