Ceci est une ancienne révision du document !
From Curtis Patranella: [What are your thoughts on the security aspects of] PGP and email encryption in this age of Big Brother and NSA infringement.
MB: Encryption to avoid eavesdroppers is a good method. The use of encryption will definitely increase in the upcoming years. There is a downside however, like legitimate email scanning (for viruses, spam) might be less efficient, as these tools can’t “read” the emails either. Next phase will be skepticism about the strength of encryption ciphers being used, especially after the NSA revelations.
From Paul Graham: Along the lines of the Flame Virus that 'tricked' Windows into thinking it was a Microsoft Update; how secure is Linux against this sort of attack?
MB: Flame used a clever method to convince users the patch was signed by Microsoft. By chaining an invalid certificate type into that of the Microsoft root certificate, updates looked to be signed by Microsoft. If an attacker would like to perform a similar method on Linux systems, the repository (or a mirror) of that distribution would most likely be targeted. Of course it is possible, especially with the help of an insider. Since most Linux distributions use signed packages, the chance for detecting an invalid package is high though. Since most software is open source, it’s a good thing that there are people around who like to dig into the security aspect of it.
From Paddy Landau: I would like to know if it's possible to easily configure Apparmor for the everyday user, rather than for a security expert. I have tried reading the manual, but I find it too complicated to follow.
MB: Like most new software, it may take a while to learn to use it. But, once configured, not much administration is needed later on. AppArmor is already considered more user friendly than its alternative, SELinux. If configuring is still too challenging, then you might want to search for ready-to-use configurations. For example, searching for “AppArmor MySQL” will provide links explaining what paths to configure and the required permissions.
From Ed Eckelmeyer: Whenever I raise the “security” issue, the response is that Linux (i.e. Unix) is stable, and has had years of finding and eliminating any backdoors for the bad guys. I suppose that another reason might be because there are so few Linux users (relative to Windows users) that it is not worth the effort for the bad guys to spend any time on writing malignant code. Am I blind, ill informed, or just plain lucky that I have not had any problems?
MB: Like Windows, there is also malware available for Linux and Mac OS. Fortunately, most of it is focused on specific software and often quickly discovered. If an attacker uses a so-called 0-day exploit, sooner or later someone will notice his system acting strange. With enough determination, the related exploit or malware will be discovered, and software can be patched against it. Most (advanced) Linux users have a different mindset and report discovered vulnerabilities. There are many security researchers scanning code for weaknesses and reporting them. Another important component – which makes Linux pretty secure by default – is the type of installation. Usually it’s a minimal (or default) installation, without unneeded services. If some service is needed, like turning the system into a web server, the user can decide to install a package like Apache. Due to so many Linux distributions being available, one might be lucky when some service is considered vulnerable. For example, an exploit on MySQL might work on a system running Fedora, while the same exploit might not work on Ubuntu. Why? Sometimes it’s just a matter of differences in how a kernel is configured, or how binaries were compiled. For the attacker, it’s almost impossible to make the perfect Linux exploit which works on all systems (32/64 bits, Intel/ARM, kernel security options, etc).