#369 Running SnakeOS on the MRT G101

[Anonymous]

Originally created by: stefansc...@googlemail.com
Originally owned by: stefansc...@googlemail.com

I've had two conrad branded of these sitting around for a while and finally wanted to see if I can get snake on them.

Specs:
SoC: CNS2183
RAM: 32MB
Flash: 1MB EN25F80

There are two rows of headers on the board:
JTAG: VCC|TRST|TDI|TMS|TCK|TDO|SRST|GND
Serial: X|DATA|DATA|X|GND

The SoC seems to be more or less the same as the STR8132/CNS2132 except with gigabit ethernet.

The flash is way too small for snake. So instead of installing the whole OS I've just put the boot loader on it, the rest is loaded from USB disk. I've ported the changes from our ancient u-boot version to the current one to get USB boot support.

I've only done limited testing, but so far this setup seems to work ok. One issue: I tried the new u-boot on my Ritmo LN-86BT and it has serious trouble detecting the builtin SD card reader, which is a bit worrying because it's also attached via USB..

Installation:

First the usual disclaimer..
DO THIS AT YOUR OWN RISK. IF SOMETIHNG GOES WRONG THE DEVICE IS PROBABLY TOAST.

The original firmware updater refuses to install a bootloader that is larger than 65k, u-boot is, so that check needs to be disabled. The main part of the firmware can't be easily modified because it's compressed. To get around this the bootloader is replaced with a modified version that disables the size check after decompressing the firmware.

Make sure the device is running firmware version G5S6070400159D0 1.59 (it's attached G5S6070400159D0.bin). Using the modified bootloader with a different one would make it randomly overwrite bytes of memory .. that would be bad..

Anyway.. the procedure:

Install g101-g5s6070400159d0-boot-sizecheckhack.bin

Install g101-u-boot.bin

Format an usb stick as FAT and put cns21xx.bin on it.

Plug usb stick into it, power up and pray.

I'm a bit disappointed by the network performance. Iperf reports ~90mb/s. It is suspicously close to 100mb/s and the results from the non-gigabit devices.. so maybe it's just a driver issue that can be fixed.

[Anonymous]

Originally posted by: stefansc...@googlemail.com

(No comment was entered for this change.)

Labels: -Type-Defect Type-Other