This week we discuss security updates in Linux Mint, Google funding Linux
kernel security development and details for security updates in BIND,
OpenSSL, Jackson, OpenLDAP and more.
Show Notes
Overview
This week we discuss security updates in Linux Mint, Google funding Linux
kernel security development and details for security updates in BIND,
OpenSSL, Jackson, OpenLDAP and more.
If using GSS-TSIG could be vulnerable to a DoS or possible RCE - this
option is not enabled by default BUT is often used when bind is
integrated with Samba or with a AD-DC. In Ubuntu we confine BIND with an
AppArmor profile by default isolates BIND quite tightly so helps to
mitigate any affect a possible RCE attack could have.
Was interesting to see upstream released 2advisories that some of
their upstream version updates (e.g. 9.16.12) for this caused some
regressions as this included some new as well features - and they
specifically ended up recommended downstreams ship the prior version
(9.16.11) with just the fix for this backported - this is what we do in
Ubuntu precisely for this reason, to minimise the chance of introducing
regressions in our security updates by only backporting the patch for
the particularly vulnerability
NULL ptr deref when parsing malicious issuer fields in X509
certificates - crash, DoS
Possible buffer overflow if some library functions were used in an
unlikely manner - had to specify an input length that was close to the
bounds of an integer size of the platform - so only if calling with a
buffer of INT_MAX or similar could this be an issue
JSON processor for Java - allows to map JSON to Java objects
Flaws in (de)serialization could expose various classes to being mapped
to the resulting input and hence allow a remote code execution attack -
fix is to deny various classes being mapped as a result
Also fixed an XML external entity issue that could also result in RCE
Possible web-cache poisoning attack - due to difference in handling of
requests between the proxy and the server - malicious requests can be
cached as they look like safe ones due to difference in interpretation
In patching previous vulnerabilities in QEMU, we backported various
patches but missed some related to riscv emulation so would cause a
possible crash in this case - fixed to add missing patches to resolve
this crash issue
Goings on in Ubuntu Security Community
Linux Mint users being slow with security updates, running old versions [06:33]
Blog post from lead developer Clem (Clement Lefebvre) discussing how
Linux Mint users seem to not be installing updates
Linux Mint is a Ubuntu derivative - uses the Ubuntu archives plus some of
their own repos - so in general all security updates for Ubuntu get
propagated to Linux Mint - cf. relationship between Ubuntu and Debian.
Interesting history in regards to security
In Febrary 2016, website was hacked and the link to the installer ISO
was modified to point to a malicious one with a backdoor -
https://blog.linuxmint.com/?p=2994
Recommend to turn of UEFI Secure Boot since their shim is not signed by
Microsoft
Update Manager would offer security updates but would rate them in
supposed terms of safety - so would in essence deter users from
installing some security updates - and also would not select to install
some updates which they deemed as more risky - but how did they assign
this safety level? Based more on if a component was critical to boot
(kernel/firmware would get rated as more risky) than anything to do
with the actual update itself. So was intended to help guide users BUT
created a system where users believed they were “safer” in terms of
stability, but in fact were less safe in terms of security.
These levels were removed in the 19.2 release but it seems users are
still wary
30% of users apply updates in less than a week (based on recent Firefox update)
30% of users are still running 17.x - EOLd in April 2019 - (based on
Ubuntu 14.04)
So it is not really surprising given their past history that their
userbase is wary of security updates and are perhaps putting themselves
at risk as a result by delaying installing security updates
But good to see they are now actively encouraging users to install
security updates
Use of timeshift is interesting as a mitigation against possible issues
with security updates
Also was interesting to see they published an emergency update just for
Firefox for the 17.x release to upgrade this from 66.0 to 78 ESR - so
this gives some protection but perhaps again lessens the incentive for
these users to upgrade to a newer supported release of Linux Mint
Google funds Linux kernel developers to work exclusively on security [14:20]