I'm getting a problem with an N800 (latest firmware) which is difficult to completely pin down but I THINK it only happens if I have an old SD card in the device.

If I leave the tablet idle for about ten minutes I come back to it and it appears to be dead - it won't respond to screen touches or button presses - the only way I can bring it back to life is to drop the battery out.

It only appears to happen if I have an older format SD card installed in either the internal or external slots. I've got 2 - a Lexar 256Mb and a Sandisk Ultra II 1Gb. If I use the 128Mb card that comes in the box then it works fine. Is the tablet fussy about the cards it uses?

I was thinking of patching the kernel and buying myself an 8Gb HDSD card anyway, and hope this might sort the problem. Has anyone seen similar symptoms before?

Tim

Can you check if the cpu is being maxed out while it appears to be dead?

It could be meta-crawler going nuts because of a dodgey file on your old SD card - what sort of files and how many files do you have on the card you believe is causing the problem? Try running the script attached to this newsgroup posting.

Does formatting the "old" card improve the situation? If meta-crawler is the problem, the SDHC kernel won't help.

Both cards were formatted in a windows PC to remove old clutter, and the 1Gb card has been completely reformatted and partitioned to install dual boot on it, with nothing in the files partition (I finally got there). I get the lock up when that card is inserted whether I boot from it or it is just sitting idle in either slot. The same thing happens with two cards.

I'm not sure what you mean by checking to see if the CPU is maxed out. The thing becomes a brick as if the battery is totally drained and the only thing I can do to get it re-started is to pull the battery.

Tim

OK - I've managed to stop it from locking up, but I might just be masking the problem. I remembered that I extended the memory partition on the card that came with it so I created a large (128Meg) virtual memory store on the FAT partition. I know that PS lists processes. Is there a switch that gives CPU and memory utilisation to find out what is going on?

Tim

@ tim

try "top"

I've tried top, if I re-direct the results to a file it seems to provide me with an ls of the /sbin folder, before the table of processes, is that correct? I'd like to post output from top (or perhaps another CPU/MEM util together with output from free on here with swap set to different sizes to see if anyone can shed any light on it. meta-crawler doesn't seem to be running. I usually only tend to have 2 or 3 apps running, and had to load up about 8 before swap even came into the occasion.

I know that I now have it working, but I'd really like to know why. It has taken me a week to even start to get to grips with a new OS (with huge amounts of help from this forum), I'm enjoying myself but realise I haven't even scratched the surface so far.

Tim

As luck would have it I find a way of capturing the information and now I can't reproduce the error. Just from casual observation it appeared there was a memory leak in opera when using it to display the slimserver webpade with the Nokia skin. How that related to swap space I don't know. I've just set up a monitor:

top -d30 > top.txt

Which should at least give me a clue as to what is going on. Are there any more appropriate tools for capturing system counters that I can perform analysis on after the fact, or am I pretty much stuck with top?

Now I've got dual booting working it is time to pick up an 8Gb card and load it up with media to take with me. Will this Adata card be suitable?

Tim

i have two of those same exact cards, from newegg.com, and although i dont have the boot from mmc option, they do work with the SDHC patched kernel.

I can't reproduce it at will yet, but there appears to be a memory leak in Opera but only under a certain set of circumstances (It was running at 49Mb a minute ago when I closed the app). The site I access most is a slightly modified version of the Nokia 770 skin for Slimserver control. It seems to only occur when I have the page loaded which controls an mp3 stream generated for playback on the N800 itself. At least at the moment I think so. It is a tricky one, I'll have to keep trying. At least now I have a way of logging memory and cpu utilisation up to the point of failure.

Any pointers gratefully received.

Tim

It could be a leak in the Slimserver skin itself, and not directly a problem with Opera - Ben Klass on the Slimserver forum is the author, should you be able to narrow it down to the skin itself. What version of Slimserver are you running - 6.5.1 or a nightly?