Netgear CG3100D-2 investigation

The Netgear CG3100D-2 is the default cable-modem you get for Telstra Cable, at least at one time. Having retired it after changing service providers, I wanted to see if it was somewhat able to be re-purposed.

In short it's hackability is low.

First thing was to check out the Netgear Open Source page to see if the source had anything interesting. There is some source, but honestly when you dig into the platform code and see things like kernel/linux/arch/mips/bcm963xx/setup.c:

 * C++ New and delete operator functions

/* void *operator new(unsigned int sz) */
void *_Znwj(unsigned int sz)
    return( kmalloc(sz, GFP_KERNEL) );

/* void *operator new[](unsigned int sz)*/
void *_Znaj(unsigned int sz)
return( kmalloc(sz, GFP_KERNEL) );

there's a bit of a red-flag that this is not the cleanest code in the world (I guess it interfaces with some sort of cross-platform SDK written in some sort of C++).

So next we can open it up, where it turns out there are two separate UARTs as shown in the following image.

UART connections on Netgear CG3100D 2BPAUS

One of these is for the bootloader and eCOS environment, and the other seems to be connected to the Linux side.

A copy of the boot-logs for the bootloader and eCOS and Linux don't show anything particuarly interesting. The Linux boot does identify itself as Linux version 2.6.30-V2.06.05u while the available source lists its version as 2.6.30- so it's questionable if the source matches whatever firmware has made it onto the modem.

We do see that this identifies as a BCM338332 which seems to be one of the many sub-models of the BCM3383 SoC cable-modem solution. There is an OpenWrt wiki page that indicates support is limited.

Both Linux and eCos boot to a login prompt where all the usual default combinations of login/passwords fail. So my next thought was to try and get to the firmware via the bootloader, which has a simple interface

BCM338332 TP0 346890
Reset Switch - Low GPIO-18 50ms
MemSize:            128 M
Chip ID:     BCM3383G-B0

BootLoader Version: 2.4.0alpha14R6T Pre-release Gnu spiboot dual-flash reduced DDR drive linux
Build Date: Mar 24 2012
Build Time: 14:04:50
SPI flash ID 0x012018, size 16MB, block size 64KB, write buffer 256, flags 0x0
Dual flash detected.  Size is 32MB.
parameter offset is 49944

Signature/PID: a0e8

Image 1 Program Header:
   Signature: a0e8
     Control: 0005
   Major Rev: 0003
   Minor Rev: 0000
  Build Time: 2013/4/18 04:01:11 Z
 File Length: 3098751 bytes
Load Address: 80004000
    Filename: CG3100D_2BPAUS_V2.06.02u_130418.bin
         HCS: 1e83
         CRC: b95f4172

Found image 1 at offset 20000

Image 2 Program Header:
   Signature: a0e8
     Control: 0005
   Major Rev: 0003
   Minor Rev: 0000
  Build Time: 2013/10/17 02:33:29 Z
 File Length: 3098198 bytes
Load Address: 80004000
    Filename: CG3100D_2BPAUS_V2.06.05u_131017.bin
         HCS: 2277
         CRC: a6c0fd23

Found image 2 at offset 800000

Image 3 Program Header:
   Signature: a0e8
     Control: 0105
   Major Rev: 0002
   Minor Rev: 0017
  Build Time: 2013/10/17 02:22:30 Z
 File Length: 8277924 bytes
Load Address: 84010000
    Filename: CG3100D_2BPAUS_K2630V2.06.05u_131017.bin
         HCS: 157e
         CRC: 57bb0175

Found image 3 at offset 1000000

Enter '1', '2', or 'p' within 2 seconds or take default...
. .

Board IP Address  []: 
Board IP Mask     []:
Board IP Gateway  []:
Board MAC Address [00:10:18:ff:ff:ff]:

Internal/External phy? (e/i/a)[a]
Switch detected: 53125
ProbePhy: Found PHY 0, MDIO on MAC 0, data on MAC 0
Using GMAC0, phy 0

Enet link up: 1G full

Main Menu:
  b) Boot from flash
  g) Download and run from RAM
  d) Download and save to flash
  e) Erase flash sector
  m) Set mode
  s) Store bootloader parameters to flash
  i) Re-init ethernet
  p) Print flash partition map
  r) Read memory
  w) Write memory
  j) Jump to arbitrary address
  X) Erase all of flash except the bootloader
  z) Reset

Flash Partition information:

Name           Size           Offset
bootloader   0x00010000     0x00000000
image1       0x007d0000     0x00020000
image2       0x007c0000     0x00800000
linux        0x00800000     0x01000000
linuxapps    0x00600000     0x01800000
permnv       0x00010000     0x00010000
dhtml        0x00200000     0x01e00000
dynnv        0x00040000     0x00fc0000
vennv        0x00010000     0x007f0000

The "read memory" seems to give you one byte at a time and I'm not certain it actually works. So I think the next step is solder some leads to dump out the firmware from the flash-chip directly, which is on the underside of the board. At that point, I imagine the passwords would be easily found in the image and you might then be able to leverage this into some sort of further hackability.

If you want a challenge and have a lot of time on your hands, this might be your platform — but practically I think the best place for this is the recycling bin.


I find it very useful to spend 5 minutes a day to keep a small log of what was worked on, major bugs or reviews and a general small status report. It makes rolling up into a bigger status report easier when required, or handy as a reference before you go into meetings etc.

I was happily using an etherpad page until I couldn't save any more revisions and the page got too long and started giving javascript timeouts. For a replacement I wanted a single file as input with no boilerplate to aid in back-referencing and adding entries quickly. It should be formatted to be future-proof, as well as being emacs, makefile and git friendly. Output should be web-based so I can refer to it easily and point people at it when required, but it just has to be rsynced to public_html with zero setup.

rstdiary will take a flat RST based input file and chunk it into some reasonable looking static-HTML that looks something like this. It's split by month with some minimal navigation. Copy the output directory somewhere and it is done.

It might also serve as a small example of parsing and converting RST nodes where it does the chunking; unfortunately the official documentation on that is "to be completed" and I couldn't find anything like a canonical example, so I gathered what I could from looking at the source of the transformation stuff. As the license says, the software is provided "as is" without warranty!

So if you've been thinking "I should keep a short daily journal in a flat-file and publish it to a web-server but I can't find any software to do just that" you now have one less excuse.

Bash arithmetic evaluation and errexit trap

In the "traps for new players" category:

things="0 1 0 0 1"

for i in $things;
   if [ $i == "1" ]; then
       (( count++ ))

echo "Count is ${count}"

Looks fine? I've probably written this many times. There's a small gotcha:

The expression is evaluated according to the rules described below under ARITHMETIC EVALUATION. If the value of the expression is non-zero, the return status is 0; otherwise the return status is 1. This is exactly equivalent to let "expression".

When you run this script with -e or enable errexit -- probably because the script has become too big to be reliable without it -- count++ is going to return 0 (post-increment) and per above stop the script. A definite trap to watch out for!

Finding out if you're a Rackspace instance

Different hosting providers do things slightly differently, so it's sometimes handy to be able to figure out where you are. Rackspace is based on Xen and their provided images should include the xenstore-ls command available. xenstore-ls vm-data will give you a handy provider and even region fields to let you know where you are.

function is_rackspace {
  if [ ! -f /usr/bin/xenstore-ls ]; then
      return 1

  /usr/bin/xenstore-ls vm-data | grep -q "Rackspace"

if is_rackspace; then
  echo "I am on Rackspace"

Other reading about how this works:

ip link up versus status UP

ip link show has an up filter which the man page describes as display running interfaces. The output also shows the state, e.g.

$ ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default
  link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
  link/ether e8:03:9a:b6:46:b3 brd ff:ff:ff:ff:ff:ff
3: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DORMANT group default qlen 1000
  link/ether c4:85:08:2a:6e:3a brd ff:ff:ff:ff:ff:ff

It is not described what the difference is between these two.

The state output is derived from IFLA_OPERSTATE as per this little bit in ip/ipaddress.c in iproute2

  print_operstate(fp, rta_getattr_u8(tb[IFLA_OPERSTATE]));

as per operstates.txt a status of UP means that the interface is up and can be used to send packets.

In contrast, ip link show up runs a different filter

if (filter.up && !(ifi->ifi_flags&IFF_UP))
  return 0;

Thus only interfaces that are in state IFF_UP are shown. This means it is administratively up, but this does not mean it is available and ready to send packets.

This may make a difference if you're trying to use the output of ip link in any sort of automation where you have to probe the available network-interfaces and know where packets are going to get out or not.

Non-stick tortilla technique

I have never had much success with tortilla making, generally finding the fragile masa would stick to everything and tear easily. However I was listening to the Cook's Illustrated podcast where they mentioned that cooking a pizza on a stone with silicone-paper makes no difference over doing it on the stone directly. For some reason I had just never thought of cooking it with the silicone-paper on; I'd always frustrated myself trying to separate the pressed tortilla in various ways.

So here's my technique:

  1. Mix the masa as on the packet; 2 cups to 2-half cups water
  2. Leave for 20 minutes or so
  3. Use plastic-wrap on the top of the tortilla press
  4. Cut a silicone-paper square for the bottom
  5. Ball the masa and press
  6. Now the plastic-wrap will peel easily off the top of the uncooked tortilla
  7. Take the silcone-paper and put the tortilla down on the pan/hotplate
  8. Give a slight press with the spatula around the edges
  9. After about 20 seconds, the silicone-paper should peel off the tortilla fairly easily.
  10. repeat!

I've tried mixing in suet, oil and copha and none of them are worth the extra calories over just plain masa in my opinion and don't help with sticking anyway. Enjoy!

Gyro Wheel Battery Pack

Several years ago I bought a Gyro Wheel for my daughter, who probably couldn't even walk at the time. The idea seemed really good; a heavy cast-iron wheel is spun up and the angular momentum keeps the new-rider upright while they learn to ride.

I don't suppose it revolutionised bike-riding, as it seems that the company has disappeared. Now my daughter can reach the pedals, when I finally dug it out the other day it was sad to find it wouldn't charge. Googling seems to show that it's a fairly common problem.

After pulling it apart, I found that it was fairly straight-forward to replace the battery-pack. Getting in requires a torx screwdriver but is otherwise straight-forward.

The battery looks like a bit of a custom job, with two-halves of a 9.6V battery-pack split into two. You won't be able to buy anything that directly replaces this, but you should be able to find individual AA-sized NiMH rechargable batteries with welded-tabs that you can use to create a new battery-pack (don't buy batteries without tabs and try to solder directly onto the battery terminals, it won't work). Unwrap the plastic from the existing packs and just follow the polarity; if you carefully de-solder the existing wires you can save yourself a lot of work making new ones. Wrap it up securely and put everything back together.

Below are some photos just to give you an idea of what to expect.

Tenvis IP391W meta-page

Recently I purchased a Tenvis IP391W-HD camera.

I would be unlikely to recommend it. The price is certainly right and the picture quality is quite good. The Android and iPhone apps do work to watch the stream live.

However, the interface is terrible and almost useless without Internet Explorer. There is a RTSP stream (rtsp://admin:password@ip) which VLC can seem to handle, but not mplayer. The recording format (.h264) is not viewable by VLC or mplayer and all I could find is a Windows .exe to convert them to an .avi.

The motion detection gets troubled by the dark. It would really only be useful for something permanently well-lit. It did send me emails via gmail.

I have got it recording to a NFS server, but I don't have a lot of confidence in the reliability of it. I think I have it configured to record in 3600-second blocks (given the interface, it's hard to tell if I've set it up to the network, or to the internal flash, etc), but it seems to intersperse 60 minute recordings with random small recordings. Given the whole idea of a security camera is to record the unexpected, you want a lot of confidence you're actually recording, which you don't get with this. You can see below it recorded 3 hour blocks, then started going a little crazy...

-rw-r--r-- 1 nobody nogroup  69M Mar 11 01:25 0-003035.v264
-rw-r--r-- 1 nobody nogroup  69M Mar 11 02:25 0-013049.v264
-rw-r--r-- 1 nobody nogroup  69M Mar 11 03:26 0-023103.v264
-rw-r--r-- 1 nobody nogroup 5.9M Mar 11 03:31 0-033117.v264
-rw-r--r-- 1 nobody nogroup 1.5M Mar 11 03:40 0-034350.v264
-rw-r--r-- 1 nobody nogroup  17M Mar 11 04:02 0-035259.v264
-rw-r--r-- 1 nobody nogroup 306K Mar 11 04:10 0-041548.v264
-rw-r--r-- 1 nobody nogroup 4.9M Mar 11 04:23 0-042457.v264

There is a support forum, where I found the following files scattered in various posts. From what I can tell, they are the latest as of this writing. I can confirm they work with my IP391W-HD, which the system tells me is GM8126 hardware and came with firmware

  • - firmware (b56f211a569fb03a37d13b706c660dcb)
  • web.pk2 - a UI update that includes dropbox support. This is really for the model that has pan and tilt, so those buttons don't work. (0e42e42bd6f8034e87dcd443dcc3594d)
  • V264ToAVIen.exe - converts the output to an AVI file that mplayer will play (with some complaints) (9c5a858aa454fed4a0186cf244c0d234) offers free limited-time Windows VM's which will work to upload this firmware. Just make sure you use a bridged network in the VM; I'm guessing the firmware ActiveX control tells the camera to TFTP the data from it, which doesn't work via NAT.

Somewhat worryingly, you can telnet to it and get a login prompt (TASTECH login). So it has a built-in backdoor you can't disable.

There have been some efforts to hack the device. did an excellent job reverse engineering the .pk2 format and writing tenvis_pack.c (no license, I'm generously assuming public domain). I used this to recreate the firmware above with a telnet daemon listening with a shell on port 2525 (no password, just telnet to it)

It's interesting to poke around, but it seems like the whole thing is really driven by a binary called ipc8126

/ # ipc8126 --help
*** Version:
*** Release date: 2013-08-05 15:48:32

In general, I'd say hackability is quite low.

Warning : any of the above might turn your camera into a paperweight. It worked for me, but that's all I can say...