Shockingly Simple Flaw Leaves 'Millions Of Home Routers Open To Attack'

[microchip]

http://www.forbes.com/sites/thomasbrewster/2015/05/19/home-routers-vulnerable-to-netusb-attack/

Not using any of the NetUSB stuff myself on my TP-Links

 

[RMerlin]

http://www.forbes.com/sites/thomasbrewster/2015/05/19/home-routers-vulnerable-to-netusb-attack/

Not using any of the NetUSB stuff myself on my TP-Links

Another day, another router flaw.

I say we nuke the entire site from orbit. Only way to be sure.

(By which I mean, time to rewrite those home gateway firmwares from the ground up, ditching all that 10 years-ish code most of them still use, which was written back when the idea of a buffer overflow was unheard of).

 

[sfx2000]

(By which I mean, time to rewrite those home gateway firmwares from the ground up, ditching all that 10 years-ish code most of them still use, which was written back when the idea of a buffer overflow was unheard of).

Agreed - many of the challenges compound themselves by having common board support packages, so it might not even be the vendor's fault, but upstream...

 

[Russ64]

Most home users are not educated on the role of a gateway router. Most home users should not be exposing anything to the Internet and if they need remote access they should know how to run a VPN. USB ports and DMZ options on ISP home routers is just a bad idea. Best is to run a UTM behind the gateway device to protect everything inside your LAN.

 

[sfx2000]

Buffer overflows are easy to introduce, and hard to find...

case in point - the first one can be bad, the second one is better, mainly due to strncpy having a built-in check...

int function(char * data){

char overflow[255];
strcpy (overflow, data);
}​

vs

int function(char * data){

char overflow[255];

strncpy (overflow, data, 255);

}​

We have to assume that the OEM's, along with the SoC vendors, are not malicious - most coders don't deliberately introduce bugs, but things creep thru - esp. late at night, during crunch time, and looking at code that has already been written - it's mentally taxing work...

 

[sfx2000]

int function(char * data){
char overflow[255];
strcpy (overflow, data);
}
vs

int function(char * data){
char overflow[255];
strncpy (overflow, data, 255);

}

Then it becomes a bug-hunt - where's Waldo? -

When looking thru millions of lines of code, these things are there...