Hi,

just managed to run Linux (DD-WRT v32 SP1 micro-generic) on a WRT54GS v5.1 (CGN8...).
Was as easy as:
- soldering JTAG to the board
- -erase:wholeflash
- -flash:cfe (modified CFE of WAP54G provided on the DD-WRT website)
- tftp the DD-WRT firmware by boot_wait/tftp
Running like a charm with lots of features.

I think OpenWRT would be an even more attractive solution for these cheap devices.
Since soldering is _not_ optional, but a must to use Linux on the devices,
it is easy to add a serial too, i tested it and it works the same way as with
the 4MB-flash WRTs. With this in mind OpenWRT for the WRT54G(S) v5.x series
would be the distribution for those of us who know not only a shell but also
soldering, electronics and ancient serial RS232 stuff. How much space would be
left for packages if a "nano-version" of OpenWRT would be build based on the
"micro-version" without:
- telnet and ssh - since we have serial, we can do anything from there
   also a very "secure" setup, no telnet and ssh at all!
- jffs2 and failsave mode, we have serial - so no need for this
- vi and other big stuff in busybox, we can edit somewhere and wget

This would also destroy Linksys/Cisco's attempt to make these cheap end-user
router useless for advanced router features like vpn, routing protocols.
I think this was one of the major drivers of the 2MB/VxWorks decision,
i could not believe that the savings on the chip in contrast to the VxWorks
license is that big saver. There also must be other drivers for this effort.

Axel

The reduction of cost in the bill of materials was in the order of US$8.00 per unit. That is very significant. 

Cisco estimated that less than 3% of their linksys product was being 'reflashed' and really does'nt worry about it. 

In this case i don't think that the big corporate was looking to kill off the open-source efforts

People just don't get it -

It's like asking if we could fit linux on a floppy disk; sure we technically could, but it would be so stripped down that there's really no point, especially considering we're really not limited to a floppy disk.

In other words - the v5 is crap and instead of supporting crap and crippling yourself just go out and buy something halfway decent.

$8 in contrast to the development effort for the move and the vxworks license per unit seems not much left for profit.

My look to the DD-WRT micro version on the piece of crap was quite impressive.
I would not agree to "really no point" with having seen this.
There is also answer to my question "how much flash would be left?" with a nano-version.
To make use of these crap things would enable us who have the knowlege and skills,
to turn OpenSource into service - exactly what the GPL wants.
Getting these cheap things to work with advanced features and use them in custom projects
is a very nice way to feed your kids legally with the help of OpenSource.

3% is a very high number IMHO. Keep in mind that Linksys volumes are much higher than for Cisco products.
All low-end Cisco-Routers, which ship >$1000 can be replaced with WRT54G/OpenWRT features.
This is definately a threat for them.

When the statement in another resent posting is true that the WRT54GSv5.1 comes with 4MB flash/16M RAM,
_and_ VxWorks, there could be only reason for doing VxWorks something like this, the reduced bill-of material was an alibi.