OESF Portables Forum

Model Specific Forums => Sharp Zaurus => Zaurus - pdaXrom => Topic started by: g33k on October 20, 2006, 11:14:52 am

Title: Gui Bails Almost Immediately - C860 / R121
Post by: g33k on October 20, 2006, 11:14:52 am

I've just done a fresh install of 1.1.0r121 - both uboot and the system itself - on my C860. Everything seemed to go well - very smooth. Eventually, the system starts and I login as root. I run startx and get the touchscreen calibration screen as expected.

Then x starts. The screen is initially rotated as if my clamshell was in portrait mode, which it isn't. As the display gets almost fully drawn - the toolbar is complete but the wallpaper and icons aren't loaded yet - the screen flashes and switches to landscape mode. The screen is drawn to the same point - toolbar but no icons or wallpaper - then I'm dumped back out to the command prompt.

I see:

Code: [Select]

# startx
Jan 1 00:02:42 sudo: root : TTY=vc/1; PWD=/home/root; =/sbin/modprobe mousedev
Jan 1 00:02:43 input.agent[1082]: ... no modules for INPUT
Removing empty element from the valid list of fontpaths
Shutting down switchevd: Jan 1 00:02:46 switchev: switchevd
Starting switchevd: Jan 1 00:02:47 switchev: Starting swit
pdaXrom [Husky]
Removing empty element from the valid list of fontpaths
This output is truncated - I guess by the switch from one screen mode to another - and I don't know how to see the full output. The above is just as it appears on my screen.

If I startx again at this point, everything goes exactly the same, except the "... no modules for INPUT" line is omitted. I see that line only once per boot.

The behavior is the same whether my clamshel is in portrait or landscape mode.

Can anyone give me any suggestions?

Title: Gui Bails Almost Immediately - C860 / R121
Post by: InSearchOf on October 20, 2006, 11:30:01 am

Did you check your MD5 checksum before burning that image to NAND?

Seems like there was a hickup... pre/during/post flash...

Late

Title: Gui Bails Almost Immediately - C860 / R121
Post by: g33k on October 20, 2006, 01:39:04 pm

Quote

Did you check your MD5 checksum before burning that image to NAND?

No, I didn't - but I have now done so and they check out.

I've just tried reflashing the kernel and root filesystem image, and - with horror - I think I see the problem.

Once the emergency system started "writing data to block," I got an I/O error. It rebooted before I could read the whole thing, but it said something about how the file didn't fit due to bad blocks!

So - I restarted the whole process, and it went without incident the second time around.

And it booted into the GUI just fine!

Is my NAND dying? Is this simply a bug or a glitch?