Florian Wesch

I give up for today. After staring at the code for the last two hours I tried replacing `idr_alloc` with `idr_alloc_cyclic`, now `vc_sm_cma_vchi_send_msg` starts returning EINTR or ENXIO after 1.2...

> I have just spotted that there is a slightly surprising loop in vchiq_queue_kernel_message. If vchiq_queue_message returns -EAGAIN, then it does an msleep(1) and tries again. I guess that's the...

So this might not even be a Linux-side bug but something in whatever is handling the mailbox replies on the other end? I would assume that the part responsible for...

Those resource leaks on program exit are another issue, unrelated to the NULL-deref, correct? I've since looked into how to avoid this issue and figured I might use a custom...

Potentially related: Happened on a Raspberry Pi 3 Model B Rev 1.2 (a02082) running the latest 6.12.21-v8+ kernel: ``` user.notice info-beamer: CRITICAL ERROR: stuck swapping surface kern.err kernel: [291076.176771] vc_sm_cma_vchi_rx_ack:...

Not the requested solution, but the boot loader config supports comments afaik, so you could embed it as a base64 blob in there like this: ``` # f00base64doobdedoob ```

> on Pi5 the space is already reserved for other features such as A/B booting. Interesting. Does that mean that the data in the EEPROM itself will have an A/B...

Here's the corresponding [forum thread](https://forums.raspberrypi.com/viewtopic.php?t=390657) about this data loss incident and here's my now [closed issue](https://github.com/raspberrypi/rpi-imager/issues/1087) predicting this would happen eventually :(

Possibly related: https://github.com/raspberrypi/linux/pull/6703/commits/bd5f8b97b2caa1ec35d3cabb5d618ee5dd8c4e22 (part of https://github.com/raspberrypi/linux/issues/6701#issuecomment-2701809352)

> Can you show the issue with a single display and two video planes? Not sure. I've not observed it with just a single display so far, but that might...