tags: all,

apache, apachecon, apacheweek, bryce, cve, fedora, fudcon, geocaching, gps, ha, jabber, metrics, microsoft, nashville, north carolina, oscon, red hat summit, security, trips

The graph below shows the total number of security updates issued for Red Hat Enterprise Linux 5 Server starting at 5.1 up to and including the 5.2 release, broken down by severity. I've split it into two columns, one for the packages you'd get if you did a default install, and the other if you installed every single package (which is unlikely as it would involve a bit of manual effort to select every one). So, for a given installation, the number of packages and vulnerabilities will probably be somewhere between the two.

So for a default install, from release of 5.1 up to and including 5.2, we shipped 46 updates to address 119 vulnerabilities. 8 advisories were rated critical, 24 were important, and the remaining 14 were moderate and low.

For all packages, from release of 5.1 to and including 5.2, we shipped 62 updates to address 179 vulnerabilities. 9 advisories were rated critical, 29 were important, and the remaining 24 were moderate and low.

The nine critical updates were in five different packages:

1. Four updates to Firefox (November, February, March, April) where a malicious web site could potentially run arbitrary code as the user running Firefox. Given the nature of the flaws, ExecShield protections in RHEL5 should make exploiting these memory flaws harder.
2. An update to the GnuTLS library (May), where a remote attacker who can connect to a server making use of GnuTLS could cause a buffer overflow. In Red Hat Enterprise Linux 5, the CUPS print server uses GnuTLS.

3. An update to MIT Kerberos (March), where a remote attacker who can conect to the krb5kdc or kadmind services could cause a buffer overflow.

4. An update to OpenPegasus (January), where a remote attacker who can connect to OpenPegasus could cause a buffer overflow. The Red Hat Security Response Team believes that it would be hard to remotely exploit this issue to execute arbitrary code, due to the default SELinux targeted policy, and the default SELinux memory protection tests.

5. Two updates to Samba (November, December) where a remote attacker who can connect to the Samba port could cause buffer overflows. In addition to ExecShield making this harder to exploit, the impact of any sucessful exploit would be reduced as Samba is constrained by an SELinux targeted policy (enabled by default).

Updates to correct all of these critical issues were available via Red Hat Network either the same day, or one calendar day after the issues were public.

To get a better idea of risk we need to look not only at the vulnerabilities but also the exploits written for those vulnerabilities. A proof of concept exploit exists publicly for one of the Samba flaws, CVE-2007-6015, but we are not aware of public exploits for any other of those critical vulnerabilities. Also of high risk was an important "zero-day" exploit affecting the Linux kernel where a local unprivileged user could gain root privileges. Red Hat Enterprise Linux 5.1 was affected and a fix was available two calendar days after public disclosure.

Red Hat Enterprise Linux 5 shipped with a number of security technologies designed to make it harder to exploit vulnerabilities and in some cases block exploits for certain flaw types completely. For the period of this study there were two flaws blocked that would otherwise have required updates:

1. A double-free flaw in CUPS. The glibc pointer checking limited the exploitability of this issue to just a crash of CUPS and not the ability to execute arbitrary code. code execution. We still issued an update, as a remote attacker could trigger this flaw and cause CUPS to crash.
2. An uninitialized pointer free flaw in unzip, caught by the glibc pointer checking. As exploitation of this flaw results in just a crash of a user application, no updates were needed.

This data is interesting to get a feel for the risk of running Enterprise Linux 5 Server, but isn't really useful for comparisons with other versions or distributions -- for example, a default install of Red Hat Enterprise 4AS did not include Firefox. You can get the results I presented above for yourself by using our public security measurement data and tools, and run your own custom metrics for any given Red Hat product, package set, timescales, and severities.

My full risk report was published yesterday in Red Hat Magazine and reveals the state of security since the release of Red Hat Enterprise Linux 4 including metrics, key vulnerabilities, and the most common ways users were affected by security issues.

"Red Hat knew about 49% of the security vulnerabilities that we fixed in advance of them being publicly disclosed. For those issues, the average notice was 21 calendar days, although the median was much lower, with half the private issues having advance notice of 8 days or less."

Using our public tool, for every Red Hat product and service, for 2007 we issued 306 advisories to fix 404 vulnerabilities. Of those 404 vulnerabilities 41 were critical (on the scale used by Microsoft and Red Hat).

Most people are not going to be using every Red Hat product, so taking just Enterprise Linux product you find 348 vulnerabilities, of which 27 were critical. A given user is going to only be vulnerable to the issues that affect the products and packages they have installed. Using the scripts on our pages you can figure it out for your own circumstances. But as an example, the default installation of Red Hat Enterprise Linux 4 AS had 172 vulnerabilities of which 4 were critical.

The Secunia report does actually make it clear you can't use their vulnerability count as a method of comparing platforms, in part due to the differences in methodology of the vendors, but I'm sure this won't stop some press from jumping to conclusions if they don't read the actual report.

I've asked Secunia how they got to their number of vulnerabilities, but in the meantime, a raw count of vulnerabilities is only a small part of the overall risk exposure in using a product. I've got some more reports that go into this in more detail for two years of Enterprise Linux 4 and Enterprise Linux 5.0 to 5.1.

Update: Coverage of this: ZDNet

Update: Secunia told me that they treat each advisory separately; so for example yesterday we issued updates for some moderate severity issues in the Apache Web server, but we did separate advisories for each affected product: Red Hat Enterprise Linux 2.1, 3, 4, 5, Red Hat Application Stack v1, v2. So in this case the same Apache vulnerability would be counted 6 times.

The graph below shows the total number of security updates issued for Red Hat Enterprise Linux 5 Server up to and including the 5.1 release, broken down by severity. I've split it into two columns, one for the packages you'd get if you did a default install, and the other if you installed every single package (which is unlikely as it would involve a bit of manual effort to select every one). So, for a given installation, the number of packages and vulnerabilities will be somewhere between the two extremes.

So for all packages, from release up to and including 5.1, we shipped 94 updates to address 218 vulnerabilities. 7 advisories were rated critical, 36 were important, and the remaining 51 were moderate and low.

For a default install, from release up to and including 5.1, we shipped 60 updates to address 135 vulnerabilities. 7 advisories were rated critical, 26 were important, and the remaining 27 were moderate and low.

• These figures include ten updates we released on the day we shipped 5.0. This was because we froze package updates some months before releasing the product. Only one of those updates was rated critical, an update to Firefox.
• The six other critical updates were:
  1. Three more updates to Firefox (May, July, October) where a malicious web site could potentially run arbitrary code as the user running Firefox. Given the nature of the flaws, ExecShield protections in RHEL5 should make exploiting these memory flaws harder.
  2. An update to the Kerberos telnet deamon (April) A remote attacker who can access the telnet port of a target machine could log in as root without requiring a password. None of the standard protection mechanisms help prevent exploitation of this issue, however the krb5 telnet daemon is not enabled by default in Enterprise Linux 5 and the default firewall rules block remote access to the telnet port. This flaw did not affect the more common telnet daemon distributed in the telnet-server package.
  3. An update to Samba (May) where a remote attacker could cause a heap overflow. In addition to ExecShield making this harder to exploit, the impact of any sucessful exploit would be reduced as Samba is constrained by an SELinux targeted policy (enabled by default).
  4. An update to the PCRE library (November). This was labelled critical because the Konqueror web browser uses PCRE to handle regular expressions in JavaScript, and therefore a user browsing a malicious site in Konqueror could trigger this issue. (Konqueror is not part of a default install, but I've left this issue as critical in the results).

• Updates to correct all of these critical issues were available via Red Hat Network within a day of the issues being public.

Red Hat Enterprise Linux 5 shipped with a number of security technologies designed to make it harder to exploit vulnerabilities and in some cases block exploits for certain flaw types completely. For the period of this study there were two flaws blocked that would otherwise have required critical updates:

1. A stack buffer overflow flaw in the RPC library in Kerberos. This flaw was blocked by FORTIFY_SOURCE which removed the possibility of remote code execution. We still issued an update, as a remote attacker could trigger this flaw and cause Kerberos to crash.
2. Another flaw in Kerberos, this time due to the free of an invalid pointer. This flaw was blocked by glibc, although a remote attacker could still cause a crash, so we issued an update.

This data is interesting to get a feel for the risk of running Enterprise Linux 5 Server, but isn't really useful for comparisons with other versions or distributions -- for example, a default install of Red Hat Enterprise 4AS did not include Firefox. You can get the results I presented above for yourself by using our public security measurement data and tools, and run your own metrics for any given Red Hat product, package set, timescales, and severities.

Since then I've been sending in corrections on a monthly basis, taking into account the worst possible score across all affected platforms (and not how Red Hat products were affected specifically).

For the five months May to September 2007 I looked at 178 vulnerabilities (across all Red Hat products and services). Only 80 were accurate. Corrections were submitted to NVD and they fixed the incorrect CVSS scores on the remaining 98 vulnerabilities.

So, before the corrections, there were 65 issues rated "High" out of 178. After the corrections there are actually only 17 rated "High".

Fortunately the number of corrections needed each month seems to be decreasing, but we'll continue to send in corrections every month. Even with the corrections, the severity rating for a given vulnerability may well vary for the version each vendor ships; so you need to be careful if you are basing your risk assesments soley on the accuracy of third-party severity ratings.

Even though I'm not a fan, NVD publish a CVSS score for every issue, security companies are using those scores in their vulnerability feeds to customers, and people are using them for metrics. So it's important that these scores are accurate.

I decided to take a look at how accurate the CVSS scores were, so for every vulnerability we fixed in any Red Hat product for June 2007 examined the CVSS score given by NVD. For each one figuring out if the CVSS base metrics were correct, and where they were not submitting the correction back to NVD. This analysis of the vulnerabilities was based on their possible worst-case threat to all platforms (I didn't adjust the CVSS scores for how the issues affected Red Hat products specifically).

There were 39 total vulnerabilities for which unfortunately only 8 scores were accurate. I submitted corrections to NVD and they fixed the CVSS scores on the remaining 31 vulnerabilities.

20 vulnerabilities ended up moving down in ranking, 6 vulnerabilities moved up, and 5 stayed the same (although the CVSS score changed).

Before the corrections there were 14 issues rated "High" out of 39, after the corrections there are just 3 rated "High".

Those corrections are now live in the NVD, and I really appreciate how quick the folks behind NVD were at checking and making the changes. I've submitted to them corrections for a couple more months too, and I'll write about those when there complete. Unfortunately it does take a lot of time to investigate each issue and do the corrections, so it will limit how far back into 2007 we can correct.

Common Platform Enumeration (CPE) is a naming scheme designed to combat these inconsistencies, and is part of the 'making security measurable' initiative from Mitre. From today we're supporting CPE in our Security Response Team metrics: we publish a mapping of Red Hat advisories to both CVE and CPE platforms (updated daily) and you can use CPE to filter the metrics. Some examples of CPE names:

cpe://redhat:enterprise_linux:5:server/firefox -- the Firefox browser package on Red Hat Enterprise Linux 5 server.
cpe://redhat:enterprise_linux:3 -- Red Hat Enterprise Linux 3
cpe://redhat/xpdf -- the xpdf package in any Red Hat product.
cpe://redhat:rhel_application_stack:1 -- Red Hat Application Stack version 1

User reports a security vulnerability (this includes things later found not to be vulnerabilities)	47 (30%)
User is confused because they visited a site "powered by Apache" (happens a lot when some phishing or spam points to a site that is taken down and replaced with the default Apache httpd page)	39 (25%)
User asks a general product support question	38 (25%)
User asks a question about old security vulnerabilities	21 (14%)
User reports being compromised, although non-ASF software was at fault (For example through PHP, CGI, other web applications)	9 (6%)

That last one is worth restating: in the last 12 months no one who contacted the ASF security team reported a compromise that was found to be caused by ASF software.

The National Vulnerability Database provides a public severity rating for all CVE named vulnerabilities, "Low" "Medium" and "High", which they generate automatically based on the CVSS score their analysts calculate for each issue. I've been interested for some time to see how well those map to the severity ratings that Red Hat give to issues. We use the same ratings and methodology as Microsoft and others use, assigning "Critical" for things that have the ability to be remotely exploited automatically through "Important", "Moderate", to "Low".

Given a thundery Sunday afternoon I took the last 12 months of all possible vulnerabilities affecting Red Hat Enterprise Linux 4 (from 126 advisories across all components) from my metrics page and compared to NVD using their provided XML data files. The result broke down like this:

Red Hat	13% Crit 24% Important 39% Moderate 24% Low
NVD	30% High 20% Moderate 50% Low

So that looked okay on the surface; but the diagram above implies that all the issues Red Hat rated as Critical got mapped in NVD to High. But that's not actually the case, and when you look at the breakdown you get this result: (in number of vulnerabilities)

NVD: High

23 Critical 24 Important 35 Moderate 8 Low

NVD: Moderate

9 Crit 18 Important 22 Moderate 12 Low

NVD: Low

7 C 32 Important 62 Moderate 51 Low

That shows nearly half of the issues that NVD rated as High actually only affected Red Hat with Moderate or Low severity. Given our policy is to fix the things that are Critical and Important the fastest (and we have a pretty impressive record for fixing critical issues), it's no wonder that recent vulnerability studies that use the NVD mapping when analysing Red Hat vulnerabilities have some significant data errors.

I wasn't actually surprised that there are so many differences: my hypothesis is that many of the errors are due to the nature of how vulnerabilities affect open source software. Take for example the Apache HTTP server. Lots of companies ship Apache in their products, but all ship different versions with different defaults on different operating systems for different architecture compiled with different compilers using different compiler options. Many Apache vulnerabilities over the years have affected different platforms in significantly different ways. We've seen an Apache vulnerability that leads to arbitrary code execution on older FreeBSD, that causes a denial of service on Windows, but that was unexploitable on Linux for example. But it has a single CVE identifier.

So if you're using a version of the Apache web server you got with your Red Hat Enterprise Linux distribution then you need to rely on Red Hat to tell you how the issue affects the version they gave you -- in the same way you rely on them to give you an update to correct the issue.

I did also spot a few instances where the CVSS score for a given vulnerability was not correctly coded. CVSS version 2 was released last week and once NVD is based on the new version I'll redo this analysis and spend more time submitting corrections to any obvious mistakes.

But in summary: for multi-vendor software the severity rating for a given vulnerability may very well be different for each vendors version. This is a level of detail that vulnerability databases such as NVD don't currently capture; so you need to be careful if you are relying on the accuracy of third party severity ratings.