Odd Shell Glitch

[Chasney913]

One of the people in my class has a TI-84+ (regular). He requested that I get him some games, to which I graciously obliged. One is Mario, and the other is Acelgoyobis (pinball, for those who don't know). They work fine on mine.
When he plays them however, the screen is screwed up, where every two pixels it moves off a line, so the whole thing is on a diagonal. It also seems to wrap around so that anything that won't fit on the screen is moved to the bottom. It does this for the aforementioned games, and if I recall correctly, TinyCarZ, but I'm not sure on that. It seems like it's doing it to anything that requires a shell.
To date, I have tried: RAM clear, OS update, and CalcUtil (I thought it might be Mirage). I have attached a fun picture of what it looks like, using a screen from RushHour (forgive me, WikiGuru!) As you should be able to see, it would be difficult to play. Mario's not too bad, but Acelgoyobis and TinyCarZ are unbearable. I'm looking for ideas on fixing it. I'm sure people have come across it before, as I've seen it on a TI-83+ previously. I have tried to do a search for something related, but it's difficult to describe. I'm sure he would be pleased.

[TheStorm]

The bug is caused by the crappy lcd drivers ti uses, someone else can give you more details.
Search for alcdfix on ticalc have him run it every time his ram is cleared it will prevent this form happening.

[Igrek]

There is also a patched version of mirageOS somewhere but I suggest using alcdfix, it does a great job.
This is the link to alcdfix:
http://www.ticalc.org/archives/files/fileinfo/366/36608.html

[brandonw]

This is a serious enough problem that the OS needs to be patched to apply this fix on boot. I don't know how people survive having to apply that patch every single time you reset RAM.

Since it's SE/84+ only, you could store the program on page 30h/70h, although you'll have to look into the OS a little to see where to apply the patch to regain control on boot, and what to do to stop it from erasing sector 30h/70h (which it does on first boot).

[Chasney913]

Is there a way to do that procedure that has been created, or are you talking theoretically? If it's been made, that'd be great (I'd assume it hasn't), but if not, how would I go about doing that? I have no experience in that sort of thing, but I'm fairly confident my friend would rather not have to run the patch every time.

[brandonw]

It doesn't exist yet. It would involve writing to the OS, which is not necessarily easy, or legal.

[Chasney913]

So it's probably not going to happen then. Is there any way to make the program provided run after the RAM is cleared? I know there's a way to run on start-up, but this would only need to run after a clear occurred.

[Liazon]

[brandonw]

My RAM is cleared every few minutes I'm using it, but that's probably because I'm crashing it through stupid code.

A calculator with this kind of problem is worthless for development without something to automate it like that.

[Taricorp]

I've never looked at an 84 ROM, but the 83+ OS has lots of free space on page 1. You could modify the code at 0056h to jump somewhere on page 1, then duplicate whatever you needed to overwrite for your jump there, and return to 0059h.

That's assuming it jumps to 0056h (where the boot code returns to) when you clear the RAM, though.

[brandonw]

The code at 0053h is a jump...it has to be, because the ID bytes and error handler are immediately after it (address 0056h and call 059h and 05Ch). You can replace the address here with one at the end of page 0, where you can place some of your own code, but just be sure to back up the original 0053h jump address, because that's where you'll be returning to.