Hello,

New user here ; a happy one so far. I came across the smartbook by pure chance and I'm glad I did.

The praise:
I like the design of the smartbook. Very well thought out, my only gripe being with the trackpad (I know I'm not the first ).

The first machine I got had the display bug (where the machine would lock up hard on a corrupted screen). André Siegel was extremely responsive and it took longer to wait for the postal services than to obtain a replacement. Congratulations to all for the exemplary customer service.

The current problem:
I have now been using the machine somewhat extensively, in part playing around with alternate distributions. Namely, debian-armhf and the latest linaro. Here again the design shines, in allowing to boot from the SD card without having to touch the preinstall.

For starters, I moved /home to a MicroSD (and updated fstab accordingly).

So, I have dd'ed the demo armhf image to a SanDisk 8GB class 4 card, toyed with it for a couple days (dist-upgrading, installing/removing packages, and probably messed the software up a bit...) without a hitch, then left it alone for a while.

More recently I dd'ed the latest linaro efikamx-ubuntu-desktop image to a no-name, made-in-korea 8GB class 10 card, rebooted...

... and only got a bunch of errors scrolling by too fast to read, with a hard power-off the only solution.

Now it seems I'm having problems with the SDcards *in general*.

Booting from the normal Ubuntu and simply putting the cards in, the linaro one would mount seemingly empty partitions. Fine, let's assume something went wrong when imaging it. Now the SanDisk one (which had been working fine) takes forever to mount anything.

Both now spurt a relatively continuous stream of Neither card is seen by fdisk/cfdisk ("unable to read") nor can be written to by dd ("input/output error") anymore.

(Looking back through syslog, it seems the problem has happened with mmcblk0 (the microSD) too, although it's not happening *now*).

I guess my question is, are there known software/driver problems with the card reader(s) that I haven't found mentioned in the forums, or have I just been slightly unlucky with crappy cards, or very unlucky with a suddenly-defective reader ?

Thanks.

As discussed personally, there is one problem known which is that the VGEN2 and VSD regulators (the ones which would usually be used to power an SD card) do not provide voltages above 3.15V which may actually be too low to drive an SD card that only supports 3.3V operation. However, the voltage is not negotiated with cards (it's left at 3.15V which is close enough) so the voltage should never be lowered.

It would be incredibly rare to find an SD card that did only support 3.3V operation, and the spec states that it must be tolerant of ~6% variance in the voltage (around 0.2V, therefore 3.15V is still perfectly valid), the current supplied to TWO cards operating on this regulator along with the low voltage may not be enough to power SOME cards in combination when both slots are in use.

What we're struggling to find is any reported data by the Linux MMC subsystem of what voltage the cards can actually support (based on card-based operating environment and information blocks) which we know it reads, know it acts on, but simply doesn't do anything about in the current driver (regulator control is a do-nothing operation). The eSDHC host controller also reports in it's capabilities register that it CAN support 3.3V operation (which is true), it's just the MC13892 regulator doesn't provide a regulator to do this with. So there's a complicated issue here of whether we do power management or not (since the other card has to be taken into account) and whether we really are hitting a problem with regulator current or voltage levels.

You didn't mention that you had tested the cards outside of the Smartbook; a USB adapter on a PC, or even plugged into the Smartbook itself. Are the cards still working fine elsewhere? If so there is definitely a problem, if not then it may be indicative of a more serious problem with the cards themselves. After regular, extreme usage of the cards like we do at Genesi (constantly imaging them with new data and testing new kernels, Linux distributions and so on) cards tend to fail with an alarming regularity, usually in the PC or USB SD reader before they fail on an Efika. It might be endemic in the industry that since they're only meant to support rare amounts of random writing in use as a camera storage or MP3 player storage medium (using rather simplistic filesystems such as FAT32) that they fail much more quickly with a desktop (NTFS, ext2/3/4, etc.) filesystem on it.

_________________
Matt Sealey

I keep a 8 GB SD and a 2 GB microSD in my SmartBook at all times. Both are visible to Disk Utility, the 8 GB SD mounts reliably to the desktop, the 2 GB microSD only sporadically mounts to the desktop. Both cards work perfectly on a USB reader on my Mac PPC machines. The 2 GB microSD works perfectly in the phone that I originally bought it for.

How can I read parameter blocks on these cards to see what voltage ranges they want? Assuming that is the reason the 2 GB microSD fails mount almost all the time.

_________________
--

Science, n. Investigation of natural phenomena.
Engineering, n. Domestication of natural phenomena.
Technology, n. Domesticated natural phenomena.

--

\"It is essential for automatic computation that, instead of containing every single instruction necessary to carry out the program, the coded program include methods for generating instructions which are needed to carry out the program. ... One measure of the extent to which an automatic digital computer is used effectively is given by the ratio of the number of instructions formed in the machine to that of the instructions inserted into the machine initially.\" Mathematical Machines Volume I Digital Computers ©1961 Francis J. Murray

If so, then why is this ratio zero on most modern computer software?

Thanks for the detailed answer.
I hadn't then, by lack of hardware. I just tested on a new external multi-card reader (Sony MRW62E) and it's becoming even weirder.

The no-name class-10 card was returned to the store and exchanged after they confirmed it was toast. I haven't used the replacement (class-10 PNY, cheaper than the no-name...) yet.

The class-4 Sandisk reads *perfectly* on the external reader connected to my desktop.

Try reading again on the smartbook: card detected, partitions seen and mounted without problem. A first-level view, i.e. ls /media/whatever/ (either with Nautilus or ls from the terminal) works fine. A second-level (i.e ls /media/whatever/bin or /media/whatever/usr hangs with the gnome-monitor reporting the CPU at 100% (and nothing else to show for it), starts spouting errors (as initally reported) in syslog, and requires a forced shutdown ; it won't even interrupt nor unmount.

Obviously I haven't retried booting off of it.

After that, retried again in the external reader: the card reads fine (ls) at any subdirectory level and unmounts without a glitch.

EDIT 2011-11-13:

Some more testing, and definitely something fishy with using both cards at the same time and/or microSD handling.

Using the armhf sandisk as a reference (mounting and reading, not booting) and the external card reader plugged directly in the Smartbook:

• * microSD /home installed, sandisk in lateral reader, r/w errors.
  * microSD installed, sandisk in external reader, no problem.
  * no microSD, sandisk mount and reads without any glitch in both lateral and external readers.
  * microSD in adapter (micro->full SD) in lateral reader glitches *at best*, unreadable/requires hard reboot at worst.
  * the same microSD in the same adapter reads fine in the external reader.

So for the time being I guess the workaround will be to not use the microSD slot and put /home back on the built-in drive while booting from/playing with distros on cards in the lateral reader.