While true. My setup was the following: I was using the mainboard 2.5 with the the first SOM and was booting from SD card which came with the Reform. Transcend 32 GB. Without any problems. I installed the mainboard 3.0 and RK3588 and I could not read the SD card anymore. With an external SD card reader I was able to get to read the SD card without any problems.
So I’ve finally got to try different SD cards. Well, every card I tried worked exept Transend 32 GB which came with the Reform2 and I was able to read with an USB SD card reader. At this point I would say my MB 3.0 is not broken. But I still can’t understand why I was able to work from this SD card with the original MB and with the MB2.5 for so long and even boot from it and with MB3.0 I can’t.
So… me and @meluzzy exchanged our sd-cards. I mailed them mine and they mailed me theirs. Long story short:
- the SD-card which was broken for them works flawlessly in my RK3588 classic reform with mainboard 3.0
- the SD-card which worked for me is broken for them with the known output on dmesg
Here is the SD-card of @meluzzy in my system doing the gnome disk benchmark without problems:
Let me quote what Lucy wrote about this issue on IRC:
15:21 <@minute> josch: ok, many thanks for spending your time debugging that for us :0 hmmmmm it might be
an alignment issue of the rk3588 module
15:23 <@minute> josch: it just occured to me that meluzzy's issue could be a mechanical alignment issue
on the rcore in the slot. this usually impacts SD card, but we have seen that mostly in
pocket QA, but it would manifest in the same way on classic
15:24 <@minute> josch: some of the rcores were manufactured with a slightly "off" (like 0.1mm) outline
and can be moved horizontally a little bit in the slot. if it's too far off then some
pins are not connected
15:24 <@minute> josch: it's a PCB right? that goes in the slot. this PCB is cut out with a CNC machine
15:25 <@minute> josch: and this cut was slightly imprecise in one batch
Hmmm I see…
Should I try disassembling the RCore and checking if there is some way of making the pin make contact? If it is such a tiny offset maybe just disassembling it and assembling it back could fix it?
(PD: Btw, my name is also Lucy so it took me a millisecond to understand who was talking to who xD)
I would try that, yes. Since the contacts are tiny I don’t think there is much chance to see where a contact is bad but the process of removing the rcore and plugging it back in will place it in a slightly different spot and that might fix your issue. Note, that the rcore is the adapter board from the firefly icore module to your mainboard via an SODIMM connector. Taking that out and putting it back in is really easy. Removing the firefly from the rcore is a bit more tricky because it sits really tightly using a mezzanine connection (you can leave the firefly icore plugged into the rcore adapter board).
I tried but, no luck.
The offset seems bigger than I thought. I guess this is not supposed to look like this:
You might want to consider writing MNT support about this. Unless Lucie has another idea, it might be necessary to send your unit in for repair under warranty 
Yeah, I will do that. Seems like there is nothing else I can do on my end.
I hope I don’t have to send the entire laptop over. Maybe I can send the R3 board or the card the CPU board sits on? Support will tell.