On Sun, 15 Feb 2009 18:06:13 +0100, Joerg Reisenweber wrote: > There are two reports about GSM activity causing I/O-errors on uSDcard. > > http://wiki.openmoko.org/wiki/Supported_microSD_cards/SD-C02G
I wrote that wiki page - but did not investigate much further because I bought a SanDisk 4 GB card that allowed stable operation at 12500 kHz and provided enough speed for daily usage. I still had trouble at full 16 Mhz clock speed. I am (and was back then) using Debian (Xglamo), running from uSD card, which makes the transfer errors very obvious when the system grinds to a halt. The kernel version I used back then was from the 2.6.24 testing series, Oct 18 being the most recent (I tried older versions, and had problems with those, too). GSM network is E-Plus in Germany. To try and see if the interference is still happening with kernel version 2.6.28, I got the old Kingston card back out of the drawer and tried to reproduce the problem before updating. Starting programs while entering the PIN usually did the trick, GPRS was even more reliable at producing transfer errors or - rarely - a screen slowly filling with colorful noise (still working on redraw then, but usually with an unusable system because of transfer errors that also occured). I failed. 16 Mhz, drive strength 0 - no problems so far. Maybe the GSM coverage at my desk has changed... I will have to try again in a train or a hilly area with bad coverage, which were the areas the problem usually occured in. It simply is hard and time consuming to get the error to appear when trying to reproduce and test. I have a feeling that random read and write access to the card (starting programs) is better at reproducing the errors than using dd to generate traffic. I will get my 4 GB SanDisk card back in, now again at 16 Mhz clock speed, with kernel version 2.6.28 - and use the FR as daily phone as usual. It will crash again when I most need it, I am certain of that. ;) Now that I know that my FR isn't the only one affected by this, I will keep you up to date with my newest findings on this list. - Tilman _______________________________________________ hardware mailing list [email protected] http://lists.openmoko.org/mailman/listinfo/hardware
