Monthly Archives: June 2012

Redirecting HTTP to HTTPS in Apache

So you have been using SSL on your Apache website for some time now; to get here you had to do a few things:

  1. You scrubbed your site content to ensure all URLs are using their relative form, e.g. “src=’//images\image.png” or explicitly reference the use of HTTPs.
  2. You have tested for certificate and SSL related problems like mixed content, appropriately tagging cookies as secure.
  3. You have ensured that you follow the best practices guidance for SSL server configuration and verified you get an A on  SSLLabs.

Are you done? Not yet there are a few things left for you to do, the most obvious being redirecting all traffic to the SSL version of your site!

This is easy enough to accomplish but before you do so you should probably monitor your CPU usage during your peak so to ensure you have some headroom. This isn’t likely to be a problem as most web-servers are not CPU bound but it’s always good to check.

Once you know you are OK then it’s just a matter of turning it on which requires:

1. Re-writing HTTP URLs from HTTP to HTTPS via .htaccess  by adding the following lines (or their equivalents):

RewriteEngine On

RewriteCond %{HTTPS} off

RewriteRule (.*) https://%{HTTP_HOST}%{REQUEST_URI} [R=301,L]

2. Restarting Apache

Now just go to your website over HTTP and you will see you are redirected to the HTTPS instance of the site.

Ryan

 

Additional Resources

RewriteRule Flags

Mod_Rewrite

 

 

Was the Flame WSUS attack caused just because of the use of MD5?

This morning I saw a number of posts on Twitter about Flame and the attacks use of a collision attack against MD5.

This flurry of posts was brought on about by Venafi, they have good tools for enterprises for assessing what certificates they have in their environments, what algorithms are used, when the certificates expire, etc. These tools are also part of their suite used for certificate management.

They published some statistics on the usage of MD5, specifically they say they see MD5 in 17.4% of the certificates seen by their assessment tools. Their assessment tool can be thought of as a combination of nmap and sslyze with a reporting module.

Based on this we can assume the certificates they found are limited to SSL certificates, this by itself is interesting but not indicative of being vulnerable to the same attack that was used by Flame in this case.

 

 

Don’t get me wrong Microsoft absolutely should not have been issuing certificates signed using MD5 but the collision was not caused (at least exclusively) by their use of MD5; it was a union of:

  1. Use of non-random serial numbers
  2. Use of 512 bit RSA keys
  3. Use of MD5 as a hashing algorithm
  4. Poorly thought out certificate profiles

If any one of these things changed the attack would have become more difficult, additionally if they had their PKI thought out well the only thing at risk would have been their license revenue.

If it was strictly about MD5 Microsoft’s announcement the other day about blocking RSA keys smaller than 1024 bit would have also included MD5 – but it did not.

So what does this mean for you? Well you shouldn’t be using MD5 and if you are you should stop and question your vendors who are sending you down the path of doing so but you also need to take a holistic look at your use of PKI and make sure you are actually using best practices (key length, serial numbers, etc). With that said the sky is not falling, walk don’t run to the nearest fire escape.

Ryan

RSA keys under 1024 bits and you

Recently Microsoft announced that they will push an update in August that will prevent the use of RSA keys with a bit-length less than 1024.

This has been a change that has been coming for some time, for example Microsoft has not allowed CAs with such small keys in their root program for quite a while. They have been proponents of the key length restrictions in the current Baseline Requirements for SSL/TLS certificates in the CA/Browser Forum.

Despite this the SSL Pulse project still shows that within the top 1 million websites there are still web servers using keys less than 1024 bits in size (5 of them). While I do not know who the issuers of these certificates are (they are not GlobalSign), the Microsoft Root Program had prohibited the issuance of smaller than 1024 bit keys and the new Baseline requirements require that all certificates 1024 bits in length expire before December 2013 so if my guess is these would have naturally expired by then.

This patch will cause those certificates to fail to validate once applied.

Microsoft is adverse to “breaking things” they do not control; it’s easy to understand why. With that in mind we can we can assume the decision to release this policy in the form of a patch is clearly based on the recent PKI issues used by Flame which was only possible because of a its use of weak crypto.

But what does this mean for you? Well in the context of getting TLS certificates from a CA who is trusted by Microsoft or Mozilla you shouldn’t be impacted at all since no CA should have been issuing certificates with such small keys.

That said in an enterprise it’s certainly possible such certificates exist and if you have an internal PKI you need to ensure they don’t in yours before this patch gets deployed in August or those systems will stop working.

Microsoft provides some guidance on how to look for these certificates but your likely better off running a tool like Venafi Accessor against your environment to find what keys and certificates you have deployed.

They release their “assessment tool” for free as a means to market to you their certificate management products, the marketing material is alarmist (and in some cases a little misleading) but the assessment tool I good and will be useful to organizations who need to understand the implications to their environment.

Ryan

Testing OCSP Stapling

So you have configured OCSP stapling and you want know if it’s actually working, it’s easy enough to check using the openssl  s_client command:

openssl s_client -connect login.live.com:443 -tls1  -tlsextdebug  -status

Loading ‘screen’ into random state – done

CONNECTED(0000017C)

TLS server extension “status request” (id=5), len=0

….

OCSP response:

======================================

OCSP Response Data:

OCSP Response Status: successful (0x0)

Cert Status: good

This Update: Jun 12 02:58:39 2012 GMT

Next Update: Jun 19 02:58:39 2012 GMT

In this example you see that the client is requesting the servers OCSP response, you then see the server providing that response successfully and openssl determining the servers certificate is good.

For another example we can query the US Mint’s website for an example of a site that has not yet (and probably won’t for some time since it’s a government site) configured OCSP stapling:

openssl s_client -connect www.usmint.gov:443 -tls1  -tlsextdebug  -status

Loading ‘screen’ into random state – done

CONNECTED(0000017C)

OCSP response: no response sent

….

Hope this helps you deploy OCSP Stapling successfully.

Ryan

OCSP Stapling in IIS

Windows Server 2008 and later support a feature called OCSP stapling. When enabled a server pre-fetches the OCSP response for its own certificate and delivers it to the user’s browser during the TLS handshake. This approach offers a privacy advantage. But, the main benefit is the browser doesn’t have to make a separate connection to the CA’s revocation service before it can display the Web page. This gives better performance and reliability.

For this pre-fetching to work the web-server certificate needs to contain a pointer to the OCSP responder, this is a best practice and a recommendation of the CA/Browser Forums baseline requirements so it’s almost certain your certificate has it.

Unlike Apache this feature is enabled by default, it’s possible your servers are already doing OCSP stapling and you do not even know it.

With that said chances are you have a firewall between your webservers and the internet; it’s also likely that firewall disallows outbound connections from your servers unless explicitly allowed. So you might need to allow your web servers to communicate with the OCSP responder before it will work.

To figure out what host and port you will need to open you will need to look at the certificates you use on your webserver; one way to do that is to browse to your current site and view the details of the certificates you are currently using, for example:

   

The value you want is in the “Authority Information Access” field, you want the ones (there may be multiple) that have the “Access Method” of “On-line Certificate Status Protocol”.

Once these two conditions are met OCSP Stapling will “Just work” there is nothing else you need to do.

Ryan

OCSP Stapling in Apache

Apache 2.3 and later support a feature called OCSP stapling. When enabled a server pre-fetches the OCSP response for its own certificate and delivers it to the user’s browser during the TLS handshake. This approach offers a privacy advantage. But, the main benefit is the browser doesn’t have to make a separate connection to the CA’s revocation service before it can display the Web page. This gives better performance and reliability.

For this pre-fetching to work the web-server certificate needs to contain a pointer to the OCSP responder, this is a best practice and a recommendation of the CA/Browser Forums baseline requirements so it’s almost certain your certificate has it.

Chances are you have a firewall between your webservers and the internet; it’s also likely that firewall disallows outbound connections from your servers unless explicitly allowed. So before you enable OCSP stapling you are going to need to allow your web servers to communicate with the OCSP responder.

To figure out what host and port you will need to open you will need to look at the certificates you use on your webserver; one way to do that is via OpenSSL, for example:

1. Get the certificate using s_client

openssl.exe s_client -connect www.globalsign.com:443


—–BEGIN CERTIFICATE—–
—–END CERTIFICATE—–

You need to copy the PEM header and footer (“—–BEGIN/END CERTIFICATE—–“) and the Base64 between them into a file.

2. Identify the OCSP responders within the server certificate

openssl.exe x509 -in globalsign.com.cer -text


X509v3 extensions:
..
Authority Information Access:
CA Issuers – URI:
http://secure.globalsign.com/cacert/gsextendvalg2.crt
OCSP – URI:
http://ocsp2.globalsign.com/gsextendvalg2

You need to find the “OCSP – URI” section, in the example certificate above the OCSP responder is http://ocsp2.globalsign.com/gsextendvalg2 there may be multiple responders specified, you should allow your servers to initiate outbound traffic to all of them.

Once your servers can request OCSP responses enabling stapling is very straight forward, there are just two directives that need to be added, these directives can be global or specific to a specific to one instance:

SSLUseStapling on
SSLStaplingCache “shmcb:logs/stapling_cache(128000)”

Their purpose is self-evident; SSLUseStapling turns the feature on while SSLStaplingCache specifies where to store the cache and how big it should be.

There are other directives also you can use but you should not need to worry about them.

As long as you are running the most recent stable versions of Apache and OpenSSL enabling this feature is safe. It is only used when the client supports it so there won’t be compatibility issues and if the server for some reason fails to populate its cache with a valid OCSP response the client will typically fall back to doing a live OCSP request on its own.

Ryan

 

Additional Resources

Overclocking Mod_SSL

Flame and Certificate Revocation

Microsoft has published patches that insert the CAs directly associated with the Terminal Services Licensing PKI into the “Untrusted” certificate store, this has the same effect as revoking the certificates for those that:

  1. Apply the patch
  2. Use Microsoft’s chain validation engine

But what about those that do not apply the patch? Or those that simply import the roots from the Microsoft store to use in another engine (like NSS or OpenSSLs, one example of such an application would be Adobe X)?

Well today those people will still trust certificates issued by those certificates, they potentially would not though if Microsoft published revocation information for those CAs.

As it stands most of the certificates in question do contain revocation pointers (a CRLdp extension in this case) but the URLs they point to are invalid, for example:

  1. “Microsoft Enforced Licensing Intermediate PCA” – No revocation information
  2. “Microsoft Enforced Licensing Registration Authority CA” – http://crl.microsoft.com/pki/crl/products/MicEnfLicPCA_12-10-09.crl
  3. “Microsoft LSRA PA” –  http://crl.microsoft.com/pki/crl/products/MicEnfLicRegAutCA_2009-12-10.crl
  4. “Terminal Services LS” – http://tk5paxprdsrv09.partners.extranet.microsoft.com/CertEnroll/Microsoft%20LSRA%20PA.crl

This may not be beneficial for the attack vector used in Flame (I do not know if it even does revocation checking – though it very probably does) but it would certainly help other cases.

Today the invalid URLs in these certificates would result in a “Unknown or Offline” response which would likely be ignored by applications due to Microsoft’s “soft-fail revocation checking” defaults.

Even with those defaults though if they published these CRLs clients would get some benefit, it would:

  1. Reduce risk until the patch could get applied.
  2. Help others who mistakenly trusted the Microsoft roots when they “imported” them from the “Trusted Root Store”.

It’s my hope Microsoft decides to publish these CRLs for relying party applications to rely on.

[3:22 PM 6/19/2012] FYI as of 6/19 the MicEnfLicPCA_12-10-09.crl is published (good), it appears to have been created June 14th several days after this post. The other two are still invalid, it’s likely they decided against publishing them since a well behaved client would fail with just one of the CAs being bad. It’s not a unreasonable decision, I am guessing the URL choices for the CRLs made it difficult to do all of the CRLs.

Ryan

Flame was just one use of the Terminal Services Licensing PKI

I wanted to do a post on how the Terminal Services Licensing PKI could be used in another attack; though the variants are endless I have one concrete example of how this was used in in an attack in 2002 in this certificate:

   
   

 

What we see in this certificate is that the attacker took the License Certificate (Terminal Services LS) and issued himself a certificate underneath it. This only worked because of another defect in the Windows chain engine, this one relating to how it handled Basic Constraints (see MS 02-050) – that was fixed in 2003.

But why would he have done this? I see two possible reasons:

  1. Bypassing the terminal services licensing mechanism.
  2. Bypassing some client authentication scheme based on X.509.

We won’t explore #1 as I am not interested in the attacks on license revenue for Microsoft but how this infrastructure designed to protect that license revenue was used to attack customers and as such I will focus on #2.

Now, before I go much further I have to be clear – I am just guessing here, I don’t have enough data to be 100% sure of what this was used for but it seems pretty likely I am right.

What I see here is the X.509 serial number is different than the Serial Number RDN the attacker put into the Subject’s DN.

This certificate also has a value in the CN of the Subject DN that looks like it is a user name (PORTATILE).

Neither of these should be necessary to attack the licensing mechanism as legitimate license certificates don’t have these properties or anything that looks roughly like them.

My theory is that the attacker found some Windows Server configured to do client authentication (possibly for LDAPS or HTTPS) and wanted to trick the server into mapping this certificate into the principal represented by that CN and Serial Number in the DN.

Why would this have worked? Well as I said there was the Basic Constraints handling defect, but that wouldn’t have  been enough the server would have had to had blindly trusted all the root certificates in the root store for “all usages”.

This was a fairly common practices back then, web servers and other server applications would be based on other libraries (for example SSLEAY/OpenSSL) and just import all the roots in the CryptoAPI store and use them with this other library.

This is actually still done by some applications today, for example Adobe Reader X:

 

The problem with this behavior is that these applications do not necessarily (read almost never) import the properties associated with these certificates, nor do they support the same enforcement constraints (for example Nested Extended Key usages).

The net effect of which is that when they import the roots they end up trusting them for more than the underlying Windows subsystem does.

The attackers in this case were likely taking advantage of this bad design of this client authentication solution as well as the bad design of the Terminal Services Licensing PKI.

The combination of which would have allowed the attacker to impersonate the user PORTATILE and perform any actions as that user they liked.

Even if I am wrong in this particular case, this sort of thing was definitely possible.

What lessons should we take away from this?

For one always do a formal security review of your solutions as you build them as well as before you deploy them; pay special attention to your dependencies – sometimes they are configuration as well as code.

Also if an issue as fundamental as the Terminal Services Licensing PKI can fly under the radar of a world class security program like Microsoft’s you probably have some sleepers in your own infrastructure too – go take a look once more for the assumptions of the past before they bite you too.

Ryan

How did the Terminal Services Licensing PKI effect you?

The other day I did a post on the age of the Microsoft PKI that was used for Terminal Services Licensing, today I thought I would talk about what that age meant in the context of the vulnerabilities it introduced.

The oldest certificate I have been able to find from the same hierarchy is from April 1999 (that’s the issuance date for the Microsoft Enforced Licensing PCA).

Based on the post from the Microsoft Security Research & Defense blog  we know that the reason the attacker had to do the MD5 collision was that as of Vista there was a change in the way critical extensions were handled.

This change made it so that Vista clients would fail when they saw a certificate that contained an unknown critical extension in a signing certificate making it an ineffective attack vector for those clients.

But what does this mean for the period of time before that? Well Windows Vista was released in November of 2006, that’s nearly 8 years in which any enterprise with a terminal services deployment could have owned a Windows PC “As Microsoft” or potentially attack a PKI based system with “trusted” but fraudulent certificate.

But did it really get better in with the release of VISTA? According to StatCounter Windows Vista received its maximum market share of 23% in October of 2009. Yes, two years after the release of Vista 77% of the Windows clients on the Internet were still vulnerable as a result of the design of the terminal services licensing solution.

Things didn’t really start to get better until XP SP3 which was released in April of 2008 as it contained the same certificate validation engine that was found in Vista.

While I do not have any public statistics I can share I can say that this service pack was picked up faster than any other service pack up until that point which says a lot since it was not a “forced” update.

If we are optimistic and say that it took only one year to get 100% penetration and we believe stat counters statistics for XPs Market share of 71% in April 2009 that it took till 2009 to get to 95% patched.

Now these numbers are just for clients on the Internet and not servers. This “fixed” chain validation engine wouldn’t have even found its way into the Windows Server code base until Windows Server 2008 which was released in February 2008 but took some time to get broadly used.

While Windows Servers are not terribly common on the Internet they are extremely common in the Enterprise, especially in 2009 where it had 73.9% market share. Again I don’t have numbers but antidotal we know that Enterprises are notoriously slow to upgrade or patch.

So where does this leave us in 2012? Still vulnerable that’s where.

The other day Microsoft released a patch that in-essence revokes the PKI in question and today WSUS announced their patch that introduces additional pinning.

You need to apply both to secure your systems.

Ryan

How one typically verifies code comes from Microsoft

Microsoft signs all code that ships from it (that is except for the case when it doesn’t); so how does one verify that the code came from them vs. anyone with a fraudulent certificate claiming to be them?

Well the answer is pinning, but the pinning for the most part does not happen at the signing certificate level as those keys change to frequently to be pinned. The pinning happens at the certificate issuer level, much like has been proposed by CAA.

So to understand how this happens in code you need to understand code signing in Windows a little. There are two types of signatures those that are “embedded” and those that are “attached”.

Authenticode would be an example of an embedded signature; Catalog Files on the other hand are detached signatures. There are other examples of both of these but in general these other formats have the same abstract properties.

So when do you use one vs. the other? Well if a user will download your binary directly (ActiveX, Setup, etc.) you would typically embed a signature.

If that’s not the case you would use a detached signature because a single signature can be placed on a list of binaries without increasing the package size (it’s one signature on n hashes). You produce these signatures with SignerSignEx.

For the two signature formats I mentioned you do your signature verifications using an API called WinVerifyTrust. You can also use this API to get information out of a signature (who signed it, when, etc.).

With that information you do a certificate validation that is accomplished with the CertGetCertificateChain API, that gives you a CERT_CHAIN_CONTEXT which you can use to call CertVerifyCertificateChainPolicy.

This API is used to do application specific verifications, for example in the case of TLS the CERT_CHAIN_POLICY_SSL provider is used – this is where the name and EKU verification logic for Windows exists.

The providers we care about for our story today are:

  • CERT_CHAIN_POLICY_AUTHENTICODE – Is this certificate appropriate for code-signing?
  • CERT_CHAIN_POLICY_AUTHENTICODE_TS – Is this certificate appropriate for time stamping code?
  • CERT_CHAIN_POLICY_MICROSOFT_ROOT – Is this certificate ultimately issued by a Microsoft Product Root?

The last one is the one in has an array of hashes in it, I don’t recall if they are key hashes or thumbprints of the certificate. In essence what it does is look at the top of the chain to see if the Root matches one of these hashes if it does it passes, if it doesn’t it fails.

Applications may choose to do additional “pinning” also, for example in the case of Windows Update Services a manifest is exchanged between the client and server, that manifest comes down over SSL they may choose to also do pinning to a set of certificates for the SSL session or apply digital signatures to the manifest and do pinning for the keys that are used to do signing.

I do not know personally for a fact that any such pinning takes place but I have been told by several people I respect that it does.

It looks as if Windows Update was not doing pinning but as of this week it will, this is of course  a best practice and I am glad to see them doing it now.

So now you know how to validate that code is coming from Microsoft, or at least how it’s supposed to work.

Issues like represented in MSRC 2718704 / Flame put some of these assumptions in question, it’s clear that pinning is an important concept though and one that should be supported more broadly.

Ryan