[Tuto] [EN/FR] Install nLoader / TI-Nspire CX

→ **MyCalcs** profile · by **critor** » 26 Feb 2018, 19:30

parrotgeek1 wrote:Also, the payload size is essentially unlimited. You're mistaken.

Possible. But when I tried weeks ago, I got payloads which suddendly stopped working over some size, which was still inferior to the ControlX size.

→ **MyCalcs** profile · by **critor** » 26 Feb 2018, 20:25

Updated nSwitchOS (fixed a bug).
Added nSwitchOS captures.
Removed the red warning.

As far as I can see, the tutorial is now complete, at least in english.

Supposed to work on all CX calculators, until TI fixes this.
I'm no reference, the newest CX CAS I have is a HW-W, and the newest CX I have is a HW-T.

But nLoader as been reported to work on HW-AC some posts ago.
So please report any strange behaviour on HW-AD+ devices. Since it wouldn't be the first time applying a tutorial on newer revision calculators made artificially incompatible would brick them.

TI already made hardware fixes for Nlaunch* (HW-J+) and nBoot+ControlX (HW-W+), by making newer devices incompatible. And since most new students are getting new calculators, the compatible devices are disappearing very quickly. Same thing each time, the exploited Boot1/Boot2 versions stopped working on the newer devices.
If they do the same thing, then we're waiting for a new revision on which Boot1.5 4.4 won't work.

The software fix is usually to wipe out and reflash the Boot2 partition.
nBoot+ControlX is not affected, as the installation tutorial was designed to work around this.
nLoader is different. It will refuse to boot any OS version it doesn't know about (can be quite annoying in my opinion, but it's the developer decision

). So you can't kill it by just sending a 4.6+ OS update.

→ **MyCalcs** profile · by **parrotgeek1** » 26 Feb 2018, 21:01

critor wrote:TI already made hardware fixes for Nlaunch* (HW-J+)

HW-J wasn't intentionally blocking Nlaunch. It had a new cheaper nand chip incompatible with old boot2

critor wrote:
parrotgeek1 wrote:Also, the payload size is essentially unlimited. You're mistaken.

Possible. But when I tried weeks ago, I got payloads which suddendly stopped working over some size, which was still inferior to the ControlX size.

the size is CURRENTLY 256kb but you can edit the base address to anything you want in the ldscript and some header files

critor wrote:It will refuse to boot any OS version it doesn't know about (can be quite annoying in my opinion, but it's the developer decision ). So you can't kill it by just sending a 4.6+ OS update.

I might change this later. I'm just afraid a new OS would delete nLoader, even though I set a4012eb4 to ffffffff, TI could check the boot2 version in NAND.

also
please do not upload the source code here archives_voir.php?id=1374509
Just upload the binary, and link to my github
The source code does not compile outside of my repo because it depends on patchfinder.

→ **MyCalcs** profile · by **critor** » 26 Feb 2018, 21:09

This wasn't a problem for nBoot+ControlX, since :
- the calculator was booting on the Diags partition which unlike the Boot2 partition isn't checked/erased/updated by the OS (since the Diags software can be quite specific to the hardware revision)
- the tutorial made you put a bootable copy of Boot2 4.0.3 in the ExtnD partition, making the modified calculators almost unbrickable whatever OS you install

I know you don't have this choice for nLoader, since Boot1.5 is hard-coded to always boot the 2nd image in the Boot2 partition if I'm not mistaken, even if you put Boot1.5 in the Diags/ExtnD partitions.

→ **MyCalcs** profile · by **parrotgeek1** » 26 Feb 2018, 21:14

critor wrote:I know you don't have this choice for nLoader, since Boot1.5 is hard-coded to always boot the 2nd image in the Boot2 partition if I'm not mistaken, even if you put Boot1.5 in the Diags/ExtnD partitions.

you are correct. It doesn't even work correctly in another partition.

→ **MyCalcs** profile · by **critor** » 26 Feb 2018, 21:23

parrotgeek1 wrote:
critor wrote:It will refuse to boot any OS version it doesn't know about (can be quite annoying in my opinion, but it's the developer decision ). So you can't kill it by just sending a 4.6+ OS update.

I might change this later. I'm just afraid a new OS would delete nLoader, even though I set a4012eb4 to ffffffff, TI could check the boot2 version in NAND.

It's not against you by the way.

Excale made a similar choice years ago with Nlaunchy, patching OSes to completely forbid OS updates.
And I probably did complain here somewhere.

I had to use the maintenance menu each time I wanted to switch to another *supported* OS version.
That's one of the reasons why I've developed nSwitchOS.

It was also preventing the USB OS dumping method to work properly (rare/unreleased OS sent to another calculator running an Ndless program not disabling interrupts).

I hated Nlaunch* for this.

→ **MyCalcs** profile · by **parrotgeek1** » 26 Feb 2018, 21:26

critor wrote:I had to use the maintenance menu each time I wanted to switch to another *supported* OS version.
That's one of the reasons why I've developed nSwitchOS.
It was also preventing the USB OS dumping method to work properly (rare/unreleased OS sent to another calculator running an Ndless program not disabling interrupts).

Neither of those things will be problems in beta 2.

→ **MyCalcs** profile · by **critor** » 26 Feb 2018, 21:28

parrotgeek1 wrote:
critor wrote:I know you don't have this choice for nLoader, since Boot1.5 is hard-coded to always boot the 2nd image in the Boot2 partition if I'm not mistaken, even if you put Boot1.5 in the Diags/ExtnD partitions.

you are correct. It doesn't even work correctly in another partition.

I could be mistaken, but I think I did test Boot1.5 images in Diags/ExtnD partitions while developing ControlX.
And while the 2nd image in your Boot2 partition is still a Boot2 image, they work.
Any version seems to be ok (Boot1.5 4.0.2 can load Boot2 4.4, Boot1.5 4.4 can load Boot2 4.0.2).

If the 2nd image in your Boot2 partition is missing or is something else (Boot1.5 for example) it'll crash.

→ **MyCalcs** profile · by **critor** » 26 Feb 2018, 22:23

parrotgeek1 wrote:also
please do not upload the source code here archives_voir.php?id=1374509
Just upload the binary, and link to my github
The source code does not compile outside of my repo because it depends on patchfinder.

Done.

But it's linked to your account. So feel free to update the file and its description whenever you want with whatever you want.

→ **MyCalcs** profile · by **Lionel Debroux** » 27 Feb 2018, 08:03

2) in beta 2, there will be support for non cas os on cas calculators. Currently there isn't.

Why not

That's not a problem, unlike the CAS OS being trivially made available on the calculators sold as non-CAS, which, as we all know, has been your driving goal from the beginning. As I wrote earlier, this would make you so toxic and damaging for the whole TI community that sadly, your binaries containing such features would have to be removed from the main sites if they were somehow uploaded there (you'll just host them somewhere else the community doesn't control, and you could even take advantage of the Streisand effect), and your advertising for this part of your work would have to be silenced, if you don't refrain yourself from doing it.
I still really hate the reasons behind that sorry, counter-productive state of fact - but should the unwanted event occur, as a staff member for one of the community's major sites, I'd have to make do with them and do what I have to do, even if it's a pain... That, or quit being a staff member, in which case performing the required moderation job wouldn't be my duty anymore

Excale made a similar choice years ago with Nlaunchy, patching OSes to completely forbid OS updates.

Well, that was the right thing to do to protect the vast majority of users from a trivial way to remove the effect of nLaunchy. It's a fact that some use cases became harder as a result.