Verizon Fios G3100 and E3200 Research
#30
@ZerBea thanks for the response and thanks for all of your work! @RealEnder did a great job with imeigen, and has added the T-Mobile hotspot further reducing the possible candidates.  I haven’t had a chance to check out the T-Mobile devices you shared to see if they have the same weakness.  

Fios-F1nDr has been updated to differentiate between ARC-XCI55AX, CR1000, CME1000, E3200 for the DC:F5:1B MAC prefix.  There is still a bit more of this type of work I have to do to the script, but overall it’s working well.

This week I was able to add 63 new entries to the password database.

Updated Data Set: 
.xlsx   router_data_FULL_072325.xlsx (Size: 898 KB / Downloads: 3)


The Dataset now contains:
G3100/E3200 - 697 entries
CR1000 A/B - 177 entries
ARC-XCI55AX - 143 entries
ASK-NCQ1338 - 165 entries
WNC-CR200A - 62 entries
G1100 - 374 entries
NVG558HX - 67 entries
Other - 143 entries
Total - 1828 entries

[Image: attachment.php?aid=1358]

We caught a new device too, the XC46BE, which is also manufactured by Arcadyan.  The device teardown shows a variety of chips. I believe the Mediatek MT6990V is the ARM CPU, but I couldn’t find much info. The device QR code and sticker provide a great bit of info.
Code:
('WIFI:S:Verizon_TC9CP6;T:WPA;P:bet9nearly8mane;;ROUTER:M:XC46BE;S:ACS44201412;D:11-26-2024;F:3.4.0.5;P:4CLBGTZS7;E:357632330053454;B:3806E60264DD;;1',)

I found the device using the data from WPA-SEC, here is the info broadcast in the packet. The device is the DRAGON  Cool
Code:
2037f022cf12 Arcadyan XC46BE ACS50602760 DRAGON bc329e001dd811b286012037f022cf10 Verizon_BJ3F49

Model: XC46BE
Manufacture: Arcadyan
Device: DRAGON
Serial Prefix: ACS
MACS: Not enough Info
MAC Prefix: 20:37:F0  38:06:E6
UUID: All entries are bc329e001dd811b28601XXXXXXXXXXXX, where X is 2 less than the broadcast MAC Address
EX: 3806e6801442 = bc329e001dd811b286013806e6801440
SSID: Verizon-XXXX

The 3 password entries I was able to find show that The SSID password is 15 characters, and follows a new format <word><digit><word><digit><word>. So far, these passwords are comprised of a 3-letter, 4-letter, and 6-letter word with single digits. The admin password is 9 character alphanumeric as we’ve seen with a lot of the other devices.

Wi-Fi Name:      Wi-Fi Password:      Len  Admin Password:
Verizon_4XZF4L wed5poem7cherub 15    6KFV7CHGS
Verizon_TC9CP6 bet9nearly8mane   15    4CLBGTZS7
Verizon_94CSGD blouse3cate7ran    15    ZK4TNDJQ6

[Image: attachment.php?aid=1359]

The CSG m106 was also caught in the scrape, which is some sort of Verizon device though it doesn’t have the Verizon/Fios SSID. There QR code is just a link to the CSG website. However, the password is 8 character hex that is actually just the end of the serial number. Unfortunately, this device does not broadcast any ESSID information. I did the normal eBay, FB, OfferUp scrape and caught 19 entries. The serial numbers appear to be a a random 16 character hex, possibly a truncated hash. So I had a script try various user input, as well as Unix Epoch time against the password. There are several hashes that produce the password, but none that produce the full serial, so I suspect they are false positives. @RealEnder found the firmware (https://connectcsg.com/pages/firmware-updates), which extracts nicely...so I checked to see how the SN is being generated.

In the file gl_init we see
Code:
uci set glconfig.general.factory_mac=$(get_default_mac_with_colon)
uci set glconfig.general.factory_sn=$(get_default_sn)

ssid=`uci get glconfig.general.factory_mac | awk -F ":" '{print $(NF-1)$NF}'`
uci set wireless.@wifi-iface[$index].ssid="CSG-${ssid}"
key=`uci get glconfig.general.factory_sn | awk '{print substr($0,9)}'`

So we see The SSID is generated from the MAC, and the key is last 8 characters of the factory_sn. Unfortunately the factory_sn is pulled from NVRAM.

[Image: attachment.php?aid=1360]

The data collected for CSG m106 all have the MAC prefix 94:83:C4, so I checked there in the WPA-SEC data. There are not any CSG entries since they don’t broadcast the information, however there are several GL-SFT1200 that overlap the address space. The firmware for this device is also available (https://dl.gl-inet.com/router/sft1200/stable), extracts cleanly,  and is very similar to the CSG m106 with some minor vendor changes.  In gl_init file for both firmware we see
Code:
ssid_prefix="GL-"${model}
uci set wireless.@wifi-iface[$index].key=goodlife

As the image above shows, devices with the SSID GL-<model> have the default password “goodlife”. The firmware shows other models this applies to AR300M, AR750, B1300, B2200, E750, MT750, S200, S1300, X750, X1200

[Image: attachment.php?aid=1361]

One of the devices that the default password doesn’t work on is the GL-MT3000 and GL-MT6000, which we see has a 10 character alphanumeric password :frown:. We have the firmware for this device too (https://dl.gl-inet.com/router/mt3000/stable and https://dl.gl-inet.com/router/mt6000/stable). The password seems to be pulled from NVRAM.
Code:
wifi_password=`lua /usr/bin/get_unique_password.lua`
lfactory.get_unique_password()

But the guest networks may still have the default password “goodlife"
Code:
set wireless.$1.key=goodlife
set wireless.$guest.device="$device"


Attached Files
.jpeg   XC46BE_2.jpeg (Size: 93.58 KB / Downloads: 40)
.jpeg   CSG_37.jpeg (Size: 122.11 KB / Downloads: 33)
.jpg   GL-SFT1200.jpg (Size: 72.91 KB / Downloads: 31)
.jpeg   GL-MT6000.jpeg (Size: 42.12 KB / Downloads: 29)
Reply


Messages In This Thread
RE: Verizon Fios G3100 and E3200 Research - by FiosFiend - 07-23-2025, 05:09 PM