On Friday we read about the Firefox security issue, CAN-2005-2871. This issue looked like it could well be a 'critical' issue potentially allowing a malicious web page to control a heap buffer overflow. We know that various technologies in Red Hat Enterprise Linux and Fedora Core are likely to reduce the chances of this being actually exploitable by an attacker -- checks foil the most usual way of exploiting heap overflows by messing with malloc control structures, and on x86 at least heap randomization makes an exploit harder. But this issue was already public and so we didn't have the luxury of time to be able to test the mitigation. So we initiated our emergency response process to get the packages through development and QA and got Firefox and Mozilla packages out via Red Hat Network within 20 hours of this issue being public (due to the awesome work from engineering folks, QA folks, and the security response team who worked late into Friday night to get this done).
The metrics from the security response team have had their monthly update at https://people.redhat.com/mjc/. This month we've also tidied up some of the XSLT used to create the web pages, so the sample reports now have the default style and contain descriptions of each vulnerability as listed at CVE.
The perl script used to analyse the raw stats has also had some updates and no longer needs to be edited to filter the vulnerabilities you are interested in. Run "perl daysofrisk.pl --help" for details.
For Red Hat Enterprise Linux 3 across all dates (20 months) we've had 13 critical vulnerabilities; of which 84% had updates available via Red Hat Network within a day of the vulnerability being public.
Back in March I wrote about a Role Comparison Report from Security Innovation which was published without involving Red Hat. Since that report they contacted and supplied their dataset in which we were able to correct some mistakes. This week Security Innovation released another report from the data, this time looking at the role of a Database Server.
Despite the report's claim to incorporate a qualitative assessment of vendor reactions to serious vulnerabilities, the headline metrics treats all vulnerabilities as equal, regardless of their risk to users.
Their headline figure is 61 days of risk for a Red Hat Enterprise Linux 3 minimal installation with the addition of MySQL server from Red Hat Enterprise Linux Extras.
That sounds like a lot of days of risk - but if you filter their dataset by severity, using the Microsoft scale for determining the severity of each issue you find the following:
** Critical issues: 3 total issues. All fixed on the same day as first public disclosure, therefore having 0 days average risk.
** Critical plus Important: 49 total, with 34 average days of risk
Red Hat prioritise all vulnerabilities and fix first those that matter the most. We publish our raw data and metrics at https://people.redhat.com/mjc/
Days of risk statistics only tell a small part of the story: studies show consumers take some time to apply patches even after a vendor has produced a security update. At Red Hat we continue to work on ways to help people keep their machines up to date. Last year we added Exec-Shield to Red Hat Enterprise Linux 3 which included support for processor EDB (execute disable bit) and NX (no execute) technology. Earlier this year Red Hat Enterprise Linux 4 shipped with Security Enhanced Linux turned on by default. These technology innovations are designed to reduce the risk of security issues.
Fedora Security
Just finished the security audit for FC4 candidate - For 20030101-20050605 there are a potential 861 CVE named vulnerabilities that could have affected FC4 packages. 759 (88%) of those are fixed because FC4 includes an upstream version that includes a fix, 8 (1%) are still outstanding, and 94 (11%) are fixed with a backported patch. I'll post all the details to fedora-devel-list later in the week. I'm also giving a keynote about Fedora and security response at FudCon later this month.
OpenSSL Security
A CSO remarked to me a couple of weeks ago that their perception was that OpenSSL had a lot of serious security issues over the years. In fact it's really only had a couple of serious issues, and in total only 15 issues in the last 4 years. So in the style of the Apache vulnerability database I did one for OpenSSL. This is now publically available and we'll keep it up to date. The page is built from a XML database of the issues.
We've not really given Apache Week any priority in the last few months -- in fact we've not posted a new issue since October 2004. So I'm glad we didn't rename it Apache Month. Time to register apachewhenthereissomethinginteresting.com.
Anyway, the most useful thing that I've kept up to date in Apache Week is the database of vulnerabilities that affects the Apache Web server v1.3 and v2.0. This list was even being linked to directly by httpd.apache.org so I made good on a promise I made a year ago and moved the database to the official site. Apache Week uses xslt for transforming the database, but the Apache site used velocity for page markup, but no one seemed to mind me adding ant-trax.jar to the site so the database gets converted from xslt to the page format that gets marked up by velocity. The end result is a couple of nice HTML pages on the official Apache site that list all the vulnerabilities that is easy for us to keep up to date.
Today a "Role Comparison Report" from Security Innovation was published which has a headline that we fix security issues less than half as fast as Microsoft.
Red Hat was not given an opportunity to examine the "Role Comparison Report" or it's data in advance of publication and we believe there to be inaccuracies in the published "days of risk" metrics. These metrics are significantly different from our own findings based on data sets made publically available by our Security Response Team.
Despite the report's claim to incorporate a qualitative assessment of vendor reactions to serious vulnerabilities, the headline metrics treats all vulnerabilities as equal, regardless of their risk to users. The Red Hat Security Response Team publish complete data sets allowing calculations to be made taking into account the severity of each flaw. Red Hat prioritise all vulnerabilities and fix first those that matter the most.
For example out of the dataset examined by the report there were only 8 flaws in Red Hat Enterprise Linux 3 that would be classed as "critical" by either the Microsoft or Red Hat severity scales. Of those, three quarters were fixed within a day, and the average was 8 days. A critical vulnerability is one that could be exploited to allow remote compromise of a machine without interaction, for example by a worm.
With the current threat landscape it is no longer sufficient for operating system vendors to just respond to security issues. As part of our overall security strategy Red Hat is continually innovating to create new technologies that proactively help reduce the risk of unpatched or as yet undiscovered vulnerabilities.
Data set and perl script to let you run your own metrics from the Security Response Team
Roy Fielding sent out a message reminding us all that the Apache web server just celebrated it's tenth birthday.
In January 1995 I found a security flaw affecting the NCSA web server and I'd forwarded my patch on to Brian Behlendorf. The flaw affected the Wired.com site he was the administrator of. He told me about the Apache project and and I was invited to join the group and share the numerous patches I'd made to NCSA httpd, so my first post was back in April 1995. I can't believe that was ten years ago!
Anyway in my official Red Hat blog I've been posting stuff about the recent comparisons of security issues in Microsoft and Red Hat, and we've published a ton of useful data. See Counting Teapots and Real Data.
Yesterday I promised that we'd publish some of the mappings that we internally use in the Security Response Team. Three of these are available now.
The first is a mapping of severity for every security advisory for Red Hat Enterprise Linux and Stronghold from release date through to Feb 15th 2005 (after Feb 15th 2005 this information is included on advisories as published).
These severities assigned to each RHSA are based on a unique assement of how each individual flaw affects the particular distribution, then rolling up the severities and picking the worst to give the overall severity rating. A second mapping therefore gives the severity rating we assigned to each individual vulnerability, by CVE name. Included in this mapping is also the date that each issue was first known publically.
The final mapping is RHSA to release date. In the majority of cases the "Issue Date" displayed on the advisory or by Red Hat Network is correct, however this file also contains fixes where the issue date was published incorrectly, was missing, or delayed. This file contains every RHSA from 2000 to date, and will get get updated from time to time.
We'll update the mappings from time to time (we keep up to date copies internally, so if you have specific questions or we've forgotten to update them in a while just drop secalert@redhat.com an email). We also have other mappings which are automatically generated from our errata system which we'll publish soon.
It's been an interesting month so far with several reports of people comparing the number of vulnerabilities in Microsoft software to those in Linux distributions. I've previously talked at length about these types of studies after last years Forrester report, but it seems that these comparisons don't get any better.
Microsoft are quoted and say that in 2005 to Feb 9th that Windows Server 2003 had 15 vulnerabilities, but in the same period Red Hat Enterprise Linux 3 had 34, more than double!
What they failed to mention was that of those vulnerabilities, 3 of the flaws affecting Windows Server 2003 were classed by Microsoft as "Critical", flaws that can be remotely exploited without user interaction to take control of a machine, for example by a worm. Of the Enterprise Linux 3 vulnerabilities quoted, using the Microsoft scale, none were Critical. Metrics like those they quoted are completely worthless if you do not take into account the risk that the vulnerabilities actually pose to users. One Critical vulnerability and a worm or remote attacker owns your machine.
So of those 15 Microsoft issues:
3 Critical
3 Important
8 Moderate
1 Low
For Enterprise Linux 3 they counted 34 issues up until Feb 9th:
0 Critical
12 Important
14 Moderate
8 Low
I'm not saying that Red Hat is immune to Critical vulnerabilities, in fact in the lifetime of Enterprise Linux 3 (Nov 2003 to date) we've had 12. I'm also not saying that I think these stats show that Linux is more secure (or safer) than Windows, just they just show how useless stats like these are.
One of the things we at Red Hat can do to help our users determine the risk of security issues is to provide some guidance on which issues Red Hat is the most worried about. Since the release of Red Hat Enterprise Linux 4 last week, the Red Hat Security Response Team has been including severity impact statements on all security advisories. find out more. We've also gone back and applied the classification to every Enterprise Linux advisory we've produced, and will publish that list shortly.
In the Red Hat earnings call last night, Matthew Szulik mentioned some statistics on the survivability of Red Hat Enterprise Linux 3. In August 2004, SANS Internet Storm Center published statistics on the survival time of Windows by looking at the average time between probes/worms that could affect an unpatched system. The findings showed that it would take only 20 minutes on average for a machine to be compromised remotely, less than the time it would take to download all the updates to protect against those flaws. See some news about the report.
Red Hat Enterprise Linux 3 was released in November 2003 and I wanted to find out what it's survival rate on x86 would likely be to compare to Windows. We'll first look at the worst case and find what flaws have been fixed in RHEL3 that could possibly be remotely exploited. Then from that work out how often they are exploited to come up with a survivability time for RHEL3.
Firstly we need to discount flaws that require user interaction as they are not included in a survivability study - for example CAN-2004-0597 or CAN-2004-0722 where a user would have to visit a malicious web page, preview a malicious email, or open a malicious file with a particular application. So we won't include CAN-2004-0006, a flaw in Gaim that requires a user to be sent a malicious packet from a trusted buddy, for example.
From the release of RHEL3 until 19th August 2004 we have the following flaws that could be triggered remotely:
CAN-2004-0493 is a memory leak in Apache. This allowed a remote attacker to perform a denial of service attack against the server by forcing it to consume large amounts of memory. This flaw could possibly be use in combination with a flaw in PHP to execute arbitrary code. However no exploit has been seen in the wild for this issue, and it looks incredibly difficult to exploit. A second memory leak affected SSL enabled Apache, CAN-2004-0113. These wouldn't allow a full installation of RHEL3 to be remotely compromised.
Flaws in OpenSSL were found that could lead to a crash, CAN-2004-0079. Any service that accepts SSL traffic using OpenSSL to decode it could be vulnerable to this issue. However for servers like Apache, a single child crashing is automatically recovered and will not even cause a Denial of Service.
A flaw in the ISAKMP daemon in racoon could lead to a DoS, CAN-2004-0403, but this daemon is not used by default.
Using one of the above flaws, remote probes could cause a service to crash or exceed OS resource limits and be terminated. These have little impact on the survivability of a machine. What affects survivability are flaws that could lead to remote code execution or a total machine crash.
Two of these type of flaws are in CVS, CAN-2004-0396 and CAN-2004-0414. These flaws could allow a remote user who has the permission to connect to a CVS server to execute arbitrary code on the server. An exploit for these flaws is widely available. However the majority of systems would not be running a CVS server, it certainly isn't default, and in order for this to be remotely exploited by an unknown attacker a system would need to have been set up to allow remote anonymous CVS access.
A flaw in Samba SWAT, CAN-2004-0600, allows remote code execution but only where the SWAT (administration) port is open to the internet. This is not the default, and not a sensible or usual configuration.
The final issue is an overflow in rsync, CAN-2003-0962. This flaw is similar to the CVS flaw in that it requires a system to be running an open rsync server to be exploited.
So a full install of a Red Hat Enterprise Linux 3 box that was connected to the internet in November 2003 even without the firewall and without receiving updates would still remain uncompromised (and still running) to this day.
It's not to say that a RHEL3 user couldn't get compromised - but that's not the point of the survivability statistuc. In order to get compromised, a user would have to have either enabled anonymous rsync, SWAT, or be running an open CVS server, none of which are default or common. Or a user would have to take some action like visiting a malicious web site or receiving and opening a malicious email.
