Here's A Funny One


Karel Jansens

Active Member
Joined
Dec 21, 2007
Messages
679
Age
62
Location
Belgium, but trying to escape
Website
Visit site
As it says in the subject line: Last night, I hung my Panda on the charger for overnight juice, but weirdly it seemed to wake up from low power about 15 minutes into the charge and I had to re-standby it. An uneventful morning passed by, until I noticed that Panda arbitrarily stopped recognizing USB drives and then the battery level began to veer wildly from 98 to 24%, finally settling at the aforementioned Kafkian number of -121% (mind you, the status light was still happily green and the Panda never shut itself down).

Then the nubs stopped working. Taking out the battery, soft-resetting, nothing helped, until I decided to recalibrate the touchscreen (that was still responsive to touch), and presto! nubs worked again. About an hour later, the battery level decided to set itself to a realistic 94% and has been behaving exemplary since (hits wood).

I didn't do anything funky, just read a few books and comics and played a session of Meritous (which, very predictably, ended in it crashing, but that's been almost a constant the last couple of days.
 
Sounds like the battery contacts might not have been aligned properly. A bit of padding on top of the battery before closing the compartment might help.
 
mali said:
Sounds like the battery contacts might not have been aligned properly. A bit of padding on top of the battery before closing the compartment might help.
As I said, I did replace the battery. What I forgot to mention was that I did so several times, with the same result (Star Trek battery reading) every time.

I somehow find it hard to believe that I managed to misalign the battery each and every time and that it decided to spontaneously align itself correctly after an hour had passed...
 
Last edited by a moderator:
The I2C subsystem failed for some reason. Nubs and the battery gauge are connected using I2C. FYI strerror says that 121 is "Remote I/O error" (not 100% sure if the error codes are compatible, but IIRC in-kernel error codes should be just negated userspace error codes), but that doesnt tell much (the dmesg from the event would have been nice for the kernel devs, but I guess its already gone.). I had an -110% (Connection timed out.) event some day, was in a hurry so i just rebooted my pandora and it worked with that, wonder why your problem persisted.

So yeah the battery gauge should be coded with a bit more logic, now it just reads a number from a special file and reports that, even if its negative (meaning to report an error code, but OTOH its very easy to see the error if you know what is happening).
 
it's especially funny that calibrating the touchscreen was the thing that fixed it. Maybe some bad data got into a config file somewhere and even on reset that data was loaded and caused the I2C controller problems; calibrating the touch screen caused it to rewrite that file to good, or something, maybe? Probably not, but I can't think of any other explanation.
 
Back
Top