diff options
-rw-r--r-- | SECURITY | 34 |
1 files changed, 23 insertions, 11 deletions
@@ -4,53 +4,65 @@ custodian-enqueue ----------------- - We open named files from the user to parse tests. The configuration file(s) - are currently opened by root, via /etc/service/custodian-enqueue. This should - be changed to run as a lower-privileged user. +We open a single named file so that we may parse tests. There seems no harm in this, +because the user must have permission to parse the file. +To successfully enqueue jobs that have been parsed the user must be able to talk to the +beanstalkd queue - but this happens over TCP, so there is no need to run the parsing +tool as root, and this is generally discouraged. + +Conclusion: The enqueuing process contains no security risks. custodian-dequeue ----------------- -One test passes arguments from the configuration file to the shell: +Only one test ultimately passes arguments from the queue/configuration file to a system command: ping -The hostname used to ping will be used assuming it matches the following regular expression: +The hostname passed to the ping-tool will initially be matched against the regular expression: -^([^\s]+)\s+ + ^([^\s]+)\s+ So the following configuration file snippet potentially allows a command to be executed by our worker: $(/home/steve/hg/custodian/exploit.sh) must ping otherwise "Owned". -Given that anybody who can talk to the beanstalkd server can submit jobs we cannot rely on catching this solely in the parser. +Given that anybody who can talk to the beanstalkd server can submit jobs we cannot rely on catching +this solely in the parser. For the moment we've solved the case of the ping-exploitation, by validating -that hostnames passed to the multi-ping command match ^[a-z0-9.-]$ - both forms -of input are validated: +that hostnames passed to the multi-ping command match ^[a-z0-9.-]$ - in both +possible forms: * Ensuring the hostname is valid prior to executing the shell command. * Ensure the hostname is valid before adding the job to the queue. +We are careful to run the system command in the Ruby array-format, which will bypass shell +exploitation. + +Conclusion: There are no obvious security implications in the dequeuing process. General ------- -We decode arbitrary jobs from the queue. We should sign them, or validate them to prevent trojan malformed lines from being added. +We pull arbitrary jobs from the queue, and it is possible an attacker could add malicious +entries. +We could sign tests to prevent trojan malformed lines from being processed, but this is +not yet done. (It isn't obvious if this would be a sensible addition either.) TODO ---- - Anything else? DoS attacks? +Anything else? DoS attacks? |