Hacker News new | ask | show | jobs
by wfjackson 4367 days ago
>They're a pretty easy target as the architecture of Windows is incredibly complicated and they're playing plug the holes rather than designing it properly to start with. For ref, I know the NT kernel, win32 and CLR inside out and no longer would I poke it with a stick.

Not sure what you mean by that. Does Linux have any protections beyond Windows to stop malware? Why does Android have a malware problem?

This is the text of a post I made yesterday in reply to a similar comment:

How can they patch it in their product without turning desktop Windows into something like iOS or Windows Phone/RT? Even Android has a ton of malware so the notion that Windows is somehow more hole ridden than other platforms stopped being true starting about 10 years ago with their Secure computing initiative. If the user can install Firefox, they can install malware.

If Firefox doesn't need to get permission from MS for their next version, Windows cannot distinguish between Firefox.exe and Codec_Flash_Shady.exe. Sandboxing will disable system level utilities.

MS is capable of making secure OSes. How many viruses and trojans do the 3 Xboxes, Windows Phone and RT have? Even Windows Server is pretty secure(atleast as secure as Linux) unless the admins start browsing on it. Malware is a real threat to any popular OS unless third party apps are entirely blocked or restricted by the use of a approval based App Store. Windows gives much more control to the user, which is why many users are able to stay away from infections.

And it's ironic that you're blaming MS here instead of the folks that propagate it(including a YC company https://www.techdirt.com/articles/20130115/17343321692/why-a...) and people who install it(users).

Remember the shitstorm that was raised against MS on here and elsewhere when they tried to secure users by preventing undetectable rootkits by enabling Secure Boot?

2 comments

1) Linux and other unixes were created with the idea of privilege separation and permission baked in. Windows had to add it later, while keeping compatibility

2) Linux has a variety of kernels and libraries versions across its base, making it difficult to exploit it uniformly

3) MS is indeed capable of making secure OSes, I don't deny it, but you should not use Xbox, Windows Phone and RT as examples, since all three of them can ONLY install approved software (less for RT, but it's not used by end users in the same way)

4) MS could have used the same approach used by Linux and Android (and partially OSX): have a central approved and monitored repository of software, but giving the possibility to add external software by jumping a few hoops, i.e. inserting a password, or checking a couple of checkboxes before allowing untrusted installations.

5) To prove the point, Android has malware almost exclusively outside of the google play store. Never heard of someone getting malware by using android, while I know an handful of people getting malware on windows (this is anecdotal experience, but I don't have any other data)

6) The shitstorm was raised because on some Secure Boot implementations it was impossible to disable it

1) IIRC, The Windows NT family had more granular level permissions than Linux. Granted before XP Windows was quite insecure, as I said in my original comment

2) Still we do see a lot of bugs and exploits that affect large swathes of Linux machines.

3) My entire point is that popular OSes that are used by nontechnical users that allow third party installs

4) Even OS X got a lot of burn for sanboxing apps and making third party apps difficult to install. They tried difficult UAC with Vista and it didn't go so well.

There isn't much stopping Linux malware in repos if the Linux desktop gets more popular. http://www.zdnet.com/blog/hardware/how-much-more-malware-is-...

Heck, even kernel.org was rooted and they still haven't revealed what happened. Not to mention other distros which were also compromised at some point.

5) http://www.pcworld.com/article/2099421/report-malwareinfecte...

6) Which ones? (apart from RT ARM machines that were a total flop in the marketplace and are like iPads)

I agree in general with all your point, apart from 4 and 5.

Malware in linux repositories is "practically" impossible. Software is most of the times peer reviewed and patched in different ways by different distros. And if a particular software becomes more popular it also comes under scrutiny by more people that want to change the source to add their own features. All the packages are checksummed and repositories have cryptographic keys to establish authenticity.

Of course bugs and security vulnerabilities exist, but the same applies to other OSes as well. And I do understand that UAC is obnoxious for users, but they didn't care about creating problems for legitimate users with the no-ip case since it was posing danger.

That android report makes two assumption: a very wide definition or malware (also installing java should be considered a malware because toolbar), and the fact that a malware doesn't usually last more than a day before being removed automatically.

> Not sure what you mean by that. Does Linux have any protections beyond Windows to stop malware? Why does Android have a malware problem?

This is the key question that I'd like to hear the answer to.

People often claim Linux/OS X/et al are more secure but they struggle to explain WHY. What technical mechanism is in place in those systems that is not in place in [current] Windows? A few years ago you could definitely name a few things (i.e. before UAC, and a few other things) but now?

I'm certain something like SELinux or AppArmor makes for a more secure system, but last time I installed a consumer distro (namely Mint and Ubuntu) they weren't shipped as standard and often broke quite a lot of default packages upon their installation.

As an aside: In my experience Windows has become less "malware ridden" since XP. Vista, 7, and 8 often survive much longer without anything bad happening. It does still happen, it just isn't as common as it used to be (e.g. 1/5 consumer PCs now instead of 3/5 or more).

Ubuntu ships with AppArmor enabled. Fedora ships with SELinux. It's been that way for a long time. Other distributions like Arch come with packages for other frameworks, including grsecurity.

But I think the primary "technical mechanism" that makes Linux more secure is the fact that users install software from distribution repositories, rather than from the web. The repos are basically impenetrable since packages are signed and contributor identities confirmed with WoT (I've never heard of there being malware in a major distribution) and security updates are deployed to everybody very quickly.

>I've never heard of there being malware in a major distribution

Not much to stop malware if desktop linux becomes more popular.

http://www.zdnet.com/blog/hardware/how-much-more-malware-is-...

I'll give it a try:

privilege separation and permission since the beginning

only super user was allowed to install new software

(simplifying) different distros and different versions created diversity making it difficult for an attack area to be widespread across all installations

typing a password for additional privileges requires more attention than clicking a button

apparmor has been enabled by default since a couple of years, it used to break some stuff but not anymore

(simplifying) new files are not executable, and they don't rely on extensions to determine the associated program

since linux is not the default it requires a learning curve that people using windows don't have, so users are more tech savvy

since the source code is available, more people COULD have a look at security vulnerabilities, and in case of emergency the don't have to wait for someone else to provide a patched binary

That said, I don't consider security on windows to be a disaster. It certainly is improving and in general they also pay a lot more attention to security.