On Thursday 25 August 2005 06:35 am, Matthias Scheler wrote:
> I've not investigated the subject further because:
> 3.) I'm not sure how well the hardware is suited for a platform bringup.
>     You usually start with an evaluation board which e.g. has a serial
>     port for debug output. I wouldn't be surprised if the units sold by
>     Netgear don't have such things to reduce costs.

To my knowledge all Netgear WGT634Us have a serial header in them. Many 
similar devices also have unpopulated serial and/or JTAG headers. This model 
uses CFE firmware. Other neat features include a USB 2.0 port, and an Atheros 
card in a MiniPCI slot. The device contains 8M flash, 32M RAM. It also has a 
crypto accelerator possibly supported by ubsec(4).

Possible areas of difficulty are the undocumented (other than non-gpl binary 
code) VLAN-tagging switch integrated into the Broadcom BCM5365P SoC (it's 
responsible for separating the WAN/LAN ports), and that bce(4) (which is the 
likely NIC) doesn't do VLAN tagging.

The general target of this port would be Broadcom 47xx and 53xx "Sentry5" 
based boards (but most would be nearly useless because of the integrated 
Broadcom wireless). These boards include the ubiquitous WRT54G and WRT54GS 
(most versions have unpopulated serial headers).

I've looked at http://www.netbsd.org/developers/new-port.html#atfirst, and I 
have a 40MHz pmax box I can use for initial development. I've written the 
console test program (http://kollasch.net/~jakllsch/brcm-devel/serialtest/). 
I await for my box (and RS232 transceiver) to arrive.

However my experience with kernel programing doesn't extend much beyond trying 
to associate a pre-existing driver with "new" hardware. I've also done some 
small dirty hacks, namely exporting some macppc obio registers to kernfs. Is 
there anyone who could either give me pointers or do the port themselves?

	Jonathan Kollasch