maemo.org - Talk

maemo.org - Talk (https://talk.maemo.org/index.php)
-   General (https://talk.maemo.org/forumdisplay.php?f=7)
-   -   Holy cow, we have been dirty for 9 years (https://talk.maemo.org/showthread.php?t=97919)

mr_pingu 2016-10-21 12:26

Holy cow, we have been dirty for 9 years
 
http://arstechnica.com/security/2016...ctive-exploit/

Quite a massive number of devices that are vulnerable to this bug as it's quite an old feature and only discovered now. Then to think many android phones won't get an kernel-update probably. I am assuming android is just as vulnerable as any other linux distro with old kernel.

meego_leenooks1 2016-10-21 16:45

Re: Holy cow, we have been dirty for 9 years
 
A distaster for the servers and a Holy Grail for the handhelds.

mikecomputing 2016-10-21 18:18

Re: Holy cow, we have been dirty for 9 years
 
AFAIU you first have to login via ssh(or similar) as normal user to the phone before you can gain root access I don't see it as critical on phone but worse on web sites.

"The exploits can be used against Web hosting providers that provide shell access"


And how many of you give out ssh access to your phone?

However I hope SFOS next release has the fix.

coderus 2016-10-21 19:13

Re: Holy cow, we have been dirty for 9 years
 
this exploit can be easily used by any malware application you install :)

eccerr0r 2016-10-22 01:51

Re: Holy cow, we have been dirty for 9 years
 
Apparently the "fix" was identified, any ideas when this will be backported to KP?

Not sure about 2.6.28, but backported to my 4.0.5 server, there had been changes so the patch in the commit wouldn't cleanly go in... but was close enough to easily figure out

Supposedly it's been around since 2.6.22 but "harder" to exploit ... and as I don't have many random binaries I run on my N900, probably somewhat safe. The regular PCs with <koff>flashplayer and any with outward facing shell access I have to be worried about...

meego_leenooks1 2016-10-22 05:32

Re: Holy cow, we have been dirty for 9 years
 
Quote:

Originally Posted by mikecomputing (Post 1517136)
AFAIU you first have to login via ssh(or similar) as normal user to the phone before you can gain root access I don't see it as critical on phone but worse on web sites.

Nope, your phone could be rooted by nice looking app which is in fact the chinese hackers' malware sneaked into Play Market / App Store e.g. http://www.ibtimes.co.uk/chinese-hac...s-risk-1520592 so it's kinda critical too.
BUT at the same time you could root your own phone if you need it and if the phone manufacturer prevents you from getting root access. So I find this vulnerability as somewhat good for the handhelds power users.

Quote:

Originally Posted by mikecomputing (Post 1517136)
"The exploits can be used against Web hosting providers that provide shell access"

And concerning the servers - it is a total disaster. You do not need the shell access to own your hosting provider's server - you just need exec/system/etc function enabled in PHP configuration and even if such functions are disabled there are plenty of other ways to run shell code where the simplest is - running a cron job from the control panel.

nthn 2016-10-22 14:23

Re: Holy cow, we have been dirty for 9 years
 
Quote:

Originally Posted by meego_leenooks1 (Post 1517154)
http://www.ibtimes.co.uk/chinese-hackers-fool-google-put-one-million-android-users-risk-1520592

"The intelligence-testing Brain Test app was discovered to be containing the malware by security firm Check Point."

Honestly, if you install "Brain Test" applications, you're bound to get hacked in some way eventually.

Every time these supposedly extreme security flaws come up, it turns out to be something you need to explicitly allow. That isn't a security problem, it's a user problem, and those have turned out to be impossible to fix.

meego_leenooks1 2016-10-22 17:02

Re: Holy cow, we have been dirty for 9 years
 
Quote:

Originally Posted by nthn (Post 1517162)
"The intelligence-testing Brain Test app was discovered to be containing the malware by security firm Check Point."

Honestly, if you install "Brain Test" applications, you're bound to get hacked in some way eventually.

That's just the first example I found in Google. I've seen news about dozens or hundreds of innocent-looking apps in Play Market / App Store with hidden ad/spy/mal-ware inside. Surely you have to explicitly allow them to run (install them yourself) but how do you know which app is clean and legitimate and which is not?

pichlo 2016-10-22 17:50

Re: Holy cow, we have been dirty for 9 years
 
9 years. So much for shallow bugs.

t-b 2016-10-22 18:09

Re: Holy cow, we have been dirty for 9 years
 
Quote:

Originally Posted by meego_leenooks1 (Post 1517167)
That's just the first example I found in Google. I've seen news about dozens or hundreds of innocent-looking apps in Play Market / App Store with hidden ad/spy/mal-ware inside. Surely you have to explicitly allow them to run (install them yourself) but how do you know which app is clean and legitimate and which is not?

Indeed, as long as those apps are proprietary / closed source you just don't know. A ton of Google Play programs ask for permissions that don't make sense. Why should a game need access to your contacts etc..
On an Android phone I assume with cyanogenmod (no gapps) and only install apps from the f-droid repo you will prevent installing any malicious software and are relatively safe.

eccerr0r 2016-10-23 22:53

Re: Holy cow, we have been dirty for 9 years
 
Well, it looks like kernel 3.10 and later are the "easy to exploit" so it sounds like we might be semi-safe on Maemo. Still worrysome.

I have yet to try this on my android phone, then again I don't use my android phone that much.

In any case you don't need permissions to run the dirty cow exploit, it's just regular code (but it looks like you may need multithreading enabled as well as being able to write to your own address space through /proc.)

In android I wish it were possible to fine grain control all the functions that they say on those "XYZ needs access to ABC"...

Just let them install, but the user gets to control whether XYZ actually gets to use feature ABC.

Even better, return garbage information.

Now that's what I wish could be done.

gerbick 2016-10-24 00:59

Re: Holy cow, we have been dirty for 9 years
 
Malware and exploits on Darwin/OS X, Windows... well, being Windows and now this.

Not a one system is without exploits.

nieldk 2016-10-24 05:35

Re: Holy cow, we have been dirty for 9 years
 
In case you want to experiment, this PoC can help identify if vulnerable.

https://github.com/dirtycow/dirtycow...ter/dirtyc0w.c

ibrakalifa 2016-10-24 07:54

Re: Holy cow, we have been dirty for 9 years
 
9 years? seriously, and some of members talk about security like God himself recently...

kinggo 2016-10-24 08:29

Re: Holy cow, we have been dirty for 9 years
 
hmmmmmm......... so that's why turing phone isn't out yet :D

ste-phan 2016-10-24 09:16

Re: Holy cow, we have been dirty for 9 years
 
Quote:

Originally Posted by eccerr0r (Post 1517259)
Well, it looks like kernel 3.10 and later are the "easy to exploit" so it sounds like we might be semi-safe on Maemo. Still worrysome.

I have yet to try this on my android phone, then again I don't use my android phone that much.

In any case you don't need permissions to run the dirty cow exploit, it's just regular code (but it looks like you may need multithreading enabled as well as being able to write to your own address space through /proc.)

In android I wish it were possible to fine grain control all the functions that they say on those "XYZ needs access to ABC"...

Just let them install, but the user gets to control whether XYZ actually gets to use feature ABC.

Even better, return garbage information.

Now that's what I wish could be done.

About your wish, I once "secured" a Note 2 where I have installed XPrivacy.
It does exactly what you say: let stuff install, pop up with questions after install allowing you to decide for yourself
A.nd to leave things working as they should, it reportedly feed garbage dummy information to the agressor app.

So I get Viber running without address book access.

If I remember correctly, the thing had to be rooted and something called Exposed framework had to be installed. It is not my favorite activity and I hope one day this becomes much easier and straight forward to install.
But once you get Xprivacy running, you feel a big relief, I can tell you that ;)

juiceme 2016-10-24 10:02

Re: Holy cow, we have been dirty for 9 years
 
Quote:

Originally Posted by nieldk (Post 1517272)
In case you want to experiment, this PoC can help identify if vulnerable.

https://github.com/dirtycow/dirtycow...ter/dirtyc0w.c

Thanks.

I just quickly tested it on my main workstation running 4.4.0-36, Jolla sbj1 running 3.4.0 and a fedora cloud image running 4.6.4-301 kernels using the above demo exploit.

On all of those devices the result was as expected, the user process was able to gain root using the leaking CoW.

Same thing applies as always; do not run unknown binaries in your system. Fortunately this is an exploit not easily used remotely.

coderus 2016-10-24 10:26

Re: Holy cow, we have been dirty for 9 years
 
https://github.com/dirtycow/dirtycow...b.io/wiki/PoCs

bandora 2016-10-24 18:39

Re: Holy cow, we have been dirty for 9 years
 
On the other hand maybe it also makes it easier for a person to "root" their Android phones without actually going through the process.. :D:D


All times are GMT. The time now is 23:04.

vBulletin® Version 3.8.8