After converting a number to a fraction, the display of my TI 84 plus got mirror inverted:
For example the cursor on the homescreen os now in the top right corner, below a picture of the "MODE"-menu:
Trying to connect the calculator some hours before the invertion didn't work (but i blame windows and its drivers...)
Resetting all ram didn't help, any ides?

Edit: Resetting all memory didn't help either...

Edit Nr.2: Reinserting the batteries did... But i wonder how that could even happen 

That would get fixed by a RAM clear though (and really shouldn't happen at all in general since the mode gets changed so often). However, newer LCD drivers have full screen mirroring capabilities built in, which can be accessed by outputting values from 0x0C to 0x0F to the control port (bit 0 controls horizontal mirroring and bit 1 controls vertical mirroring). Of course, nowhere in the OS is there any code to output 0x0C to turn mirroring off, so the only way to turn it off without writing ASM code or using Calcsys is to force the LCD driver to reset, i.e. remove a battery.

At first I thought someone created an inversion program and I was gonna say that some ASM programs can already do that but horizontally rather than vertically. Then I saw it happen vertically

Does it work on every 84+ tho? I recall that certain revisions didn't do that. It's even more shocking that it just happened with no single line of ASM code

erm it's caused by port 10 bit x0E ( I think) getting set. but how the heck could that happen on its own O.o

Because TI is an idiot and they left the LCD auto-increment mode in the wrong setting?

It's a bug on TI's latest MathPrint OS. It happens to me all the time. To fix this witouth data loss, turn off your calc, take out one battery (NOT the backup battery, a regular one!), and reinsert it. When you turn your calc back on, the screen driver was reset. Unfortunately, calculating an other fraction flips the screen again. To fix this, Thepenguin77 once releaced a patch. Contact him for it.

It's not TI's fault, it's actually brandonW's (though he should receive no blame at all for it).

The trouble comes from the universal flash unlock exploit. The one that's been passed around for about 5 years actually had a critical flaw in it that overwrote part of the OS. In the past, this write never made a difference, but in 2.55 it does.

Here (link to post) is an "immunization" for 2.55 that makes the OS overwrite do nothing.

And here (link to file) is the new universal unlock that should be used.