Microdrive Woes!

[evilyoda]

HI!
I got a Hitachi 4 GB CF Microdrive and can't get it to work with my Zaurus SL-5600, ROM 1.32. When I connect the microdrive to my computer using a Kingston CF reader, I can read/write/format/repartition to my heart's delight in Linux or Windows. However, when I put the card into the Z, the little white "CF" icon appears in the task tray, but nothing gets mounted. dmesg makes it look like nothing really gets recognized. My friend's Axim couldn't see the microdrive, either.
I've tried it in the factory default fat32, and also repartitioned (1 single partition) ext2 and vfat32.

I believe this came out of a muvo. dmesg shows some pcmcia activity, but nothing to indicate that it recognizes the device as anything. Do I need a different kernel?

Thanks!

- Josh

[freizugheit]

Have you tried reformatted the drive using FAT32?

Post the content of /etc/fstab here.

[evilyoda]

Yes, I've done FAT32 format from windows and also I believe from linux. Neither worked.

Here's an observation: when I insert a regular 16 MB CF card, which my Z recognizes and I can read/write from/to, no kernel modules are loaded. However, when I insert the microdrive, the memory_cs module is loaded.

My fstab:

/dev/mtdblock2    /   ...
/dev/mtdblock3   /home   ...
none   /dev/shm   ...
/dev/hda1   /mnt/cf   auto   noauto,owner   0   0
/dev/sda1   /mnt/md   vfat   defaults   0   0
/dev/mmcda   /mnt/card   auto   noauto,owner   0   0
none   /dev/pts   ...

The /mnt/md one is leftover from my attempts at getting this thing working.
Are there any /proc files that would be useful to look at?
Thanks!

[evilyoda]

Oh, one more interesting thing:
My brother also owns a Hitachi 4GB CF microdrive, and when I popped it into my Zaurus it worked just perfectly! It must be this issue:
https://www.oesf.org/forums/inde...?showtopic=5006

I must have a new and "crippled" microdrive.
However, that still doesn't seem to make sense. Could my PC read the microdrive if it is new and non-standard?

[roderickv]

I just recently purchased a Hatachi 4GB (OEM) version micro drive also. Mine is working great! I know that there are susposed to be some models that will not work, but I would suggest the following before you give up, because there are some steps you need to go through before you format the drive in your Windows Machine, according to the post by Omicron. I followed his step by step information
( found here  http://www.zaurususergroup.com/index.php?n...3779&highlight=  ) and it worked perfectly for me. As far as I know it will/should work on the SL-5500 also, so maybe it will work for you and your SL-5600.
Hope this helps.

P.S. If the link doesn't work for you, PM me and I will send you the info in a Word document.

-Rod-

[freizugheit]

Some of proc info in my SL-5600 for reference:

/proc/ide/hda/capacity:7999488
/proc/deviceinfo/revision: 1.32 E
/proc/deviceinfo/filever: 1.0

[evilyoda]

I just recently purchased a Hatachi 4GB (OEM) version micro drive also. Mine is working great! I know that there are susposed to be some models that will not work, but I would suggest the following before you give up, because there are some steps you need to go through before you format the drive in your Windows Machine, according to the post by Omicron. I followed his step by step information
( found here  http://www.zaurususergroup.com/index.php?n...3779&highlight=  ) and it worked perfectly for me. As far as I know it will/should work on the SL-5500 also, so maybe it will work for you and your SL-5600.
Hope this helps.

P.S. If the link doesn't work for you, PM me and I will send you the info in a Word document.

-Rod-

Thanks for your suggestion. I followed the instructions at the link you listed, but still my Z doesn't recognize the card. All I get from 'dmesg' output is:

pxa_pcmcia_init(0)
<<<< Init_Module PASS >>>

According to posts at dpreview.com microdrives with a barcode # ending in 44C or 3Ca are set to work only in ATA mode, not CF+ mode. People there report also being able to read their MD from their PC over a reader, (which perhaps uses ATA), but not on any other CF devices.

Would there be a way to set the linux kernel on the Z to use ATA rather than CF when dealing with the crippled version of this device?

[tz]

You might need to manually load the ide_cs.o and any other ide modules required.

You might also want to check if something else (e.g. the memory_cs or cfxga_cs) is being loaded instead.  The cfxga card is basically a dumb memory card and I think it uses an ATA style interface - but if you need the real memory_cs to access it as ATA, they might not be there (and do a strings even if they are called memory_cs - they may be cfxga drivers as that card also does not have any ID information).

modprobing ide_cs.o and seeing the dmesg might be enlightening, as would cardctl info and lsmod.

Pure ATA style should work, but I don't think the /etc/pcmcia configuration is set up properly.  I don't have such a card, but a check in the linux pcmcia area should give the right info.

e.g. section 4.8 (near the very bottom) of

http://pcmcia-cs.sourceforge.net/ftp/doc/PCMCIA-HOWTO-4.html

[evilyoda]

This is exactly what I want to try, tz. My current kernel doesn't have the ide_cs module, though. I think I mentioned it above, but what IS being loaded is memory_cs.

'cardctl status' with the drive in yields:
Socket 0:
   3.3V 16-bit PC Card
   function 0: [ready], [bat dead]
Socket 1:
   no card

So, the next step is definitely getting a kernel onto my machine that does have the ide_cs module compiled. I have no experience with this, and the largest currently-working CF device I have is 16 MB. Perhaps if I had the kernel image and the modules I could just copy them over to the Zaurus manually? How would you suggest doing this?
Thanks!

[tz]

You may also want to try REMOVING the memory_cs.o since that might be a CFXGA driver (use the strings command or less or something - or even grep -l for the string CFXGA).

You may just need ide.o, though I think ide_cs.o sets up the PCMCIA chip with the right register mapping.  That is set to 'Y' and not 'M' in my .config for my kernel which I think is the "stock" value.

The (wrong) memory driver  may want to load ahead of any internal ide-cs, so removing it (and maybe tweaking something in /etc/pcmcia) might make things work.