Linux Project

---

Does anyone here have an update on the Linux progress?

Second that.

The mailing list never seems to send anything

MmF
:?:

As far as I know not much is happening on that front.

To safely develop a linux port, a safe way to reflash the device is needed. Normally this is done with JTAG. Because the location of the JTAG pads on the XDA are not known, JTAG flashing is not yet possible as far as I know.

Now if anyone does have this info somewhere that would help to kickstart Linux porting.

Finding the JTAG pads would probably need hardware tracing and thus killing an XDA. If anyone has a dead one laying around please donate it to the cause.

You don't need JTAG for flashing, you only need it for developing the bootloader itself (which isn't necesary as the serial port and ROM should work the same way as on iPAQ afaik). The bootloader exists and it's developers seem quite capable. But the hard part would be hacking the linux kernel to support the radio part (signal monitoring, call handling etc.) and modifying it to be realtime-capable (because of the radio). Then, phone integration to the environment (userland), but that wil be quite easy...

Personaly, I doubt there will be a linux port ever unless HTC provides some real support (do they? :)), but it would be soooo cooooooooool :))

I know JTAG flashing is not nessecary per se. Lets say it this way, I know noone that is prepared ****ing up his XDA buy destroying the bootloader. As far as I have seen the flashing process is not fail safe. If the bootloader has a bug or the flashing is corrupt the XDA is dead. The only remedy is JTAG flashing then. To experiment with something like new bootloaders JTAG flashing is preferred but not nessecary.

I wanted to say that you can use the bootloader that exists, as it will already support serial port-talking and ROM flashing (probably RAM formatting etc. as well). If you don't mess with those features, it is quite fail-safe. If you implement something (GSM reset) in it, and it doesn't work, you can always disable it at runtime.
But I won't argue, I'm just a power user :)

Does that mean then that you can flash the thing without writing to ROM?

So to do lots of experiments, you just load the info straight into RAM...
Then just do a hard reset and the ROM reloads...

This all sounds much safer. But too good to be tue...

We all look forward to the progression of Linux. So long as HTC or somebody else in the chain help us out...

MmF

In the end it is just about how much risk you are prepared to take. Currently the XDA contains a bootloader in the first blocks of Flash ROM. There doesn't seem to be any other ROM that can take over in case this first block is corrupted -> the XDA is dead.

Now one can experiment with RAM bootloaders that are started like our Wallaby patch tool. This way the orginal bootloader is not touched. It is not clear to me if this would be sufficient for Linux development.

For the Linux on Ipaq development bootloaders were used that were written over the orginal bootloader. This could result in corrupting the booting process. The only way I know to recover from this is to write the Flash ROM back with JTAG as this needs no running bootloader.

Currently I (and others) are not prepared to risk killing an XDA this way, cause our experience has been that things will go wrong once you start such projects. We first would like to solve the JTAG flashing problem with an already dead XDA. Thats also why we called for people with dead XDA's to donate them.

If anyone knows more about a good way to kick of this Linux porting process we'd be happy to look into it or publish work in progress reports.

Update

---

Ok, here is a little update. I might try this at some point; I am currently too busy with other stuff. As discussed flashing the device can be risky. However a combination of booting from SD/CF and running from there might be a possibility to start porting the major parts.

These are informative links:

http://www.handhelds.org/z/wiki/Boot...ithoutFlashing

Also this describes a bit of what we already know too. If we figure out where the wallaby boot loader kicks off the image, we could consider just flashing the image and not the bootloader:

http://www.handhelds.org/z/wiki/LoadingFromCF

Linux sets its sights on the PDA market

---

for more details

http://www.linuxdevices.com/articles/AT8728350077.html


Othman