A Pi and a broken card sensor

Hi everyone,

I’m sure that some of you have run into the infamous “card sensor” issue with the Pi, where it’ll boot the ROM from the SD card but will then complain that the drive is empty. I figured that the easiest way around this would be to leave the ROM on the SD card but move FileCore onto a USB drive.

Thanks to these instructions over on the Pi forums, I mostly have it working. After removing and reinserting the SD card countless times I managed to get to a working desktop. I then ran HForm on the USB drive, copied everything across, and performed the two Configure steps. A test reboot was successful, and all was well.

Almost.

I removed the SD card and reinserted it, and powered up again. It loaded the ROM, finished booting from USB, showed the desktop… and then gave me “Please insert RISCOSpi”. It seems that it still wants the card to be present even though the boot sequence ran from a SCSIFS drive.

As a workaround, I tried unplugging SDFSFiler and SDFS. It now boots up with no errors, but as you may expect, CMOS changes don’t persist. Is anyone aware of a better way to make the system “forget” the SD card, or is this the best I can do until I get a new Pi?

Thanks :)

I removed the SD card and reinserted it, and powered up again. It loaded the ROM, finished booting from USB, showed the desktop… and then gave me “Please insert RISCOSpi”.

Sounds like the old chestnut of specific commands in the disc based boot sequence including a named rather than generic path.

If you do a search for RISCOSpi in the files in your boot sequence using Steve Fryatt’s rather nice Locate utility you should find the offending item.

It could just be an item you added to the Run at Boot.
Check in $.Boot.Choices.Desktop (and PreDesktop) for RISCOSpi lines
this is the correct form

|Start RISCOS BootRun 0.01 Run

IfThere Boot:^.Utilities.!SparkFS Then Filer_Boot Boot:^.Utilities.!SparkFS

Edit: The nice bit being the “IfThere” which means if it can’t find the item in question it doesn’t stop the boot and sit and sulk – although a small utility in the boot to log what did and didn’t load would be nice. Rick should pop up soon and mention one that does just that simple task.

Thanks for that; I’ll take a look at it after work :)

Once again I’ve proven that I shouldn’t mess with things that I don’t fully understand; PackMan now errors on launch, saying something about SDFS (I didn’t note down the exact text). But fear not! The local Pi distributor is no longer adding a huge markup to the model B+ so I’ve bought a new one. It should be here tomorrow so I can survive until then :)

Thanks for the suggestions though!

PackMan now errors on launch, saying something about SDFS (I didn’t note down the exact text)

Same issue essentially: either the configuration of Packman has an absolute path, or it knows about an application it installed on the SDFS mounted drive (the card) again with an absolute path. Probably the latter.
Check the config files in Packman (does it use Choices: ?)

Actually this whole experiment was with a freshly-imaged SD card (don’t want to ruin my good one!) so I’d never run Packman before. There was therefore also nothing obvious in $.Boot.Choices.Desktop, so the whole thing is a bit bizarre.

When my B+ arrives I might try it again just for the fun of it, and will see whether I get the same results. I didn’t end up using Locate but I may do next time. Having a reliable SD slot in the first place should also help :)

Having a reliable SD slot in the first place would also help

Have you tried this fix – I believe Mastercard and others also work!

The problem with PackMan as set up in the RC12a SD Card image has been identified and will be corrected in any subsequent image. It is #382 in the bug tracker.

When my B+ arrives I might try it again just for the fun of it

Well, as it happened, I ended up being too lazy and didn’t take another look :)

Have you tried this fix – I believe Mastercard and others also work!

Not bothering at this point; I’ll just use the B for Linux since it doesn’t care about the card sensor, and use the B+ for RISC OS.

The problem with PackMan as set up in the RC12a SD Card image has been identified and will be corrected in any subsequent image. It is #382 in the bug tracker.

Thanks; good to know that it’s not just me :)