Rpi lockup during file access

[Braincell1973]

The RPI on the mystery shield 20k with the mystery nano loaded on the tang 20k locks up solid when accessing some directories.

It's like it can't see above a certain number of files.

The RPI locks up solid with the keyboard and mouse becoming unresponsive, resetting the tang with the S1 button works but the pad remains as it was, frozen.

Using the latest release, and I'm unable to get any output from the rpu using putty over the usb (would only connect to com1 at 115200, blank screen)

Lee

[Braincell1973]

Here's a video showing the problem with the mysterynano core

https://youtu.be/aEqpf66Kx8A?si=2vIM4PCS-SnTy_fJ

[Braincell1973]

Here's a video showing the issue on the C64 core

(49 files causes fatal lockup !!)

https://youtu.be/aEqpf66Kx8A?si=F-yM05vA9ncAIV_J

[harbaum]

We are aware of this.

If you want to help, you can investigate the debug output via the debug header using a USB uart.

[Braincell1973]

It doesn't output via the usb port ?

How does this work with the Pico (I'm not a programmer) on the mist it was just a usb to ttl type device, does it need something special to direct the output to putty ?

[harbaum]

See

https://github.com/MiSTle-Dev/FPGA-Companion/blob/main/DEBUGGING.md

It's not using the USB port. It should actually be possible to use that as well whenever the USB port isn't needed otherwise.

Not being a programmer is something one can change and these hobby projects are often the perfect targets for first experiments. None of us was born as a programmer.

[Braincell1973]

The debugging link you supplied above doesn't help me with how I need to connect to the Pico, you said above the debug output but that apparently needs a 'debug probe' unless it's outputting serial directly on a pin ? And if it is a pin it's not clear which pin to use ?

[harbaum]

It does indeed output serial directly on a pin:

https://github.com/MiSTle-Dev/FPGA-Companion/tree/main/src/rp2040#pin-usage

These pins are available in pin headers on the board you built.

You can use a pico debug probe or any other USB uart adapter that supports 921600 bit/s.

[Braincell1973]

Hopefully this should help someone, poxy pinout on the shield was NOT as labelled so had to bell it out to get debug.....

Anyway , looks like a memory leak...... But it's beyond my skills to fix

[terracide303]

The problem is that it's running out of ram, I made experimental code that fixes it using pages. I can share it but I'm not very familiar with github so maybe fork it first?
https://youtu.be/0I-yO5Qa7eQ

[harbaum]

Welcome to the world of "machines without insane amounts of memory". These microcontrollers have kilobytes of ram. Not megabytes or even gigabytes.

Running out of memory doesn't necessarily mean there's a leak. It can simply mean that what you are trying to do exceeds the capabilities of the device being used.

I'll look into this. The paging approach indeed isn't very fast, but it may be the only option to circumvent this problem. I may also add some small "busy" animation to the OSD, so the user sees that something is happening.

[Braincell1973]

Understood , I did some work on the mist 10years ago but never learnt c++ , I tried to look at the code and could only understand some of it, but with only 264k the poor Pico can't hold a directory full of characters I suspect. It looks like it tries to read the whole dir then sorts it but I will stick to my day job of PLCs and ladder code :)

#terra Are you in the UK ? I think your the same person that commented on the video about a PCB, If your in the UK I can send you one , my email should be attached to the captain commodore YT channel

Lee

[terracide303]

Understood , I did some work on the mist 10years ago but never learnt c++ , I tried to look at the code and could only understand some of it, but with only 264k the poor Pico can't hold a directory full of characters I suspect. It looks like it tries to read the whole dir then sorts it but I will stick to my day job of PLCs and ladder code :)

#terra Are you in the UK ? I think your the same person that commented on the video about a PCB, If your in the UK I can send you one , my email should be attached to the captain commodore YT channel

Lee

Hey lee yeah I commented on it but i'm in the Netherlands. I guess i can just solder one on a protoboard for now it does not matter much I have all the materials here... but a pcb looks so nice!

[harbaum]

Hmm ... I have just put 300 files into one directory and can easily browse it fast and smooth. So it's not just the number of files. Maybe it's related to long file names or unsorted directories ...

[Braincell1973]

Yes I suspect it's the number of bytes in the directory and filenames additively Sent from Outlook for Android<https://aka.ms/AAb9ysg>

…

________________________________ From: Till Harbaum ***@***.***> Sent: Thursday, January 15, 2026 10:29:31 AM To: MiSTle-Dev/FPGA-Companion ***@***.***> Cc: Braincell1973 ***@***.***>; Author ***@***.***> Subject: Re: [MiSTle-Dev/FPGA-Companion] Rpi lockup during file access (Issue #117) [https://avatars.githubusercontent.com/u/9105084?s=20&v=4]harbaum left a comment (MiSTle-Dev/FPGA-Companion#117)<#117 (comment)> Hmm ... I have just put 300 files into one directory and can easily browse it fast and smooth. So it's not just the number of files. Maybe it's related to long file names or unsorted directories ... — Reply to this email directly, view it on GitHub<#117 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AG62KOWSP3IAW5LJAEH3TZL4G5TYXAVCNFSM6AAAAACPPJQ4QCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTONJUGA3DCMRSGU>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[harbaum]

Ok. It sounded like you already found out that it's the number of files that causes the problem. I wasn't aware that this was just a guess and that the root cause itself still needs to be investigated.