Some reverse engineering on the E-TEN glofiish DX900

Today I've found a bit of time to start my reverse engineering efforts on the E-TEN glofiish DX900. In case you don't know what that is: It's E-TEN's latest PDA-phone model, 2.8" full-VGA touch-screen, GPS, WiFi, Bluetooth, 3.5G and dual-sim (it has a 2G modem next to the 3.5G modem). It runs the Samsung S3C6400 application processor. It is so new, that it doesn't even yet have FCC approval. Luckily it was available in Taiwan at PDA king for NT$ 16,900.

It seems like a great device. There's basically only one big flaw: It runs Windows Mobile. And it does that in a really bad way. From how sluggish and unresponsive the UI is, you can clearly see that they don't use any of the 2D acceleration functions of the S3C6400 hardware.

Now in order to get rid of Windows Mobile, we need to discover more about the device hardware. The first step for that is HaReT, the Hardware Reverse engineering Tool. Unfortunately the S3C6400 is so new that HaReT doesn't yet have support for it. So I had to dig into the code and add support for it. You can find the preliminary patch here.

The information that I was able to dig out using the first round HaReT can be found at this DX900 wiki page. As expected even before touching the device:

  • the 2G modem connects to a UART
  • the 3.5G modem is SPI slave just like in the M800
  • the wifi is still Marvell 8686 connected to SPI
  • the GPS is still SIRF3 connected to a UART
  • the buttons still are the same, some connected to GPIO some not

I won't have much time to work on this right now, as too many other higher priority tasks are pending. But it seems like the DX900 is a nice s3c6400 based device to play with, and a Linux port to it is not really further away than for the M800 or X800