Posted: Sun Aug 24, 2008 11:22 Post subject: Re: Up and running!
cherry31415 wrote:
I'm back and I rebuilt the rootfs and now linux is working. Right now I'm using OpenWrt. I'll be documenting install process here.
Don't get too exited, there's still a lot of work to do.
Welcome back! Just a few quick questions if I may, once you have up and working how does the firmware get applied? I see with all the different routers each device presents its own challenges as with this one. I was able to flash a Linksys WTR54G but it looks that was one of the easier installs because of the TFTP utility. I assume the ActionTec will not be that simply, I know Sash stated earlier that it will be difficult. So I’m assuming it’s going to be nothing like the Linksys upgrade. I know this is premature but is there anything I could do to try and read up on how to apply it?
Just got verizon FiOS installed a couple of days ago. Found the hardware of the ActionTec MI424-WR to be quite nice, but the firmware's crap! Waste of bandwidth. I wonder if this is Verizon's doing since they are control freaks with all their services and hardwares. Guess I'll have to bridge my WRT54GL to the MI424 until someone in this forum saves us! Thanx for all the hard work.
Joined: 08 May 2008 Posts: 25 Location: New Jersey, USA
Posted: Sun Aug 31, 2008 20:53 Post subject:
I've created a python script that makes a copy of the existing flash image via the telnet interface and then burns the redboot.bin image using tftp. No JTAG needed. However, if something goes wrong then JTAG will be required to fix it.
This doesn't seem like the right forum to discuss details as this is very experimental and based on openwrt.
Clearly what you are doing to make this work is beyond my scope however I’m very much interested in seeing your crack it. If you could come with a step-by-step instruction on how to install it I’ll give it a shot or if you don’t have time to instruct me maybe someone else could show me how,, or maybe someone at work next week could give me a hand. In other words I’ll test it and I’m not worried about bricking the board if someone could walk me though. I just don’t want to brick it because of something stupid I did. FWIW I’ve got the Atheros card too... if the board does get bricked I’ll order another off eBay or wherever becasue I really like the dd-wrt firmware and I'd really like to see it work on the A/T. Thanks for your efforts!
Cherry, sounds like you've had a productive labor day weekend! Thanks for the update, I've looked over the OpenWRT wiki page you've done up for the MI424-WR and the flashing instructions... a bit much for me to take on without additional understanding of redboot and OpenWRT in general. Will be reading up on this though. Still hoping for an eventual DD-WRT port for this hardware as it's what I'm familiar with but will be happy to learn the ropes of a new software package if it allows ditching of the Verizon software.
Providing you some feedback here since I didn't immediately find a thread topic for this at the OpenWRT site and I know we've got people here looking for any signs of advancement. I did try out your python script without the -w option and have this info:
Line 77-
change def telnet_timeout()
to def telnet_timeout():
Received python error when I ran as provided, had to make this change to run it.
Line 30-
change password = "password"
to password = "password1"
Doesn't match the defaults per -h help option text.
Had to run with -v verbose option else got "NameError: global name 'verbose' is not defined" for line 53. Not familiar enough with python to fix this part of the script.
After I figured out the above I did get what appears to be a successful dump of the existing firmware. From a file list of my working directory at the time I ran the script...
-rw-r--r-- 1 pow pow 8388608 2008-09-02 22:30 mi424wr-xxxxxxxxxxxx.bin
where xxxxxxxxxxxx was the MAC address extracted (I assume) by the script from the router. Haven't confirmed this is a valid image, but at lest something that appears to be the correct size.
Any feedback on how OpenWRT is working so far on the MI424-WR? Is the coax bridge operational? Will this work for people that have the Ethernet for the WAN as well as those that have the coax WAN? I know there are a few different hardware versions out there, between the one I got from Verizon and the two I purchased separately (one ebay and one geeks.com... maybe they'll get more in stock again- http://www.geeks.com/details.asp?invtid=MI424WR-PB-R) I've got 3 different hardware revs. Know of any anticipated problems between revisions?
This is definitely looking promising. I think sometime in the future while I'm passing through NJ I'll have to track you down and buy you a beer!
Joined: 08 May 2008 Posts: 25 Location: New Jersey, USA
Posted: Wed Sep 03, 2008 13:28 Post subject:
Pow,
Thanks for trying the script. I've updated with your changes.
You bring up an interesting point about the different mi424-wr models. I've got 4 (from geeks.com) that are identical. So, I can't really comment on the other models.
Regarding the coax (MoCA) interface, I have no plans to work on that. The HW interface is proprietary, and without additional information there's not much hope of progress.
One final note, this is my first experience with any *wrt software. Just got the basics up. I'm not quite sure how to setup things at the user level! I'll focus on that next.
As I start in on further testing on my own hardware of what you've accomplished I'll use the MI424-WR I got from geeks.com, if I'm lucky I'll be on the same hardware version as you while learning this unit. It's either rev A or B, the older "brick" style shell. I'll confirm which one later.
I still have hope that someone will be able to figure out the MoCA interface, cracking that would really open this up for more people. In my current setup my VZ internet comes in on Ethernet, so I've got a Linksys WRT54G in front of the MI424-WR with the MI424-WR's WAN port connected to one of the LAN ports of the WRT54G. The VZ provided router is just acting as a glorified bridge for the coax network allowing the STBs access to the internet. For the most part this works, but video on demand for the STBs seems to try and suck more bandwidth than the WRT54G's more limited hardware can handle. (just a guess, haven't tried to prove this theory yet.) With a *WRT software package for the MI424-WR I could replace the WRT54G with another MI424-WR, hopefully the more robust hardware would be better able to channel the data. For people that just have the one unit and need it for the STB bridging or have the VZ connection coming in on coax they're still stuck.
Thanks for trying the script. I've updated with your changes.
You bring up an interesting point about the different mi424-wr models. I've got 4 (from geeks.com) that are identical. So, I can't really comment on the other models.
Regarding the coax (MoCA) interface, I have no plans to work on that. The HW interface is proprietary, and without additional information there's not much hope of progress.
One final note, this is my first experience with any *wrt software. Just got the basics up. I'm not quite sure how to setup things at the user level! I'll focus on that next.
FWIW, it appears Actiontec uses the Entropic chipset for its MoCA implementation.... and just doing a search on Entropic's web site, I was able to come accross this:
"The c.LINK Home Network Software Development Kit (SDK) provides the software and hardware necessary for developers to incorporate reliable, effective c.LINK technology into networking, media server, set-top box, audio, video and gaming devices that meet the MoCA standard. Supporting a Linux OS, the kit provides c.LINK driver source code that developers can compile, build and integrate into their own designs"
So it would appear that Entropic's MoCA chipset has Linux drivers available in their SDK. So I'm hopeful that supporting the MoCA interface shouldn't be too terribly challenging after all?
So it would appear that Entropic's MoCA chipset has Linux drivers available in their SDK. So I'm hopeful that supporting the MoCA interface shouldn't be too terribly challenging after all?
Of course they provide source and/or drivers to people evaluating their chip. That SDK is targetted at settop box makers, and is not generally available to download. They are not likely to provide it to anyone who is not in discussions with them about using the chip in a new design. You certainly would not be able to republish the code from it.
If people want to get MoCA working the best bet is to copy the module Actiontec is using. If it is built against a different kernel version than the WRT distro you are trying to use you may need to do unsavory things to get it to load.
It appears that Actiontec used Entropic's PCI solution, not directly hooking up an Entropic PHY to a MAC on the processor, correct? Reverse engineering an unknown PCI MAC is pain in the neck, figuring out what registers to bang via MDIO to setup and use a weird PHY with a well understood MAC is much simpler.
There's little hope of getting info from Entropic. Their SDK is quite expensive!
Looking at the mi424-wr, the clinkdrv module is around 180K, and it's for kernel 2.4; no source is provided in the GPL sources. Interestingly /proc/sys/kernel/tainted reports 0.
There's little hope of getting info from Entropic. Their SDK is quite expensive!
Looking at the mi424-wr, the clinkdrv module is around 180K, and it's for kernel 2.4; no source is provided in the GPL sources. Interestingly /proc/sys/kernel/tainted reports 0.
So looking at the public spec sheets does not make me particularly hopeful. While they can operate in PHY only mode, that is not done through a seperate chip, it just bypasses the MAC in their controller!! That means there is no cost savings in going MII/GMII over PCI based, so on a platform that has spare PCI interfaces it seems unlikely anyone would do that.
I had been hopeful they were hooking one of the MoCA controllers to NPE-A via MII, but that seems very unlikely 8-(
